Poster Abstracts
(in alphabetical order by presenters last name)
Quality of Chronic Kidney Disease Care in Primary
Care
Adrienne Allen, M.D., John Forman, M.D; E. John
Orav, Ph.D.; David Bates, M.D.; Bradley Denker,
M.D.; Thomas Sequist, M.D.
Presented by: Adrienne Allen, M.D., General
Medicine Fellow, Division of General Medicine,
Brigham and Women's Hospital, 1620 Tremont
Street, Boston, MA 02120; Phone: (617) 525-7541;
Email: aallen0@partners.org
Research Objective: To evaluate the quality of
chronic kidney disease(CKD) care for
cardiovascular risk reduction, prevention of
metabolic bone disease, and drug safety at a large
multi-site group practice and to identify predictors of
high quality CKD care.
Study Design: We used electronic medical record
data to identify patients with stages 3 or 4 CKD,
defined as two recent estimated glomerular filtration
rates (eGFR) between 15 and 60, separated by 90
days. We evaluated the quality of CKD care for
cardiovascular risk reduction, prevention of
metabolic bone disease, and drug safety (Table) for
the 12 month period prior to July 2009. Drug safety
was examined via electronic prescription rates of
potentially inappropriate medications (NSAIDs,
glyburide, metformin, nitrofurantoin) within the last
12 months. Primary care physician (PCP)
awareness was measured by documentation of
CKD on the electronic problem list, and nephrology
comanagement was defined as a nephrology visit
within the past 12 months. We fit hierarchical
logistic regression models to identify predictors of
appropriate CKD care after adjusting for patient
age, sex, income, education level and insurance;
presence of diabetes or hypertension; and
clustering within clinics and PCPs.
Population Studied: Primary care patients with
chronic kidney disease in the metro-Boston area.
Principal Findings: Among 11,760 patients treated
by 166 PCPs across 15 clinics, 66% had
hypertension and 29% had diabetes. PCP
awareness of CKD was low (24%), and only 10%
were co-managed with nephrology. Most patients
were not receiving appropriate CKD care. Blood
pressure control was achieved in 54% of patients;
only 13% were screened for metabolic bone
disease. Twenty percent had been prescribed one
or more inappropriate drug within the year. Patients
co-managed with nephrology were less likely to be
prescribed an inappropriate drug (OR 0.77; 95% CI
0.64-0.93). PCP awareness and nephrology comanagement were consistently associated with
improved effectiveness in microalbumin
measurement, ACE/ARB use, and screening for
metabolic bone disease.
Conclusions: Significant gaps in quality of CKD
care exist. These might be addressed by increasing
primary care physician awareness of CKD,
addressing knowledge deficits, and encouraging comanagement with nephrology.
Implications for Policy, Delivery or Practice: To
improve quality of CKD care, practices should
consider using alerts for patients with stage 3 or 4
CKD as well as early nephrology referral.
Funding Source(s): NRSA #T32HP10251-02
Poster Number: 1
Targeted Services to Improve Mental Health in the
Context of Substance Abuse Treatment: Outcomes
for Women with a History of Intimate Partner
Violence
Christina Andrews, M.S.W.; Jeanne Marsh, Ph.D.
Presented by: Christina Andrews, M.S.W.,
Doctoral Candidate, School of Social Service
Administration, University of Chicago, 969 E. 60th
Street, Chicago, IL 60637; Phone: (773) 308-5123;
Email: cmandrews@uchicago.edu
Research Objective: Intimate partner abuse is a
widespread phenomenon among women in
substance abuse treatment. While prevalence
estimates in the general population range from 13%
to 30%, rates for women in substance abuse
treatment are approximated to be much higher,
spanning from 25% to 57%. Moreover, women who
enter treatment with a history of intimate partner
violence are more likely to develop mental health
disorders stemming from abuse and trauma.
Despite these statistics, little is known about the
mental health services provided to this group of
women. The present study responds to this gap in
the literature by examining targeted services to
address mental health for women with and without
a history of intimate partner violence in substance
abuse treatment.
Study Design: Generalized linear mixed modeling
was used to assess the association between receipt
of mental health services and change in mental
health distress from pre- to post-treatment.
Interaction variables were used to test for
moderating effects of intimate partner violence.
Descriptive comparisons were made using chisquare tests for categorical variables, and analysis
of variance (ANOVA) for continuous variables.
Population Studied: The study uses data from the
National Treatment Improvement Evaluation Study
(NTIES), a prospective, cohort study of substance
abuse treatment facilities and clients. The analytic
sample consists of 1,123 women from 58 treatment
facilities. Missing values were specified for these
comparisons.
Principal Findings: Women with a history of
intimate partner violence (IPV+) had higher average
levels of mental health distress than women without
such a history (IPV-) at pre-treatment (p <.001) and
post-treatment (p <.001). However, IPV+ women
made greater reductions in mental health distress
from pre-treatment to post-treatment (p <.001).
IPV+ women were significantly more likely than
IPV- women to receive mental health services (89%
v. 67%, p <.001). Yet, receipt of mental health
services was associated with similar improvement
in mental health for all women who received them
(p =.010); no interaction between mental health
service receipt and IPV status was found (p =.601).
Conclusions: In this sample, women with a history
of intimate partner violence entered treatment with
high levels of mental health distress, and the vast
majority received mental health services. Over the
course of treatment, women with a history of
intimate partner violence made greater reductions
in mental health distress than women without such
a history. Receipt of mental health services is likely
to have played a role in their improvement—receipt
of such services was associated with significant
reductions in mental health distress for all women
who received them.
Implications for Policy, Delivery or Practice: The
study findings suggest that women with a history of
intimate partner violence are more likely than
women without such a history to need, use and
benefit from mental health services within the
context of substance abuse treatment.
To assist women with a history of intimate partner
violence to recover from addiction, providers should
strive to offer mental health services in tandem with
substance abuse treatment.
Funding Source(s): NIDA
Poster Number: 2
The Non-Physician Clinician Workforce of Nebraska
Bettye Apenteng, B.Sc.; Keith Mueller, Ph.D.;
Preethy Nayar, Ph.D.
Presented by: Bettye Apenteng, B.Sc., Student,
Health Services Research and Administration,
University of Nebraska Medical Center, 984350
Nebraska Medical Center, Omaha, NE 68198;
Phone: (206) 291-3132; Email:
bapenten@unmc.edu
Research Objective: To examine the demographic
characteristics and practice location of nurse
practitioners (NPs) and physician assistants (PAs)
practicing in the state of Nebraska in 2008.
Study Design: The study is a cross-sectional
descriptive analysis of the 2008 University Of
Nebraska Medical Center’s Health Profession
Tracking Service survey data on nurse practitioners
and physician assistants in Nebraska (N = 1175).
Health Professional Shortage Area (HPSA)
designations and the 2003 Federal Office of
Management and Budget definition of metropolitan
counties were used to characterize practice location
and rural/urban location of practice.
Population Studied: All Non Physician Clinicians
(NPCs) or mid-level providers (Physician Assistants
and Nurse Practitioners) actively practicing in the
state of Nebraska in 2008.
Principal Findings: Physician assistants and nurse
practitioners in Nebraska are predominantly female
and Caucasian with 66.5% of mid-level providers
practicing in urban areas. Despite the fact that
training programs for mid-level providers were
originally inspired by the need to fill in gaps in
primary care delivery, only 48.9% of Nebraska’s
mid-level providers are in primary care. Of those
who opt for specialty care, only 10.6% take up
practice in mental health. A higher proportion of
NPs are females, compared to PAs (92.6% vs.
61.0%). PAs in Nebraska are younger than NPs,
(42yrs vs. 47yrs) and this is true for both primary
care and specialty care, except for mental health.
Among mental health mid-level professionals, there
is no difference in the average age between the two
professions. Psychiatric mid-level providers tend to
be older than other mid-level providers (53yrs vs.
44yrs). Among primary care mid-level providers, a
higher proportion of PAs than NPs practice in a
non-metro location (? = 0.005). Also, a higher
proportion of male mid-level providers practice in a
non-metro area compared to female mid-level
providers ((? = 0.01). In specialties other than
primary care however, no association is seen
between gender, license type and practice location.
Over half of psychiatric mid-level providers (54.7%)
practice in federally designated mental health
HPSAs. Only 14.8% of Nebraska’s primary care
mid-level providers practice in HPSA designated
single counties, geographic areas, rural clinics,
health centers and clinics serving special
populations. These providers however, serve 39 of
the 51 designated HPSA counties in Nebraska.
Conclusions: There is a shift in interest, away from
primary to specialty care as well as the preference
for urban practice, this could serve as a threat to
effectively addressing primary care and mental
health access issues in underserved and rural
areas.
Implications for Policy, Delivery or Practice:
Assessment of the demographic characteristics and
practice location of NPCs as presented is crucial in
workforce planning. The willingness of PAs and
males to practice in rural areas should be factored
into strategies for the recruitment and retention of
health care professionals to rural areas. Policy
should also address the lack of an adequately sized
and aging mental health workforce.
Poster Number: 3
Medicare Advantage Extra Payments in 2010
Grace Arnold, B.A.; Brian Biles, M.D., M.P.H.
Presented by: Grace Arnold, B.A., Research
Assistant, Department of Health Policy, George
Washington University, 2021 K Street, NW, Suite
800, Washington, DC 20036; Phone: (202) 9944203; Email: gca@gwmail.gwu.edu
Research Objective: Over the past decade,
Medicare Advantage (MA) payment policy has
shifted from one which encourages private plans in
areas where they can provide benefits more
efficiently than fee-for-service (FFS) Medicare to
one which encourages plans in all areas of the
country. In order to encourage plans in all areas of
the country to participate in the MA program, plans
in some areas are paid more than Medicare FFS
costs. President Obama and Congressional leaders
have argued that this is wasteful spending and
eliminating these overpayments to MA plans was a
major source of Medicare savings to make health
reform legislation deficit neutral. We quantified
overpayments to Medicare private plans in 2010 by
geographic area and legacy benchmark category.
Study Design: County-level payment benchmarks,
enrollment and penetration data were downloaded
from the Center for Medicare and Medicaid services
website in November, 2009. Enrollees in Medicare
"cost" and private fee-for-service (PFFS) plans
were not included in the analysis. County-level
payments were calculated by applying MedPAC's
analysis of average bids in relation to FFS costs to
county-level benchmark data in order to estimate
bids. Seventy-five percent of the difference between
the benchmark and the estimated bid was added to
the bid to account for the Medicare rebate in the
final payment figure. Average payments for
geographic areas are enrollee-weighted.
Population Studied: Medicare Advantage plan
enrollees and Medicare beneficiaries
Principal Findings: Nearly 10 million Medicare
beneficiaries in the US participate in the MA
program in 2010. Nationally, Medicare will pay
plans more than $7 billion and an average of 110%
more than local FFS costs in 2010. Counties that
are paid at the legacy urban floor are home to 43%
of MA enrollees and will receive 54% of the total
national overpayments. Plans in these counties are
paid an average of 114% of local FFS costs.
Counties paid at the legacy floor of 100% of 2004
FFS costs will receive 28% of over payments and
are home to 33% of plan enrollees. Plans in these
counties are over paid an average of 108% of local
FFS costs. Counties paid at the legacy blend
benchmark are paid an average of 118% of FFS
costs and those receiving a minimum payment
update are paid at 116% of FFS costs.
Conclusions: Medicare Advantage plans
continued to receive payments in excess of
Medicare FFS costs in 2010. The amount and
distribution of extra payments varied with
geographic area and county benchmark type.
Implications for Policy, Delivery or Practice: If
health reform legislation is enacted, the elimination
of overpayments will have a significant impact on
the MA program. This analysis will provide a
baseline against which the effects of health reform
will be measured.
Funding Source(s): CWF
Poster Number: 4
Why Do Individuals Receive Acupuncture: A Cross
Sectional Analysis Based on U.S. Adult Sample?
Shamly Austin, M.H.A., Ph.D ; Zo
Ramamonjiarivelo, M.B.A.; Richard Shewchuk,
Ph.D.
Presented by: Shamly Austin, M.H.A., Ph.D.
Candidate/Graduate Research Assistant,
Department of Health Services Administration,
University of Alabama at Birmingham, 1320 11th
Ave South, Birmingham, AL 35205; Phone: (225)
252-8650; Email: shamly@uab.edu
Research Objective: Over the years, the use of
complementary and alternative medicine (CAM) has
increased in the U.S from 36% in 2002 to 38.3% in
2006. Acupuncture is considered as a type of CAM
with its origin in the traditional Chinese medicine.
An estimated 3.1 million adults used acupuncture in
2007. Previously no studies analyzed the reasons
for the use of CAM. Our study examines the
reasons for the utilization of acupuncture services
among U.S adults.
Study Design: A cross sectional analysis was done
based on 2007 National Health Interview Survey.
The analysis includes 1,532 U.S adults who have
reported the use of acupuncture services in their
lifetime. The variables were recent visits measured
as “saw a provider for acupuncture during the last
12 months”, reasons for the use of acupuncture,
and socio-demographic variables including age,
race/ethnicity, sex, education, and income.
Univariate and bivariate analyses were performed
to analyze the reasons for utilization of acupuncture
services.
Population Studied: Non-institutionalized U.S
civilian population >=18 years, who reported the
use of acupuncture services.
Principal Findings: Of the 23,393 adults who
completed the survey on alternative health
questionnaire, 6.5% (n=1,532) reported use of
acupuncture in their lifetime. Of these 1,532
individuals, 22.5% (n=344) reported recent use of
acupuncture services, 64 % were females, 38%
resided in the West, and 27% in the South, 65%
were Whites, 14 % Hispanics, and 11% Asians,
78% were <65 years old, 45% were married, 43%
had > high school education. 47% reported arthritis
and body pain as the specific health condition for
which they used acupuncture. The major reason for
use was recommendations from family and friends
(21%). This was followed by the belief that
conventional medical treatment was of no help
(20%), and another 20% used it for general
health/wellness, 12% used it as it was
recommended by their health care provider, and
another 11.8% used it to enhance energy level.
10.4% used it to enhance immune function, and
4.6% felt that conventional medical treatment was
too expensive. Additional analysis was done to
determine whether reasons for acupuncture use
were different for both sexes. Female responses
indicated that the major reason for acupuncture use
was that conventional medical treatment was of no
help to them in curing illness (20%), and another
20% were recommended by their family/friends,
followed by 19.5% reporting general health and
wellness as the reason for use. Among men, the
major reason for use was recommendation from
family/ friends (24.5%), followed by general
health/wellness (21.6%), and 19.6% used it as
conventional medical treatment was of no help to
them.
Conclusions: The major reason for acupuncture
use was recommendations from family/friends
followed by the reason that conventional medicine
was of little help.
Implications for Policy, Delivery or Practice: The
study helps to understand the reasons for emerging
migration of individuals from conventional medicine
to CAM and may be helpful for health insurers to
make decisions on expansion of coverage to CAM
users.
Poster Number: 5
Physician Self-Referral of MRI and Care for
Patients with Low Back Pain
Jacqueline Baras, M.D., M.S.; Laurence Baker,
Ph.D.
Presented by: Jacqueline Baras, M.D., M.S.,
Medical Student, School of Medicine, Stanford
University, 1070 Mercedes Avenue Apartment #11,
Los Altos, CA 94022; Phone: (727) 742-8229;
Email: jbaras@stanford.edu
Research Objective: Physician self-referral of
advanced imaging such as magnetic resonance
imaging (MRI) is increasingly common. While selfreferral offers convenience and better continuity of
care for patients, such arrangements produce
strong financial incentives for physicians to order
MRI scans that may not be necessary or
appropriate. Additional scans may influence
treatment patterns and total costs of care, but the
size of the impact is uncertain. This paper
investigates the relationship between physician selfreferral, use of MRI, receipt of surgery, and total
spending for patients with low back pain.
Study Design: Panel data difference-in-difference
analysis, comparing changes in treatment patterns
of self-referring physicians before and after they
begin self-referring to changes in treatment patterns
of non-self-referring physicians over the same time
period. We use claims from a 20% sample of
traditional Medicare beneficiaries to identify nonradiologist physicians who see low back pain
patients and appear to begin self-referral
arrangements for MRI between 1999 and 2005, as
well as their patients who have a new episode of
low back pain care during this time. We also identify
episodes of low back pain treatment with physicians
who never self-refer for MRI. For each episode, we
track use of MRI, use of back surgery, and total
Medicare spending associated with low back care
claims, including all Medicare payments for
physician, inpatient, and outpatient claims. We use
regression analysis to identify changes in use of
MRI, use of surgery, and spending for episodes
before and after physicians begin self-referring,
using episodes with non-self-referring doctors to
control for overall time trends. The regressions also
adjust for an array of demographic and
socioeconomic covariates and include month, year,
and physician fixed effects.
Population Studied: 487,234 ninety-day episodes
of care for low back pain among fee-for-service
Medicare patients over age 65, associated with
79,971 physicians.
Principal Findings: Physicians ordered more MRI
scans for their patients with low back pain once
they began to engage in self-referral of MRI. The
unadjusted average of MRI procedures billed within
90 days increased from 11.4 per 100 episodes
before self-referral to 16.3 per 100 episodes with
self-referral; non-self-referring physicians ordered
MRI considerably less often, at an average of 7.0
MRI scans per 100 episodes. These additional
scans are statistically significantly associated with
increases in surgery receipt and low backassociated spending. After stratification on
physician specialty, the effect of self-referral of MRI
was strongest among orthopedic surgeons and
weakest among primary care physicians.
Conclusions: Non-radiologist physicians who
participate in self-referral arrangements of MRI
appear to change their practice patterns and order
more MRI scans for patients with low back pain.
These increases in MRI use lead to increases in
surgery receipt and total low back-related spending,
but the strength of this effect varies widely across
physician specialty.
Implications for Policy, Delivery or Practice:
Initiatives aimed at strengthening prohibitions on
physician self-referral have the potential to
influence overuse of care and spending, particularly
if they address self-referral among subgroups of
physicians and patients.
Funding Source(s): Stanford University, California
HealthCare Foundation
Poster Number: 6
Qualitative Research and Racial Disparities in
Screening Behaviors: Understanding Barriers and
Facilitators of Colorectal Cancer Screening
Arden Morris, M.D., M.P.H.; Elizabeth Becker,
M.P.H.; Heather Elliott, B.A.; Gwen Alexander,
Ph.D.; Derek Griffith, Ph.D.
Presented by: Elizabeth Becker, M.P.H., Doctoral
Student, Health Behavior/Health Education,
University of Michigan School of Public Health,
1415 University Heights, Ann Arbor, MI 48109;
Phone: (612) 590-1431; Email:
eabecker@umich.edu
Research Objective: Relative to Whites, African
Americans with colorectal cancer experience a 20%
higher cancer-specific mortality rate; however, even
among patients with medical insurance, African
Americans are less likely than Whites to undergo
colorectal cancer screening. Traditional theoretical
models of the risk-screening relationship may miss
patients who delay or avoid screening for colorectal
cancer in spite of their clinical status as high risk.
Our objective was to understand racial differences
in decision making factors around colorectal cancer
from the patient’s perspective, and to create a new
framework for understanding how African
Americans make decisions regarding colorectal
cancer screening.
Study Design: Using trained race-concordant
interviewers, we conducted semi-structured
interviews with a convenience sample of 29
individual African American and White patients who
had undergone colorectal cancer surgery.
Population Studied: White and African American
colorectal cancer surgery patients at a major urban
hospital.
Principal Findings: Thematic analysis of the data
suggested a wide range of influences on patients’
screening behaviors. While some patients were
proactive in the face of increased risk, others were
not, despite high levels of social support and
adequate insurance and access to care. Patients
exhibited a wide range of patterns of interactions
among relevant factors including social support,
medical insurance, access to care and disease risk.
These patterns differ from those established in the
literature and that have informed screening
interventions to date. For example, in contrast to
established beliefs, some patients equated disease
presence with illness, and failed to follow screening
guidelines because they felt healthy. Others report
delays in screening due to the burdens of other
medical conditions, fearing both the screening
procedure itself as well as facing yet another
disease.
Conclusions: Patient accounts of their decision
making processes surrounding screening behaviors
vary dramatically, suggesting the need for a
multifaceted approach to screening promotion.
Traditional models of decision making may not be
adequate to capture the broad range of factors
involved, as revealed by the qualitative data in this
study. Neither risk, social support, access to care
nor insurance has as strong of an influence on
decision making as what the literature suggests.
Using these data, we propose an alternative model
of screening behavior in both low and high risk
individuals accounting for variation within and
between racial groups as well as commonalities
across the study population. Patient accounts of
their own decision making behaviors offer a unique
lens through which to examine this process, and
can help institutions and practitioners tailor their
screening promotion messages.
Implications for Policy, Delivery or Practice: To
effectively address health and health care
disparities, we need to understand underlying
mechanisms of decision making and behavior.
Utilizing qualitative research methods can help
shed light on the ‘black box’ of decision making
processes that patients undergo as they consider
the burdens and benefits of colorectal cancer
screening.
Funding Source(s): American Cancer Society
Poster Number: 7
Relation of Health Insurance to Weight Status
Jean Bernhardt, Ph.D., M.S.N.
Presented by: Jean Bernhardt, Ph.D., M.S.N.,
M.H.S.A., Student, University of Massachusetts;
Phone: (978) 465-3989; Email:
jbernhardt@partners.org
Research Objective: The major purpose of this
study was to examine the relationship of health
insurance to weight status in 19-26 year olds using
body mass index (BMI) as the surrogate measure of
body weight. It was hypothesized that having health
insurance would predict lower BMI. The study was
conceptualized within an adapted social-ecological
framework with eM.P.H.asis on the interactions of
individuals and their environments.
Study Design: This secondary data analysis study
included data from 3804 young adults in the 2004
Medical Expenditure Panel Survey. Multivariate
logistic regression analyses (using STATA 8.0)
tested a non-recursive causal/path analytic model
and structural equations to explain the likelihood of
being obese (BMI=>30kg/m2). Youths’ race,
gender, and education, along with parental
education and obesity, were primary explanatory
variables. Being on a restricted diet was considered
a response to being obese. Sensitivity analysis was
conducted on causal effects of BMIs >35 kg/m2.
Population Studied: 19-26 year old young adults
with private or public health insurance
Principal Findings: Being Black (OR 1.90, CI 1.46,
2.48), having a high school diploma or less (OR
2.07, CI 1.40, 3.07), with at least one obese parent
(OR 1.99, CI 1.36, 2.92), all significantly increased
the probability of being obese. Physical activity (OR
0.59, CI 0.47, 0.73) reduced obesity likelihoods but
was less common in females and minorities. Private
insurance was associated with lower BMIs yet was
statistically insignificant (OR 0.82, 0.63, 1.05)
whereas young adults on public insurance (OR
1.40, CI 1.01, 1.95) had higher BMIs, suggesting
unobserved differences in socioeconomic status
related to type of insurance. However, at BMIs
greater than or equal to 35 kg/m2, private insurance
(OR 0.57, CI 0.38, 0.86) was negatively associated
and statistically significant at influencing lower BMI
levels.
Conclusions: Insurance is a complex measure
which reflected unmeasured characteristics
associated with the type of insurance individuals
had. Most importantly, having insurance did not
offset unmeasured factors that contributed to being
obese for those with public insurance.
Implications for Policy, Delivery or Practice:
Targeting health education to mitigate obesity,
prevent future intergenerational transfer of obesity,
and address the factors associated with obesity and
public insurance in young adults may be effective.
Understanding the relationship of health insurance
to weight status in young adults has the potential to
inform and guide public and health policy decisionmaking related to the utility of insurance to address
the epidemic of obesity and support clinical
interventions.
Poster Number: 8
The Hidden Beneficiaries: Population
Characteristics, Health Care Cost, and Health Care
Access for the Five Subpopulations Within
Medicare
Liz Blodgett, M.H.P.A.
Presented by: Liz Blodgett, M.H.P.A., Graduate
Research Assistant, Health Policy and
Administration, Washington State University, PO
Box 1495, Spokane, WA 99210; Phone: (509) 9533667; Email: lizblodgett@wsu.edu
Research Objective: The objective of this study is
to assess variation in population characteristics,
healthcare cost, and healthcare access for
Medicare beneficiaries in 2007 divided both by
eligibility and age.
Study Design: This study was a secondary data
analysis of the Medicare Current Beneficiary Survey
2007 Access to Care database. Beneficiaries were
sorted into one of five groups by entitlement and
age: end-stage renal disease-eligible beneficiaries
under 65, ESRD over 65, Social Security Disability
Insurance-eligible beneficiaries under 65, formerly
SSDI-eligible beneficiaries over 65, and retirees.
Group classification served as the independent
variable.Dependent variables were derived from the
MCBS Access to Care Codebook and grouped into
four major areas: population characteristics, health
and disability status, coverage and access, and
cost.
Population Studied: Full-year Medicare
beneficiaries interviewed in the 2007 Medicare
Current Beneficiary Survey.
Principal Findings: The groups were found to
differ significantly (p<.001) on population measures
including race/ethnicity, income level, education
level, and urban vs. rural residence; health and
disability status measures including number of
chronic conditions, number of ADLs, self-rated
health status, rates of diagnoses for mental
disorders and depression, and BMI; coverage and
access measures including rates of dual eligibility,
enrollment in Medicare HMOs and Medicare Part D,
and rates of reporting trouble accessing healthcare
or delaying care due to cost; and annual healthcare
expenditures. This study found that former SSDI
beneficiaries over age 65 - like younger SSDI
beneficiaries - are sicker, poorer, and more likely to
experience cost and access problems than retirees.
These results suggest that Medicare population
comparisons based on age or eligibility alone are
overlooking major subpopulation differences.
Conclusions about ESRD beneficiaries both under
and over age 65 were limited due to small sample
size. However, these beneficiaries have high rates
of comorbidity, poor self-rated health, and the
highest healthcare costs of any Medicare
subpopulation. For example, ESRD beneficiaries
had significantly higher annual levels of inpatient
reimbursement than SSDI beneficiaries of any age
or retirees; on average, $21,991 for ESRD under 65
and $10,291 for ESRD over 65 as compared to
$2,239 for SSDI under 65, $3,233 for SSDI over 65,
and $2,179 for retirees.
Conclusions: In differentiating Medicare
subpopulations, researchers and policymakers
traditionally contrast younger and older
beneficiaries. This division neglects the fact that
younger beneficiaries become older beneficiaries
and remain within the program. When subsumed
into the category of older beneficiaries, the cost and
use patterns of these sicker, poorer beneficiaries
may actually obscure the true degree of difference
between retirees and younger SSDI beneficiaries.
Implications for Policy, Delivery or Practice: The
health and population characteristics of SSDI
beneficiaries over age 65 make them more reliant
on Medicare than retirees, and therefore more
vulnerable to policy changes that restrict or reduce
coverage. As healthcare reform affecting Medicare
is enacted, policymakers should consider the needs
of these beneficiaries carefully. Medicare
beneficiaries with ESRD are typically removed from
comparisons entirely, due to their relatively small
numbers. However, the exceptionally high cost of
this subpopulation suggests that they have a
disproportionate impact on Medicare relative to their
population size.
Funding Source(s): National Institute on Disability
and Rehabilitation Research
Poster Number: 9
Making Sense of Online Patient Conversations
Karen Born, B.A., M.Sc.; Carlos Rizo, M.D.; Neil
Seeman, L.L.B., M.P.H.
Presented by: Karen Born, B.A., M.Sc., Ph.D.
student, Health Policy, Management and
Evaluation, University of Toronto, 155 College
Street West Suite 425, Toronto, M5T3M6; Canada;
Phone: (416) 710-2018; Email:
karen.born@utoronto.ca
Research Objective: To listen to patients' stories
online to complement current approaches to
gathering patient satisfaction data. Understand how
listening to patients' stories through online social
media can be part of the quality improvement
process of health care organizations. Test the
implementation of best practices in social media at
health care provider organizations and understand
the organizational context in which these practices
are being implemented.
Study Design: The study includes an
environmental scan of organizations within and
beyond healthcare who use social media to listen to
clients and customers. Next, case studies will be
developed using organizational partners. Case
studies will glean data through online reviews of
patient and public-generated data, as well as
interviews with senior leaders.
Population Studied: Health care organizations
within Ontario. Current partners include a
rehabilitation organization, a division of the Ontario
Ministry of Health and Long-Term care and
potential a mental health and addiction health care
provider.
Principal Findings: Preliminary research of patient
data online has found that patients express their
satisfaction with care providers through online
forums. Organizations who 'listen' to these online
voices can feed this information into quality
improvement cycles. Organizations, however are
cognizant of risk management and have concerns
about the privacy and influence of social media
forum. Patient conversations online, however, can
be measured and tools used in this study
demonstrate the influence and penetration of these
conversations.
Conclusions: Case studies developed will test
emerging best practices and guidelines on quality
improvement and patient-centred care using social
media. Case studies will also detail organizational
journeys in using this kind of information to inform
quality improvement, and will describe the barriers,
enablers and challenges in this journey.
Implications for Policy, Delivery or Practice:
This work will inform the development of an
eToolkit, aimed at sharing insights and lessons from
patient and caregiver stories online and offers ideas
and guidance to other care providers on how to
leverage online patient data to support
organizational improvement.
Funding Source(s): The Change Foundation
Poster Number: 10
Patient- and Family-Centered Care - The Impact on
Patient Safety: A Comparison Study of Intensive
Care Units at an Academic Medical Center
Barbara Brumbaugh, M.P.H., Patricia Sodomka,
M.H.A.; Janice Probst, Ph.D.
Presented by: Barbara Brumbaugh, M.P.H.,
Health Services Policy and Management, University
of South Carolina-Arnold School of Public Health,
122 Twin Creek Farm Road, Aiken, SC 29805;
Phone: (803) 640-7763; Email:
blbrumbaugh@bellsouth.net
Research Objective: Patient- and Family-Centered
Care (PFCC) was called out by the Institute of
Medicine in 2001 as one of the six aims in
improving the quality of health delivery. Despite the
fact that there is growing recognition that this model
of care results in better outcomes, decreased costs,
and greater clinician and patient/family satisfaction,
there is still limited research that demonstrates
these results. This study examines performance
and patient safety measures and outcomes in Adult
ICUs at an academic medical center.
Study Design: Data was mined from the University
HealthSystem Consortium (UHC), Clinical Data
Base (CDB). The first study cohort, n=705, was
derived from the General Neurology and
Neurosurgery Product Line and the second study
cohort, n=440, from all other product lines excluding
general neurology, neurosurgery and pediatrics.
Using these cohorts facilitated comparisons of
outcomes for patients in the Neurosciences ICU
with those for patients in the Shock Trauma ICU. To
acknowledge and to better address the remaining
differences in the study populations in these two
intensive care units, UHC data were used to
compare the observed outcomes in each of the two
populations to outcomes predicted by UHC models
for those specific populations. Length of stay (LOS),
expected LOS per Medicare Severity Diagnosis
Related Group (MS DRG) observed costs
(calculated using costs to charge ratios), expected
costs per MS DRG, potentially avoidable postprocedure complications, and safety indicators were
analyzed by statistical methods, examining and
comparing frequency, mean, and other measures
utilizing SAS version 9.2. P values less than or
equal to 0.05 were considered statistically
significant.
Population Studied: The target population was all
adult patients discharged between January 1, 2007
and December 31, 2007 at MCGHealth with 1 to 75
ICU days and attesting physician specialty of
neurology, neurosurgery, or trauma.
Principal Findings: Costs compared to expected
costs per MS DRG were less for patients spending
time in the Neurosciences ICU. The observed LOS
as compared to the expected LOS was found to be
less for patients having stays in the Neuroscience
ICU as compared to patients having stays in the
Trauma ICU. Patients in the Neurosciences ICU
experienced fewer complications than those in the
Trauma ICU. Patients in the Neuroscience ICU had
less safety issues relative to the AHRQ safety
indicators than those in the Trauma ICU.
Conclusions: This study suggests that Patientand Family-Centered Care has a positive effect on
performance measures such as LOS and costs for
ICU patients. It further suggests that patients
receiving PFCC experience less complications and
safety issues than those patients who stayed in
units where PFCC is not universal.
Implications for Policy, Delivery or Practice:
While further research is needed this area, the
findings strongly suggest that the patient- and
family-centered care model should be implemented
across the continuum of healthcare, even in the ICU
to improve patient safety and reduce costs.
Poster Number: 11
Growing Old Together? Exploring the Relationship
Between Social Capital and Nursing Home
Institutionalization among Elderly in the United
States
Rebecca Cadigan, M.Sc.
Presented by: Rebecca Cadigan, M.Sc.,
Graduate Student, Health Policy Ph.D. Program,
Harvard University, 14 Story Street, 4th Floor,
Cambridge, MA 02138; Phone: (617) 484-0848;
Email: rcadigan@fas.harvard.edu
Research Objective: As Americans live longer,
elders experience increasingly diverse long-term
care needs. While people with significant
impairments require the prosthetic physical
environment and around-the-clock medical
monitoring of a nursing home (NH), many people
with lesser impairment can remain in their homes
with minimal support. Past research has shown that
there are numerous social and economic benefits to
allowing elders to select long-term care services
that enable aging-in-place. Home and communitybased services have the potential to allow elders to
receive care while preserving ties with family,
friends, and neighbors, and maintaining
connections to religious institutions, social groups,
and other community organizations. While social
integration is a well-studied correlate of healthy
aging at the individual-level, less is known
regarding the relationship between area-level social
capital and patterns of long-term care utilization.
The objective of the present study is to explore the
relationship between social capital and risk of
institutionalization of elderly across all U.S.
counties. It is hypothesized that areas with lower
levels of social capital will have higher rates of
elders admitted to nursing homes unnecessarily
(referred to as “low-care”). This hypothesis is based
on the theory that areas with higher levels of social
capital are more likely to have resources that allow
elders to age in place, such as active civic groups,
robust religious communities, accessible
transportation networks, generous social services,
safe neighborhoods, and opportunities for
volunteerism.
Study Design: The percentage of low-care NH
residents in each county was obtained from Brown
University’s Long-Term Care: Facts on Care in the
U.S. Website. County-level social capital data were
obtained from the Northeast Regional Center for
Rural Development. The addresses of all licensed
NHs were obtained from the Centers for Medicare
and Medicaid Services Nursing Home Compare
database, and geocoded using ArcGIS. Univariate
and multivariate linear regression were used to
explore the relationships between social capital
indicators and prevalence of low-care NH residents
by county. Maps were created using ArcGIS to
depict the geographic variations in social capital
and prevalence of low-care NH residents across the
U.S.
Population Studied: The population of interest is
person over age 64 residing in the U.S. The unit of
analysis for the present study is all U.S. counties.
Principal Findings: Number of civic and
professional organizations per capita were both
inversely associated with percentage of low-care
NH residents at the county-level, controlling for
median income, proportion of the population over
age 65, and population density. This suggests that
counties with a greater relative number of civic and
professional organizations have significantly lower
rates of unnecessary NH admission. It is unclear
whether the aging-in-place of elders is contributing
to higher social capital in these areas, or if elders
living in communities well endowed with social
capital are less likely to enter NHs unnecessarily. It
is also possible that there are exogenous factors
mitigating these relationships.
Conclusions: Several social capital indicators were
associated with prevalence of low-care NH
residents at the county level, namely number of
civic and professional organizations. This suggests
that there may be area-level predictors of aging-inplace, and further research is needed to identify the
direction and mechanisms of these relationships.
Implications for Policy, Delivery or Practice: The
identification of area-level characteristics
associated with prevalence of low-care NH
residents may help to inform policy decisions
regarding long-term care services and financing.
Funding Source(s): AHRQ
Poster Number: 12
State’s Budget Cut Matters for Uncompensated
Care Spending?
JongWha Chang, M.A., M.S.P.H.; SunJung Kim,
B.S.; Isha Patel, M.S.; Rajesh Balkrishnan, Ph.D.
Presented by: JongWha Chang, M.A., M.S.P.H.,
Graduate Fellow, Social and Administrative
Science, The University of Michigan at Ann Arbor,
428 Church Street, Ann Arbor, MI 48109; Phone:
(734) 936-1505; Email: jochang@umich.edu
Research Objective: The study examined the
impact of state budget cuts on uncompensated care
of general acute care hospital organizations. This
study capitalized on the variations in the state of
Texas, and California to form a natural experiment
testing the joint impact of budget cut status on
uncompensated care costs, as well as specific
charity care costs and bad debt expenses from
indigent patients.
Study Design: The budget cuts in the state of
Texas occurred in the year 2004. Information was
obtained from the Texas Department of Health
(TDH) and California Department of Health
Services (CDHS) regarding financial characteristics
of hospitals, and the American Hospital Directory
(AHD) annual survey regarding hospital
organizational characteristics. We created three
dependent variables; RUCGPR (the ratio of the
total uncompensated care costs to gross patient
revenue), RCHRGPR (the ratio of charity care to
total patient revenue), and RBDGPR (the ratio of
bad debt expenses to gross patient revenue). Using
a two-period panel data and individual hospital fixed
effects, we captured hospital uncompensated care
spending that could also have influenced budget cut
status. Additionally, the impact of the state budget
cut status on hospitals’ uncompensated care
spending, charity care spending, and bad debt
expenses were also similarly estimated.
Population Studied: The number of public, not for
profit and for profit hospitals who completed the
annual survey in the sample periods, 2002-2005
[416 (Texas) and 352 (California)].
Principal Findings: For state of Texas, results
from the fixed effects model confirmed that the year
2005 was directly related to increasing RUCGPR,
and RCHRGPR. The coefficients of year dummy of
2005 were significantly and positively associated
with RUCGPR (0.4282, p-value <0.05) and
RCHRGPR (0.29, p-value <0.1). These results
support the findings that the RUCGPR (RCHRGPR)
would be more positively associated with 2005 than
any other year with other things being equal.
However, for the state of California, even though
the coefficient of year dummy of 2005 was
significant and positively associated with
RCHRGPR (0.319, p-value <0.05), the overall
coefficient of uncompensated care spending was
not statistically significant for 2005.
Conclusions: The healthcare industry is
characterized by increased regulation, a growing
number of uninsured patients, increasingly stringent
reimbursement, and competitive practices among
hospitals and other providers. Federal and state
healthcare agencies are increasing the standards of
eligibility for outlier payments and uncompensated
care provisions. Tax exempt status of many NFP
hospitals is being examined and tied to specific
performances, particularly the provision of
uncompensated care. This study provides empirical
evidence of the impact of budget cut status as
reported by Texas and California general acute
care hospitals. Characteristics of those hospitals
using RUCGPR, RCHRGPR, RBDGPR ratios and
other market factors are presented and analyzed.
Implications for Policy, Delivery or Practice:
This study contributes to prior healthcare finance
research for hospitals’ uncompensated care
spending as impacted by state regulations. This
study finds that if the state has budget cut crisis at
specific period, RUCGPR and RCHRGPR
significantly impact the budget cut crisis in the same
direction. However, if state does not have budget
cut crisis at specific period, RUCGPR and
RCHRGPR do not significantly impact the budget
cut status. The study finds that there is evidence to
support hypothesis that the state budget cuts
reduce the levels of uncompensated care.
Funding Source(s): The University of Michigan at
Ann Arbor, College of Pharmacy
Poster Number: 13
Shopping Behavior of Colon and Rectum Cancer
Patients for Health Care Services
Ruei-Yi Chang; Chih-Liang Yaung; Shang-Jyh
Chiou, Ph.D.
Presented by: Ruei-Yi Chang, Student,
Healthcare Administration, Asia University, 500,
Lioufeng Road, Wufeng, Taichung 41354, Taiwan;
Phone: +8864233234561742; Email:
j76926@hotmail.com
Research Objective: Shopping behavior for
healthcare services is a special phenomenon in
Taiwan because patients have moderate accesses
to low costing healthcare under the National Health
Insurance (NHI) program. The malignant tumor in
colon and rectum cancer has occupied the 3rd
place as cause of death since 1995. The mortality
rate of colon and rectum cancer was 14.2/10000 in
1999, female had higher rate. The goal of this study
was to investigate the status of shopping behavior
for healthcare services of newly diagnosed colon
and rectum cancer patients.
Study Design: The study used the cancer dataset
from National Health Research Institutes (NHRI)
which included all claim information, registry for
contracted medical facilities and registry for
beneficiaries in 2006. The definition of shopping
behavior in this study is defined as inquiring into the
services more than 2 providers for test and
confirmation before the regular treatment in the first
time which determined from the surgical code (V57,
V58), radio therapy and chemotherapy.
Population Studied: New cases of colon and
rectum cancer were extracted using the Principal
Diagnosis of ICD-9-CM codes 153-154 and A-code
A093 and A094, with no relative diagnosis record
before 13 months from the cancer dataset. There
were 2,580 cases in this study.
Principal Findings: There were 139 cases fit the
definition of shopping behavior, male was 44.6%
(n=62). The majority was 40~64years old (56.12%).
The average duration of shopping behavior was
165 days as compared to 72 days from those who
did not. The average times of visits were 25, and
those who did not were 7. The average medical
expenditures of shopping and not shopping
behavior were US$2,842 and $730, respectively.
The average number of hospital visiting was 3.44.
Most cases occurred in medical centers because
they believe that medical center can offer reliable
services.
Conclusions: This study finds that shopping
behavior incurs extraordinary expenses and is a
torment problem for NHI in Taiwan when the cost
control pressure roars with years. Those patients
also take high risk in the delaying cancer treatment
especially the better prognosis in the regiment
beginning early.
Implications for Policy, Delivery or Practice:
Further investigations need to explore the impact
factors related to the shopping behavior. More
research needs to realize the relevant factors either
from providers or patients and the influence to the
survival rate.
Funding Source(s): Asia University
Poster Number: 14
Suicide Mortality Trends by Sex, Age, Methods,
Season and Region in Taiwan, 1986-2007
Wei-ting Chang, M.S ; Chien Wu Chien, Ph.D.
Presented by: Wei-ting Chang, M.S, Graduate,
Public Health, Natinal Defence Medical Center,
No.161, Section 6, Min-Chuan East Road, Taipei,
114 Taiwan; Phone: +886287923100; Email:
moyen1985@msn.com
Research Objective: To examine the trends in
suicide mortality in Taiwan by sex, age, methods,
season, and region between 1986 and 2007.
Study Design: Analysis of routine mortality data
from the official publications of vital statistics from
1986 to 2007 to determine trends by sex, age,
method, season and region. Linear regression was
used to test the trends.
Population Studied: Population base
Principal Findings: A total of 47,148 people in
Taiwan died as a result of suicide between 1986
and 2007, and 66.65% of them were males.
Hanging was the most common method of suicide,
especially among more than 65-year-old elderly.
However, the most common method of suicide
among 25-49 year of age was poisoning by other
gases and vapors. In one year, May to September
were higher months of suicide mortality rate. During
study period, suicide mortality rate of Taitung
County was the highest and decreased, but Nantou
County, Koahsiung City and Keelung City
increased.
Conclusions: Total suicide mortality rate rose, and
male was higher than female. Suicide mortality rate
of middle age increased significantly. Methods of
suicide changed from poisoning by solid or liquid
substances into jumping from high place and
poisoning by other gases and vapors.
Implications for Policy, Delivery or Practice: The
prevention programs should focus on male and who
suicide by jumping from high place and poisoning
by other gases and vapors.
Poster Number: 16
The Impacts of Depression and Anxiety on Elderly
COPD Patients in Taiwan
Ling-hsuan Chang; Ying-Chun Li, Ph.D.
Presented by: Ling-hsuan Chang, Graduate
Student, Institute of Health Care Management,
National Sun Yat-Sen University, 70 Lien-Hai Road,
Kaohsiung, 804 Taiwan; Phone: +88675252000
ext. 4870; Email: ksamuel0917@gmail.com
Research Objective: Chronic obstructive
pulmonary disease (COPD) is a major cause of
disability and death worldwide. The prevalence of
depression and anxiety in COPD patients is
generally higher than those reported in other
chronic illnesses, and these mental diseases
significantly affect the health expenditures and the
course of the diseases on elderly COPD patients.
The COPD patients with co-morbid depression,
anxiety or other chronic diseases are more impaired
and have 50% to 100% higher medical costs than
those without depression and anxiety. COPD
affects approximately 1.8 million people in Taiwan
every year, however, little studies have examined
the impacts of mental diseases on elderly COPD
patients. This study aims to investigate the
depression and anxiety status on COPD patients’
medical care utilizations and health outcomes.
Study Design: This study analyzed the nationally
representative research database from National
Health Insurance program between 2002 and 2006
in Taiwan, by exploring outpatient visits,
hospitalizations, and medical expenditures of
COPD and its associated co-morbidities. The data
analyses were carried out by Chi-square test,
multiple linear regression and multivariate logistic
regression.
Population Studied: Patient aged over 65 with
primary diagnosis of COPD and co-morbidity
diseases based on ICD-9-CM in Taiwan. To
investigate the complications in COPD patients,
four diseases such as diabetes, hypertension, heart
failure and ischemic heart disease were included in
the analyses.
Principal Findings: Among the 131,846 COPD
patients (63.3%, male), most of them were aged
from 65 to 84 (65.9%) and lived in northern Taiwan
(44.8%). The results indicated that those who were
suffered from depression or anxiety increased
medical utilization and costs. Multiple linear
regression model showed that age, locations, and
hospital level significantly affect the health
expenditures of COPD patients with depression or
anxiety (p<0.001). Multivariate logistic regression
model presented that length of stay in hospital and
number of complications significantly affect the
probability of hospitalization among the COPD
patients with mental diseases (p<0.001).
Conclusions: COPD symptoms were usually
getting worsen over time and depression or anxiety
affect the COPD patients significantly, though the
mental diseases are usually under-estimated. The
study results demonstrated that COPD patients
suffered from depression and anxiety increased the
risks on longer length of stay, higher medical
expenditures and more complications.
Implications for Policy, Delivery or Practice:
WHO predicts that COPD will become the third
leading cause of death worldwide by 2030. Since
mental diseases may worsen elderly COPD
patients’ health outcomes, physicians are advised
to pay more attention to these types of
patients.With on time and appropriate treatments,
thus may help to improve quality of care and reduce
medical expenditures for these patients.
Funding Source(s): National Science
Council,Taiwan
Poster Number: 15
Linking RNs’ Perception of the Work Environment
to Job Satisfaction in Nursing Homes
JiSun Choi, M.S.N .; Linda Flynn, Ph.D.; Meg
Johantgen, Ph.D.
Presented by: JiSun Choi, M.S.N., Doctoral
student, University of Maryland School of Nursing,
655 West Lombard Street, Room 475, Baltimore,
MD 21201; Phone: (443) 320-3079; Email:
jchoi008@son.umaryland.edu
Research Objective: Consistent with the nursing
practice environments in acute care settings, work
environment is assumed to be associated with job
satisfaction among nurses working in nursing
homes. Yet the specific characteristics of the
practice environment that impact job satisfaction
have not been identified. The objective of this study
was to quantify the effects of various dimensions of
the practice environment on job satisfaction among
Registered Nurses (RN) in nursing homes
controlling for individual and nursing home
characteristics.
Study Design: A descriptive correlational design
was used to conduct a secondary analysis of data
from two sources: the parent study that conducted a
statewide survey of registered nurses in New
Jersey and the Online Survey, Certification, and
Reporting (OSCAR) data. Two-level hierarchical
modeling was used to examine the relative impact
of 5 dimensions of the nursing practice environment
while accounting for the clustering of nurses in
facilities.
Population Studied: Of the approximately 25% of
all New Jersey RNs (n = 22,406) that participated in
the survey, 340 RNs working as direct care
providers in nursing homes where there were at
least 4 RN respondents, were used for this study.
These represent 63 skilled nursing facilities.
Principal Findings: Controlling for individual and
nursing home characteristics, nurse participation in
facility affairs as well as staff and resource
adequacy were positively associated with RNs’ job
satisfaction. Ownership status was significantly
related to job satisfaction; RNs working in for-profit
nursing homes were less satisfied.
Conclusions: As in acute care settings,
opportunities to participate in facility affairs and
having enough staff contribute significantly to higher
job satisfaction among RNs working in nursing
homes.
Implications for Policy, Delivery or Practice:
Unlike other nursing home characteristics,
characteristics of the practice environment can be
modified through administrative actions.
Administrative initiatives to enhance RN job
satisfaction by creating a more supportive practice
environment should include increasing RNs'
opportunities to provide meaningful input into
organizational decision-making, and ensuring the
adequacy of available resources. Embarking on the
Magnet Accreditation journey is one mechanism by
which hospitals have dramatically enhanced all
dimensions of their practice environments, and is
an initiative that should be seriously considered by
nursing home administrators.
Funding Source(s): RWJF, The National Institute
Nursing Research (# NR004513)
Poster Number: 18
Epidemiology of Hospitalized Burn Patients in
Taiwan, 2007
Wu-Chien Chien, Ph.D.; Chi-Hsiang Chung; Lu Pai,
Ph.D.
Presented by:, Student, Graduate Institute of Life
Sciences, National Defense Medical Center, School
of Public Health 4209R, No. 161, Section 6, MinChuan East Road, Taipei, 11490 Taiwan; Phone:
+886287923100 ext. 18445; Email:
g694810042@ndmctsgh.edu.tw
Research Objective: The objectives of this study is
to find the high-risk groups of death caused by burn
inpatient and medical services utilization of
hospitalized burn patients, and the factors
associated with hospital mortality caused by burn
injuries.
Study Design: This research brought “inpatient
expenditures by admissions (DD)” and “registry for
contracted medical facilities (HOSB)” from National
Health Insurance Research Database in 2007 into
analysis by using SPSS 14.0 software. We defined
burn cases as ICD-9-CM N Code 940-949.
Principal Findings: The total inpatient rate was
31.04 per 100,000, and males were higher than
females (males 37.58 per 100,000, females 31.04
per 100,000). The inpatient rate of child form 0
through 4 years old was higher than any other age
groups. Chiayi City had the highest inpatient rate
(73.97 per 100,000), followed by Ilan County (61.90
per 100,000), Hualien County (50.98 per 100,000)
and so on. There were 7,126 inpatient cases
caused by burn injuries (males 61.23%, females
38.77%), the average age of them were 39.73
years old, and each person had 2.5 diseases or
injuries. Among hospitalized burn patients, most
parts of “site of burn injuries” were limbs (48.12%),
“TBSA (total body surface area)” were <10%
(39.29%), “degrees of burn injuries” were second
degree (57.08%), and “causes of burn” were
unintentional burn (89.87%). Among hospitalized
burns patients, 53.79% (3,833 people) were
consulted at plastic surgical department, 41.36%
(2,947 people) were hospitalized in the medicine
center, and they received 1.06 surgeries or
operations in average; the mean of length of stay
was 13.07 days, and the median of medical
expenditures was 33,744.5 NT dollars. The number
of deaths caused by hospitalized burn was 138
people. The factors associated with hospital
mortality caused by burn injuries were ages, the
number of other diseases or injuries, sites of burn
injuries, intentional burn, hospital level, and the
number of operations and treatments by
multivariate logistic regression analysis; the
explanation of overall model was 43.5%.
Conclusions: All factors associated with
hospitalized burn injury mortality included ages, the
number of other diseases or injuries, sites of burn
injuries, intentional burn, hospital level, and the
number of operations and treatments by using
multivariate logistic regression analysis.
Implications for Policy, Delivery or Practice: In
order to reduce burn mortality rates effectively, the
government and NGOs should continue to promote
the burn prevention programs together, especially
for the high-risk of death caused by burn inpatient
groups (0-4 age groups)
Poster Number: 19
Trends of Fall Mortality in Taiwan, 1986-2007
Wu-Chien Chien, Ph.D.; Chi-Hsiang Chung; PingHsun Wu; Chien-Cheng Lai; Lu Pai, Ph.D.
Presented by: Chi-Hsiang Chung, Student,
Graduate Institute of Life Sciences, National
Defense Medical Center, School of Public Health
4209R, No. 161, Section 6, Min-Chuan East Road,
Taipei, 11490 Taiwan; Phone: +886287923100 ext.
18445; Email: g694810042@ndmctsgh.edu.tw
Research Objective: The objectives of this study
were to describe the trends in mortality rates due to
fall injury by sex, age, causes, and occupation in
Taiwan, 1986-2007.
Study Design: Data were obtained from official
Vital Statistics System for the period from 1986 to
2007. The types of “fall” include “fall from different
level (E880, E881, E882, E884)”, “fall on the same
level (E885, E886)”, and “other and unspecified fall
(E883, E887, E888)” by using ICD-9-CM. The
trends of mortality rates were tested with “curve
estimation” by using SPSS 14.0 software.
Principal Findings: There were 26,973 deaths
caused by fall in Taiwan in the 22-year period, and
the predominant cause of fall was “fall from one
level to another (E884)” (30.26%). The mortality
rates in males were higher than females (males
8.05 per 100,000; females 3.23 per 100,000), and
without significantly change through the years.
Elders aged over 70 years have the highest
mortality rate (49.03 per 100,000). Among people of
70-year old or older, the mortality rates of “fall on
the same level” descended in both genders,
whereas; the mortality rates of “fall from different
level” ascended in males, but not in females. The
fall mortality rates varied with types of occupational
groups. The highest rate was found with “nonphysical strength labors”, followed by “physical
strength labors”. The “managers or administrators”
had the lowest fall mortality rate.
Conclusions: The mortality rates in males were
higher than females, and without significantly
change through the years. The fall mortality rates
varied with types of occupational groups, and The
highest rate was found with “non-physical strength
labors”.
Implications for Policy, Delivery or Practice: In
order to reduce fall mortality rates effectively, we
should continue to promote the fall prevention
programs for the elderly. However, different target
groups should be differentiated for “fall from the
same level” and” fall from different level”.
Furthermore, safety education for non-physical
strength labors should also be enhanced.
Poster Number: 20
Medicare Advantage Enrollees in Florida: Are They
Healthier than Traditional Medicare Enrollees?
Meg Comins, M.P.A .; Barbara Orban, Ph.D.;
Etienne Pracht, Ph.D.
Presented by: Meg Comins, M.P.A., Instructor,
Health Policy and Management, University of South
Florida, 13201 Bruce B Downs Boulevard MDC56,
Tampa, FL 33612; Phone: (813) 313-7732; Email:
mcomins@health.usf.edu
Research Objective: The Medicare Modernization
Act of 2003 (MMA) increased plan payments to
private Medicare Advantage (MA) plans. Opponents
claim that these plan enrollees are healthier than
traditional Medicare (FFS) enrollees, contributing to
higher costs for these plans. Proponents claim that
MA plans lead to a reduction in racial and SES
health disparities. MA plans have lower premiums
and out-of-pocket expenses than FFS. A reduction
in MA payments of $400 billion over the next
decade is recommended as part of health care
reform. This paper examines trends in patient
clinical and demographic characteristics over time
for Medicare enrollees to determine if healthier
beneficiaries are enrolling in MA plans.
Study Design: The primary source of data is
hospital discharge and financial reports from the
Florida Agency for Health Care Administration.
Primary diagnosis, admission source, length of
hospital stay, and age of FFS and MA enrollees are
compared. ANCOVA was used to test for
differences between the groups.
Population Studied: Florida Medicare and
Medicare Advantage enrollees were included in the
study. Counties with more than 40% or less than
5% of Medicare eligibles enrolled in MA plans were
excluded. Broward and Miami-Dade counties were
the highly concentrated counties, and 23-35 rural
counties had few or no MA enrollees over the study
period. The access to MA plans for rural eligibles
has not increased since the legislation.
Principal Findings: More MA patients are 79 or
younger. MA patients are more likely to be admitted
to an Emergency Department (p<.001), with the
difference between groups decreasing over time.
MA patients are more likely to spend less than 2
days, and less likely to spend 7 or more days in the
hospital, p<.001.
Conclusions: More admits from the emergency
department may be explained by more urgent
conditions for MA patients, but could also point to
limited access to care. The increase in admissions
for MA patients reflects the growth in private fee-orservice (PFFS) MA plans, where traditional
managed care tools like selective contracting are
not used. These plans are more attractive to sicker
patients that would have chosen FFS in the past to
keep their provider. As more eligibles switch over to
MA plans, the differences in health status will likely
decline even more.
Implications for Policy, Delivery or Practice:
Medicare without supplemental insurance covers
about one half of an enrollees medical expenses.
MA plans are attractive to lower SES groups that
can’t afford supplemental premiums. Higher ED
admissions for MA enrollees may support past
research on the association between lower SES
and reliance on the ED as a source for routine
health care. Problems with access to care can be
partly attributed to low payments to physicians from
the MA plans. A solution may be to require that a
portion of payments to MA plans be used to
increase plan payments to primary care physicians;
this would likely increase access to care and reduce
emergency department use. Future research will
address practice patterns for lower SES enrollees
to determine how to make primary care as
accessible as ED care.
Poster Number: 21
Does Receiving Primary Care from a Geriatrician
Affect Health Care Use and Expenditures for Frail
Elders and Elders at the End of Life?
Laura D'Arcy, M.P.A.
Presented by: Laura D'Arcy, M.P.A., Ph.D.
student, Health Policy and Management, University
of North Carolina at Chapel Hill, CB #7411, Chapel
Hill, NC 27599-7411; Phone: (330) 606-0392;
Email: ldarcy@email.unc.edu
Research Objective: Older adults generally rely on
internal medicine or family medicine physicians for
primary care; a small share receives primary care
from geriatricians. Geriatricians have a number of
skills that distinguish them from other primary care
providers (e.g., expertise in managing geriatric
syndromes and multiple medications, assisting endof-life decision-making). Primary care from a
geriatrician may be particularly beneficial for frail
elders and those at the end of life. Evidence on the
effect of geriatricians comes from small randomized
controlled trials of geriatrics evaluation and
management programs that offer conflicting results.
Studies also suggest that primary care plays an
important role in health care use and expenditures,
but existing analyses combine specialties. The
objective of this study is to separately estimate the
effect of geriatrician use on health care use and
expenditures from internal medicine and family
medicine physician use.
Study Design: Using 2002-2007 Medicare claims
data, I compare the associations between primary
care provider specialty and health care use and
expenditures among six groups of elderly
beneficiaries: high geriatrician, medium geriatrician,
low geriatrician/high internal medicine, low
geriatrician/high family medicine, no geriatrician/
high internal medicine, and no geriatrician/high
family medicine use. Outcome variables include
emergency department use, inpatient admissions
for ambulatory care sensitive conditions, in-hospital
death, and inpatient and total expenditures. Since
primary care provider specialty may be related to
unobservable factors that affect use and
expenditures (e.g., supplemental health insurance),
I use instrumental variables to exogenously predict
provider specialty based on the share of primary
care doctors in the area that are geriatricians.
Analyses are conducted separately for frail elders
and those in the last year of life.
Population Studied: The sample is drawn from
nearly one million elderly Medicare beneficiaries
who had an inpatient admission for acute coronary
syndrome in 2003-2004. Approximately 125,000 of
these beneficiaries are considered to be frail, and
more than half a million died during the study
period. In 2003 and 2004, nearly 20,000
beneficiaries had at least one geriatrician visit.
Principal Findings: Analyses are presently in
process. Based on the literature and descriptive
results, I expect that elders who use a geriatrician
may be less likely to have an emergency
department visit or inpatient admission for an
ambulatory sensitive condition and lower
expenditures than those who rely on an internal
medicine or family medicine physician.
Conclusions: Primary care from a geriatrician may
play an important role in reducing preventable
health care use and expenditures for frail elders
and those at the end of life. Treating primary care
provider specialty as endogenous to the outcome
equation is likely important to obtaining an unbiased
estimate of the relationship between provider
specialty and health care use and expenditures.
Implications for Policy, Delivery or Practice:
Estimates indicated that by 2030, 36,000
geriatricians will be needed; current graduation
rates from geriatrics fellowship programs suggest
there will be only 7,750 geriatricians. Examining
whether elders who receive primary care from a
geriatrician have lower and less costly health care
use than those who rely on other specialties is
critical to determining whether and how drastically
to change incentives to encourage more physicians
to pursue geriatric certification.
Funding Source(s): NIA
Poster Number: 22
Social Capital and Mental Health: An Examination
of the Relationship among Neighborhood
Conditions, Primary Caregiver Depression, and
Childhood Behavioral Problems
Ayesha Delany-Brumsey, M.A.,Vickie Mays, Ph.D.,
M.S.P.H.
Presented by: Ayesha Delany-Brumsey, M.A.,
C.Phil., Graduate Student, Clinical Psychology,
University of California, Los Angeles, 1285 Franz
Hall, Box 951563, Los Angeles, CA 90095; Phone:
(301) 928-8408; Email: adelanybrumsey@ucla.edu
Research Objective: Previous research has
established the link between where we live and our
mental health. However, relatively little is known
about the particular aspects of neighborhoods
which may be important in determining this
relationship. The primary purpose of this study is to
examine how the environment of the neighborhood,
both economic and social, can be a contributing
factor in the development of child behavior
problems. The study also explores the contribution
of individual level characteristics including child
demographics, caregiver demographics, and
caregiver depression to child behavior problems.
The proposed model acknowledges that children’s
mental health is influenced both by their caregivers,
and by the context in which their relationship with
their caregiver takes place.
Study Design: The data is from the Los Angeles
Neighborhood and Family Survey (L.A. FANS), a
study of Los Angeles County families and their
neighborhoods. The data used here is from Wave 1
of the study which was collected from April 2000 to
January 2002 (Peterson et al., 2004). The data
were analyzed using two-level hierarchical linear
models with robust standard errors in which
children are clustered within neighborhoods. The
dependent variable is child behavior problems as
measured by the Behavior Problems Index
(Peterson & Zill, 1986). The individual level
predictors are child and caregiver demographic
characteristics including family income, child
gender, child and caregiver race/ethnicity, caregiver
marital status, caregiver educational attainment,
and family income. Caregiver depression was also
an individual level predictor and was measured
using the CIDI-SF (Kessler et al., 1998). The
neighborhood level predictors are neighborhood
socioeconomic disadvantage, residential instability,
and social capital. The questions used to measure
neighborhood social capital were first published in
Sampson et al. (1997) and Sampson et al. (1999).
Population Studied: The participants are 1,535
children between the age of 3 and 17 (M = 8.31, SD
= 5.14). The demographic makeup of the sample is
as follows: 49% of the children are male and 51%
female; 54.62% are Latino, 25.75% White, 9.81%
Asian/Pacific Islander, 8.31% Black and 1.51%
Native American.
Principal Findings: Child behavior problems were
not significantly associated with any of the
neighborhood level socioeconomic characteristics
or neighborhood level social capital. However,
behavior problems were significantly predicted by
primary caregiver depression. As the number of
symptoms endorsed on the CIDI-SF increased so
do the number of child behavior symptoms. In
addition, younger children displayed more behavior
problems than older children. Also, children of
caregivers with less than a high school education
had more behavior problems than children of more
educated caregivers.
Conclusions: Child behavior problems was not
associated with any of the neighborhood level
variables. However, behavior problems were
significantly associated with the family level variable
of primary caregiver depression.
Implications for Policy, Delivery or Practice: The
results of this study suggest that one way to reduce
the incidence of child behavior problems is to
provide supportive services to the caregivers.
Services such as mental health counseling, or
parent training may reduce the incidence of
depression among caregivers which in turn could
reduce child behavior problems among the children
they are responsible for.
Poster Number: 23
Quality Trend in Nursing Homes
Nailya DeLellis
Presented by: Nailya DeLellis, Ph.D. Student,
Health Administration, Virginia Commonwealth
University; Email: delellisno@vcu.edu
Research Objective: Does change in quality of
care depends on nursing home organizational
characteristics?
Study Design: The study featured a longitudinal
comparison of quality measures provided by the
Online Survey Certification and Reporting (OSCAR)
database
Population Studied: The population observed was
taken from the OSCAR database. Excluded from
analysis were nursing homes that were hospital
based and size outliers.
Principal Findings: Regarding process and
outcome quality, there was a positive correlation
between quality measures and staff (RNs LPNs
aides) to patient ratios, confirming what has been
long established in the literature. A second finding
was that belonging to a system, larger size, and
higher occupancy had a negative correlation with
quality. Definitions for process and outcome quality
were those found in the National Quality
Clearinghouse website. There were 5 process
measures (use of catheters, use of restraints, flu
and pneumonia vaccination and pain management)
and 6 outcome measures (presence of ulcers,
diagnosis of depression, bowel and bladder
incontinence, weight loss and bedfast). For all
measures average values were calculated. If a
nursing home facility had a measure in the upper
quartile, it received a score of 1. So each nursing
home had 2 scores, one for process and one for
outcome, and changes in those score will
compared.
Conclusions: to be described later based on the
findings
Implications for Policy, Delivery or Practice: to
be described later based on the findings
Poster Number: 25
Implications of Applying New WHO Growth
Standards on Assessment of Nutritional Status
among Under-Five Children in a Rural Health
Center in India.
Avika Dixit, M.B.B.S.; Uzma Khan, M.B.B.S.; Jane
Philip, M.B.B.S.
Presented by: Avika Dixit, M.B.B.S., Student,
Harvard School of Public Health, 199, Park Drive,
Apartment 133, Boston, MA 02215; Phone: (617)
834-8576; Email: adixit@hsph.harvard.edu
Research Objective: World Health Organization,
after a review of NCHS anthropometric references
of 1977 concluded that these did not adequately
represent early childhood growth and that new
growth curves were necessary. In view of this,
WHO generated new curves based on the
Multicentre Growth Reference Study. These
standards provide a more robust tool for assessing
child growth. The design characteristics provide a
wider array of references for expanded uses, such
as monitoring of childhood obesity and
management of early lactation. Since the NCHS
growth charts were still in use, we did a study to
assess the nutritional status of under-five children
using NCHS and WHO standards with the weightfor-age and height-for-age charts, to compare the
two standards in detecting malnutrition among
children and study the implications of adopting new
WHO standards in assessing prevalence and
detection of malnutrition.
Study Design: We conducted a cross – sectional
study where we retrieved from hospital records the
anthropometric measurements of 100 under five
children in rural Karnataka, India in September
2008. The heights and weights of children were
plotted on the weight-for-age and height-for-age
charts separately for males and females on the
growth charts based on the NCHS reference. These
were then plotted on the WHO growth charts and
the nutritional statuses according to the two
references were compared.
Population Studied: Children under five years of
age who attended the out-patient department of a
community health center in village Karkala,
Karnataka, India were selected for the study.
Principal Findings: Our study showed that the
WHO growth standards pick up more underweight
children as compared to NCHS standards (19% vs.
11%). The prevalence rates were higher with the
WHO standards for the 0-6 months (n=30), 6-12
months (n=26) and 2-5 years age groups (n=14)
than with the NCHS standards (26.7% vs. 6.7%,
15.4% vs. 7.7% and 24.1% vs. 7.1% respectively).
However, in the 1-2 years age group (n=30), the
prevalence was less with the WHO standards than
with the NCHS standards (13.3% vs. 20%). The
prevalence was higher with the WHO standards in
the first half of infancy. Regarding height-for-age,
the prevalence of stunting was less with the WHO
standards as compared to NCHS standards (11%
vs. 13%). However, the prevalence was higher upto
6 months of age (n=30) with the WHO standards
(13.3% vs. 6.7%). Amongst 6-12 months (n=26)
and 1-2 year age group (n=30), WHO standards
showed lower prevalence (7.7% vs. 15.4% and
13.3% vs. 20% respectively). The prevalence
(7.1%) was same in the 2-5 years age group
(n=14).
Conclusions: Our study, though small in sample
size, showed that adoption of the new standards
would help in picking up more malnourished
children and probably at an earlier stage.
Implications for Policy, Delivery or Practice: The
new approach is important to improve early
detection and proper management of malnourished
children and prevent life-long sequelae, which
prevent them from achieving their full potential of
physical development.
Poster Number: 26
Cholera Vulnerability in a Rural Community in
Western Kenya
Avika Dixit, M.B.B.S.
Presented by: Avika Dixit, M.B.B.S., Student,
Harvard School of Public Health, 199, Park Drive,
Apartment 133, Boston, MA 02215; Phone: (617)
834-8576; Email: adixit@hsph.harvard.edu
Research Objective: The study was done in July
2009 to assist UNICEF Kenya in identifying areas
requiring intervention for cholera prevention and
control.
Study Design: A cross – sectional study of ninetysix households, who were interviewed on water
handling practices, hygiene and sanitation, and
knowledge, regarding acute watery diarrhea
including cholera. Water containers, hand washing
techniques and condition of latrines were observed.
Inpatient and outpatient records were obtained from
two health facilities. Interviews were held with key
informants.
Population Studied: Households in rural area of
Muhuru Division, Kenya, located at the shore of
Lake Victoria near the Tanzanian border.
Principal Findings: During the dry season,
majority (79.2%) used lake water and during rainy
seasons 76% used rainwater. The respondents said
that they washed containers for carrying water
everyday (77.1%) and cleaned containers for
storage adequately (81.2%). However, both
containers were found dirty in almost 75% of the
households on inspection. Almost 40% used
chlorine to treat drinking water. The technique of
hand washing at critical times was found to be
inadequate for more than 70% though more than
80% of households had soap. Almost 70% thought
use of soap unnecessary. Majority (59.4%) bought
cooked food from the market sometimes. Latrine
coverage was 27.1%. Amongst those, 65.4% of the
latrines’ condition was found to be fair. Majority
(81.4%) said they didn’t have enough money to
construct one. Amongst the households with under
fives (n=85), 47.1 % buried children’s feces.
Regarding knowledge about cholera, 81.3%
identified the main symptoms; majority (70.8%)
knew how it was transmitted, though about 30% did
not know how it could be prevented. If a family
member had diarrhea, 72.9% said that they would
take them to the health facility. Other findings
included a lack of national policy and strategy on
diarrheal disease control, absence of ORT corners
in health facilities, lack of laboratory services, lack
of staff capacity in appropriately treating and
controlling cholera and inappropriate use and
promotion of salt solution instead of sugar-salt
solution by the community health workers (CHWs).
Conclusions: Due to the lack of culture facilities,
Muhuru Bay has never been identified as an area
with cholera. Over use of intravenous fluids lead to
increased hospitalizations and made cholera
treatment more expensive. Poor community
awareness with regards to adequate hygiene and
sanitation practices, ORS use and water safety was
evident. Low latrine coverage due to poverty and
collapsing soils, lack of a safe water source and
shortage of chemicals for treatment of water
compounded the community’s vulnerability to
cholera.
Implications for Policy, Delivery or Practice:
Development of national diarrheal disease control
policy and strategy should be expedited and
implemented in Kenya. ORS and Zinc should be
made available to CHWs on a regular basis, not
only during an outbreak. This will enable use of
ORS soon after onset of diarrhea. To improve ORS
usage as the first line of treatment, establish ORT
corners in all health facilities. Develop skills at the
community level to identify the need for and bring
about improvement and maintenance of community
water points.
Funding Source(s): UNICEF
Poster Number: 27
A Study of Personal Health Records: A Pillar of
Meaningful Use for Oregon’s Medicaid Population?
Oliver Droppers, M.S., M.P.H., Sherril Gelmon,
Ph.D.; Jill Rissi, Ph.D.; Aasta Thielke
Presented by: Oliver Droppers, M.S., M.P.H.,
Research Assistant, Hatfield School of
Government, Portland State University, PO Box
751, Portland, OR 97207-0751; Email:
Odropper@pdx.edu
Research Objective: The State of Oregon,
supported by a federal Medicaid Transformation
Grant is implementing a personal health record
system (PHR) for Medicaid enrollees using a health
record bank model. A research team from Portland
State University was contracted by the State to
evaluate the utility of the PHR based on the
experience and perspective of Medicaid enrollees.
The study will assess awareness, enrollment, and
utilization patterns among active and non-active
PHR users, and determine which population groups
are more or less likely to initiate and continue using
a personally controlled and self-managed PHR.
Study Design: The study is quasi-experimental,
employing mixed methods through a communityengagement research strategy including use of
online surveys, interviews, focus groups, direct
observations, and documentation review. Data will
be collected during multiple stages of PHR
adoption: implementation, ongoing enhancements,
and maintenance. An evaluation conceptual
framework of key concepts and indicators will guide
the research, helping to illustrate barriers to and
facilitators of PHR adoption.
Population Studied: Potential participants include
adults age 18 to 64 enrolled in Medicaid who
establish an online PHR during the study period. As
of October 2009, Oregon Medicaid covered
approximately 384,000 individuals. The priority
populations include low-income children in foster
care, pregnant and postpartum women, individuals
enrolled in disease management, disabled and
senior adults. Adults are targeted for enrollment in
the first phase; followed by children and their
associated caregivers, case managers and/or foster
parents.
Principal Findings: Preliminary findings indicate
success as measured relative to the following
indicators: number of registered and active PHR
users and providers; user satisfaction and concerns
with use of the PHR; the degree of system use by
clinicians and professional service providers; and
the extent to which participants are willing to allow
data collected by the Health Record Bank of
Oregon to be used for research and public health
purposes.
Conclusions: It is critically important that we
evaluate the functionality and use of PHRs among
vulnerable population groups; and in particular,
identify potential benefits and perceived value
among Medicaid enrollees and clinical providers.
Findings can help inform ongoing and future
activities intended to improve enrollment and
continuity of use of PHRs among priority Medicaid
populations. In addition, findings will help enhance
our understanding of the overall feasibility of PHRs
among Medicaid clients and assess potential future
application and use of the Health Record Bank of
Oregon in the context of meaningful health
information exchange.
Implications for Policy, Delivery or Practice:
Medicaid beneficiaries experience high levels of
“churn” between public, private and no health
insurance coverage. Consequently, increasing their
engagement and participation in a PHR through
effective outreach, enrollment and utilization of a
PHR offers the opportunity to potentially improve
health information exchange across the continuum,
and enhance the coordination and efficiency in the
provision of health services. An enhanced
understanding of factors that mediate the decision
to establish and use a PHR can guide future efforts
to establish, maintain and expand PHRs, and
possibly contribute to a broader statewide strategic
planning progress designed to achieve "meaningful
use" of health IT in Oregon.
Funding Source(s): CMS
Poster Number: 28
Predictors of Atomoxetine Initiation in Medicaid
Children and Youth Newly Diagnosed with Attention
Deficit Hyperactivity Disorder
Lobna Eldasher; Scott Bilder, M.S.; Stephen
Crystal, Ph.D.
Presented by: Lobna Eldasher, Research Intern,
Institute for Health, Health Care Policy, and Aging
Research, Rutgers University, 30 College Avenue,
New Brunswick, NJ 08901; Phone: (609) 914-5647;
Email: lmeldasher@gmail.com
Research Objective: Atomoxetine was first
introduced in late 2002 as an alternative to the
stimulant medications methylphenidate and
aM.P.H.etamine that are widely used in the
treatment of Attention Deficit Hyperactivity Disorder
(ADHD). This study focused on the identification of
variables that predict the initiation of the
nonstimulant atomoxetine vs. a stimulant
medication in children and youth after they received
a new diagnosis of ADHD. For various reasons,
including concerns about contraindications, side
effects, and adverse events; as well as the abuse
potential of stimulants; it was expected that children
and youth initially prescribed atomoxetine after
diagnosis would differ in important and predictable
ways from those prescribed stimulants. The results
of this study provide insight into some of the patient
characteristics evaluated by clinicians in their initial
selection of an ADHD drug.
Study Design: This was a retrospective
observational study of fee-for-service Medicaid
claims using 2002-2004 Medicaid Analytic Extract
(MAX) data for eight states. A series of multivariate
logistic regression models were estimated in order
to identify patient variables associated with initiation
of atomoxetine versus a stimulant following initial
diagnosis with ADHD. Predictors included type of
ADHD diagnosis (hyperactive vs. inattentive); sex,
age, and race/ethnicity; year of diagnosis (2003 or
2004); and the following pre-existing comorbid
conditions: drug abuse or dependence, tics, bipolar
disorder, conduct disorder, emotional disorders,
and developmental delays. Results from the full
model containing all variables are reported.
Population Studied: The study sample included
children and youth aged six through seventeen who
had continuous Medicaid eligibility for at least 12
months and who had initiated pharmacotherapy
with atomoxetine, methylphenidate, or
aM.P.H.etamine following a first diagnosis of ADHD
(N=32,110).
Principal Findings: Twenty-two percent of newly
diagnosed children and youth received atomoxetine
and 78% received a stimulant. Beneficiaries with a
drug-related disorder had 34% greater odds of
being prescribed atomoxetine than persons without
such a disorder and beneficiaries with tic disorders
had over three times the odds of those without tics.
hose with developmental delays had approximately
22% lower odds of being prescribed atomoxetine
vs. a stimulant after diagnosis. White beneficiaries,
older youth, and females were more likely to initiate
treatment with atomoxetine than their nonwhite,
younger, and male counterparts. Beneficiaries first
diagnosed with ADHD in 2004 (vs. 2003) had
approximately 25% lower odds of starting
pharmacotherapy with atomoxetine. Beneficiaries
diagnosed with the inattentive-only type of the
disorder had 30% higher odds of receiving
atomoxetine compared to those with hyperactivity.
Conclusions: A prior diagnosis of a tic disorder
greatly increased beneficiaries’ odds of receiving
atomoxetine, reflecting ongoing concern about the
use of stimulants in this population. It also appears
that choice of ADHD drug reflects prescribers’
concerns about the abuse potential of stimulants;
older youth and those with diagnoses of drug abuse
or dependence were significantly more likely to
have received atomoxetine in place of a stimulant.
These results provide some insight into the patient
characteristics taken into account by prescribers
deciding on an initial choice of ADHD drug.
Funding Source(s): RWJF
Poster Number: 29
Exploring the Effects of Treatment Processes on
the Ischemic Stroke Patients in Taiwan
Pei-Hsuan Fang; Ying-Chun Li, Ph.D.
Presented by: Pei-Hsuan Fang, Graduate
Student, Institute of Health Care Management,
National Sun Yat-Sen University, 70 Lien-Hai Road,
Kaohsiung, 804 Taiwan; Phone: +88675252000
ext. 4875; Email: pangel0927@gmail.com
Research Objective: Ischemic stroke is a
commonly seen chronic disease in aged population
and at high risk of stroke recurrence, physical and
intellectual disability, long-term institutionalization,
and death. Antiplatelet agents, such as aspirin and
clopidogrel, are the first-line antiplatelet therapies
for prevention of recurrent stroke in patients with
ischemic stroke. The diverse pharmacological
effects of the different antiplatelet drugs not only
influence the patient outcomes but also affect the
health-care utilizations. International studies have
broadly discussed the impacts of antiplatelet agents
on ischemic stroke patients, however, little studies
have examined such issue in Taiwan. Taiwan
implemented national health insurance program
and covered over 98% of its residents since 1995.
This makes Taiwan an excellent study subject of
outcome research because it reduces the potential
study biases from health care coverage variations.
Therefore, this study aims to explore the
hospitalizations and direct medical expenditures of
antiplatelet treatments in the aged population for
secondary prevention of ischemic stroke.
Study Design: This study examined three
antiplatelet treatment processes for the aged
population in Taiwan –none antiplatelet therapy,
aspirin, clopidogrel–by using the National Health
Insurance Research Database between 2003 and
2006. The data sets included patients’ demography,
prescriptions, and health care history, plus
outpatient records, inpatients records, and
information about the medical facilities they visited
during the study period. The data analyses included
Chi-square test, Anova, multiple linear regression
and multivariate logistic regression.
Population Studied: Patients aged 40 or older who
were new cases of ischemic stroke based on ICD9-CM of hospital admissions and readmissions from
2003 to 2004 in Taiwan. Patients were followed for
two years and excluded if they shifted antiplatelet
therapy during research period.
Principal Findings: Among 1422 patients of
ischemic stroke, 52.67% were with none antiplatelet
therapy, 44.44% with aspirin and 2.88% with
clopidogrel. Among 20% of the patients who had
hospital readmissions because of the ischemic
stroke, 29.63% were readmitted within 14 days after
discharge and 12.96% were readmitted within 1530 days after discharge. More health care
utilizations were observed in the clopidogrel group
than aspirin and none antiplatelet therapy groups
(p<0.05). Age and hospital level were significantly
affected health care utilizations and length of stay.
But the probability of hospital readmissions among
three antiplatelet treatments presented no
statistically significant difference.
Conclusions: Clopidogrel were not significantly
effective in the prevention of recurrent stroke in
patients with ischemic stroke, but were associated
with the increased health care utilizations in
Taiwan. By understanding more about factors
affecting hospitalization and health care utilization
in different antiplatelet treatments of ischemic
stroke, we may modify treatment process, reduce
unnecessary costs, and improve the health status
of patients.
Implications for Policy, Delivery or Practice:
How to design an optimal drug reimbursement
system while ensuring the quality of care and
controlling medical cost growth is an important
study issue worldwide in recent years. This study
provides useful reference for policymakers and
health care professionals in care of the ischemic
stroke patients.
Funding Source(s): National Science Council
Poster Number: 30
Social Context and Alcohol Use Disparities in a
Racially Integrated Community
Ruth Fesahazion, B.S.H.S.; Caryn Bell, B.S.;
Natieka Green; Roland Thorpe, Jr., Ph.D.; Thomas
LaVeist, Ph.D.
Presented by: Ruth Fesahazion, B.S.H.S.,
Doctoral Student, Health Policy and Management,
Johns Hopkins Bloomberg School of Public Health,
624 N. Broadyway, Baltimore, MD 21205; Phone:
(520) 272-5345; Email: rfesahaz@jhsph.edu
Research Objective: National estimates of race
differences in alcohol use suggest that nonHispanic whites are more likely to be current
drinkers; however, African Americans suffer a
greater burden of alcohol-related problems.
However, findings from national studies fail to
account for the social and environmental risk
exposures resulting from residential segregation.
Thus, little is known about race disparities in alcohol
use among individuals who share similar social and
environmental conditions. The objective of this
study is to determine if racial differences exist in
alcohol use between African American and nonHispanic white adults residing a racially integrated
community.
Study Design: We compared race disparities in
1,408 individuals from the Exploring Health
Disparities in Integrated Communities-Southwest
Baltimore (EHDIC-SWB) Study with 29,372
individuals from the 2003 National Health Interview
Survey (NHIS 2003) to determine if race disparities
in alcohol use were attenuated in EHDIC-SWB, a
racially integrated urban community without race
differences in socioeconomic status.
Population Studied: EHDIC (Exploring Health
Disparities in Integrated Communities) is an
ongoing multisite study of race disparities within
communities where African Americans and nonHispanic whites live together and where there are
no race differences in SES (as measured by
median income). The National Health Interview
Survey (NHIS) is nationally representative
household survey of non-institutionalized U.S.
citizens conducted by the National Center for
Health Statistics.
Principal Findings: In the NHIS 2003 sample,
compared to non-Hispanic whites African
Americans had a lower odds of being a current
drinker (OR=0.56, 95% CI=0.49-0.64); former
drinker (OR=0.84, 95% CI=0.72-0.99); or binge
drinker (OR=0.68, 95% CI= 0.58-0.79) independent
of covariates. However in the EHDIC-SWB sample,
African Americans had similar odds of of being a
current drinker (OR=0.95, 95% CI=0.67-1.34);
former drinker (OR=0.71; 95% CI=0.50-1.01); or
binge drinker (OR=1.02, 95% CI=0.77-1.35)
compared to non-Hispanic whites after adjusting for
covariates.
Conclusions: Among individuals living in a lowincome, urban, racially integrated community with
similar exposure to social and environmental risk,
race disparities in alcohol use were eliminated.
These results suggest that the social environment
should be accounted for when developing research,
policies, and interventions that address health
disparities.
Implications for Policy, Delivery or Practice:
These results suggest that the social environment
should be accounted for when developing research,
policies, and interventions that address health
disparities.
Poster Number: 31
A Comparison of U.S. Army and Civilian Registered
Nurses: Influence of Organizational and Personal
Factors on Intent to Leave Their Current
Organizations
Linda Fisher, M.H.A., R.N.; Mary Dietrich, Ph.D.
Presented by: Linda Fisher, M.H.A., R.N., Nursing
Research, Nursing Research Service, U.S. Army,
Madigan Army Medical Center, Tacoma, WA
98431-1000; Phone: (253) 968-2289; Email:
linda.w.fisher@vanderbilt.edu
Research Objective: The purposes of this study
were to measure and compare organizational,
personal, and economic factors in two groups of
nurses (military and civilian) to: a) determine if and
how much difference exists between the two
group’s intent to leave their current organizations,
b) determine how structural empowerment,
psychological empowerment, and job satisfaction
factors predict intent to leave, and c) explore the
mediating and moderating effects of personal and
economic factors on the intent to leave. This is
important because poor nurse retention in the US
Army at the mid-career level (losses up to 20%) can
have devastating implications in future Army Nurse
Corps leadership.
Study Design: A targeted comparative group
design with population sampling of US Army
Nurses. Data were collected through use of
Vanderbilt University’s REDCap Survey service (A
service that provides secure internet survey
completion from anywhere in the world). Theoretical
and conceptual frameworks of organizational
structure and empowerment (including those of
Kanter and Laschinger) guided survey content
which included the Conditions of Work
Effectiveness Questionnaire-II (CWEQ-II), the
Psychological Empowerment Instrument (PEI), job
satisfaction scale, additional economic and
demographic questions, and intent to leave items.
Population Studied: The sample included two
groups: a) US Army nurses between the ranks of
2nd Lieutenant and Major (n=623) and, b) civilian
registered nurses (n=210), randomly selected from
the total population of registered nurses employed
at Vanderbilt University Medical Center (VUMC)
because the VUMC registered nurse population
was demographically similar to the US Army
population. he ranks of 2nd Lieutenant to Major
were targeted because the individuals in that group
have the lowest retention rate. Response rates
were 19.7% and 24.4% respectively.
Principal Findings: Data collection closed in late
December 2009 and the study is in the analysis
phase. Preliminary analysis indicates no coding
problems and a high percentage of completeness of
cases. Findings will be available prior to the start of
the Annual Research Meeting in June 2010.
Conclusions: Conclusions are pending and will be
available prior to the start of the Annual Research
Meeting in June 2010.
Implications for Policy, Delivery or Practice:
High nurse turnover can negatively affect nurse
work environments, healthcare quality, nurse job
satisfaction, and ultimately nurse recruitment. There
is value in identifying civilian solutions already in
use and solutions that may be unique to military
nurses. The findings from this study will provide
greater understanding of organizational factors that
influence nurses’ intent to leave and provide insight
into possible areas for military healthcare
organizations to concentrate limited resources to
improve nurse work environments that may result in
increased military nurse retention.
Poster Number: 32
An Examination of Rural-Urban Differences in
Prepregnancy Body Mass Index and Gestational
Weight Gain
Presented by: Alexa Gallagher, M.S.P.H.,
Student, Rural Health Research Center, University
of South Carolina, 220 Stoneridge Drive, Suite 204,
Columbia, SC 29210; Phone: (803) 251-6317;
Email: GALLAGHA@mailbox.sc.edu
Co-authors: Alexa Gallagher, M.S.P.H.; Jihong Liu,
Sc.D.; Amy Martin, Dr.PH.; Janice Probst, Ph.D.
Research Objective: Having an unhealthy
prepregnancy weight and/or gaining an
inappropriate amount of weight during pregnancy
are known to increase the risk for poor pregnancy
and birth outcomes. To date, no studies have
examined differences in prepregnancy body mass
index (BMI) and gestational weight gain (GWG) in
women who reside in rural versus urban areas.
Because rural residents tend to have a lower
socioeconomic status and poorer access to health
care than urban residents, an examination of
residence-based differences in prepregnancy BMI
and GWG is warranted. We sought to examine
prepregnancy BMI and GWG patterns in women
residing in rural versus urban areas of South
Carolina.
Study Design: Data from the 2004-2006 South
Carolina birth certificates (n=132,795) were used.
Rural/urban residence was determined using RuralUrban Commuting Area (RUCA) codes. Mothers
were categorized into 4 groups according to their
prepregnancy BMI: underweight (<=18.5 kg/m2),
normal weight (18.5-24.9), overweight (25.0-29.9),
and obese (>=30.0). GWG was then defined as
inadequate, adequate or excessive according to the
Institute of Medicine’s 2009 weight gain guidelines.
Multinomial logistic regression was used to examine
urban and rural differences in prepregnancy BMI
and GWG after adjusting for potential confounders
(maternal age, race, marital status, education,
parity and tobacco use).
Population Studied: This study population
(n=132,795) included women residing in South
Carolina who delivered a liveborn, singleton infant
between 2004-2006.
Principal Findings: In 2004-2006, 34.7% of
women who were pregnant in SC resided in rural
areas. Compared to urban pregnant women, rural
pregnant women were more likely to be younger,
African American, unmarried and have less than a
high school education. Rural pregnant women were
also more likely to be obese (30.7% v. 25.8%). In
multinomial logistic regression analysis, they had
higher odds of being obese (adjusted odds ratios
(AOR) = 1.21, 95% CI = 1.18, 1.25), overweight
(AOR= 1.08; 95% CI = 1.05, 1.11), and underweight
(AOR=1.06, 95% CI=1.00, 1.12) than urban
women. This relationship was found to be partially
explained by the higher proportion of minorities
living in rural areas. Regarding GWG, a slightly
higher proportion of rural compared to urban
women gained inadequate weight (30.0% v. 27.2%)
and a slightly lower proportion gained excessive
weight (47.0% v. 49.3%). The relationship between
GWG and rurality was found to differ by
prepregnancy BMI class. Specifically, there were no
residence-based differences in GWG in
underweight women. Among normal weight women,
rural women had an increased odds of inadequate
GWG (AOR=1.09, 95% CI=1.04, 1.15). Among
overweight or obese women, compared to their
urban counterparts, rural women had decreased
odds of excessive GWG (AOR=0.92, 95% CI=.87,
.98) and a decreased odds of both inadequate
(AOR=.85, 95% CI=.80, .91) and excessive GWG
(AOR=.86, 95% CI=.81, .92), respectively.
Conclusions: Small, but significant differences in
prepregnancy BMI and GWG were found between
rural and urban women that were not fully explained
by socio-demographic differences. Future
researchers should further explore reasons for
these findings and the resulting health
consequences for the rural population.
Implications for Policy, Delivery or Practice:
Confirmation of these findings may help health care
providers better serve and advise their pregnant
population.
Poster Number: 33
Prescription Drug Plan Enrollment and CostRelated Nonadherence in Medicare Part D
Beneficiaries with Diabetes
Presented by: Kristin Geonnotti, Doctoral
Candidate, Health Policy and Management,
University of North Carolina Chapel Hill, CB# 7400,
Chapel Hill, NC 27599-7400; Email: klg@unc.edu
Co-authors: Mary Roth McClurg, Pharm.D.,
M.H.S.; Tim Carey, M.D., M.P.H.; George Pink,
Ph.D.; Marisa Domino, Ph.D.; Morris Weinberger,
Ph.D.
Research Objective: Medicare Part D aims to
provide seniors with affordable prescription drug
coverage. Annual out-of-pocket costs can vary
widely between plans. This study will determine the
proportion of individuals who are enrolled in lowest-
cost plans, characteristics associated with being in
a lowest-cost plan, and whether lowest-cost plan
enrollment and differential costs are associated with
experiencing cost-related nonadherence.
Study Design: This observational, cross-sectional
study combines data from participant surveys,
medical records, and publicly-available CMS plan
data. Study participants are surveyed about their
Medicare Part D plan, cost-related nonadherence,
and plan enrollment decisions. For each
beneficiary, we used the CMS Prescription Drug
Plan Finder tool to calculate: (1) actual out-ofpocket medication costs for one year based on their
current plan and (2) lowest out-of-pocket
medication costs for one year if they were enrolled
in a plan with the least out-of-pocket spending.
Differential costs are the difference between the
actual and lowest out-of-pocket costs if all
prescriptions are filled for one year. These
calculations were made twice: once assuming that
prescriptions were filled as written, and again with
generic substitutions. Medical record data include
patient characteristics, diagnoses, prescribed
medications, clinical outcomes, and health services
utilization. Descriptive and multivariate logistic
regression analyses will be used to determine the
proportion of the sample in lowest-cost plans and
the association between lowest-cost plan
enrollment/differential costs and cost-related
nonadherence.
Population Studied: Study participants are drawn
from diabetes registries at the UNC-Chapel Hill
Internal Medicine and Family Medicine practices.
Participants include Medicare Part D beneficiaries
who: (1) have diabetes; (2) are age 65 and older;
(3) are community-dwelling; (4) have had = 1
primary care visit in the past 12 months; and (5) are
able to complete a telephone interview.
Principal Findings: Of 228 participants, 61.8% are
female and 16.7% are dually-eligible for Medicare
and Medicaid. Participants take an average of 9.2
medications annually, and 33.6% report
experiencing cost-related nonadherence. 22.6% of
participants have switched plans in the past year.
Lowest-cost plan enrollment and differential costs
have been estimated for the 162 participants who
granted medical records access. Only 27.2% of
beneficiaries are enrolled in a plan that is within
10% of their lowest cost plan, regardless of generic
substitution. Differential costs are substantial (mean
± SD = $1140.73 ± $1475.14; median = $707.50).
When assuming that generic substitutions were
allowed, differential costs are smaller, but still
considerable (mean ± SD = $641.10 ± $795.03;
median = $449.50).
Conclusions: Preliminary results suggest that the
majority of beneficiaries are not enrolled in a
lowest-cost plan, and are therefore paying
substantially more than is necessary to obtain their
medications. We are also examining the
relationship among differential costs, medication-
related non-adherence, clinical outcomes, and
health services utilization over a one-year period.
Implications for Policy, Delivery or Practice:
Policy interventions can target plan enrollment to
increase the proportion of beneficiaries in lowestcost plans. For instance, the random autoassignment of dually-eligible individuals could be
replaced with a beneficiary-centered assignment
process, which considers medication coverage
needs. Additional strategies could be developed to
simplify the complex enrollment decision-making
process facing other beneficiaries and their
caregivers.
Funding Source(s): NRSA Pre-doctoral
Traineeship from AHRQ, sponsored by the Cecil G.
Sheps Center for Health Services Research, UNC
Chapel Hill, Grant No. T32-HS-000032-19 and the
UNC Graduate School Dissertation Completion
Fellowship
Poster Number: 34
Contributing Factors of Count Discrepancy with
respect to Operating Room (OR) Processes
Presented by: Hyo Geun Geun, M.P.H., R.N.,
Doctoral Student, School of Nursing, University of
Michigan, 400 North Ingalls Building, Ann Arbor, MI
48109; Phone: (818) 636-2982; Email:
geunhyo@umich.edu
Co-authors: AkkeNeel Talsma, Ph.D., R.N.;
Christine Anderson, Ph.D., R.N.
Research Objective: Since the publication of the
Institute of Medicine (IOM) report, “To Err is Human
(2000),” a variety of organizational characteristics
has been examined in relation to patient safety.
Accordingly, research focuses have moved toward
identifying processes or human factors that
contribute to adverse events. In Operating Room
(OR) particularly, surgical incorrect counts occur
relatively rarely, but the consequences are serious.
This lack of a complete count of sponges and
instruments has been identified as possible risk
factors for retained objects after surgery. Although
studies have shown that all the processes and
nurse staffing patterns in OR are connected as a
system and impact patient outcomes, studies to
date about surgical errors have not been examined
well. The objectives are to describe the systems
factors associated with count discrepancy in acute
care settings.
Study Design: An exploratory descriptive study
was performed. Due to the lack of patient-specific
demographic data, only variables related to OR
process and staffing pattern were included for
analysis. Initial step focused on describing overall
surgical count procedures in OR. Then, the
distributions of correct and incorrect counts in
relation to system and human factors were
included.
Population Studied: Of surgical cases performed
at a hospital in 2008, a total of 7,791 cases with 110
different high risk procedures were included. All
data were obtained from the OR electronic record
and analyzed using SPSS.
Principal Findings: Compared to correct cases (all
counts are correct in this group), the other incorrect
cases which include at least one count discrepancy
show significantly higher percent of selected system
factors within incorrect cases when: operations
were performed on emergency (16% vs 24 %),
weekend (8% vs 13%), in-patient (21% vs 30%),
with multiple procedures at a time (4% vs 11%), in
cases exceeding the operation time to the next shift
(29% vs 62%), begin shifts in the evening (16% vs
17 %) or at night (3% vs 6%) and end shifts in the
evening (39% vs 61%) or at night (4% vs 9%). In
addition, procedures with higher volume of staff
participating and longer operation times appear to
show more count discrepancy (p<.001).
Conclusions: The count process involves multiple
counts at different stages of the procedure, creating
the opportunity for systems errors and adverse
events. Numerous systems factors have contributed
to a count discrepancy including: cases with high
complexity, emergency procedure, multiple
services, weekend surgeries, longer surgeries (>
4hrs), and higher volume of staff.
Implications for Policy, Delivery or Practice:
Even if in case counts were documented as correct,
surgical counts are not always sufficient for zeroerror surgery. Thus, identifying the relationship
between surgical incorrect counts and actual
retained objects are needed. In addition, staffing
patterns in consideration of fewer rotations and
appropriate work time also needs to be developed.
Last, further studies should explore whether specific
strategies and technologies for screening of highrisk patients in peri-operative process need to be
established to reduce count discrepancy.
Funding Source(s): Medline
Poster Number: 35
Discharge Against Medical Advice: Are the Risks of
30-day Readmission and Mortality Greater?
Presented by: Justin Glasgow, M.S., Graduate
Student, CRIISP, Iowa City VA 152, 601 Highway 6
West, Iowa City, IA 52246-2208; Phone: (319) 3380581, ext. 3538; Email: justin.glasgow@va.gov
Co-authors: Mary Vaughn-Sarrazin, Ph.D.; Peter
Kabolie, M.S., M.D.
Research Objective: With 1-2% of patients leaving
the hospital against medical advice (AMA), there is
concern that these patients may suffer increased
rates of adverse health outcomes post-discharge.
The objective is to determine if patients leaving
AMA from a Veterans Administration (VA) hospital
experience greater adverse health outcomes as
measured by 30-day hospital readmission and
mortality rates.
Study Design: Case-control study of all patients
discharged from VA hospitals between 2004 and
2008. Primary outcome measurements are 30-day
mortality and all-cause hospital readmission.
Bivariate analyses were used to compare cases
and controls on demographics and comorbidities.
Separate multivariable Cox proportional hazard
models were generated to identify independent risk
factors associated with 30-day mortality and
readmission.
Population Studied: Over the 5 year period, there
were 2,204,703 medical admissions to the 129 VA
hospitals. Patients that were discharged to a
nursing home (6.13%), transferred to another
hospital (2.33%), discharged to a non-standard
setting (1.28%) and those who died in-hospital
during the index admission (2.68%) were excluded
allowing for a comparison of AMA patients (cases)
and those with a standard discharge to the
community (controls).
Principal Findings: In the final sample of
1,930,947 admissions, 32,819 (1.70%) were
discharged AMA. Over the five year study period
AMA rates were relatively constant ranging
between 1.65% – 1.74% of discharges. There was
variation observed across the 129 hospitals with
AMA discharge rates at the 1st quartile of hospitals
ranging from 0.42% – 1.29%, while 4th quartile
ranges were 2.40% – 5.46%. AMA patients were
younger, more likely to be African American, have
low income, and have co-morbid alcohol abuse [For
all, X2 df = 1, p<0.001]. Readmission rates (17.7%
vs. 11.0%, p<0.001) and mortality rates (0.75% vs.
0.61%, p=0.001) were higher in AMA patients. Cox
proportional hazard modeling of readmission
adjusting for age, income, hospital discharged from,
and comorbidity indicated that discharge AMA was
the largest hazard for readmission within 30 days
(HR = 1.36, 95% CI 1.32 – 1.40). In the 30-day
mortality model discharge AMA was a significant
hazard (HR = 1.12, 95% CI 1.00 – 1.26).
Conclusions: Patients leaving AMA are at
increased risk for adverse health outcomes
including hospital readmission and death.
Multivariable modeling adjusting for demographic
characteristics and the presence of co-morbidities
reveals that discharge AMA is an independent risk
factor for these adverse health outcomes
Implications for Policy, Delivery or Practice:
Considering the increased risk of adverse outcomes
in AMA patients, hospitals should target these
patients for discharge transition interventions.
Potential interventions include phone follow-up,
home visits, or mental health counseling.
Alternatively, identifying and ameliorating the
factors that lead to AMA discharges may help to
reduce the number of patients who leave before
achieving clinical stability. This may include factors
related to the healthcare team such as provider
communication style, access to social services
support, or involvement of family in care decisions.
Whether targeting patients before or after
discharge, clear communication is likely to play a
significant role in improving outcomes.
Funding Source(s): University of Iowa Clinical and
Translational Science Program
Poster Number: 36
Switching Statins in a Medicaid Population: The
Health Outcomes and Policy Implications of
Formulary Restrictions
Presented by: Amie Goodin, M.P.P., Research
Assistant, Martin School of Public Policy and
Administration, University of Kentucky, 3554
Creekwood Drive #17, Lexington, KY 40502;
Phone: (606) 499-0940; Email: ajgood3@uky.edu
Research Objective: Two incidences of formulary
restrictions on statins by Kentucky Medicaid were
analyzed to determine whether or not the
preferential coverage and the act of switching
statins led to adverse health outcomes in patients
that were switched to the preferred drug as
compared to patients that continued with their
originally prescribed statin therapies.
Study Design: Medicaid claims data was utilized to
collect a sample of Kentucky Medicaid patients
affected by a formulary change (“switch”) in 2002
and in 2008. Inclusion criteria were parameterized
as the amount of a statin that the patient had prior
to the switch and after the switch, as measured by
the medication possession ratio. Information
regarding whether or not the patient had a stroke,
angina, MI, bypass surgery/angioplasty,
hypertension, or died before or after the switch date
was collected, and concomitant medications as well
as demographic characteristics were controlled for
in the analysis. Logistic regression and marginal
effects analysis were conducted for each of the two
time periods.
Population Studied: Kentucky Medicaid patients
who refilled a prescription for simvastatin,
atorvastatin, rosuvastatin, or simvastatin/ezetimibe
at least once in the 60 days prior to the date of the
formulary change (“switch date”) in 2002 or in 2008.
Principal Findings: The difference in age
distribution between 2002 and 2008 is striking. It is
clear that statins are being prescribed increasingly
to younger patients. Patients are typically female
and have about a 40% chance of having diabetes.
Incidents of MI and stroke are very low across 2002
and 2008 patients, though death appears more
frequently in patients that stopped taking all statins
after the switch date. The percentage of patients
who died after the 2002 switch is much higher than
that of 2008 and this is most likely due to the longer
period of data collection for 2002 patients. When
controlling for demographic characteristics and
comorbidities, there is a statistically significant
relationship between stopping all statins and death.
Age was a significant factor in both the 2002 and
the 2008 cohorts, though other risk factors varied
between the two time periods. The act of switching
statins was only found to be statistically significantly
associated with death in 2008, whereas nearly all
patients suffering from MI (in 2002), stroke, and
angina died within the time frame of the analysis.
Conclusions: Formulary restrictions on statins
have an impact on health outcomes and the
Medicaid population exemplifies many
characteristics that make these patients more
vulnerable to changes in prescriber behavior than
the general population.
Implications for Policy, Delivery or Practice:
Statins are the top selling class of medications in
the United States and costs to tax payers must be
balanced with the program’s objective of providing
healthcare to the poor and medically indigent.
Enacting preferential coverage, though potentially a
money saving strategy, can backfire if patients
suffer worse health outcomes and need additional
services or hospitalizations as a result of
inconsistencies in the continuity of treatment.
Poster Number: 37
Community Help-Seeking among Part D
Beneficiaries
Presented by: Melissa Hensley, M.S.W., M.H.A.,
Doctoral Candidate, George Warren Brown School
of Social Work, Washington University in St. Louis,
One Brookings Drive, Campus Box 1196, St. Louis,
MO 63130; Phone: (314) 725-6283; Email:
mahensle@wustl.edu
Research Objective: To use qualitative research
techniques to explore mentally ill Medicare
beneficiaries' use of community helpers such as
pharmacists, nurses, and social workers to interpret
and utilize their Part D benefits.
Study Design: The study used the technique of
focused ethnography to obtain in-depth interview
information from 26 Medicare beneficiaries with
mental illness.
Population Studied: Adults with mental illness who
were receiving Medicare benefits and were
participating in a psychosocial rehabilitation
program.
Principal Findings: Medicare beneficiaries with
mental illness reported a high degree of reliance on
community helpers in order to choose an
appropriate Part D plan, utilize their benefits, and
obtain and utilize their prescription medications.
Adults with mental illness reported particularly
intense and important relationships with their
pharmacists.
Conclusions: Community helpers such as
pharmacists, nurses, and social workers play an
integral role in helping adults disabled by mental
illness to utilize their prescription drug benefits.
Many adults with mental illness would have great
difficulty utilizing their benefits if they did not have
these sources of assistance.
Implications for Policy, Delivery or Practice:
Public sources of funding, such as the Medicare
Medication Therapy and Management Program, as
well as Medicaid and community mental health
funding for nurses and social workers, should
continue to support these important community
helpers as they work to assist adults disabled by
mental illness to get the most out of their benefits.
Funding Source(s): National Institutes of Health
Poster Number: 38
Cost-Sharing for Colorectal Cancer Screening in
Medicare Managed Care Plans, 2001-2008
Presented by: Priya Hirway, M.S., Student,
Community Health, Brown University, 121 South
Main Street, Providence, RI 02912; Phone: (401)
863-3172; Email: priya_hirway@brown.edu
Co-authors: Amal Trivedi, M.D., M.P.H.
Research Objective: Although colorectal cancer
(CRC) screening reduces deaths from colorectal
cancer, less than half of patients receive intervalappropriate screening. Increasing out-of-pocket
costs for preventive services has been
demonstrated to reduce their use. We therefore
determined the cost-sharing amounts for CRC
screening services among all Medicare managed
care (MMC) plans from 2001-2008.
Study Design: Using benefits data from all MMC
plans, we determined each plan’s required
copayment or coinsurance for CRC screening
services (e.g. screening flexible sigmoidoscopy or
colonoscopy) within all insurance products offered
by the plan. For plans with a range of cost-sharing
amounts, we assigned the median value. We tested
whether the number of plans requiring cost-sharing
and the amount of the mean copayment or
coinsurance increased over time using the chisquare and t-tests.
Population Studied: All Medicare managed care
beneficiaries per year from 2001-2008. The sample
size ranged from approximately 6.5 million
beneficiaries in 2001 to over 10.5 million in 2008.
Principal Findings: The proportion of MMC plans
that charged a copayment or coinsurance for
colorectal cancer screening services in any of their
insurance products increased from 13% to 38%
from 2001 to 2008 (p<0.0001). The proportion of
MMC enrollees in cost-sharing plans also
increased, from 9% in 2001 to 35% in 2008,
(p<0.0001). While no plans required coinsurance in
2001, 180 plans (29%) required coinsurance in
2008. Among plans charging a copayment, the
mean copayment for colon cancer screening
services increased from $12 in 2001 to $50 in 2008,
(p<0.0001). The mean coinsurance remained stable
at 20% throughout the study period. For enrollees
of plans with cost sharing, the mean age was 71.2
years, and 72.1 years for those in plans without
cost sharing (p<0.01). The proportion of enrollees
who were black or living in areas of low income and
education were similar in plans with and without
cost-sharing.
Conclusions: The number of MMC plans requiring
cost-sharing for CRC screening services and
proportion of MMC enrollees exposed to such costsharing increased substantially from 2001 to 2008.
Among plans that imposed cost-sharing, the use of
coinsurance increased in frequency.
Implications for Policy, Delivery or Practice:
Given the importance of screening in the early
detection and treatment of colorectal cancer, the
increase in cost-sharing for this preventive service
may have negative health consequences.
Poster Number: 39
What Can We Expect from the Meaningful Use of
Health Information Technology?
Presented by: Michael Hoaglin, B.S., Policy
Analyst / Medical Student at University of
Pennsylvania, Office of the National Coordinator for
Health Information Technology, Department of
Health and Human Services, 1425 Locust Street
Unit 2F, Philadelphia, PA 19102; Phone: (202) 5560077; Email: mhoaglin@med.upenn.edu
Co-authors: Melinda Beeuwkes Buntin, Ph.D.
Research Objective: The Centers for Medicare
and Medicaid Services (CMS) recently issued a
proposed rule defining the "meaningful use" of
health information technology (HIT) which will
qualify providers for Medicare incentive payments.
We use the lens of meaningful use to review the
literature on the benefits of electronic health records
(EHRs). Specifically, we systematically review the
HIT literature to determine whether EHR systems
meeting more of the proposed meaningful use
criteria produce greater health gains and cost
savings than more basic EHRs. We also develop a
framework to test a set of hypotheses about which
patients will benefit the most from EHRs. For
example, we hypothesize that EHRs will have
greater positive effects for patients with complex
care needs and those treated in multiple settings -but may not significantly improve patient outcomes
where available evidence does not lend itself to
clinical decision support (CDS) or during acute
episodes of illness addressed in a single patient
encounter.
Study Design: We updated the systematic HIT
literature review conducted by Goldzweig et al.
(Health Affairs 2009) and are stratifying the studies
by setting, patient population, and the functionalities
of the HIT systems used.
Population Studied: Searches of the literature
indexed in PubMed turned up 5776 additional
articles published since Goldzweig et al.'s review:
we expect 200 of these to ultimately meet our
review criteria. We will also conduct a search of
industry and consultant reports, the Cochrane
Database of Reviews of Effectiveness, and popular
periodicals.
Principal Findings: Preliminary review of the
recent literature appears to confirm some of our
hypotheses: studies examining the effects of HIT
coupled with secure health information exchange
(HIE) show better outcomes in terms effectiveness
and cost. In the same spirit, settings using
computerized provider order entry (CPOE) with a
well-designed CDS system show quality and cost
gains, as opposed to CPOE alone. In acute medical
situations, HIT enhances outcomes for conditions in
which timeliness of therapy and data-driven
decision-making are determinants of success. At
the same time, there are published examples of
systems which were implemented poorly or lacked
sophistication, contributing to greater inefficiencies
and risks to patients.
Conclusions: The widespread use of advanced,
interoperable HIT systems has the potential to yield
greater gains in clinical effectiveness and efficiency
than would be expected from the earlier literature
examining EHR systems lacking decision support
features and the ability to exchange data. In
addition, these gains can be expected to be greater
among patients with more complex illnesses.
Implications for Policy, Delivery or Practice: The
programs mandated under the 2009 federal
stimulus bill as part of the Health Information
Technology for Economic and Clinical Health
(HITECH) Act are intended to drive increases in the
adoption and meaningful use of interoperable
electronic health records. This review indicates that
giving providers incentives to achieve meaningful
use can be expected to yield health gains and cost
reductions -- but challenges remain in the areas of
EHR implementation and the development of
products that are easier and less expensive to
adopt.
Funding Source(s): DHHS/ONC
Poster Number: 40
Unintentional Poisoning Mortality Trends by Sex,
Age and Cause in Taiwan, 1986-2007
Presented by: Yu-Chen Hung, M.S, Graduate,
Public Health, National Defense Medical Center,
No.161, Section 6, Min-Chuan East Road, Taipei,
114 Taiwan; Phone: +886287923100 ext. 18445;
Email: style2498@hotmail.com
Lead Author: Wu-Chien Chien, Ph.D
Research Objective: To examine the trends of
unintentional poisoning mortality in Taiwan by sex,
age, and cause between 1986 and 2007.
Study Design: We analyze routine mortality data
from the official publications of vital statistics from
1986 to 2007 to determine trends by sex, age and
cause of death. Linear Regression was used to test
the trends.
Population Studied: population-based
Principal Findings: A total of 12,916 people in
Taiwan died as a result of unintentional poisoning
between 1986 and 2007. It showed the downward
trend in total mortality rate, a decline of 73.6%. The
unintentional poisoning mortality rate of male was
higher than female (rate ratio: 2.07). Unintentional
poisoning mortality rate increased by age, but the
mortality rate of all age groups declined. Pesticide
(38.3%) and drug (33.7%) were two dominant
poisoning agents, yet both the mortality rates
declined. The former had the largest decrease
(96.1%), then, the later has become the first leading
cause of death since 2000, rising 71% from 2000 to
2007. Carbon monoxide mortality rate, the third
poisoning agent (9.42%), increased 27%, but the
change wasn’t significant during the 22-year period,
then, as the second agent since 2002.
Conclusions: Although unintentional poisoning
mortality rate declined, we should still make more
efforts on developing prevention programs to
reduce drug or carbon monoxide mortality in the
future.
Implications for Policy, Delivery or Practice:
Basic preventive approaches include eliminating the
manufacture or sale of an agent, reducing the
length of exposure or the concentration of the
poison to non-injurious doses, preventing access to
the substance, changing its chemical formulation,
and providing appropriate emergency and
therapeutic measures.
Poster Number: 41
Mental Health Diagnosis and Treatment on AsthmaRelated Services in Medicaid Youth with Asthma
Presented by: Eric Jamoom, M.P.H., M.S., Health
Services Research, Mangement and Policy,
University of Florida, PO Box 100195, Gainesville,
FL 32610; Phone: (352) 246-5957; Email:
jamoom@phhp.ufl.edu
Co-authors: Jeffrey Harman, Ph.D.; Allyson Hall,
Ph.D.; Marilyn Dumont-Driscoll, M.D., Ph.D.; David
Janicke, Ph.D.; Paul Duncan, Ph.D.
Research Objective: An association between
asthma and mental health has been well
documented. However, quantifying the association
of mental health comorbidities, such as depression,
anxiety, and attention-deficit hyperactivity disorder
(ADHD) on asthma-related health services use and
expenditures in youth with asthma and the factors
which may mitigate asthma-related utilization in this
group remains an area that requires more attention.
Therefore, the study aims to understand the impact
of specific mental health diagnosis (e.g., ADHD,
depression, anxiety) on asthma-related use and
expenditure and to understand the role mental
health treatment plays on the relationship between
mental health and asthma-related use and
expenditure.
Study Design: This specific study design uses a
quasi-experimental, interrupted-time series over 36
months with no-treatment control group. A fixed
effect (FE) method was used to quantify the
differences in asthma-related cost and utilization
among those with a mental health diagnosis of
depression, anxiety, and ADHD. Also, a fixed
effects approach was used to determine whether
mental health treatment mediated asthma-related
use and expenditures for those with a mental health
diagnosis. Mental health treatment was measured
using any treatment and continuous months of
treatment (chronicity of treatment). Dependent
variables included asthma-related services (ED
visits, med/physician claims, inpatient admissions,
outpatient visits), asthma-related costs (inpatient,
outpatient, med/physician, pharmacy, and total). All
analyses were controlled for sex, race/ethnicity,
month, and ssi status.
Population Studied: Youth with asthma on
between age 5 years and 15 years enrolled in
Florida Medicaid over three years from 2002
through 2005.
Principal Findings: Precursory findings of this
population suggest that specific groups of youth
with asthma had different asthma-related
experiences by mental health diagnosis and mental
health treatment.
Conclusions: Limitations associated with this study
include the common underreport of depression,
anxiety, and ADHD in claims data, inability to
assess asthma severity, limited framework, external
validity limited to Florida Medicaid youth 5-15 years
of age, and ICD-9 coding in claims data are often
unreliable. Further research is warranted to
investigate the impact of asthma in the general nonMedicaid population, as well as examine the impact
of mental health diagnosis on other common
chronic conditions.
Implications for Policy, Delivery or Practice:
Effective screening and treatment protocols for
youth with asthma may improve health quality and
reduce medical costs.
Poster Number: 42
Examining Racial Disparities in Persistent
Depression and Mental Health Treatment Utilization
Presented by: Audrey Jones, Ph.D. Student,
Health Services, University of California, Los
Angeles, 650 Charles E. Young Drive, 31-269 CHS,
Los Angeles, CA 90095; Email: jonesa@ucla.edu
Co-authors: Vickie Mays, Ph. D., M.S.P.H.
Research Objective: The purpose of this study is
to examine how access to different modes of
psychiatric treatment is related to racial disparities
in the persistence of Depression and Dysthymia.
Study Design: Data were obtained from the
Collaborative Psychiatric Epidemiologic Surveys;
three national surveys that examined the
prevalence and correlates of DSM-IV psychiatric
disorders. DSM-IV diagnoses were assessed using
the World Mental Health Composite International
Diagnostic Interview. Persistence was defined as
meeting the criteria for Depression or Dysthymia in
the last year among those with a history of the
disorder. Participants were asked about their history
of mental health treatment including medication,
counseling or psychotherapy, speaking with
medical providers, using alternative treatments,
speaking with spiritual or religious provider, and
self-help strategies such as hotlines, support
groups, or the internet. Logistic regression models
were run to examine disparities in treatment
utilization for each of the six treatment modalities.
Logistic models were run to examine disparities in
the persistence of Depression or Dysthymia after
controlling for treatment utilization. Finally, stratified
models were used to examine the differential
effects of treatment utilization across the racial /
ethnic groups. All of the models controlled for
sociodemographic characteristics, insurance, age of
onset, suicide history, and psychiatric comorbidity.
Population Studied: Analyses were restricted to
1,791 Non-Hispanic White, Black, Latino, and Asian
adults who met the DSM-IV criteria for Depression
or Dysthymia at least two years ago.
Principal Findings: All three minority groups were
less likely than Non-Hispanic Whites to use formal
treatment services (speaking with a medical
provider, receiving counseling or psychotherapy, or
medication treatment). Minorities with a disorder
were also less likely to use informal treatment
services (religious provider, alternative treatment,
and self-help) though the results varied by
treatment and racial group. All minority groups were
more likely than Non-Hispanic Whites to meet the
criteria for a persistent disorder. These effects
remained after controlling for differential use of
formal and informal mental health treatment. Use of
counseling / psychotherapy and self-help strategies
were associated with lower persistence. Use of
alternative treatments was associated with greater
persistence. Speaking with a medical provider,
speaking with a religious provider, and use of
medication treatments were not associated with
persistence after controlling for the other treatment
modalities. In the stratified analyses, counseling /
psychotherapy, self-help strategies, and alternative
treatments were associated with persistence among
White adults. However, none of the treatment
modalities were associated with persistence in the
models for Black, Latino, or Asian adults.
Conclusions: Racial minorities with a psychiatric
disorder were less likely to receive treatment and
more likely to than Non-Hispanic Whites to meet the
criteria for a persistent disorder. Some forms of
treatment were associated with lower persistence
among White adults, but most forms of treatment
were not associated with the course of Depression
or Dysthymia among the minority groups.
Implications for Policy, Delivery or Practice:
Recent reports have highlighted racial differences in
persistence. More information is needed regarding
treatment effectiveness for minority groups as well
as cultural factors that may influence the course of
depression.
Funding Source(s): AHRQ
Poster Number: 43
Factors Influencing the Presence of Nursing Home
Based Certified Nurse Aid Training Programs
Presented by: Hye-Young Jung, B.A.,
Department of Community Health, Brown
University, 121 South Main Street, Box G-S121,
Providence, RI 02912; Phone: (401) 863-3172;
Email: hye-young_jung@brown.edu
Co-authors: Denise Tyler, Ph.D.; Zhanlian Feng,
Ph.D.; Vincent Mor, Ph.D.
Research Objective: The Omnibus Budget
Reconciliation Act of 1987 mandated that all
certified nursing assistants (CNA) receive at least
75 hours of training. Previous research indicates a
significant decline in facility-based CNA training
programs over last decade despite an overall
increase in CNA staffing levels. However, no
studies have attempted to identify influences
leading to the decline. The objective of this study is
to identify the determinants of NH based CNA
training programs.
Study Design: This was an observational,
longitudinal design testing for the annual presence
of a NH based CNA training program. Fixed effects
were included in each model to adjust for time
invariant state characteristics and year to year
trends. Generalized estimating equations were
utilized to account for the longitudinal nature of the
data and to adjust the standard errors for repeated
observations. Using a discrete-time logit model has
the benefit of accommodating both the binary
outcome and repeated observations associated
with the same facility over time, in addition to
accounting for duration effects.
Population Studied: Data came from the On-Line
Survey Certification and Reporting System. Freestanding NHs with 20 or more beds in the period
1998 through 2008 were included. A total of 15,230
NHs (144,344 facility years) were used in logistic
regressions. For the discrete time logistic models,
42,231 facility-years (8720 NHs) were identified as
at risk of dropping CNA training programs, whereas
83,311 facility-years (13,488 NHs) were recognized
as at risk of adopting programs.
Principal Findings: The prevalence of NH based
CNA training programs fell from 34.8% in 1998 to
22.8% by 2008. Estimates from the discrete time
logit model indicate that a high deficiency score was
a strong contributor to NHs dropping CNA training
programs (OR 1.56, p < .001), as was being located
in an urban area (OR 1.29, p < .001). Large NHs
(OR 1.34, p < .001) were more likely to adopt
programs, whereas facilities with more CNA hours
per resident day were less likely to adopt them (OR
0.68, p < .001). Other organizational factors, such
as profit status, chain affiliation, and market
competition were also shown to be significant
determinants of program status. Logistic estimates
were consistent with these results.
Conclusions: NHs’ total deficiency score, a
measure of regulatory compliance, is an important
determinant of whether a NH maintains a CNA
training program. Strict state regulations
presumably lead to involuntary program
termination. A relatively abundant supply of
alternative training sites in urban areas may explain
why facilities in urban areas are more likely to drop
CNA programs, while NHs with insufficient CNA
staffing levels are more likely to adopt them.
Implications for Policy, Delivery or Practice: The
majority of CNA training programs are facilitybased, but their prevalence is steadily declining.
Other training sites are increasingly supplying
CNAs. Understanding both the organizational and
market forces leading to CNA training programs
shifting away from NHs provides insight for proper
regulatory and oversight adjustments. More
importantly, it raises questions about the quality of
CNA training and its potential influence on the
quality of care received by residents.
Funding Source(s): NIA
Poster Number: 44
Using Administrative Data to Identify Patients at
Risk of Hospital Readmission within 30 Days: The
Development of an Algorithm to Identify High Risk
Patients
Presented by: Jillian Jweinat, B.S., Rush
University, Chicago, IL; Phone: (650) 303-6423;
Email: Jillian_Jweinat@rush.edu
Co-authors: Tricia Johnson, Ph.D.; David Ansell,
M.D., M.P.H.; Raj Behal, M.D., M.P.H.; Susan
Huerta, Ph.D., M.S., R.N.; Shannon Sims, Ph.D.,
M.D.
Research Objective: The need to systematically
identify and intervene on patients at high risk for
readmission is critical as a means to improve
quality and decrease costs. Little evidence exists
about how to measure the risk of readmission.
Existing publicly available algorithms for patients at
risk for readmission have been developed in other
countries or on narrowly defined subsets of U.S.
patients using administrative and patient data.
When existing algorithms are applied to a broader
U.S. patient population, different factors may be
important predictors of 30-day readmission, and the
costs savings trade-offs will likely differ. In response
to this need, Rush University Medical Center
(RUMC) committed to developing and validating a
predictive algorithm to assess the likelihood for
readmission within 30 days of hospital discharge for
acute care patients.
Study Design: Data for all adult hospital
discharges between July 2008 and June 2009 were
obtained from the RUMC Financial Data Mart, an
administrative database of all hospital discharges.
All adult patients age 18 or older who were
discharged alive were followed for 30 days to
determine whether they were readmitted to RUMC.
A binary logistic regression model was fit to
estimate the relationship between 30-day
readmission and sociodemographic factors, carerelated factors and severity of illness. The
performance of the model was tested using
sensitivity and specificity. The algorithm was
developed using 75% of the sample and validated
on the remaining 25%.
Principal Findings: A total of 17,265 unique
patients were discharged from RUMC during the
study time period, with 1,789 (10.4%) readmitted at
least once within 30 days of discharge. Of those
readmitted, 1,478 (8.6%) were readmitted once;
265 (1.5%) were readmitted twice; and 46 (0.3%)
were readmitted three or more times. Predictors
included being male; discharged to an inpatient
rehabilitation facility or home under the care of an
organized home health service organization;
general surgery and cardiovascular/thoracic surgery
attending physician department; length of stay
greater than 4 days; and specific comorbidities
including coronary heart failure, renal failure,
metastatic cancer, coagulation deficiency and
deficiency anemia. The discriminatory power of the
algorithm was modest as determined by area under
the receiver operating characteristic curve for the
development (c=0.690) and validation (c=0.709)
samples. Using a cut value of 0.1, the sensitivity
was 61.9% and the specificity was 64.7% on the
development sample and 68.6% and 61.3% on the
validation sample.
Conclusions: Hospitals can develop a risk
stratification methodology based on their own
patient populations using electronic data. Routine
administrative hospital episode statistics can be
used to identify patients who are at high risk of
suffering from multiple readmissions.
Implications for Policy, Delivery or Practice:
Factors common to those readmitted within 30 days
can be determined and interventions around
discharge planning and care coordination at the
hospital level to keep high-risk patients from being
readmitted again can be developed. Systematic
identification of patients at high risk for readmission
has the potential of reducing avoidable
rehospitalizations and unnecessary costs in the
health care system. These results are particularly
important, with health care reform efforts focused
on reducing avoidable readmissions.
Poster Number: 45
Diabetic Care in General Practice: An Assessment
of Who Guidelines Followed by General
Practitioners
Presented by: Rukhsana Khan, M.B.B.S., M.P.H.,
M.Sc., Doctor, Epidemiology, Health Services
Academy, Health Services Academy, Park Road,
Chak Shehzad, Islamabad, Pakistan; Phone:
+00923445509238; Email:
drrukhsanakhan@hotmail.com
Lead Author: Hameed Durrani, M.B.B.S., M.S.P.H.
Research Objective: The aim of the study is to
improve the outcome of diabetics by assessing the
proportion of the general practitioners following
WHO guidelines for the management of diabetes
mellitus in Rawalpini city of Pakistan.
Study Design: Descriptive- Cross Sectional
Population Studied: General Practitioners of
Rawalpindi City were selected and interviewed for
their knowledge regarding signs, symptoms,
laboratory diagnosis, treatment, complications and
education of the diabetics as per WHO guidelines
Principal Findings: reasonable number of GPs
have knowledge of signs, symptoms, and
complications of the disease but surprisingly only
17% of the total responded correctly regarding
laboratory diagnosis, 61% regarding treatment and
40% educate the patient about the disease. As far
as application of knowledge to manage the patient
is concerned, only 5% of the total GPs were found
following complete (100%) of the WHO guidelines
for diabetes management.
Conclusions: the general practitioners underdiagnose under-educate the diabetics as per WHO
guidelines
Implications for Policy, Delivery or Practice:
This study provides preliminary but important
information about the standards of diabetes care in
Pakistan, a country which is already facing steep
increase in diabetes prevalence and complications.
Improving quality of life for persons with diabetes
and reducing morbidity and mortality are major
health-care challenges for providers and
governments in developing countries.
Funding Source(s): Health Services Academy
Poster Number: 46
Logic Modeling for Policy Analysis: A Case Study of
HIV Testing Policy Change at the VA
Presented by: Erika Langer, Department of Health
Policy and Management, Boston University School
of Public Health, 715 Albany Street, Boston, MA
02118; Phone: (617) 414-6991; Email:
elanger@bu.edu
Research Objective: Since 2006, both the Centers
for Disease Control and Prevention (CDC) and the
Department of Veterans Affairs (VA) have issued
revised HIV testing policies. These agencies
operate within a U.S. context where 1.2 million
people are estimated to be living with HIV, and
more than 25% of those infected do not know their
status. Despite widespread cost-effectiveness of
HIV screening in a majority of health care settings,
there are a significant number of missed
opportunities for testing, and HIV-positive
individuals continue to be identified late in the
disease process. This study evaluates the extent to
which a recent change in VA’s HIV testing policy
reflects the current CDC guidance on HIV testing.
Using logic modeling, the study compares
components of the final rule, “Elimination of
Requirement for Prior Signature Consent and Preand Post-Test Counseling for HIV Testing”,
published by the VA in the Federal Register in July
2009, and the 2006 Revised Recommendations for
HIV Testing from CDC. Agencies' use of research
evidence in justification of policy components is
also examined.
Study Design: This case study is a policy analysis
of VA HIV testing policy adoption in 2009. Logic
modeling was conducted to examine the sequence
of events thought to bring about desired change, as
well as the evidence base used to justify linkages
between policy components. CDC and VA logic
models were analyzed for the extent of policy
overlap.
Principal Findings: VA has largely replicated the
policy logic underlying CDC guidance formulation.
The agency draws from a setting-specific evidence
base to justify components of its HIV testing policy
change. While evidence was given in support of
policy resources, activities, outputs, and outcomes,
arguments about the longer-term impact of this
policy lacked a clear basis of research evidence
support.
Conclusions: Research utilization differs between
policy formulation and adoption stages as
contextualized justifications become more relevant
to policy argument. By identifying research
utilization along the policymaking process,
researchers may begin to evaluate how the
evidence base for policy changes over time and
across settings. This is significant to our
understanding of the conditions for evidence-based
health policymaking.
Implications for Policy, Delivery or Practice:
This study aims to identify instances of applied
research evidence within policy formulation and
adoption decisions as a mode of planning for future
policy implementation and evaluation. The use of
logic modeling also allows for explicit examination
of the actors, context, and content which interact
along this process. Findings from this study may
inform future evaluation of VA HIV testing policy
change, and have additional utility to other health
care settings that may be considering adoption of
CDC HIV testing recommendations.
Poster Number: 47
Patient-Provider Language Concordance and
Colorectal Cancer Screening
Presented by: Amy Linsky, M.D., Department of
Public Health, Health Services Research, Boston
University, 64 Perry Street, Brookline, MA 02446;
Phone: (617) 277-0148; Email:
mcintnm@gmail.com
Lead Author: Nathalie McIntosh, M.Sc.
Co-authors: Amy Linsky, M.D.; Howard Cabral,
Ph.D.; Lewis Kazis, Ph.D.
Research Objective: Colorectal cancer (CRC)
screening rates are lower in racial and ethnic
minorities, and patient-provider language barriers
may play a role. Professional interpreters and
language-concordant providers may mitigate this
disparity. The purpose of this study is to assess the
association of the presence of patient-provider
language concordance with CRC screening in
patients who do not speak English at home
compared to rates in those who do.
Study Design: We did a retrospective cohort study
using the 2002, 2004 and 2006 Medical
Expenditures Panel (MEP) Surveys. MEP surveys
are nationally representative surveys of noninstitutionalized US civilians. The primary
independent variable created three cohorts:
English-Concordant (spoke English at home), Other
Language-Concordant (did not speak English at
home but someone at their provider’s spoke their
language or offered translation services), and
Language-Discordant (did not speak English at
home and no one at their provider’s spoke their
language or offered translation). The primary
outcome of current CRC screening was defined as
receipt of fecal occult blood testing within the past 2
years or endoscopy within the past 5 years.
Covariates included: race/ethnicity, age, sex,
education, marital status, income, employment,
time since last check-up, insurance, physical
component score, year, and region. Using survey
methods, multivariate logistic regression assessed
the association between language concordance
and CRC screening.
Population Studied: US adults 50 years and older
who had no history of colorectal cancer and who
completed Medical Expenditures Panel Surveys in
2002, 2004 and 2006.
Principal Findings: The final study sample of
23,297 represents 222 million individuals. The
majority spoke English at home (96%). English
Concordant (n=21,594) were likely to be white nonHispanic (84%) and have a high school education,
high income and private insurance. Other
Language-Concordant (n=1,463) were likely to be
Hispanic (61%), less educated, poor and have
public insurance. Other Language-Discordant
(n=240) had the highest percentage of Asians
(29%) and were likely to be older, unemployed and
uninsured. Compared to English speakers, nonEnglish speakers had lower adherence to CRC
screening [30.7% vs. 50.8%; Odds Ratio (OR),
0.63; 95% Confidence Interval (CI), 0.52-0.77]. In
adjusted analyses, compared to EnglishConcordant, Other Language-Discordant had
similar adherence [adjusted OR, 0.83; 95% CI,
0.58-1.20], while Other Language-Concordant had
lower adherence [adjusted OR, 0.58; 95% CI, 0.460.72].
Conclusions: Rates of CRC screening are lower in
adults who do not speak English at home and were
not mitigated by having someone at the provider's
who spoke the patient's language.
Implications for Policy, Delivery or Practice:
Racial and ethnic disparities in CRC screening
rates can not be attributed soley to patients' access
to someone at their provider's who speaks their
language. Other social determinants of health, such
as culture, acculturation, and the quality of patientprovider communication may need to be addressed
to overcome this health disparity.
Poster Number: 48
Factors Associated with Proton Pump Inhibitor
Discontinuation in Nursing Home Residents
Presented by: Amy Linsky, M.D., Fellow, General
Internal Medicine, Boston Medical Center, 801
Massachusetts Avenue, 2nd Floor (GIM), Boston,
MA 02118; Phone: (617) 414-2792; Email:
amy.linsky@bmc.org
Co-authors: John Hermos, M.D.; Elizabeth Lawler,
D.Sc.; James Rudolph, M.D., S.M.
Research Objective: Continued use of
medications, beyond initial indications or targeted
endpoints, contributes to over-medication in elderly
patients. Proton pump inhibitors (PPIs) are
commonly prescribed drugs that are often initiated
and continued without clear indication. As nursing
home admission is an opportunity for review of
medication appropriateness, this study sought to
identify factors associated with PPI discontinuation
among patients admitted to nursing homes.
Study Design: Retrospective cohort study of
veteran nursing home patients prescribed PPIs on
admission. Baseline exposure to oral PPIs required
an active inpatient prescription within the first three
days of admission. The primary outcome was PPI
usage (“continuation” or “discontinuation”) at 90
days. Covariates included patient sex, age, year of
nursing home admission, comorbidities, and prior
PPI use, defined as a VA prescription in the 30
days prior to admission. Multivariable logistic
regression and 95% confidence intervals (CI)
estimated the associations of patient characteristics
with PPI discontinuation.
Population Studied: The study population included
patients, age 40 years and older, admitted to
Veterans Affairs (VA) Community Living Centers in
2005-2006 for a minimum stay of 90 days. We
identified 26,743 patients, of whom 6,416 (24%)
were prescribed a PPI at nursing home admission.
Principal Findings: Of baseline PPI users, 2,006
(31%) had the medication discontinued by 90 days
while the remaining 4,440 (69%) patients were
continued on PPIs. Compared to those with
continued PPI use, patients discontinued were
similar in age (median age 66 vs. 67 years, p=0.8)
and sex (97.4% vs. 96.9% male, p=0.3). There
were no statistically significant differences between
continuers and discontinuers in prevalence of ICD9-CM coded comorbidities, including diabetes,
hypertension, cardiovascular disease, chronic
kidney disease, liver disease, malignancy and
substance abuse. Those who discontinued were
more likely to have prior PPI use (79% vs. 72%,
p<0.0001), be admitted in 2006 rather than 2005
(53% vs. 34%, p<0.0001) and have dementia (7.2%
vs. 5.7%, p=0.01). Discontinuers were less likely to
have chronic obstructive pulmonary disease
(COPD) (33% vs. 38%, p=0.0005) and gastric acid
or reflux related upper gastrointestinal disorders
(34% vs. 42%, p<0.0001). In an adjusted model,
factors associated with increased odds of PPI
discontinuation were dementia [odds ratio (OR) 1.4;
95% CI 1.1-1.7] and prior PPI use (OR 1.6; 95% CI
1.4-1.8). Diagnoses of upper gastrointestinal
disorders (OR 0.68; 95% CI 0.60-0.76) and COPD
(OR 0.85; 95% CI 0.76-0.96) were associated with
a decreased odds of discontinuation.
Conclusions: Within 90 days of admission to a VA
nursing home, PPI prescriptions were more likely to
be discontinued in patients with prior PPI use - that
is, not initiated in the nursing home - and those with
dementia. Patients with documented upper
gastrointestinal diseases and COPD were less
likely to have PPIs discontinued.
Implications for Policy, Delivery or Practice:
These findings would suggest a reasoned
evaluation of PPI needs at early stages of care, but
the extent to which physician and system level
factors, as well as patient functional status and
concurrent medication use, impact decisions of
medication discontinuation remains to be
elucidated.
Funding Source(s): VA
Poster Number: 49
The Relationship Between a Physician Incentive
Plan and the Hospital Performance
Presented by: Nai Yng Liu, M.S., student,
Accounting, National Taipei University, 2F.-8, No.5,
Fuhe Road., Yonghe City, Taipei County 234,
Taiwan (R.O.C.), Taipei, 23449 Taiwan; Phone:
+011886960111392; Email: naiyng@gmail.com
Research Objective: In this study, we investigate
the appropriateness of incentive plans that link
physicians’ compensation to the measurement of
their performance in a large Taiwan private not-forprofit hospital.
Study Design: The main research question in our
study is investigated the relationship between a
physician incentive plan and the hospital
performance. Hospital performance is measured by
profitability and efficiency. We use the multiple
regression model to examine the relationship
between the size of cash bonuses and hospital
profitability. The Data Envelopment Analysis (DEA)
model is used to measure the operational efficiency
of each department in the case hospital. Then, a
multi-factor tobit model is used to examine factors
that might influence hospital efficiency to examine
the effectiveness of the incentive plan.
Population Studied: Monthly financial data and
related information for twenty-six departments
during the period from January 2007 to June 2009
were provided by the case hospital.
Principal Findings: When evaluating the
relationship between physicians and cash bonuses,
we found that physicians’ cash bonuses are
significantly negative to hospital performance and
efficiency. Using current incentive plan
measurement, the case hospital’s per employee
productivity contribution is significantly (p<0.001)
negative to physicians’ cash bonuses. Meanwhile,
the directed costs, most of which can be controlled
by physicians, are significantly (p<0.001) negative
to the incentive payment too. In addition, using
multi-factor tobit model to evaluate the efficiency,
we learnt that the case hospital is less efficiency
when implementing cash bonuses.
Conclusions: Case hospital’s current incentive
plan, which based on physicians’ medical practice
quantity, jeopardizes its financial situation and
decreases its efficiency. The case hospital should
design an incentive plan that includes multidimension measurement, such as cost control and
efficiency improvement. The case hospital needs to
modify its current incentive plan; this inappropriate
incentive plan will decay its competition and even
might let case hospital facing bankruptcy.
Implications for Policy, Delivery or Practice: Our
study contributes to the literature by providing
empirical evidence that examines the
appropriateness of performance measures in
physician incentive plans for the healthcare
industry. The study also offers useful lessons for
practitioners concerning pitfalls (e.g., dysfunctional
behavior) to avoid when they design incentive plan
for physicians.
Poster Number: 50
Impact of Financial Incentives on Physician EMR
Implementation and Functionality Use: A Statewide
Survey
Presented by: Chang Liu, Student, Community
Health, HSR Section, Brown University, Box GS121, 121 South Main Street, Providence, RI
02912; Phone: (401) 523-7299; Email:
chang_liu@brown.edu
Co-authors: Amal Trivedi, M.D., M.P.H.; Rosa
Baier, M.P.H.; Rebekah Gardner, M.D.
Research Objective: In the context of an economic
downturn and the soaring cost of health care, and
with limited evidence linking EMRs to improvements
in health outcomes at a national level, the federal
government is planning an ambitious program to
align financial incentives on the implementation and
functionality use of computerized patient records.
The objective of my research is to justify the costeffectiveness of this approach before a substantial
investment is enacted.
Study Design: Based on the 2009 Physician HIT
Survey of Rhode Island, an annual survey
conducted by the Rhode Island Department of
Health, we assessed the current state of EMR
implementation and functionality use among
physicians in active practice in Rhode Island, and
compared these results to national and regional
data derived from previous studies. Moreover, in
this cross-sectional study, we examined the
relationship between financial incentives and the
possibility of adoption and utilization of EMRs by a
two-step regression model.
Population Studied: In early 2009, the Rhode
Island Department of Health (HEALTH) conducted
and administered the 2009 Physician HIT Survey to
all 3,248 physicians licensed in Rhode Island.
Rhode Island was the first state in the nation to
report individual-level measures of health
information technology adoption for all licensed
physicians. In our study, we excluded respondents
who did not have a record in HEALTH’s licensure
file, who were not in active practice or providing
direct patient care, and whose licensure state was
not Rhode Island, Connecticut, or Massachusetts.
The overall response rate of the survey was 58.1%
(n=1,888).
Principal Findings: Based on responses from
58.1% of licensed physicians in both office-based
and hospital-based settings, we found that in 2009,
67.6% of the physicians in Rhode Island had some
EMR components in their practice, 12.5% of them
had “qualified” EMRs (22.7% if CCHIT certification
is excluded), and 41.2% of them were eprescribing. On average, responding physicians
scored 63.6 (out a possible 100 for maximum use)
for “basic” EMR functionality use, and 44.1 for
“advanced” EMR functionality use. These results
are much higher than the average performance of
physicians nationwide. After controlling for the
physician and organizational characteristics (e.g.,
practice setting, practice size, and hours of direct
patient care), we found that physicians who were
eligible for financial incentives were more likely to
adopt and use EMRs than physicians who were not
eligible.
Conclusions: Our study suggests that aligning
financial incentives to physicians may play a crucial
role in accelerating the implementation and
functionality use of electronic medical records.
Implications for Policy, Delivery or Practice: We
support the proposed federal program that provides
incentive payments to physicians to adopt and
“meaningfully use” certified EMRs.
Funding Source(s): Rhode Island Department of
Health
Poster Number: 51
Validating the Post Operative Hip Fracture PSI in
the VA
Presented by: Julie Lynch, R.N., M.B.A.,
Research Assistant and Student, Nursing,
University of Massachusetts, Boston, 100 Morrissey
Boulevard, Boston, MA 02125; Phone: (617) 6995032; Email: julie.lynch001@umb.edu
Co-authors: Ann Borzecki, M.D.; Amy Rosen,
Ph.D.; Sally MacDonald, R.N.
Research Objective: To examine the Agency for
Healthcare Research and Quality(AHRQ) post
operative hip fracture Patient Safety Indicator (PSI)
in the Veterans Health Administration for criterion
validity. Secondary analysis was conducted to
analyze presence of multisite pain and diagnosis of
osteoarthritis in patients.
Study Design: The AHRQ’s PSI software was
applied to 2003 to 2007 inpatient administrative
data from a representative sample of 28 VA
hospitals to identify flagged cases of post operative
hip fracture. Fourty-six cases were identified.
Trained nurse abstractors conducted a
retrospective chart reviews of the electronic medical
record to determine rates of true positives (TPs)
and false positives (FPs). Interrater reliability testing
of 2 nurse abstractors was performed on 10% of
charts. Cases were examined with respect to
baseline clinical characteristics, operative
procedures and completeness of a fall risk
assessment. Medical records of all cases were then
reviewed to analyze reports of multisite pain and
diagnosis of osteoarthritis in patients with hip
fracture.
Population Studied: Surgical admissions between
fiscal years 2003-2007 in 28 acute care Veterans
Administration Hospitals
Principal Findings: Of the 46 cases analyzed, 13
(28%) were true positives (positive predictive value
of 28%, and 33 (72%) were false positives. Of the
33 false positives , 24 (70%) were determined to be
present upon admission; 5 (15%) patients had falls
that occurred in hospital but prior to an operative
procedure; 3 (9%) patients were inaccurately
coded; 1 (6%) patient had a medical event (stroke)
which excluded him. Findings related to presence of
multisite pain and diagnosis of osteoarthritis is
ongoing.
Conclusions: Limiting hip fracture to a secondary
diagnosis does not exclude many cases with a hip
fracture present upon admission. Low numbers of
posterative hip fracture cases identified in this study
supports findings by Iezzoni et al. (1994) that
postoperative hip fracture continues to be a rare
event. 69% of true positives had no fall assessment
or were rated as being at no fall risk depsite having
many obvious risk factors such as mobility
impairment or prior history of falls.
Implications for Policy, Delivery or Practice: A
present upon admission indicator is needed to
identify hip fractures that exist on admission. Chart
review of administrative data identified cases of
postoperative hip fracture PSI remains a necessary
and recommended step for hospitals prior to
reporting this measure. Although a rare event,
postoperative hip fracture is a complication that is
sensitive to nursing interventions to reduce fall risk.
Therefore fall risk assessment should be standard
of care. Fall risk assessment policies, procedures,
and tools should be evaluated by all hospitals to
ensure tools are meeting the objective and nurses
are adequately trained to conduct fall risk
assessment and implement interventions to reduce
fall risk.
Funding Source(s): VA
Poster Number: 52
Racial and Mental Health Disparities in Safe Sex
Counseling for Individuals with Substance
Dependence
Presented by: Meredith Manze, M.P.H., Research
Coordinator, Section of General Internal Medicine,
Boston Medical Center, 801 Massachusetts
Avenue, #2098, Boston, MA 02118; Phone: (617)
414-8441; Email: meredith.manze@bmc.org
Co-authors: Debbie Cheng, Sc.D.; Donald
Allensworth-Davies, M.Sc.; Jeffrey Samet, M.D.,
M.A., M.P.H.; Richard Saitz, M.D., M.P.H.,
F.A.S.A.M., F.A.C.P.
Research Objective: Despite a higher prevalence
of sexual risk behaviors among people with
substance dependence than in the general
population, little is known about racial/ethnic and
mental health disparities in safe sex counseling and
behavior among such people. We sought to identify
whether these disparities exist, accounting for
sociodemographic variables and quality of patientclinician relationship.
Study Design: This study was a cross-sectional
analysis using self-reported interview data.
Race/ethnicity and mental health were our key
independent variables. Three mental health
variables were evaluated: depression (9-item
Patient Health Questionnaire (PHQ-9) score > 10),
psychotic symptoms (?2 symptoms from the BASIS
24 selected items), and any manic/hypomanic
episodes (Mini International Neuropsychiatric
Interview). Patient self-report of their regular doctor
having 'ever talked to them about safe sex' and
'ever practiced safer sex because of their doctor’s
advice' were the two dependent variables (Primary
Care Assessment Survey (PCAS)). The PCAS
“trust” and “communication” scales were used to
assess the quality of the patient-clinician
relationship. We performed multiple logistic
regressions, controlling for age, education, gender
and PCAS scale scores, to test associations
between mental health and race/ethnicity with each
outcome. Each model contained only one mental
health variable and one PCAS scale due to
correlation among variables.
Population Studied: We analyzed data from 275
participants with alcohol and/or drug dependence
who indicated they had a regular doctor.
Principal Findings: The majority of participants
were male (68%) and had at least a high school
education (77%); mean age was 40. Most
participants had a PHQ-9 score greater than 10
(53%), and reported at least two psychotic
symptoms (56%); 40% reported any
manic/hypomanic episodes. The mean scores for
the trust and communication scales were 73 and
75, respectively (0-100 scale). The majority of
participants reported having their regular doctor talk
to them about safe sex (71%) and practicing safer
sex because of their doctor’s advice (61%).
Compared to non-Hispanic Whites (reference
group), being Hispanic or non-Hispanic Black
race/ethnicity was associated with significantly
greater odds of receiving safe sex counseling
(adjusted odds ratio (AOR) 2.67, 95% CI:1.01, 7.09;
AOR 2.74, 95% CI:1.43, 5.27, respectively) and
practicing safer sex due to the advice (AOR 3.76,
95% CI: 1.47, 9.63; AOR 2.26, 95% CI: 1.26, 4.07,
respectively). Patients who reported any
manic/hypomanic episode had more than two times
the odds of reporting having a doctor ever talk to
them about safe sex (AOR=2.54, 95% CI: 1.38,
4.70), but no significant association with practicing
safer sex as a consequence. We did not detect
associations between depression and psychosis
with reports of safe sex counseling or practicing
safer sex.
Conclusions: Minorities and patients with manic
episodes appear more likely to receive safe sex
counseling and, for minorities, to practice safer sex
as a result. The increased counseling for minorities
may be in response to their higher rates of
unintended pregnancy and sexually transmitted
infections.
Implications for Policy, Delivery or Practice:
Future research should explore provider
perceptions of risk of unsafe sex, particularly for
minorities and people with mental illness, and why
more counseling may not translate into safer sex
behaviors for people with manic episodes.
Funding Source(s): NIAAA
Poster Number: 53
Comparative Effectiveness of Surgery for Lumbar
Spinal Stenosis in Older Adults: Trends, Major
Medical Complications, and Charges Associated
with Surgical Invasiveness
Presented by: Brook Martin, Ph.C., M.P.H.,
Health Services Research Scientist, Comparative
Effectiveness, Cost, and Outcomes Research
Center, University of Washington, 325 Ninth
Avenue; Campus box 359736, Seattle, WA 98104;
Phone: (206) 744-1803; Email:
bim@u.washington.edu
Lead Author: Richard Deyo, M.D., M.P.H.
Co-authors: Brook Martin, Ph.C., M.P.H.; Jeffrey
Jarvik, M.D., M.P.H.; William Kreuter, M.P.A.;
Sohail Mirza, M.D., M.P.H.
Research Objective: The fastest growth in lumbar
surgery has occurred in older patients with spinal
stenosis. Trials indicate that for selected patients,
decompressive surgery offers an advantage over
non-operative treatment, but surgeons often
recommend more invasive fusion procedures.
Comorbidity is common in elderly patients, so
benefits and risks must be carefully weighed in the
choice of surgical procedure. We examine trends in
use of different types of spinal stenosis operations
and the association of complications, healthcare
use, and costs with surgical complexity.
Study Design: Retrospective cohort analysis of
Medicare claims data for 2002-2006 was used to
examine surgical trends. We then focused on 2005
to assess complications and resource use.
Operations were grouped into 3 gradations of
invasiveness: decompression alone, simple fusion
(one or two disc levels, single surgical approach) or
complex fusion (more than 2 disc levels or
combined anterior and posterior approach). We
used multivariate logistic regression, controlling for
age, comorbidity, previous spine surgery, and other
patient characteristics, to examine the association
between surgical invasiveness and serious
complications, rehospitalizations within 30 days,
and nursing home discharges. Costs were
estimates by applying Medicare cost-to-charge
ratios to reported charges.
Population Studied: Medicare recipients
undergoing surgery within U.S. acute care hospitals
for lumbar stenosis. We examined trends in overall
surgical rates using data from 2002-2006, and then
focused on 2005 data (n=36,165) to assess
complications and costs associated surgical
invasiveness.
Principal Findings: Overall surgical rates changed
little from 2002-2006, but the rate of complex fusion
procedures increased 14-fold, from 1.3 per 100,000
in 2003 to 18.2 per 100,000 in 2006. The risk of lifethreatening complications increased with increasing
surgical invasiveness. Compared to decompression
alone, the adjusted odds ratio (OR) for life
threatening complications among patients
undergoing complex fusion was 3.04 (95% CI 2.583.59). A similar pattern was observed for
rehospitalization within 30 days (OR 1.98; 95% CI
1.77-2.21) and probability of discharge to a nursing
home (OR 2.67; 95% CI 2.42-2.95). Adjusted mean
hospital charges for complex fusion procedures
were $24,517 compared to $6,593 for
decompression alone.
Conclusions: Use of complex procedures for
spinal stenosis among the Medicare population is
increasing, and is associated with greater risk of
major complications, mortality, and resource use.
Implications for Policy, Delivery or Practice:
These findings suggest a need to improve
indications for more invasive lumbar spinal stenosis
surgery; to involve patients in informed decisionmaking; and to determine the safest and most
effective surgical techniques.
Funding Source(s): NIAMS
Poster Number: 54
An Intervention to Improve Medication Adherence in
a Vietnamese Population Under Care for One or
More Chronic Conditions
Presented by: Jennifer Mason, B.A.; B.S.; M.P.H.,
Graduate Student, Public Health, San Diego State
University, 5500 Campanile Drive, San Diego, CA
92182; Phone: (619) 594-6317; Email:
jaymay6@gmail.com
Research Objective: The aim of the study is to
learn whether increasing patient knowledge about
their medications through individual counseling and
the use of compliance aids (pill boxes) improves
pill-taking behaviors.
Study Design: The study design was a prospective
randomized pilot study with an enrollment of 8
weeks.
Population Studied: The participants were
recruited from the Linda Vista Health Care Center in
San Diego, California with the inclusion criteria that
they were at least 21 years of age, Vientamese and
being treated for hypertension, high cholesterol,
diabetes and/or asthma.
Principal Findings: Significant improvements in
pill-taking behaviors were found by self-report with
the introduction of pill box usage and baseline
education on medicine regimen delivered by a
pharmacy student. Additional outcomes on
biological measures including lipid panels, glucose
levels and blood pressure readings are currently
being analyzed.
Conclusions: Final write-up and conclusions are
currently being drafted.
Implications for Policy, Delivery or Practice:
Potential benefits are improving chronic disease
management for high-risk populations in a
community health clinic setting. Further, potentially
improve patient-provider relations due to increased
knowledge of decision making between traditional
values and Western medicine.
Poster Number: 55
The Impact of Mental Health/Cognitive Disability
Upon Healthcare Utilization: Perspectives from
Patients and Healthcare Providers
Presented by: M. Maya McDoom, M.P.H., B.S.,
Ph.D. Candidate, Health Policy and Management,
Boston University School of Public Health, 715
Albany Street T3W, Boston, MA 02118; Phone:
(617) 414-1362; Email: mmcdoom@bu.edu
Co-authors: Mari-Lynn Dranioni, Ph.D., M.Ed.
Research Objective: The US Census estimates
nearly one in five persons have some type of
disability, comprising 54.4 million people. Of the
one in five persons, approximately 16.1 million have
some level of mental health or cognitive disability .
Persons with disabilities often face undue barriers
in accessing health services and medical treatment.
Differences in perceived barriers to healthcare
access between healthcare providers and persons
with disabilities can influence the health care
experience. This study evaluated the perceptions of
barriers to healthcare access between persons with
disabilities and healthcare providers. Study
objectives were to: 1. assess the impact of disability
upon healthcare service utilization; 2. examine the
relationship between disability and barriers to
healthcare access; 3. determine if a relationship
exists between healthcare provider type and
perceptions of barriers to healthcare; and 4.
compare perceptions of persons with disabilities
about their barriers to healthcare to providers’
perceptions.
Study Design: Retrospective database review of
de-identified survey data examining patientperceived barriers to care compared with
healthcare provider perceptions of barriers to care.
Population Studied: Persons with disabilities
receiving Medicaid benefits enrolled in an HMO and
healthcare providers in the state providing care to
the population grouped into five categories:
outpatient mental health, inpatient mental health,
outpatient medical, inpatient medical, and dentists
Principal Findings: Persons with mobility
limitations had higher odds of utilizing overnight
hospital services (OR 1.48, p 0.0298), lower odds of
outpatient mental health (OR 0.648, p 0.015) and
inpatient mental health services (OR 0.566, p
0.040) compared to those who do not have mobility
limitations. Persons with mental health and
cognitive disabilities had higher odds of utilizing
outpatient mental health, dentists, and inpatient
mental health services. Persons with mobility and
cognitive limitations had higher odds of
experiencing barriers related to need and
understanding. Over half of providers indicated
daily care for persons with mobility, mental illness,
or cognitive limitations. Strikingly, there were
significant differences in perception of barriers by
provider type. Dentists (OR 4.00, p 0.0012) and
outpatient mental health/substance abuse providers
(OR 2.52, p 0.0102) were more likely to perceive
inadequate insurance coverage compared to
outpatient medical providers. Inpatient medical (OR
0.626, p 0.037) and inpatient mental
health/substance abuse providers (OR 0.581, p
0.023) were less likely to perceive inadequate
insurance coverage compared to outpatient medical
providers. Dentists also had higher odds of
perceiving barriers related to entering buildings,
exam rooms, and restrooms.
Conclusions: Health care access is an important
issue, but persons with mental health or cognitive
disabilities can face barriers to accessing
healthcare beyond what can be physically
observed. Healthcare providers’ often have very
limited awareness and understanding of the barriers
of persons with cognitive or mental health
disabilities.
Implications for Policy, Delivery or Practice: The
perception gap between healthcare providers and
persons with mental health or cognitive disabilities
highlights the importance of addressing the specific
needs of persons with different types of disabling
conditions rather than assuming that all persons
with disabilities face the same challenges in
accessing healthcare. There is a need to reduce
barriers to care through improvements in medical
training, health care practice settings, and health
care coverage.
Poster Number: 56
Dental Utilization and Edentulism Across Three
County Types: Urban, Rural, and High Poverty
Rural
Presented by: Jordan Mitchell, M.B.A., Research
Associate, Health Services Policy and
Management, South Carolina Rural Health
Research Center, 220 Stoneridge Drive, Suite 204,
Columbia, SC 29210; Phone: (919) 760-5650;
Email: jpm0901@gmail.com
Research Objective: Issues with utilization and
access to care, such as lower incomes, lower
employment, and the low-density of dentists
contribute to people in rural areas being more
vulnerable to oral health problems as compared to
their urban counterparts. In all income ranges,
dental care is deemed as a lower priority than other
medical care; thus with the added employersponsored dental insurance variable added to the
equation, preventative dental visits are utilized at a
much lower rate in rural areas than other medical
visits. While there is little literature available on SES
and racial impacts on dental visits and tooth loss,
there is less information on the rural/urban impact
as well as the combination of high-poverty and rural
and its impact on dental visits and tooth loss; this
paper seeks to address this gap of information.
Study Design: Analyses were conducted using
data from the Behavioral Risk Factor Surveillance
System (BRFSS) from the Centers for Disease
Control (CDC) and the Area Resource file (ARF)
from Health Resources and Services Administration
(HRSA). HRSA is the funding source for this
project. All analyses were carried out by SAS and
SAS-callable SUDAAN, in order to account for
weighting and the complex sample design. Weights
were assigned based on the probabilistic factors
affecting the selection of a residential phone
number by state, adjusting for differences by area
codes/regions, number of adults in each household,
and non-coverage and non-response. The U.S.
Bureau of Census provided data to add weights for
the variables of age, sex, race/ethnicity, and
household size.
Population Studied: The population included
individuals 18 years of age and older in the BRFSS
survey.
Principal Findings: Lower dental visits were
significantly (p<.05) related to: males, ages 25-44,
Blacks, fair/poor health status, unemployed, some
high school – some college attainment,
divorced/separated, presence of a chronic disease,
cost deferrals, and per-capita incomes of less than
$29,972. Higher dental visits were significantly
related to: being insured, having a usual source of
care, and not being edentulous, and having at least
one dentist per a population of one thousand.
Higher edentulism rates were significantly
associated with: Black, “other” races, fair/poor
health status, unemployed, some high school –
some college educational attainment,
divorced/separated, presence of a chronic disease,
cost deferrals, and per-capita incomes of less than
$29,972. Lower edentulism rates were associated
with: all ages less than 65, being insured, and
having at least one dentist per population of one
thousand. No significant associations were found by
county-type and dental visits/edentulism.
Conclusions: While controlling for all variables
studied, county type was not significantly
associated with differences in dental visits or
edentulism at the 95% confidence level. This being
said, there are associations at a lower confidence
level that influence dental visits and edentulism in
the different county types, which need policy
changes to ameliorate.
Implications for Policy, Delivery or Practice:
Possible policy changes could include focusing on
dentist availability within rural and especially highpoverty rural areas, dental education in rural
counties and incentives for dentists to practice in
these disparate areas.
Funding Source(s): HRSA
Poster Number: 57
Provision of Rehabilitation Therapy in Home Health:
Does Payer Matter?
Presented by: Tracy Mroz, M.S., Graduate
Student, Health Policy and Management, Johns
Hopkins Bloomberg School of Public Health, 149
Archimedes Court, Pikesville, MD 21208; Phone:
(484) 226-0174; Email: tmroz@jhsph.edu
Research Objective: 1) To examine the
relationship between primary payer and receipt of
home health occupational and physical therapy
services; and 2) to examine the relationship
between primary payer and number of home health
occupational and physical therapy visits.
Study Design: This study was a secondary
analysis of cross-sectional survey data from the
2007 National Home and Hospice Care Survey
(NHHCS) publically-available patient file. Outcomes
included dichotomous variables for receipt of OT,
PT, or any therapy services and continuous
variables for number of OT and PT visits. The key
independent variable was primary payer: Medicare,
Medicaid, or private insurance. Logistic and linear
regression models adjusted for current episode
length, age, gender, race, marital status, living
arrangement, caregiver presence, cognitive status,
functional status, and number of diagnoses.
Population Studied: The population studied was
adult home health patients in the US. The study
sample included 3,647 home health patients aged
35 and over. Using NHHCS-provided weights to
account for sampling design yielded a weighted
sample of 1,070,682 home health patients, of which
approximately 40 percent received either OT, PT, or
both therapy services.
Principal Findings: Patients with Medicaid as
primary payer were significantly less likely to
receive any therapy services compared to Medicare
patients (adjusted OR (95% CI): .34 (.17-.69);
p<.01). Patients with private insurance as primary
payer did not differ significantly in receipt of any
therapy services with respect to Medicare patients
(adjusted OR (95% CI): .72 (.46-1.12); p=.15). A
similar pattern of findings persisted when examining
receipt of OT and PT services separately. Of the
patients who were receiving home health therapy
services, Medicaid patients received on average 1.5
fewer PT visits and 2.0 fewer OT visits than
Medicare patients, though these differences were
not statistically significant (p=.08 and .06
respectively). Privately insured patients received
more similar numbers of visits compared to
Medicare patients, on average .98 fewer PT visits
(p=.15) and .11 fewer OT visits (p=.91).
Conclusions: Provision of home health therapy
services appears to vary to some extent by primary
payer. Medicaid patients are less likely to receive
occupational and/or physical therapy services
compared to Medicare patients, while privately
insured patients receive therapy at rates more
similar to Medicare patients. When home health
therapy services are received, however, there does
not appear to be significant variation in the number
of visits per patient by primary payer.
Implications for Policy, Delivery or Practice:
Changes in Medicare and Medicaid payment policy
may help reduce variation in receipt of home health
therapy services with respect to primary payer;
however, whether therapy services are over-utilized
by Medicare or under-utilized by Medicaid remains
unclear without linking these data to patient
outcomes.
Poster Number: 58
Use of Trigger Tools to Detect Adverse Events in
Ambulatory Surgery
Presented by: Hillary Mull, M.P.P., Research
Health Scientist, Research, VA Boston Healthcare
System, 150 S. Huntington Avenue, Boston, MA
02130; Phone: (857) 364-2766; Email:
hjmull@bu.edu
Co-authors: Kamal Itani, M.D.; Haytham Kaafarani,
M.D., M.P.H.; Kathleen Hickson, R.N., M.N.; Amy
Rosen, Ph.D.
Research Objective: Numerous healthcare
systems in the U.S., including the Veterans Health
Administration (VA), use the National Surgical
Quality Improvement Program (NSQIP) to detect
surgical adverse events (AEs). The current NSQIP
sampling methodology excludes many routine
ambulatory surgeries from review. Given the low
prevalence of ambulatory surgical AEs, NSQIP may
not detect enough events to highlight areas in the
ambulatory setting to focus quality improvement
initiatives. Trigger tools, algorithms derived from
clinical logic to flag cases where AEs have most
likely occurred, could potentially complement
NSQIP by detecting a higher yield of ambulatory
surgeries with a true surgical AE.
Study Design: We developed four ambulatory
surgical AE trigger algorithms based on clinical
input and empirical testing. We compared the
number of trigger-flagged surgeries from Oct. 1,
2007 to Sept. 30, 2008 (FY08) using both inpatient
and outpatient administrative data from VA Boston
Healthcare System (HCS) to the 322 surgeries
reviewed in FY08 by VA Boston NSQIP. Using the
NSQIP chart review results as the gold standard for
AE detection, we refined the triggers to improve
sensitivity and specificity. To determine whether our
triggers could reliably detect surgical AEs, we ran
the triggers on the 783 ambulatory surgeries VA
Boston NSQIP excluded from review. A trained
nurse abstractor reviewed the trigger-flagged cases
using the VA electronic medical record and NSQIP
chart abstraction guidelines. We calculated the ratio
of true events over flagged events (i.e., the positive
predictive value (PPV)), and the 95% confidence
interval (CI) for each trigger.
Principal Findings: Of the 29% of ambulatory
surgeries reviewed by VA Boston NSQIP in FY08,
nine had a surgical AE. Five of the nine AEs were
detected by at least one of the ambulatory surgical
AE triggers. A total of 198 (25.3%) of the 783
surgeries excluded by VA Boston NSQIP in FY08
were flagged by one or more of the four triggers
and underwent nurse review. Final trigger
algorithms and results were as follows: (1) flagged
surgeries with at least one emergency department
or urgent care visit within 30 days: PPV = 19% (CI:
11-29%); (2) Flagged surgeries where the patient
was admitted to the hospital with a length of stay
greater than 24 hours: PPV = 33% (CI: 12-62%); (3)
Flagged surgeries where the patient had a hospital
admission within 30 days: PPV = 33% (CI: 19-51%);
and (4) Flagged surgeries where the patient had at
least one primary care visit and one specialty care
visit within 30 days: PPV = 6.2% (CI: 2-12%).
Conclusions: Compared to 2.8% of surgeries
found by NSQIP, 14.1% of the charts flagged by
one of the four ambulatory surgical AE triggers
were found to have a surgical AE. Because our
results are based on a relatively small sample,
further research is necessary to confirm these
findings.
Implications for Policy, Delivery or Practice: Our
results suggest that trigger tools have the potential
to complement NSQIP’s current sampling design by
increasing the yield of ambulatory surgeries with
surgical AEs. Improved detection of AEs in the
ambulatory surgery setting is critical to ensure
patient safety and quality of care.
Funding Source(s): VA
Poster Number: 59
Racial Differences in Patients' Preferences about
Care-Seeking
Presented by: Saira Nawaz, M.P.H., Student
Researcher, Health Services Research, The
Dartmouth Institute for Health Policy and Clinical
Research, 30 Lafayette Street, 1st Floor, Lebanon,
NH 03766; Phone: (603)-653-3268; Email:
saira.nawaz@dartmouth.edu
Lead Author: Denise Anthony, Ph.D.
Co-author: Laura Yasaitis
Research Objective: Disparities in health service
use between Black and White patients have been
well documented. Explanations for these disparities
include lack of health insurance and access to care,
variations in income and education, and poor
communication and mistrust in the patient-physician
interaction. Preferences in care may also drive
these differences. The study’s objective was to
examine the association between race and patients’
stated preferences for seeking outpatient and
specialty care.
Study Design: Analyzed survey items that elicited
stated preferences for care in two clinical scenarios:
chest pain upon walking up the stairs and a
lingering cough after an episode of the flu.
Outcomes included whether respondents preferred
to see a doctor right away, whether they would want
cardiac tests or chest X-ray and if they would want
to see a specialist. Odds ratios of preference for
care-seeking in black vs. white respondents were
adjusted for age, sex, median household income,
self-reported health, education, self-efficacy and
average number of physician office visits in the
respondent’s Hospital Referral Region (HRR).
Population Studied: Random sample of 4000
Medicare Beneficiaries and an over-sample of 700
Blacks among community-dwelling beneficiaries
who were age 65 or older as of July 1, 2003, were
entitled to Part A and/or Part B, and who were
residents of a US Hospital Referral Region in 2003
and 2004.
Principal Findings: Blacks (n=489) were more
likely to be younger, female, report that finances
were ‘important’ in healthcare decisions, live in
areas with lower median household income, have
less education, rate their health as ‘fair’ or ‘poor’
and live in HRRs with a higher number of yearly
physician office visits per beneficiary, compared to
Whites (n= 2110). If experiencing a lingering cough,
Blacks were more likely to want to see a doctor at
two days rather than wait (45.9% vs 26.6%,
adjusted OR=2.03), to request an x-ray even if the
doctor told them it was unnecessary (72.3% vs
54.9%, adjusted OR=1.98) and to want to see a
specialist even if the doctor told them it was
unnecessary (54.1% and 38.1%, adjusted
OR=2.36). Blacks were more likely to want to see
the doctor right away if they experienced chest pain
(59.9% vs 43.3%, adjusted OR=1.94) and to want
to see a specialist even if they were told it was
unnecessary (70% vs 52.8%, adjusted OR=1.90).
Conclusions: Given evidence that Blacks receiver
fewer health services compared to Whites, and may
avoid or resist health services, we might have
expected that Blacks would be less likely to say that
they would seek doctor visits or tests when
experiencing new symptoms. Yet our results show
that Black Medicare beneficiaries are significantly
more likely than White beneficiaries to say they
would seek visits and tests, even after adjusting for
socio-demographic factors.
Implications for Policy, Delivery or Practice:
Physicians should undergo training in shared
decision-making and cultural competency, which
have been shown to effectively elicit patient
preferences for care. Using Medicare Claims data,
our future research will assess whether Black or
White patients are more able to assert their stated
preferences.
Poster Number: 60
Effect of Increased Copayments on Use of
Outpatient Mental Health Services among Medicare
Managed Care Enrollees
Presented by: Chima Ndumele, M.P.H., Graduate
Student, Community Health, Brown University, 121
South Main Street, Providence, RI 02912; Phone:
(617) 947-0205; Email:
chima_ndumele@brown.edu
Co-authors: Amal Trivedi, M.D., M.P.H.
Research Objective: The impact of increased costsharing on use of mental health services is not well
understood, particularly among the elderly.
Furthermore, whether vulnerable populations
exhibit selective sensitivity to changes in cost
sharing for mental health treatment is also unclear.
We examined the effects of increased mental health
copayments on use of outpatient mental health
services among a nationwide sample of elderly
individuals in managed care plans, as well as within
racial subgroups.
Study Design: We used a difference-in-differences
(DID) design to assess changes in the proportion of
enrollees who received outpatient mental health
services in Medicare plans that increased mental
health copayments compared to the concurrent
trend in matched control Medicare plans that did not
increase copayments. We selected all Medicare
plans that increased copayments by at least 25%
after two consecutive years of stable copayments.
We matched concurrent control plans on the basis
of US census region, tax status, and model type.
The study period was from 2001 to 2007. Analyses
were restricted to continuously-enrolled
beneficiaries. We constructed generalized linear
models that adjusted for age, sex, race, area-level
socioeconomic status, clustering within health plans
using fixed-effects and repeated measurement of
enrollees using GEE. We conducted stratified
analyses for white and black enrollees.
Population Studied: 601,378 Medicare managed
care beneficiaries age 65 and older enrolled in 14
Medicare plans that increased mental health
copayments and 404,633 beneficiaries enrolled in
14 matched control plans.
Principal Findings: Among the 14 case plans,
mean mental health copayments increased from
$14.43 to $21.07. In control plans, mean mental
health copayments were unchanged at $21.43.
Among case plans, the proportion of enrollees that
received outpatient mental services declined from
2.4% to 2.3% (P=0.16). Among control plans, the
concurrent change was from 1.4% to 1.3%
(p=0.07). The adjusted DID was 0.006 percentage
points (95%CI -0.002-0.14; p=0.11) Among the 14
case plans, only 4 experienced declines in the
proportion of enrollees receiving outpatient mental
health services relative to the concurrent trend in
the matched control plan. In stratified analyses, the
adjusted DID for black enrollees was 0.001
percentage points (95%CI -0.003-0.005; p=0.63).
Conclusions: Only a small percentage of Medicare
Managed Care enrollees use outpatient mental
health services. Among this group, cost sharing
increases of approximately 50% had a limited effect
upon the proportion of patients using outpatient
mental health services. This relationship holds true
for both black and white enrollees.
Implications for Policy, Delivery or Practice: The
low prevalence of mental health service use among
the elderly may be driven by factors other than costsharing.
Poster Number: 61
Rural/Urban Comparison of Nebraska’s Health
Care Workforce: A County-Level Analysis
Presented by: Anh Nguyen, M.S.P.H., Health
Data Analyst, Health Services Research and
Administration, University of Nebraska Medical
Center, 984350 Nebraska Medical Center, Omaha,
NE 68198-4350; Phone: (402) 559-3295; Email:
anguyen@unmc.edu
Co-authors: Preethy Nayar, M.D., Ph.D.; Keith
Mueller, Ph.D.
Research Objective: This study uses Nebraska
county-level data to examine the rural versus urban
distribution of eight classes of health care
professionals and the correlates of county-level
health care provider supply.
Study Design: Data for eight classes of health care
professionals – primary care providers, primary
care physicians, physicians, dentists, general
dentists, pharmacists, mental health providers, and
psychiatrists – were obtained from the 2007/2008
University of Nebraska Medical Center’s Health
Profession Tracking Service survey data. Primary
care providers include physicians, nurse
practitioners, and physician assistants specialized
in primary care specialties. Using the 2003 RuralUrban Continuum Codes, the rural versus urban
distribution of professionals were examined with
descriptive statistics. Correlations were examined
between county-level provider-to-population ratios
and county characteristics.
Population Studied: The study population includes
health professionals practicing in all Nebraska
counties in 2007-2008.
Principal Findings: In non-metropolitan counties,
the average provider-to-population ratios for various
actively practicing providers to 100,000 population
are: primary care providers (89.3), primary care
physicians (119.1), physicians (86.2), dentists
(35.5), general dentists (34.6), pharmacists (62.0),
mental health providers (34.7), and psychiatrists
(1.2). The average provider-to-population ratios for
physicians, pharmacists, and psychiatrists are
below national average or federal health
professional shortage area designation benchmark
ratios in non-metropolitan counties. In metropolitan
counties, the average provider-to-population ratios
for various actively practicing providers to 100,000
population are: primary care providers (68.3),
primary care physicians (80), physicians (100.2),
dentists (40.4), general dentists (36.8), pharmacists
(61.6), mental health providers (51.9), and
psychiatrists (4.5). The average provider-topopulation ratios for physicians and pharmacists
are below national average benchmark ratios in
metropolitan areas. Health professionals tend to be
actively practicing in more urban counties, highpopulation counties, high-income counties, counties
with lower proportions of an aging population,
counties with higher hospital admissions-topopulation ratios, and counties with higher infant
mortality rates.
Conclusions: Overall, rural counties have lower
provider-to-population ratios for physicians,
dentists, general dentists, mental health providers,
and psychiatrists.
Implications for Policy, Delivery or Practice:
Rural areas in Nebraska face significant challenges
in recruiting and retaining health care professionals.
The findings from this study have implications for
quantifying the need and demand for health care
professionals in workforce planning and policy
analysis.
Poster Number: 62
Determinant of Access to the Systems of Care for
Mentally Ill Juvenile Offenders in Texas
Presented by: Gul Nowshad, M.B.B.S., M.P.H.,
Graduate Research Assistant, Management, Policy
and Community Health, School of Public Health, UT
Houston, 1200 Hermann Pressler, Houston, TX
77030; Phone: (713) 500-9489; Email:
gul.nowshad@uth.tmc.edu
Co-authors: Sartaj Alam, M.S; Caesa Nagpal,
M.B.B.S.
Research Objective: This study examined the
socio-demographic, clinical characteristics of
juvenile offenders coming in contact first time either
with mental health system or with juvenile justice
system, and defined the determinant of their first
time access to the system of care in Houston
Texas.
Study Design: We did a secondary analysis of the
data; consist of records of Harris County Juvenile
Probation Department, of all children and
adolescents referred to juvenile probation from
1990 until 2005 and Mental Health records between
1992 and 2006, matched through probabilistic
matching method. About 25% of individuals present
in the juvenile justice database were also present in
the mental health database. The sample size of our
study was 23,866.
Population Studied: Eligible study population,
were mentally ill offenders who were 17 years and
younger.
Principal Findings: In our sample population 30%
of the juvenile offenders, first time come in contact
with this pool of care (juvenile probation department
and mental health system) is through the door of
mental health system while 70% of them first got
through the juvenile justice system. Females were
25% of the sample population, and more then third
(37%) were White. Female adolescents were 22%
depressed compared to males (9%), while female
adolescents were 2% suffering from adhd and
males were 8%. Age, gender, ethnicity, depression,
mood disorders, ADHD, impulse disorders,
behavioral disorders and substance abuse
disorders were found statistically significant with
mentally ill juvenile offenders entering in the pool of
care through the portal of Juvenile Probation
Department. Juvenile offenders who were one year
older were 71% and those who used substance
abuse were 80% more likely to enter through the
justice portal. Male adolescent were two times more
likely to enter through the justice portal. Compared
to White, African American were 28% and
Hispanics were 22% more likely to enter through
the Justice door. Similarly kids who were suffering
from depression, impulse disorder, mood disorders,
ADHD were more likely to get treatment early from
mental health system before getting involved with
the justice system.
Conclusions: Ethnicity, substance abuse, mental
illness like depression, impulse disorder and ADHD
will predispose our youth to encounter the justice
system as a symptom of their disease instead of
getting treatment for their illness.
Implications for Policy, Delivery or Practice:
These finding will enable researcher, judicial
personnel and policy makers to understand the
determinants and disparity in access to mental
health care system.
Funding Source(s): Houston Endowment inc.
Poster Number: 63
Association Between Survey-Based NurseReported Healthcare-Acquired Infection Frequency
and Hospital-Reported Rates
Presented by: Danielle Olds, R.N., M.S.N.,
M.P.H., Pre-doctoral Research Fellow, Center for
Health Outcomes and Policy Research, University
of Pennsylvania, Claire Fagin Hall, #388, 418 Curie
Boulevard, Philadelphia, PA 22203; Phone: (215)
898-5673; Email: danielle.m.olds@gmail.com
Co-authors: Linda Aiken, R.N., Ph.D.; Dr. Jeannie
Cimiotti, R.N., D.N.Sc.
Research Objective: The collection of adverse
outcomes data by caregivers is increasingly
burdensome. The objective of this research is to
determine whether surveys of nurses about
infection frequency provide similar estimates to
more burdensome hospital reporting requirements.
Study Design: This study is a cross-sectional,
correlational secondary analysis of nurse survey
data and publicly available data on hospital
infection rates. Hospital-reported healthcareacquired infection (HAI) rates were obtained from
the Pennsylvania Health Care Cost Containment
Council’s (PHC4) publicly available report,
"Hospital-acquired Infections in Pennsylvania,
Calendar Year 2006". In 2006, Pennsylvania
hospitals were mandated to report to the PHC4
rates of HAIs including: surgical site,
gastrointestinal, multiple, and “other” infections, and
device and non-device related pneumonia,
bloodstream infection, and urinary tract infection.
Thirteen hospitals used electronic surveillance; the
other hospitals used traditional clinical surveillance
methods. Nurse survey data were obtained in 2006
from a random sample of registered nurses
licensed and residing in Pennsylvania. Nurses
reported the names of their primary hospital
employer so that survey results could be linked to
hospitals. The survey queried nurses on the
incidence of healthcare-acquired urinary tract
infections (UTI), central-line associated
bloodstream infections (CLABSI), and ventilatorassociated pneumonia (VAP). Responses were
scored on a 7-point Likert-type scale ranging from
no infection to every day occurrence. The hospitallevel mean nurse survey response was calculated
for each hospital. The appropriateness of
aggregating nurse responses at the hospital-level
was determined using the intraclass correlation for
each infection type. Spearman’s rho correlation was
used to compare nurse-reported frequency of HAIs
with hospital-reported rates.
Population Studied: The study sample included
nurses working on inpatient units of 156 acute care
hospitals in Pennsylvania that submitted data to the
PHC4. Analyses for VAP were restricted to nurses
working on intensive/critical care units (146
hospitals).
Principal Findings: Intraclass correlations were
calculated for each infection. The lowest was UTI
(ICC(1,k) = 0.45); and the highest, CLABSI
(ICC(1,k) = 0.71). VAP had an ICC (1,k) of 0.46. In
order to obtain an adequate intraclass correlation
(ICC(1,k) = .60) and aggregate nurse responses at
the hospital-level, correlational analyses were
limited to hospitals with at least 30 respondents for
UTI (52 hospitals) and at least 7 respondents for
VAP (126 hospitals). Nurse estimates of HAI
frequency were highly correlated with hospitalreported rates for UTI (n = 52, r = 0.40, p = 0.0034);
CLABSI (n = 156, r = 0.53, p < 0.0001); and VAP (n
= 126, r = 0.34, p = 0.0001).
Conclusions: Nurse estimates of HAI frequency
are highly correlated with hospital-reported infection
rates for UTI, CLABSI, and VAP. UTI and VAP had
wider intra-hospital respondent variation, compared
to CLABSI.
Implications for Policy, Delivery or Practice:
Survey-based nurse reports of HAI frequency may
be valid estimates when patient data are
unavailable, or in combination with shorter periods
of intensive reporting that are burdensome for
clinicians. These findings also suggest that nurse
reports on patient care may be a valuable
component of quality improvement.
Funding Source(s): National Institute for Nursing
Research
Poster Number: 64
Correlates of Perceived Risk of Acquiring HIV in a
High HIV/AIDS Prevalence Population: Does Age at
First Sexual Debut Matter?
Presented by: Elijah Onsomu, M.S., M.P.H.,
Teaching/Research Assistant, Department of Public
Health Sciences, University of North Carolina at
Charlotte, 9201 University City Boulevard,
Charlotte, NC 28223; Phone: (704) 649-6240;
Email: eoonsomu@uncc.edu
Co-authors: DaKysha Moore, M.S., Ph.D.; James
Kimani, M.A., Ph.D.; Ahmed Arif, M.D., Ph.D.,
F.A.C.E.; James Studnicki, Sc.D., M.B.A., M.P.H.
Research Objective: In 2007, there were 33.2
million people living with HIV worldwide, with 85%
of Sub-Saharan Africa (SSA) countries
experiencing pandemic situations. Of those living
with HIV, 22.5 million (>68%) reside in SSA despite
only representing 12% of the world population with
women and children experiencing the most impact
of the disease (UNAIDS, 2007). Two main
objectives for this study are: 1) To determine the
association between perceived risk of acquiring HIV
and age at first sexual debut among men and
women, and 2) To identify correlates associated
with perceived risk of acquiring HIV among men
and women after first sexual debut.
Study Design: Retrospective cross-sectional data
from Kenya Demographic and Health Survey-2003
were used. Multivariate multinomial logistic
regression analysis was performed. Data were
weighted for stratum and primary sampling units
attaining linearized standard errors. The main
outcome variable is perceived risk of acquiring HIV,
while the main independent variable-age at first
sexual debut-both categorical variables. Parametric
testing using Univariate and Multivariate
multinomial logistic regression was performed.
Stata version 10.1, utilized for all analyses with
statistical significance, was set at p<0.05.
Population Studied: Two datasets were used: 1)
the female dataset (8,195) 94% response rate; and
2) the male dataset (3,578) 86% response rate. The
sample included women aged 15-49 years
(reproductive age) and men aged 15-54 years.
Principal Findings: Fifty four percent of males
aged 15-19 perceived themselves to be at high-risk
of acquiring HIV compared to females 47%,
p<0.001. Before adjusting for confounding, women
who had their first sexual debut at 15-19 years were
9.77 times more likely to perceive themselves to be
at high-risk of acquiring HIV compared to those who
have not had sex at all (95% CI 6.53-14.62,
p<0.001). Men who had their first sexual debut at
15-19 years were 4.34 times more likely to perceive
themselves to be at high-risk of acquiring HIV
compared to those who have not had sex at all
(95% CI 2.17-8.690, p<0.001). After adjusting for
confounding, women who had their first sexual
debut at 15-19 years were 4.14 times more likely to
perceive themselves to be at high-risk of acquiring
HIV compared to those who have not had sex at all
(95% CI 2.47-6.94, p<0.001). Men who had their
first sexual debut at 15-19 years were 3.88 times
more likely to perceive themselves to be at high-risk
of acquiring HIV compared to those who have not
had sex at all (95% CI 1.76-8.55, p<0.001).
Conclusions: Women and men who had first
sexual debut at an earlier age perceived
themselves to have a high-risk of acquiring HIV,
which increased with decreasing age.
Implications for Policy, Delivery or Practice:
HIV/AIDS prevention programs need to pay special
attention on age at first sexual debut, particularly
children and youth when trying to influence and
control HIV infections in Kenya. If first sexual debut
is not paid enough attention to HIV infections will
continue to increase among this age group.
Poster Number: 65
Sexually Transmitted Diseases and Circumcision in
Kenya: What Role Do They Play in HIV
Transmission?
Presented by: Elijah Onsomu, M.S., M.P.H.,
Teaching/Research Assistant, Department of Public
Health Sciences, University of North Carolina at
Charlotte, 9201 University City Boulevard,
Charlotte, NC 28223; Phone: (704) 649-6240;
Email: eoonsomu@uncc.edu
Co-authors: Ahmed Arif, M.D., Ph.D., F.A.C.E.;
James Studnicki, Sc.D., M.B.A., M.P.H.; Judith
Cornelius, M.S., Ph.D.; Ayana Moore, Ph.D.,
R.A.C.; James Laditka, D.A., Ph.D.
Research Objective: CDC has found 2-5 folds of
HIV infection if one has STD(s) compared to those
without an STD(s). Circumcision as a protective
factor for HIV has been documented elsewhere.
Two objectives were explored in this study: 1) To
determine predictors for HIV, and 2) To determine if
there is any relationship between STDs,
circumcision and HIV acquisition.
Study Design: Retrospective cross-sectional data
from the Kenya Demographic and Health Survey2003 was used. Data were weighted for stratum
and primary sampling units attaining linearized
standard errors. The main outcome variable was
HIV test results, a dichotomous variable (negative
“0” or positive “1”). The main independent variables
were STD(s), a dichotomous variable (does not
have STD(s) “0” or has STD(s) “1”), and
circumcision, a dichotomous variable (circumcised
“0” or uncircumcised “1”). Parametric testing using
Univariate and Multivariate logistic regression was
performed. Stata version 10.1, utilized for all
analyses with statistical significance, was set at
p<0.05.
Population Studied: Three datasets (female, male
and HIV datasets) were merged with a sample size
of 6,190 retained for the analysis. These comprised
those individuals who were offered an HIV test,
accepted and took the test. The sample included
women aged 15-49 years (reproductive age) and
men aged 15-54 years.
Principal Findings: 9% (p<0.001) of men who had
STD(s) tested positive for HIV compared to 8% of
women, p<0.002. Also, 48% of men who tested
positive for HIV were not circumcised, p<0.001.
STD(s): After adjusting for confounding, women
who had an STD(s) were 3.05 times more likely to
test positive for HIV than those who did not have a
STD(s) (95% CI 1.34-6.96, p<0.008). Men who had
an STD(s) were 2.01 times more likely to test
positive for HIV than those who did not have a
STD(s) (95% CI 0.92-4.40, p<0.08). Circumcision:
After adjusting for confounding, uncircumcised men
were 4.23 times more likely to test positive for HIV
than those who were circumcised (95% CI 2.517.13, p<0.001).
Conclusions: STD(s) among women and men,
uncircumcised men in Kenya increased ones
likelihood of being diagnosed positive for HIV.
Implications for Policy, Delivery or Practice:
There is need to continuously monitor and test for
STD(s) as a routine among those seeking
healthcare. Individuals should also be encouraged
to seek medical services if they suspect STD(s)
symptoms. Cultural barriers need to be addressed
in order to encourage circumcision in communities
that do not practice circumcision among men.
Circumcision should be related to important
community cultural health practices. However,
participants need to be made aware that being
circumcised does not prevent them from being
infected with HIV; hence, condom use should be
encouraged.
Poster Number: 66
Effect of Practitioner on Chronic Opioid Use in
Fibromyalgia Syndrome Patients
Presented by: Jacob Painter, Pharm.D., M.B.A.,
Fellow, Pharmaceutical Outcomes and Policy,
University of Kentucky College of Pharmacy, 789 S.
Limestone, Lexington, KY 40536-0596; Phone:
(859) 260-1960; Email: jtpainter@gmail.com
Co-authors: Jeff Talbert, Ph.D.
Research Objective: Fibromyalgia syndrome
(FMS) is an idiopathic, functional disorder
characterized by widespread pain. Fibromyalgia
syndrome affects approximately 2-4% of the
general population of the United States. Patients
suffering from fibromyalgia have been shown to
have average total healthcare costs about three
times higher than sex and age matched controls.
According to a 2004 Journal of the American
Medical Association clinical review titled
Management of Fibromyalgia Syndrome,
pharmacologic therapies of FMS can be divided
according to the level of evidence for efficacy that
exists. This review found that opioids did not have
evidence of efficacy for use in the treatment of
FMS. Moreover, evidence since that review
indicates the possibility of opioid-induced
hyperalgesia in patients being treated chronically
with opioids. Our previous research shows that
within our population there is an elevated level of
chronic opioid use in FMS patients. The goal of our
current research is to investigate what practitioner
factors influence the use of opioids chronically
within this population.
Study Design: Our overall inclusion criteria are:
age 18-64, continuous eligibility for at least 18
consecutive months, two or more healthcare
encounters with a diagnosis of fibromyalgia (ICD-9
code 729.1) within a fiscal year. Patients with
malignancy were excluded from this analysis, due
to differences in treatment patterns between these
patient populations. With these criteria we
investigate the characteristics of practitioners that
lead to the chronic use of opioids in this population.
We describe chronic opioid use in this population as
having a day supply of opioids greater than 180
days in a single fiscal year. Attributes we are
interested in include physician or other, primary
care or specialist, type of specialty, practice type,
practitioner location type and geographic location.
Population Studied: In order to investigate the
prescribing behaviors of practitioners treating FMS
we examined patient records for drug utilization
patterns and cost information. Data is an extract
from a health claims data warehouse consisting of
eligibility, pharmacy, institutional, and medical
services. The database represents the entire
medical and prescription claim history for Kentucky
Medicaid recipients from January 1, 2000 to June
30, 2009. This database contains medical
information such as diagnoses as indicated by
International Classification of Diseases (ICD) codes
and procedures as indicated by Current Procedural
Terminology (CPT) codes. It also contains
information on prescription medication utilization
such as drug name, strength, day supply, and
prescribing physician. Economic cost data includes
Medicaid reimbursement rates for all procedures,
services, and pharmaceuticals incurred by
recipients.
Principal Findings: Although research is ongoing,
preliminary results show that division of our sample
according to whether they have received chronic
opioid therapy results in differences in distribution of
diagnosing practitioner according to primary care
versus specialist and according to specialty type.
Conclusions: Our preliminary findings suggest that
this research will illuminate differences among
practitioner characteristics that can contribute to
explaining what leads to chronic opioid use within
this disease population.
Implications for Policy, Delivery or Practice:
Knowledge concerning practitioner characteristics
that contribute to chronic opioid use in fibromyalgia
patients will be valuable both to policy makers and
to clinical practice, providing insight into evidencebased practice adoption trends.
Poster Number: 67
Stakeholders’ Views on Medicare’s Policy of
Nonpayment for Hospital-Acquired Urinary Tract
Infections
Presented by: Jennifer Palmer, M.S., Research
Assistant, Center for Health Quality, Outcomes, and
Economic Research, ENR Memorial VA Hospital,
200 Springs Road, Bedford, MA 01730; Phone:
(781) 687-2855; Email: jennifer.palmer1@va.gov
Co-authors: Grace Lee, M.D., M.P.H.; Timothy
Hoff, Ph.D.; Maya Dutta-Linn, M.P.H.; Peter Wroe,
B.A.; Christine Hartmann, Ph.D.
Research Objective: The Centers for Medicare
and Medicaid Services (CMS) instituted a new
policy in 2008 that limits reimbursement for
hospitalizations associated with preventable
complications, including certain hospital-acquired
infections (HAIs). Some of the goals of the policy
include increasing awareness of HAIs and
promoting HAI prevention. We explored some
implications of this policy by exploring stakeholders’
impressions of one of the selected HAIs: catheterassociated urinary tract infections (CAUTIs).
Study Design: Because our study was exploratory,
we used qualitative methods to investigate
participants’ impressions. We purposively sampled
hospitals on a national level, stratifying by bed size,
geography, and nurse staffing levels. We sought to
attain wide representation of variation in each
stratification category. We interviewed infection
preventionists (IPs) and intensive care unit (ICU)
charge nurses using a semi-structured interview
guide. The guide included questions about
participants’ perceptions of the policy, about how it
is being implemented at their site, and about any
related consequences they have observed. For
data analysis, we employed components of content
analysis with adapted tools from grounded theory.
Population Studied: Between September 2009
and January 2010, we interviewed 35 IPs and 5
ICU charge nurses from 35 hospitals. We chose
these representatives of the field due to their solid
understanding of infection control activities in their
facilities and their close involvement in institutional
response to the new CMS policy.
Principal Findings: Participants discussed CAUTIs
in relation to the CMS nonpayment policy both in a
positive and in a negative light. Positive comments
focused on how the policy increases awareness of
and adherence to infection prevention practices
around CAUTIs. For example, in some facilities the
policy prompted changes in the duration of catheter
use or in the criteria for catheter insertion, thereby
minimizing unnecessary use. Negative comments
focused on the limited utility of: 1) prioritizing efforts
to prevent CAUTIs, which were perceived to be a
minor complication; and 2) diverting administrative
and financial resources away from prevention of
other infections with higher rates of morbidity and
mortality. Furthermore, some participants felt that
nonpayment for CAUTIs did not have a large
financial impact on their hospitals’ bottom lines.
Finally, in larger hospitals, complete prevention of
CAUTIs was thought to be difficult in a high-risk
patient population.
Conclusions: We conducted an in-depth analysis
of stakeholders’ views on the selection of CAUTIs
as an HAI in CMS’ new nonpayment policy. IPs and
ICU charge nurses did not uniformly consider the
inclusion of this particular hospital-acquired
condition to be either completely founded or
unfounded. Further research should examine
insiders’ perspectives on other targeted HAIs in the
policy and assess the overall impact of this policy
on patient outcomes to be sure there are no
unintended negative consequences.
Implications for Policy, Delivery or Practice:
Stakeholders’ varying support for targeting CAUTIs
through Medicare nonpayment has larger
implications. CMS might benefit from continued
careful review of the targeted HAIs since the policy
may divert already limited hospital resources away
from alternatively important infections and patient
populations.
Funding Source(s): National Institute of Allergy
and Infectious Diseases
Poster Number: 68
Racial/Ethnic, Gender, and Socioeconomic
Disparities in Diabetes-Related Hospitalizations
among Medicare Beneficiaries
Presented by: Felipe Pereira, M.H.S., student,
Department of Health Policy and Management, The
University of Kansas Medical Center, 3901
Rainbow Boulevard, Kansas City, KS 66160;
Email: fpereira@kumc.edu
Co-authors: Michael Fox, Sc.D.; David Kingsley,
Ph.D.; Megha Ramaswamy, Ph.D., M.P.H.
Research Objective: Through the implementation
of Medicare Part D in 2006, individuals with
diabetes gained broader access to coverage for
previously limited items, such as insulin and
medical supplies associated with injection. This
increased access to coverage should result in a
decrease in diabetes-related hospitalizations, with
the concurrent effect of reducing disparities. Little
has been done previously to demonstrate
disparities in the risk of diabetes-related
hospitalization among the Medicare population over
time. The goal of this study is to examine trends in
the risk of diabetes-related hospitalizations among
Medicare beneficiaries and to determine whether
the implementation of Part D was associated with
any reduction in disparities.
Study Design: We analyzed data from the 20032007 Agency for Healthcare Research and Quality's
Healthcare Cost and Utilization Project Nationwide
Inpatient Sample. We conducted binary logistic
regression to examine risk of being hospitalized due
to diabetes mellitus with complications.
Hospitalizations with diabetes listed as the primary
diagnosis were identified in HCUP through the use
of the Clinical Classifications Software, or CCS,
which collapses ICD-9-CM codes into smaller
number of clinically meaningful categories.9 The
CCS code utilized was 50, diabetes mellitus with
complications, which contains ICD-9-CM codes
from 249.xx and 250.xx. The factors for which we
controlled in the analysis were race, income quartile
where patients reside, and gender.
Population Studied: We examined all
hospitalizations from the 2003-2007 HCUP NIS for
which Medicare was listed as the primary payer.
Principal Findings: Racial and ethnic disparities in
the risk of diabetes-related hospitalizations remain
largely unchanged between 2003 and 2007,
although minority groups, particularly Blacks and
Hispanics, remain at significantly higher risk of
hospitalization compared to Whites. However,
gender and income disparities in risk have
increased between 2003 and 2007, with males
showing significantly higher risk than females, while
the risk of hospitalization for those in the lower
three income quartiles significantly increased
relative to the highest income quartile between
2003 and 2007.
Conclusions: The data demonstrates that
disparities in diabetes-related hospitalizations
persist among racial/ethnic minorities and are
growing among men and the poor. The evidence
suggests that greater availability of insulin coverage
through the implementation of Part D has not
reduced disparities in the risk of hospitalization, and
in fact may be exacerbating certain disparities.
Implications for Policy, Delivery or Practice:
Medicare needs to place greater focus on
measuring and tracking disparities in high cost,
preventable hospitalizations such as those related
to diabetes. Additionally, confronting the sources of
these disparities is a challenge that Medicare, as a
payer of health services, must address through
targeted interventions.
Poster Number: 69
Attitudes Toward Drug-Eluting Stent Use and
Distribution of Regulatory Focus Type among
Interventional Cardiologists in New York State
Presented by: Feng Qian, M.D., M.Sc., Ph.D.
candidate in Health Services Research and Policy,
Community and Preventive Medicine, University of
Rochester, 601 Elmwood Avenue, Box 644,
Rochester, NY 14642; Phone: (585) 275-9496;
Email: feng_qian@urmc.rochester.edu
Co-authors: Charles Phelps, Ph.D., M.B.A.;
Frederick Ling, M.D.; Edward Hannan, Ph.D.; Peter
Veazie, Ph.D.
Research Objective: The safety of drug-eluting
stent (DES) use was called into question in 2006.
Literature in psychology suggests that regulatory
focus type (promotion vs. prevention) orients
physicians toward using different strategies to
approach a task: ensuring against errors of
omission (promotion) vs. ensuring against errors of
commission (prevention). However, the attitudes
toward DES use after safety concerns were
expressed and the distribution of regulatory focus
type among interventional cardiologists are
unknown. The objectives of this study are: (1) to
examine the attitudes toward DES use among
interventional cardiologists in New York State
(NYS) after DES safety concerns were expressed,
and (2) to investigate the distribution of regulatory
focus type among these interventional cardiologists.
Study Design: We conducted a questionnaire
survey of interventional cardiologists in New York
State from October 2008 to April 2009 by mail,
email and fax. The questionnaire included face valid
items to measure the attitudes toward DES use,
valid Regulatory Focus Questionnaire to measure
the regulatory focus type, and items collecting
demographic information.
Population Studied: Interventional cardiologists in
New York State.
Principal Findings: We received total 119 valid
returned questionnaires and the response rate was
47%. There are no statistically significant
differences regarding the demographic factors
between the respondents and the non-respondents.
The vast majority of interventional cardiologists
(92%) agreed or strongly agreed that DES is a
revolutionary technology and that DES use will
increase in the future (70%). The regulatory focus
type of interventional cardiologists was
predominantly of the promotion type (89%). We
found that age and gender were significant
predictors of the promotion score, but we did not
find any demographic factors were significant
predictors of the prevention score.
Conclusions: Despite the DES safety issues
reported in 2006-2007, interventional cardiologists
have a very positive attitude regarding DES use
and most of them predict the continual growth of
DES use in the future. Such information is valuable
in predicting the modality change in coronary
revascularization therapy. Furthermore, the vast
majority of interventional cardiologists were found to
be promotion focused, which means that they are
concerned about achieving positive outcomes and
they want to prevent errors of omission.
Implications for Policy, Delivery or Practice: To
the best of our knowledge, this is the first study to
report the distribution of regulatory focus type
among physicians. As suggested by the theory in
psychology, this finding can lead to more effective
interventions in conveying medical messages to
physicians and changing their behavior.
Poster Number: 70
Primary Coronary Revascularization Utilization
Pattern Change and Practice Variation in DrugEluting Stent Era in New York State
Presented by: Feng Qian, M.D., M.Sc., Ph.D.
candidate in Health Services Research and Policy,
Community and Preventive Medicine, University of
Rochester, 601 Elmwood Avenue, Box 644,
Rochester, NY 14620; Phone: (585) 275-9496;
Email: feng_qian@urmc.rochester.edu
Co-authors: Charles Phelps, Ph.D., M.B.A.;
Frederick Ling, M.D.; Edward Hannan, Ph.D.; Peter
Veazie, Ph.D.
Research Objective: Drug-Eluting Stent (DES), the
newest medical device in percutaneous coronary
intervention (PCI), was perceived as a revolutionary
medical technology due to its substantial reduction
of the restenosis rate when it was approved by FDA
in April 2003. However, the safety of DES use was
called into question in 2006 due to increased late
stent thrombosis. Further, literature suggests that
practice variations in DES use occurred during the
initial adoption period in 2003. The objectives of the
study were: (1) to describe the relationship between
DES adoption and coronary revascularization
utilization pattern change in New York State, and
(2) to test the hypotheses: (i) DES played a strong
substitution role for Bare Metal Stent (BMS) in 2003
and 2004, (ii) BMS played a strong substitution role
for DES in 2006, and (iii) there were medical
practice variation in DES use across the study
years 2003, 2004 and 2006.
Study Design: We used New York State’s
Statewide Planning and Research Cooperative
System (SPARCS) data, a comprehensive and
accurate data base administered by New York
State Department of Health, to describe the
coronary revascularization pattern change in the
DES era and assess the above hypotheses. Since
we were not sure how Coronary Artery Bypass
Graft (CABG) utilization was affected since DES
adoption, we examined the relationship between
the CABG and the PCI procedures.
Population Studied: Patients who was
hospitalized and received primary coronary
revascularization during the study period in New
York State.
Principal Findings: Total 102,078 patients who
received the first isolated coronary revascularization
procedures were included in the study. Our findings
confirmed above hypotheses for the Pearson
correlation coefficient r equaled -0.93 (p=0.0003), 0.79 (p=0.002), and -0.77 (p=0.004) for DES and
BMS in 2003, 2004, and 2006 respectively. We
found that the coefficient of variation (COV) for DES
use in primary coronary revascularization in New
York State were 0.30, 0.24, and 0.25 for 2003,
2004, and 2006 respectively. However, CABG
volume remained stable and was positively
correlated with the PCI volume across the above
years.
Conclusions: There are significant coronary
revascularization utilization pattern changes and
consistent practice variation in DES use in New
York State.
Implications for Policy, Delivery or Practice: It
suggests corresponding changes in hospital
reimbursement and physician reimbursement
policies to achieve an optimal rate of
revascularization treatments. Future research is
needed to identify the causes of practice variation.
Poster Number: 71
Medical Tourism: A Systematic Review of the
Literature
Presented by: Zo Ramamonjiarivelo, M.B.A.,
Doctoral Student, Health Services Administration,
University of Alabama at Birmingham, 603, WEBB
Nutrition Sciences Bldg., 1675 University
Boulevard, Birmingham, AL 35294; Phone: (205)
239-3462; Email: zoramam@uab.edu
Lead Author: Anantachai Panjamapirom, M.B.A.,
M.S.
Co-authors: Gouri Gupte, Ph.D. Candidate,
M.H.A.; Zo Ramamonjiarivelo, M.B.A.; Shamly
Austin, Ph.D. Candidate, M.H.A.; David Au, B.A.
Research Objective: Medical tourism is the
concept of travelling to a foreign country to seek
medical care. It has grown at a rapid pace during
recent years. This study has systematically
reviewed the extant literature on medical tourism
published up to December, 2009. Specifically, this
study contributes to the literature by (1) Specifying
the methodology used (2) Listing number of articles
published by database (3) Examining number of
articles published based on keywords, and (4)
Determining the streams in the medical tourism
literature and their related sub-streams.
Study Design: This study systematically reviewed
articles that met the following inclusion criteria:
articles that were entirely published in English peerreviewed journals, and articles and abstracts that
were available in online databases. The following
keywords were used: medical tourism, health
tourism, wellness tourism, spa tourism, health
seeking travelers, reproductive tourism, dental
tourism, fertility tourism, medical outsourcing, and
transplant tourism. These 10 keywords were
searched in 8 databases: PubMed, ABI/Inform
Complete, JSTOR, Science Direct, Springer Link,
Wiley InterScience, CINAHL, and Web of
Knowledge. The articles’ publication dates were not
part of the inclusion criteria to allow a
comprehensive analysis of research trends in
medical tourism.
Principal Findings: Our search yielded a total of
1137 items. Of which, 738 were not journal articles,
33 were duplicates within the same databases and
127 were similar articles in different databases.
These were excluded from our analyses. Thus, the
study had a final sample of 239 articles. The articles
included in our analyses were published in 141
different journals from 1990 to 2009. 30.3% articles
were found in Web of Knowledge, 26.8% in
PubMed, 17.8% in Science Direct, 9.6% in CINAHL,
6.6% in Wiley InterScience, 4.8% in ABI/Inform
Complete, 2.5% in Springer Link, and 1.9% in
JSTOR. The keyword search found that 45.6% of
the articles could be attributed to medical tourism,
followed by transplant tourism (18.4%), health
tourism (11.7%), reproductive tourism (10.0%),
medical outsourcing (5.0%), dental and fertility
tourism (3.3% respectively), spa tourism (1.3%),
wellness tourism (0.8%) and health seeking
travelers (0.4%).
Conclusions: Publication trends in medical tourism
reflect a recent increase in interest among
researchers. A significant increase in articles
published was observed after the year 2000. 2009
saw a sharp increase in the number of articles
attributable to medical tourism, transplant tourism,
and health tourism. Although publication trends in
medical tourism indicate an increase in number of
published articles in recent years, the concept has
not acquired a clear and universal definition. Our
analyses observed an overlap between the 10
keywords used in this study. The exhaustive
literature review determined that a dominant and
formal journal classification scheme, disciplinedetermined categories, and consensus among
medical tourism contributors do not currently exist.
Implications for Policy, Delivery or Practice:
Lack of a universal definition and understanding of
medical tourism creates confusion not only among
the recipients and providers of healthcare, but also
among scholars, policy makers and third-party
payers. An explicit and mainstream definition of
medical tourism and a comprehensive analysis on
the available literature will benefit these
stakeholders.
Poster Number: 72
A Study of the Effect of Parental Charateristics and
Child Safety Interventions on Childhood Injury
Rates
Presented by: Leslie Rideout, F.N.P., Pediatric
Trauma Nurse Coordinator, Kiwanis Pediatric
Trauma Institute, Floating Hospital for
Children/Tufts Medical Center, 800 Washington
Street, Box 344, Boston, MA 02111; Phone: (617)
636-1380; Email: lrideout@tuftsmedicalcenter.org
Research Objective: Injuries are a major health
problem for young children. “An estimated one half
of all unintentional, nonfatal injuries occur in the
home environment” (Scheidt, Harel, Trumble,
Jones, Overpeck, & Bijur’s study as cited in Gielen,
et al., 2007). The purpose of this study is to assess
those factors which influence the utilization of child
safety interventions and the resultant impact that
this may have on early childhood injury reduction.
Hypothesis: Implementation of child safety
interventions result in a decrease in injury rates
when controlling for other factors among children
followed in primary care practices.
Study Design: Secondary data analysis utilizing
the Healthy Steps for Young Children Program in
the Inter-University Consortium for Political and
Social Research (ICPSR) database. A path analysis
to assess those factors which influence the
utilization of child safety interventions and the
resultant impact that this may have on early
childhood injury reduction was undertaken.
Variables for the analysis were identified in the
codebooks that accompany the data. The selected
data was downloaded into the statistical software,
STATA. Descriptive and multivariate regression
analyses were completed.
Population Studied: The data in this study
pertains to children 30-33 months of age. The
population studied includes only children who are
followed by one of 15 pediatric primary care
providers and who speak English or Spanish
fluently. The majority of the respondents surveyed
were mostly middle income, with some post high
school education, and a mean age of 30, whose
child was followed by a pediatric primary care
provider.
Principal Findings: Study suggests that parental
characteristics are associated with the
implementation of childhood safety interventions.
When education, marital status and age are held
constant, income is not associated with safety
intervention implementation. This study did not
show a correlation between safety interventions and
injury rates.
Conclusions: This study assessed for a small
number of safety interventions, however, there are
many different causes attributable to childhood
injuries. The probability of a child not incurring an
injury could be due to factors suggested in the
literature such as parental supervision, the child’s
temperament and personality, and other behavioral
characteristics (Abboud Dal Santo, et al., 2004;
Schwebel & Gaines, 2007). This study suggests
that injury prevention involves more than just the
implementation of child safety measures. Future
research needs to be continued in the area of injury
prevention outcomes to better ascertain the impact
of these initiatives on childhood injury rates. The
active involvement of parents and caregivers as
well as passive environmental approaches and
legislative enforcement are probably more
encompassing of child safety interventions
necessary to prevent childhood injuries.
Implications for Policy, Delivery or Practice:
Implications for practice include the provision of
evidence-based anticipatory guidance and teaching
for parents/caregivers about early childhood injury
prevention.
Poster Number: 73
Cost-Effectiveness Analysis of a Cooling Protocol to
Induce Therapeutic Hypothermia in Out-of-Hospital
Cardiac Arrest
Presented by: Monica Rochman, B.S.N., R.N.,
Ph.D. Student, Nursing Business and Health
Systems, Univeristy of Michigan, 400 North Ingalls
Room 4170, Ann Arbor, MI 48109; Phone: (734)
846-6077; Email: mrocwal@umich.edu
Co-authors: Comillia Sasson, M.D.
Research Objective: Over 300,000 cases of outof-hospital cardiac arrest (OHCA) occur annually in
the United States and survival estimates to hospital
discharge are generally less than 10%. Anoxic
neurological injury is a major cause of morbidity and
mortality in patients who are resuscitated after a
cardiac arrest. The 2005 American Heart
Association Guidelines recommend induction of
hypothermia for out-of-hospital cardiac arrest to
improve neurological outcomes after reperfusion.
OHCA patients have significant costs to the U.S.
health system and hypothermia treatment adds to
those costs. It is unclear whether the benefits of
therapeutic hypothermia justify these additional
costs. The aim of this study was to evaluate
whether implementation of a hypothermia protocol
justified the costs accrued with OHCA patients.
Study Design: A decision tree was constructed to
analyze the costs and outcomes of an intention-to –
treat analysis comparing pre and postimplementation phases of an intravenous
hypothermia protocol. A cost effective analysis was
conducted with key parameters including
conditional probabilities of patients in a shockable
or non-shockable rhythm, health states based on
cerebral performance categories (good and poor
neurological outcomes) of patient data and median
reimbursements obtained from actual billing data
(CPT codes) collected. The main outcomes
measured were quality adjusted life years, which
were obtained from recent literature.
Population Studied: The study sample included
245 emergency department patients at a University
hospital who experienced OHCA between the years
2005-2008.
Principal Findings: Preliminary analysis showed
that males comprised 68% of the sample (n=166).
The median age of all patients was 57 years. The
median length of stay in the hospital was 8.96 days.
The primary payor was Medicare 46.9% (n=115),
followed by private insurance, 35.1%, (n=86). Posthypothermia protocol implementation resulted in
0.73 quality adjusted life years compared to 0.47 in
the pre-hypothermia protocol phase. The
incremental cost effectiveness ratio was $31,113
per quality adjusted life year.
Conclusions: From a societal perspective,
preliminary results demonstrate the implementation
of a hypothermic protocol could justify the costs and
allocation of healthcare dollars towards therapeutic
hypothermia for OHCA patients. Limitations include
lack of randomization, and a small sample size of
patients limited to a single tertiary care facility.
Results eM.P.H.asize the importance of
investigating this issue with larger, more definitive
studies.
Implications for Policy, Delivery or Practice:
Preliminary results demonstrate that the
hypothermia protocol improved survival and
outcomes after a cardiac arrest and costs less than
$100,000/QALY gained. As a result, quality of life
for survivors could be improved with less costs
attributed to quality of life years gained.
Poster Number: 74
Prevalence of Previously Recognized Alcohol and
Other Substance Use Disorders among VA Patients
Who Screen Postivie for Alcohol Misuse
Presented by: Anna Rubinsky, M.S., Research
Health Science Specialist, Services Research and
Development, VA Puget Sound Health, 1100 Olive
Way Suite 1400, Seattle, WA 98122; Phone: (206)
324-2146; Email: Anna.Rubinsky@va.gov
Co-authors: Gwen Lapham, M.S., M.S.W.; Daniel
Kivlahan, Ph.D.; Emily Williams, Ph.D., M.P.H.; Eric
Hawkins, Ph.D.; Katharine Bradley, M.D., M.P.H.
Funding Source(s): VA
Poster Number: 75
Research Objective: Brief alcohol counseling
interventions are effective for reducing alcohol
consumption among patients who drink above
recommended levels (i.e. alcohol misuse). Further,
more intensive treatment is often recommended for
patients with alcohol misuse who have alcohol use
disorders (AUD), other substance use disorders
(SUD) or history of addictions treatment. However,
the prevalence of previously recognized AUD/ SUD
among patients who screen positive for alcohol
misuse is unknown. Among Veterans Affairs (VA)
patients who screened positive for alcohol misuse,
we estimated the proportion who had previously
recognized AUD/ SUD, including diagnoses or
treatment documented in the year prior to
screening.
Study Design: The AUDIT-C is a validated screen
for alcohol misuse scored 0-12 points, with higher
scores reflecting greater consumption. A positive
screen for alcohol misuse was defined as =3 points
for women and =4 points for men. AUD/SUD
diagnoses and outpatient addictions treatment
documented in the year prior to alcohol screening
were obtained from VA administrative records. The
prevalence of previously recognized AUD/ SUD
was evaluated across risk groups based on AUDITC scores: 3 or 4-5, 6-7, 8-9 and 10-12 points.
Population Studied: This study included VA
patients who screened positive for alcohol misuse
on the Alcohol Use Disorders Identification Test Consumption (AUDIT-C) questionnaire between
July 2006 and March 2009 and were randomly
selected for standardized medical record review for
the VA’s Office of Quality and Performance.
Principal Findings: Among 23,607 eligible VA
patients who screened positive for alcohol misuse,
26.5% had an AUD diagnosis, 10.6% had a SUD
diagnosis and 10.3% had VA outpatient addictions
treatment, documented in the past year. The total
prevalence of previously recognized AUD/ SUD
(including treatment) was 28.7%; and increased as
AUDIT-C scores increased: 15.8%, 35.7%, 49.2%
and 73.5% for AUDIT-C scores 3 or 4-5, 6-7, 8-9
and 10-12, respectively.
Conclusions: Many patients who screen positive
for alcohol misuse on the AUDIT-C have previously
recognized AUD/SUD. Patients with higher AUDITC scores are very likely to have had prior AUD/DUD
diagnoses or treatment, including nearly 75% of
those in the highest risk group.
Implications for Policy, Delivery or Practice:
Patients with higher AUDIT-C scores may benefit
from more intensive interventions and/or
medications for alcohol dependence, and health
care systems implementing alcohol screening and
brief intervention need to consider the resources
necessary to address the needs of patients with
AUD/SUD.
Modeling Causes of Aggressive Behavior in
Patients with Dementia
Presented by: Kavita Sail, M.S., Graduate
Research Assistant, Management Policy and
Community Health, University of Texas, School of
Public Health, Houston, 1200 Hermann Pressler
Drive, Houston, TX 77054; Phone: (713) 500-9483;
Email: Kavita.R.Sail@uth.tmc.edu
Co-authors: Robert Morgan, Ph.D.; A. Lynn Snow,
Ph.D.; Jessica Davila, Ph.D.; Mark Kunik, M.D.
Research Objective: Aggression is among the
most distressing and dangerous of symptoms
experienced by individuals with dementia, yet
uncertainty remains about its incidence and causes.
Our prior work has demonstrated higher levels of
baseline caregiver burden, pain and decline in
mutuality over time to be associated with increased
risk of aggression in patients with dementia.
However, we believe that onset of aggression is a
consequence of complex interrelations between
various psychosocial factors in patients with
dementia. In this longitudinal study, we used a
structural equation modeling approach to test a
conceptual model of associations between
constructs predicting the development of
aggression in non aggressive patients newly
diagnosed with dementia.
Study Design: Patients already aggressive,
residing in a nursing home or having a caregiver
less than 8 hours a week were excluded. Patients
and caregivers were assessed over 24 months
starting at baseline and then at months 5, 9, 17, 21
and 25. Aggression was evaluated using the
Cohen-Mansfield Agitation Inventory. Independent
variables were assessed using the Hamilton Rating
Scale for Depression, Philadelphia Geriatric Center
Pain Intensity Scale, Burden Interview, Mutuality
Scale and Pleasant Events and Activities Schedule
Alzheimer’s Disease. Cognitive impairment was
assessed with The Dementia Rating Scale. Yearly
incidence of new aggression was calculated, and a
structural equation model was used to analyze
associations between independent variables and
onset of aggression.
Population Studied: Study participants were VA
patients over age 60, newly diagnosed with
dementia.
Principal Findings: Of the 215 enrolled patients,
89 (41%) developed aggression (unadjusted
incidence of aggression of .37/year at risk). Mean
dementia severity and mutuality scores at baseline
were significantly lower among aggressive patients
than among nonaggressive patients; mean scores
for depression, hallucinations, delusions, and
caregiver burden were higher. Our structural model
for predicting onset of aggression showed a good fit
to the data (ChiSquare = 8.12, p > 0.05; CFI =
0.954; TLI = 0.915; RMSEA = 0.046). Higher levels
of baseline caregiver burden (p < 0.05) and decline
in mutuality over time (p < 0.05) were our primary
predictors of increased risk of aggression in our
SEM model. The significance of the relation
between pain and onset of aggression was
dependent on whether presence of aggression or
time to development of aggression is modeled,
while the relation between depression and
aggression appeared to be mediated jointly by pain
and caregiver burden.
Conclusions: We found potentially mutable factors
internal and external to the patient that were
associated with development of aggression. Our
SEM findings clarify the relationships found in our
prior work and highlight their significance in
predicting the development of aggression in
dementia patients.
Implications for Policy, Delivery or Practice:
Antipsychotics remain the predominant treatment of
aggression, but have limited efficacy and can have
serious side-effects. Consequently, our findings
have important preventive and treatment
implications for the almost 40% of dementia
patients that become aggressive each year.
Funding Source(s): VA
Poster Number: 76
Expansion of Newborn Screening Panels: A Formal
Program Evaluation of Krabbe Disease Screening
in New York State
stakeholder groups described in the CDC
Framework: 1. those involved in program
operations, 2. those served or affected by the
program, and 3. those in a position to make
decisions about the program.
Principal Findings: Research in progress expected completion: May 2010
Conclusions: This research will provide objective,
scientific information for public health decision
makers in all states considering inclusion of Krabbe
disease in their newborn screening panel.
Implications for Policy, Delivery or Practice:
Newborn screening is recognized as a successful
part of the preventative health system in all states
and territories of the United States. Advancement in
technology has increased the number of disorders
that could be included in these programs, however,
it is not clear that having the ability to screen for a
disorder justifies inclusion in a screening program.
In 2006, New York became the only state to add
Krabbe disease,a progressive fatal neurological
diseasee, to the newborn screen. To date, there
has been no formal evaluation of this program.
Funding Source(s): Sigma Theta Tau - Alpha Zeta
chapter
Poster Number: 77
Impact of the Medical Home on the Safety and
Quality of Health Care in Canada
Presented by: Roberta (Bobbie) Salveson,
Doctoral Candidate, Center for Health Policy,
Columbia University School of Nursing, 617 West
168th Street GB-239, New York, NY 10032; Phone:
(212) 241-7055; Email: rls2128@columbia.edu
Co-authors: Paticia Stone, Ph.D.
Presented by: Andrea Scobie, M.H.S.A., Ph.D.
Student, College of Pharmacy, Dalhousie
University, 5968 College Street, Halifax, B3H 3J5
Canada; Phone: (902) 494-6023; Email:
andrea.scobie@dal.ca
Co-authors: Neil MacKinnon, Ph.D.; Sean Higgins,
B.Sc.; Holly Etchegary, Ph.D.; Rhonda Church,
M.D.
Research Objective: The purpose of this study is
to conduct a formal program evaluation of newborn
screening for Krabbe disease in New York State
using the CDC Framework for Program Evaluation
in Public Health. The aims of the study are: 1.
assess if Krabbe disease screening is meeting the
stated goals of the New York State newborn
screening program for each of the stakeholders
using this program, 2. assess the Krabbe disease
test characteristics with the most recent data
available, and 3. assess the cost to identify one true
positive Krabbe disease screening result.
Study Design: The CDC Framework for Program
Evaluation in Public Health provides the guideline
for collection of data. Data will be collected from a
variety of sources in New York State, including
published brochures, public health law, technical
manuals, stakeholder interviews, annual newborn
screening reports, meeting minutes,coding, and
charge reports. Data will be analyzed using
qualitative and quantitative methodologies.
Population Studied: Using purposive sampling,
participants will be recruited to represent
Research Objective: To explore the relationship of
a medical home to self-reported risk factors for
medical error and self-reported indicators for quality
of care.
Study Design: The Commonwealth Fund's 2007
International Health survey in seven countries was
the primary data source. We focused only on the
Canadian data. Using SPSS, we explored the
relationship between a medical home and a number
of variables based on leading indicators for quality
primary health care as developed by the Canadian
Institute for Health Information, including: access,
medical error, coordination, confidence, emergency
department use and patient-provider relationship.
Chi-square tests were used for categorical variables
and Mann-Whitney U tests for ordinal variables.
Population Studied: As part of a seven-country
health policy survey, the Commonwealth Fund
surveyed a sample of 3003 people 18 years and
older in Canada’s 10 provinces and 3 territories
between 6 March and 7 May 2007.
Principal Findings: Of Canadians surveyed, 51%
did not have a medical home. Based on the 2006
census, this extrapolates to nearly 13 million adult
Canadians. Overall, the presence of a medical
home was associated with self-reported improved
access to health care services, coordination of the
services received, confidence in the services
received and provider knowledge. Self-reported
medical and medication error rates were higher
among those without a medical home. However,
emergency department use patterns were similar
among those with and without a medical home.
Conclusions: The presence of a medical home is
associated with perceived safer and higher-quality
patient care. Ensuring that Canadians have a
medical home through primary health care reform
may be an effective means to mitigate medical and
medication errors while increasing patient
satisfaction and strengthening patient–provider
relationships.
Implications for Policy, Delivery or Practice:
Greater eM.P.H.asis on primary health care through
the promotion of medical homes may offer patients
and providers a more effective means to navigate
the Canadian health care system. Given Canada's
history of slow change within the primary care
arena it is unlikely that change will take place
quickly. Nonetheless, there is mounting evidence
that concept of medical homes will help to increase
the quality and safety of care received by
Canadians.
Poster Number: 78
Pharmacy Staff Perceptions of Medication Incident
Reduction and Reporting in Nova Scotia
Presented by: Andrea Scobie, M.H.S.A., Ph.D.
Student, College of Pharmacy, Dalhousie
University, 5968 College Street, Halifax, B3H 3J5
Canada; Phone: (902) 494-6023; Email:
andrea.scobie@dal.ca
Co-authors: Neil MacKinnon, Ph.D.; Todd Boyle,
Ph.D.; Heidi Deal, B.Sc.; Tom Mahaffey, Ph.D.
Research Objective: Medication incidents (MI) can
have serious consequences for the health of
Canadians and the perceived safety of community
pharmacy practice. The objective of this research
was to assess pharmacy staff perceptions of
various strategies for MI reduction and reporting in
community pharmacies, and how such perceptions
differ between staff and ownership type.
Study Design: The Pharmacy Perceptions of
Safety Survey (PPSS) contained twenty-three
questions aimed at assessing the state of MI
identification and disclosure in Nova Scotia
pharmacies as related to the role of continuous
quality improvement (CQI) and safety culture.
Multivariate analysis of variance (MANOVA) was
used to examine the overall difference of means for
each group. Significance for all calculations was set
a priori at the .05 level. All calculations were
performed using SPSS (version 15.0).
Population Studied: Community pharmacy staff
were recruited on a volunteer basis from thirteen
pharmacies in Nova Scotia representing rural and
urban locations, independent and chain ownership
and high and low prescription volume. Staff
positions included Staff Pharmacist, Pharmacist
Manager and Pharmacy Support Staff (i.e.,
technician, student and intern). Pharmacy
ownership encompassed Independent, Corporate
and Franchise pharmacies.
Principal Findings: Respondents indicated that
the use of pharmacists to help select drug therapies
would be the most effective strategy to reduce MIs,
whereas sharing learnings with colleagues and
assuring anonymity when reporting would increase
MI reporting. Celebrating reporting of errors and
limiting the use of abbreviations were seen as the
least effective strategies. MANOVA results
indicated that practitioner type played an overall
significant role in staff preferences to increase MI
reporting (F=1.775, p=0.027), indicating significant
differences between pharmacists, managers and
support staff in how they view strategies for MI
reporting. Ownership type did not have an overall
significant impact on either MI reduction (F=1.520,
p=.095) or reporting (F=1.357, p=.152).
Conclusions: The results of this study indicate that
some MI reduction and reporting strategies may be
more successful than others within the community
pharmacy context. Staff perceptions may rely
heavily on the actions of their colleagues, and
creating a comfortable environment to share error
experiences may lessen hesitance to report.
Respondents also viewed anonymity for
pharmacists and pharmacy technicians as an
important strategy to increase MI reporting and this
finding coincides with literature that suggests that
ensuring anonymity can improve error reporting by
reducing staff fears of blame and punishment.
Implications for Policy, Delivery or Practice:
This study has shown that there is a great deal of
variation among pharmacy staff with regards to their
perceptions of MI reduction and reporting
strategies. This presents challenges for improving
MI reporting at both the community pharmacy and
regulatory authority levels. Ensuring that community
pharmacy staff perceptions and expectations are
incorporated into MI reporting and reduction
strategies will ultimately increase the likelihood of
widespread adoption.
Funding Source(s): Social Sciences and
Humanities Research Council
Poster Number: 79
Self-Reported Medical, Medication and Lab Error in
Eight Countries: Risk Factors for Sicker Adults
Presented by: Andrea Scobie, M.H.S.A., Ph.D.
Student, College of Pharmacy, Dalhousie
University, 5968 College Street, Halifax, B3H 3J5
Canada; Phone: (902) 494-6023; Email:
andrea.scobie@dal.ca
Co-authors: Neil MacKinnon, Ph.D.
Research Objective: The objective of this study
was to identify risk factors associated with selfreported medical, medication and laboratory error
and explore their role in the likelihood of
experiencing an error.
Study Design: The Commonwealth Fund’s 2008
International Health Policy Survey of chronically ill
patients in eight countries was the primary data
source for this research. Bivariate analysis was
used to determine significant explanatory variables
(p <.01) for exclusion in the binary logistic
regression model. Odds ratios were calculated to
explore which risk factors most greatly explain the
likelihood of experiencing an error.
Population Studied: The Commonwealth Fund’s
2008 International Health Policy Survey of
chronically ill patients was conducted in Australia,
Canada, France, Germany, the Netherlands, New
Zealand, the United Kingdom and the United
States. The final sample consisted of 9,944 adults
18 years of age and older who were considered to
have chronic and ongoing health needs. Eligible
adults had to have self-reported at least one of the
following criteria: (1) fair or poor health condition,
(2) a serious or chronic illness, injury or disability
that has required a lot of medical care in the past
two years, (3) hospitalization (for something other
than uncomplicated delivery of a baby) in the past
two years or (4) major surgery in the past two
years.
Principal Findings: Results of the bivariate
analysis resulted in eight variables being included in
the logistic regression analysis: age, education
level, number of chronic conditions present, number
of doctors seen in past two years, presence of care
coordination problem, poor communication with
doctor, prescription drug in the past two years, and
emergency room use within the past two years. The
inclusion of all eight variables was justified by the
Omnibus tests of model coefficient significance (chisquare = 933.456, p=.000) and the model was able
to accurately predict 77.4% of the cases. The
Hosmer and Lemeshow Test was non-significant
(chi-square=12.503, p=.130), indicating that the
model fit the data well.
Conclusions: The final regression model indicates
that there are a number of risk factors associated
with the likelihood of experiencing a health care
error among the eight countries studied. Although
some demographic factors, including age and
education level do play a role, risk factors with the
greatest ability to predict experiencing an error
encompassed issues with coordination and
continuity of care and provider knowledge of a
patient’s medical history.
Implications for Policy, Delivery or Practice:
Ensuring that providers have access to necessary
patient information, and educating patients as to the
importance of keeping their providers up-to-date,
could greatly reduced the risks associated with
experiencing an error. Greater development and
use of electronic health records and organizational
practices aimed at continuity of care can help to
achieve this. Furthermore, greater understanding by
patients of the risks associated with health care can
help to engage patients in participating in error
prevention strategies. Overall, the identification of
risk factors help policymakers and organizations to
proactively reduce the likelihood of patients
experiencing an error through greater examination
of system- and organization-level practices.
Poster Number: 80
Family Health History Discussion in PatientProvider Encounters
Presented by: Nina Shiang, B.S., Graduate
Student Associate, Verdict, 7400 Merton Minter
Boulevard, San Antonio, TX 78229; Phone: (210)
617-5238; Email: shiang@uthscsa.edu
Co-authors: Joann Seo, B.A.; Eva Shipp, Ph.D.;
Nedal Arar, Ph.D.
Research Objective: This study will attempt to
characterize discussion(s) of family health history in
the clinical setting and also detail potential
inadequacies when this topic is discussed.
Study Design: Cross-sectional, Observational
Study
Population Studied: Patients served by primary
physicians at the Audie L. Murphy Memorial
Veterans Hospital (VHA) in San Antonio, TX from
2003 to 2005
Principal Findings: Of the 131 total encounters
transcribed among the 37 patients, only 24
encounters (18.3%) included discussion of FHH. In
new patients, 16 encounters (66.7%) included
discussion of FHH, while in established patients
only 8 encounters (33.3%) included discussion of
FHH. In new patients, discussion of FHH most
frequently occurred in the first encounter (62.5%).
In established patients, no observed pattern could
be ascertained. Providers initiated discussion on 14
of these encounters (58.3%) and patients initiated
discussion on 10 of these encounters (41.7%). In
provider initiated FHH discussion, satisfactory
levels of discussion were most frequently achieved
(>3 elements of FHH discussed) while in patient
initiated FHH discussion, moderate levels of
discussion were most frequently achieved. FHH
discussion included topics of diabetes
management, cancer, and cardiovascular disease
(CVD). In provider initiated FHH discussion,
exchange on cancer and CVD were most frequently
observed, 58.3% and 41.7% respectively.
Additionally, the relationship between FHH
discussion and discussion of self-care management
(SCM) topics were assessed. The average number
of SCM topics discussed was 6 (std = 2.9; range: 213). Patient-provider discussion of SCM was related
to discussion regarding FHH information (p< 0.05).
Conclusions: In this study, providers were more
likely to initiate discussion on family health history
among new patients. In provider initiated FHH
discussion, satisfactory levels of discussion were
most frequently achieved (>3 elements of FHH
discussed) while in patient initiated FHH discussion,
moderate levels of discussion were most frequently
achieved. Additionally, FHH discussion was highly
correlated with topics of cancer and CVD.
Implications for Policy, Delivery or Practice:
Family health histories could be one means to this
end through the implementation of personalized
preventive care strategies. The publication of these
results may benefit both the clinical and scientific
community. Social scientists may use this data to
design clinical interventions aimed to best facilitate
and improve patient-provider communication on the
topic of family health history.
Funding Source(s): VA
Poster Number: 81
Trends in Length of Stay among Patients Who
Leave Against Medical Advice from U.S. General
Hospitals Between 1988 and 2003
Presented by: Rima Tawk, M.S., M.P.H.,
Graduate assistant, Health Policy and
Administration, University of Illinois at Chicago,
1603 W Taylor Street, Room 722, Chicago, IL
60612; Phone: (312) 413-1468; Email:
rtawk1@uic.edu
Research Objective: The aims of the study are
twofold: to examine the national trend in hospital
resource utilization among patients who leave
against medical advice (AMA) from US general
hospitals between 1988 and 2003; and to explore
changes in the healthcare system and the factors
that have precipitated the change.
Study Design: The study is a secondary data
analysis from the National Hospital Discharge
Survey (NHDS) conducted by the National Center
for Health Statistics. NHDS provides nationally
representative estimates on the characteristics of
patients, lengths of stays, diagnoses, and surgical
and nonsurgical procedures in hospitals of different
bed sizes, ownerships, and geographic regions of
the US. Utilizing a complex multistage stratified
probability sampling design, the NHDS annually
collects data from a sample of approximately
300,000 inpatient medical records from a national
sample of about 500 nonfederal short-stay care
hospitals.
Population Studied: This study is limited to AMA
adults (18 yrs and over) between 1988 and 2003.
We used a sample of 35,616 AMA patients
representing a weighted total of 3,986,936. Years of
data were collapsed into intervals to reflect the
changes that have shaped the healthcare system.
To date, there is a paucity of published literature on
the hospital resource utilization of AMA patients and
the factors that predict their length of stay (LOS).
Most of the previous studies on AMA discharges
have shown that mental illness was an important
predictor of such discharges.
Principal Findings: Males leave about a quarter a
day before females (p=0.0008). Patients whose age
category ranged between 18 to 44 years and 45 to
64 years leave approximately before 0.9 days
(p<.0001) and 0.6 days (p=<.0001) respectively
than the 65+ age category. Blacks stay more than
half a day than whites (ß=0.61; p<0.0001), other
race stay less than a quarter of a day than whites
(p=0.0134). Singles stay more than a quarter a day
than married (p=0.0317). With respect to region,
patients discharged from the Northeast stay 0.4
days longer than patients discharged from the
South region (p=<.0001) while patients discharged
from the West stay 0.3 days longer than patients
discharged from the South region (p=<.0030).
Regarding insurance coverage, patients on
Medicare stay approximately 0.7 days more than
patients on private insurance (p<0.0001). Patients
on Medicaid stay approximately more than half a
day than patients on private insurance (p<0.0001).
Self-pay patients leave a quarter a day earlier than
patients on private insurance (p=0.0285). Patients
discharged from hospital bedsize (6-99) leave about
one day earlier than patients who are discharged
from 500 + bedsize category (p=<.0001). Patients
from hospital bedsize (100-199) leave 0.88 day
earlier than patients who are discharged from 500 +
bedsize category (p=<.0001). Patients from hospital
bedsize (200-299) leave 0.71 day earlier than
patients who are discharged from 500 + bedsize
category (p=<.0001). Patients from a bedgroup
(300-499) leave about half a day a day earlier than
patients who are discharged from 500 + bedsize
category (p=<.0001). The ownership of hospitals
did not yield significant results in all categories. On
average, patients with mental illness stay half a day
more than patients without mental illness
(p=<.0001).Gender and Marital Status did not
change across intervals.The significant predictors
that changed across time were African Americans,
Medicaid insurance and the West region.
Conclusions: LOS has significantly decreased
across time intervals (p<0.0001). Hospital
ownership influences LOS less over time. Data
reveal substantial regional variation in particular the
West had a substantial decrease in LOS. Patients
covered by Medicaid had shorter LOS.
Implications for Policy, Delivery or Practice:
AMA patients pose problems for service delivery
systems because they experience short
hospitalizations but multiple readmissions and
therefore increased healthcare costs. It is important
to examine factors that either facilitate or impede
utilization of services among patients who do not
adhere to medical treatment. LOS can reveal useful
information about the performance of hospitals and
it has been accepted as a robust index of system
performance. Its magnitude can be influenced by
system as well as patient –related factors. The
results may offer insights to reduce AMA
discharges.
Poster Number: 82
The Impact of Length-of-Stay on Hospital Acquired
Adverse Drug Events: A Regression Discontinuity
Approach
Presented by: Mark Unruh, M.S., Ph.D.
Candidate, Department of Community Health,
Brown University, 121 South Main Street, Box G S121, Providence, RI 02912; Phone: (831) 4207039; Email: mark_unruh@brown.edu
Co-authors: Vincent Mor, Ph.D.
Research Objective: To quantify the effect of
inpatient length-of-stay on hospital acquired
adverse drug events.
Study Design: Using the 2006 New York State
Inpatient Database, we employ a regression
discontinuity design to assess the influence of
length-of-stay (LOS) on the likelihood of
experiencing a hospital-acquired adverse drug
event (ADE). Medicare’s prospective payment
system creates an incentive for hospitals to
discharge patients as soon as possible, which is
partially offset by reduced inlier payments for short
stays. Since Medicare bases LOS on the number
midnights spent in the hospital, the number of hours
constituting LOS will vary by time of admission. For
patients to have the same LOS, those admitted
near, but after, midnight will need to spend more
hours in the hospital than will admissions just prior
to midnight. Admission through the emergency
department (ED) shortly before or after midnight
should be a random process, creating a treatment
group (post-midnight admissions) and the proper
counterfactual (pre-midnight admissions). Our
empirical strategy utilizes two-stage least squares
estimation with heteroskedastic-robust standard
errors adjusted for clustering within hospitals. An
assignment rule, equal to one for post-midnight
admissions and zero for pre-midnight admissions, is
incorporated as an instrumental variable. The firststage estimates the relationship between time of
admission and LOS. The reduced-form derives the
relationship between LOS and hospital acquired
ADEs.
Population Studied: The study population includes
patients 65 and older admitted through hospital EDs
in the state of New York, in 2006, with Medicare
listed as the primary payer. Only individuals
admitted plus or minus two hours of midnight were
included.
Principal Findings: The observable characteristics
of the two groups are well-balanced. In the firststage we find that, on average, individuals admitted
in the hour following midnight have a LOS that is
approximately 10 hours longer compared to
admissions in the hour prior to midnight. Likewise,
those admitted after midnight are 8% more likely to
experience an ADE. The ratio of the reduced-form
and first-stage estimates indicates that each
additional hour in LOS adds 0.8% risk of
experiencing an ADE, on average. After adjusting
for clustering within hospitals, estimates remain
significant at the 5% level. These results were
robust to samples comparing admissions that took
place from 10:00 to 10:59 PM to those entering the
ED from 12:00 to 12:59 AM. Additionally, estimates
obtained using samples composed of nondeferrable conditions had consistent results. Nondeferrable conditions were identified by diagnoses
with similar admission rates for each day of the
week, including weekends, consistent with recent
studies. This indicates that any potential
confounding due to unusual treatment patterns near
midnight, such as those associated with staff shift
changes, are unlikely to account for our results.
Conclusions: We conclude that hospital acquired
ADEs are common and that longer LOS presents
additional risk for acquiring an ADE.
Implications for Policy, Delivery or Practice: Our
results indicate that LOS is amenable to policy,
putting some patients at increased risk of
experiencing an ADE. Since many hospital acquired
ADEs are preventable, reducing the likelihood of a
patient experiencing one provides an opportunity to
simultaneously improve the quality of care and
reduce costs.
Funding Source(s): NIA
Poster Number: 83
Understanding the Cultural Impact of Familismo on
Mental Health Services Utilization among Latinos in
the U.S.
Presented by: Alice Villatoro, B.A., Ph.D.
Student, Health Services, UCLA School of Public
Health, 650 Charles E. Young Drive, 31-269 CHS,
Los Angeles, CA 90095-1772; Phone: (310) 8252594; Email: avillato@ucla.edu
Co-authors: Vicki Mays, Ph.D., M.S.P.H.
Research Objective: There is strong empirical
evidence that untreated mental disorders are
associated with substantial morbidity. Hence, it is
important to examine the underlying mechanisms
that may deter help-seeking behaviors for mental
health services within the Latino community since
this population generally underutilizes mental health
specialty services. The aim of this study was to
examine how familismo, a prominent value on
familial loyalty and solidarity in the Latino culture,
may be a cultural determinant of mental health
service utilization among Latino adults.
Study Design: Data utilized came from the 20022003 National Latino and Asian American Study
(NLAAS), a nationally representative community
household survey examining the prevalence of
mental health disorders and services utilization of
Latinos and Asian Americans. Familismo was
examined as the combined effect of family-provided
emotional support and level of familial cultural
conflict. Logistic regressions were performed to
examine the relationship between familismo and
lifetime mental health service utilization, controlling
for sociodemographic, access, need, and immigrant
experience characteristics. Sample design variables
and sample weights developed by Heeringa et al.
(2004) were utilized in the analyses to correct for
clustering, unequal probabilities in selection during
the sampling stage, and nonresponse patterns of
underrepresented populations in the sample.
Population Studied: The sample included 2554
non-institutionalized Latino adults (18 years of age
or older) residing in the U.S.
Principal Findings: Latinos reported high levels of
family-provided emotional support and low levels of
familial cultural conflict. Low rates of lifetime mental
health service utilization were also observed. The
results showed that higher levels of both familyprovided emotional support (OR=1.17, 95% CI:
1.03-1.35) and familial cultural conflict (OR=2.13,
95% CI: 1.41-3.22) were associated with higher
likelihoods of lifetime utilization among Latino
adults. However, familial cultural conflict appeared
to be a more robust predictor of service use than
family emotional support. A significant interaction
between lifetime diagnosis status and perceived
need was also observed (OR=0.16, 95% CI: 0.080.32). Though individuals with clinical need for
mental health services showed high rates of
utilization, Latinos with both a history of mental
illness and subjective need for mental health care
were significantly less likely to utilize services in
comparison to Latinos with no prior diagnosed
mental illness. Other significant predictors of
utilization included years residing in the U.S, years
of education, language, and subethnic identification.
Conclusions: The results of these analyses
illustrate the importance of family, most especially
family conflict, on mental health services utilization
among Latinos in the U.S. Though Latino families
have been traditionally characterized as supportive
entities, the results of this study imply that the use
of mental health services for emotional or
substance-related concerns becomes an important
outlet for Latinos with high levels of familial conflict
as these services may provide the help and support
they lack in their family immediate environments.
Additionally, the interaction suggests that the
question of utilization appears to be most relevant
to those in need for services.
Implications for Policy, Delivery or Practice: The
findings from this study can assist health care
providers in the development of culturally
competent strategies aimed to improve access to
and utilization of mental health services for Latinos.
Poster Number: 84
The Sensitivity of Adverse Event Cost Estimates to
Diagnostic Coding Error in Administrative Data
Presented by: Gavin Wardle, M.B.A., B.Sc., Ph.D.
Candidate, Health Policy, Management, and
Evaluation, University of Toronto, 95 Bertmount
Avenue, Toronto, M4M 2X8 Canada; Phone: (416)
220-2210; Email: gavin.wardle1@gmail.com
Co-authors: Walter Wodchis, Ph.D.; Geoff
Anderson, Ph.D.; G. Ross Baker, Ph.D.; Audrey
Laporte, Ph.D.
Research Objective: Previous research has
established a consensus that diagnostic data
elements are recorded with error in hospital
administrative data. However, as yet, there is no
published study that has quantified the impact of
coding error on estimates of the costs of adverse
events. Given the increasing importance of these
estimates for hospital payment policy and for
assessments of the business case for patient
safety, this is an important gap in the literature. This
study aims to fill this gap by examining the impact
of coding error on estimates of the excess cost of
each of the eleven nursing sensitive adverse events
studied by Needleman et al. (NEJM 2002).
Study Design: A retrospective cohort analysis of
originally reported hospital administrative data and
corresponding records from a large reabstraction
study. Analysis revealed that adverse event coding
in the original data had low sensitivities and low
positive predictive values and that the reabstraction
data were biased against adverse events that had a
low impact on hospital resources. A hybrid data set
was constructed to redress these problems and to
establish a reference standard. In each data set,
propensity score matching and multivariate
regression analysis were used to estimate the
excess cost of each adverse event. Coding errorinduced variation was examined by comparing the
cost estimates derived from the three data sets.
Population Studied: All adult, non-elderly medical
and surgical admissions (n=9,670) included in a
reabstraction study of complex cases from 16
Ontario hospitals. Patient specific case costs were
obtained from the Ontario Case Costing Initiative
database.
Principal Findings: Estimates of the excess costs
of the adverse events derived from the reference
standard data ranged from $24,286 (metabolic
derangement) to $48,897 (hospital-acquired
sepsis). Coding errors in the originally reported data
caused the excess costs of the adverse events to
be underestimated, on average, by 27 percent. In
contrast, the bias against low cost adverse events
in the reabstraction data caused the excess costs to
be overestimated, on average, by 34 percent. At the
institutional level, coding induced misestimation of
the cost attributable to adverse events ranged from
48 percent underestimated to 2 percent
overestimated.
Conclusions: Estimates of adverse event costs are
highly sensitive to coding error. Adverse event
costs are likely to be significantly underestimated if
the error is ignored.
Implications for Policy, Delivery or Practice: The
finding that coding error in administrative data can
cause the costs of adverse events to be
significantly underestimated, coupled with the
observation that the likelihood of error is ignored in
most studies, suggests that previous assessments
of the business case for patient safety may have
been biased against the cost effectiveness of
patient safety improvements. Furthermore, the
observed extent of institutional level variation in
adverse event coding indicates that administrative
data are inadequately reliable as a basis for
adverse event payment policies or for public
reporting of adverse event rates.
Funding Source(s): Health System Performance
Research Network
Poster Number: 85
Does High Cost Sharing in Physician Care Reduce
Health Care Utilization and Expenditure Differently
for People with Severe Disease than Those
Without?
Presented by: Haichang Xin, Ph.D. student,
Department of Health Services Research,
University of Florida, 101 S. Newell Drive, PO Box
100195, Gainesville, FL 32610-0195; Phone: (352)
273-6627; Email: hsppyxhc@phhp.ufl.edu
Research Objective: This study examines the
potential differential impact of cost-sharing in
physician care on health care utilization and
expenditure for those with and without severe
health conditions. By providing empirical evidence,
this study’s findings will inform policies that help
protect and prevent the severely ill from
exacerbating their poor health status and financial
burden, as well as save cost for the society at the
aggregate level. This present study fills the
following gaps in the literature by focusing costsharing on physician care, and using representative
sample of the whole population.
Study Design: The study adopted a crosssectional study design from the 2007 Medical
Expenditure Panel Survey data. The Hausman test
eliminated the concern of the potential endogenous
problem between cost-sharing and health care
utilization or expenditure (p=0.12). Negative
binomial regressions and two-part models were
employed to estimate the utilization and
expenditures.
Population Studied: Subjects were included with
an age of 18 or older with the full year of insurance
coverage in the United States. Disease severity is
measured by physical component score (PCS), and
a dummy severity variable was then created with
two groups of individuals whose PCSs fell within the
bottom tertile and the top tertile. Cost-sharing was
calculated by dividing out-of-pocket expenditures by
total expenditures for each person, then a dummy
cost-sharing variable was created with the cut off
points of 20% in Medicare and private plans and
mean values in Medicaid.
Principal Findings: The data was analyzed with
STATA. Among the 13,020 selected subjects,
although an insignificant differential impact was
found in physician care utilization (p=0.34), costsharing, severity and their interaction had expected
signs. Specifically, severely ill individuals reduced
MD visits less than the general population in
response to high cost-sharing. As a component of
MD visits, primary care physician visits analysis
demonstrated the significant differential impact
(p=0.00). In downstream inpatient care utilization,
although an insignificant differential impact was
found (p=0.34), cost-sharing, severity and their
interaction demonstrated expected signs.
Specifically, severely ill individuals increased
inpatient care more than the general population,
which suggested that the suppressed MD visits due
to high cost-sharing were released in the
downstream inpatient services. As for expenditures,
in all physician care, inpatient care and overall
services the differential impact demonstrated
consistent positive directions for both parts, but was
significant in part one (p=0.04, 0.00 and 0.00
respectively) and insignificant in part two (p=0.36,
0.79 and 0.48 respectively).
Conclusions: In conclusion, the differential impact
of cost-sharing was basically confirmed.
Specifically, it demonstrated that severely ill people,
in response to high cost-sharing, tended to reduce
less health care utilization and expenditures that
were crucial to maintain their health.
Implications for Policy, Delivery or Practice:
Therefore, this vulnerable people should be treated
differently. Specifically, a low cost-sharing policy
should be designed to reflect and match their
situation. This study will highlight the necessity and
importance of insurance policy design in terms of
differentiation and specification for its target
population. Furthermore, it will inform the current
debate on health care reform, especially how a
universal plan, rather than specialized plans for
subpopulations, would complement the existing
public and private designs.
Poster Number: 86
Digital Divide in a Personalized Health Record
Presented by: Cyrus Yamin, B.S., Medical
Student, Brigham and Women's Depatment of
General Medicine, Harvard Medical School, 156
Allston Street, Cambridge, MA 02139; Phone:
(617) 413-4714; Email: cyamin@gmail.com
Co-authors: Srinivas Emani, Ph.D.; Jonathan
Wald, M.D.; Andrew Karson, M.D., M.P.H.; Stuart
Lipsitz, Deborah Williams, Sc.D., M.H.A.; David
Bates, M.D., MSc.
Research Objective: Electronic personal health
records (PHR) offer the potential of enhancing
patient-practice communication and streamlining of
administrative functions-such as appointment
requests, referral requests, and prescription refills.
At the same time, the “digital divide,” the
population-level gap in Internet access, hinders the
role a PHR plays in the delivery of care. This study
assesses the existence of a "digital divide"
associated with a PHR.
Study Design: Retrospective analysis of a PHR
implemented at over 50 practices at Partners
Health Care in Massachusetts.
Population Studied: Rates of PHR and degree of
usage of PHR among adopters based on total
unique usage requests. Adopters were those
defined as registering online for the PHR.
Principal Findings: 37,599 patients adopted the
PHR since its inception in 2002. Compared to a
random sample of non-adopters, adopters were
more likely to be female (OR, 1.32, 95% CI, 1.281.36) and have greater than two chronic conditions
(OR=1.396, CI 1.34-1.482). Blacks and Hispanics
were less likely to be adopters (OR .397, 95% CI
.294-.351, OR .321 95% CI .294-.351, respectively).
Adopters were stratified based on unique usage
requests in 2007 and 2008: 0-1 requests comprised
61.5% of usage; 1-4 requests comprised 25.1%; 59 requests comprised 9.1%; > 10 requests
comprised 4.2%. To develop an area-based
socioeconomic measure for adopters, we geocoded
adopter’s home address to the census block level,
and linked it to the 2000 Census poverty levels.
After adjusting for practice level clustering and
fitting to a Poisson Model the degree of usage was
best predicted by the number of patient
comorbidites (with more comorbidities increasing
usage), followed by patient's race (whites over
blacks and hispanics), and insurance status (private
insurance over Medicare). The area-based
socioeconomic measure was not associated with
degree of usage.
Conclusions: Despite increased rates of Internet
access, those population groups most likely to
benefit from PHRs still are less likely to initiate use.
Among those who do adopt, however, higher usage
does not appear to be associated with
socioeconomic status.
Implications for Policy, Delivery or Practice:
Perhaps once a person has joined a PHR, thereby
demonstrating some level of Internet access,
variations in usage might instead reflect overall
heath service utilization trends, while the “digital
divide” poses less of a barrier than once suggested.
Funding Source(s): Other, Harvard Medical
School, Brigham and Women's Hospital
Poster Number: 87
Medical Savings Account in China and the Effect of
Its Balances on Outpatient Utilization
Presented by: Vivienne Hui Zhang, M.Sc., Ph.D.
Student, Department of Management and
Marketing, The Hong Kong Polytechnic University,
Hung Hom, Kowloon, Hong Kong; Phone:
+85261805323; Email: chefyzhang@hotmail.com
Research Objective: Medical Savings Account
(MSA) is established for insured individuals to pay
for their own medical expenses to control costs, and
save money for future medical needs. Some
literature in China suggested that the presence of
MSA has a negative effect on equity in health care
financing, and recommended to reduce the role of
personal MSA. However, these studies were only
based on the estimated MSA balances, and
impossible to distinguish the previous year’s
balances from current year’s contributions. This
study will fill this gap by using real data of MSA
balances, and examines the effect of MSA balances
on outpatient services utilization among different
income groups.
Study Design: Drawing on theoretical framework
from Andersen’s behavioral model of health
services use, we investigated individual enabling
factors measured by MSA balances to evaluate
health policy in China. The city in this study,
Guangzhou is a typical large city in southern part of
China, and it is representative with respects to
population size, age structure and income levels.
The study first divided MSA account-holders into
four groups, based on the contributions reflecting
income levels. Age-gender adjustment mean
expenditures in 2007 were then calculated to see
whether income affects the expenditure of MSA
with similar medical needs. Then, within each
group, two-part model was applied to study the
effect of MSA balances in December 2006 on the
expenditures in 2007. The first part used logit model
to estimate the effect on probability of outpatient
usage, and a log-linear model in the second part
was employed to study the incurred level of its
expenditures for users only, holding other factors
fixed. This stratified analysis was used to control
the income effect.
Population Studied: Sample (N=93,752) is the first
group of insured employees since MSA scheme
was launched in the city in 2002. Six years data
include each MSA account-holder previous years’
balances and current years’ contributions,
expenditures between 2002 and 2007.
Principal Findings: Age-gender adjustment mean
expenditures in 2007 were significantly different
among these groups after controlling health needs.
MSA account-holders in different income groups
had different health behaviors. In the lowest income
group, MSA balances were positively related to
probability of outpatient usage and its expenditures.
However, for the highest income group, more MSA
balances were significantly associated with lower
usage probability and less medical spending. For
the two medium income groups, the MSA balance
was also significantly associated with probability of
usage and its expenses when holding other factors
fixed. But the relationship was non-linear.
Conclusions: The impact of MSA balance on
outpatient utilization is different among four income
groups. Low income insured employees use health
services as long as they have funds in MSA. They
may not have enough balances to cover their
necessary health services. But high income groups
have lower utilization as MSA balances increase.
More savings in MSA are available for high income
people. Among two medium income groups, MSA
can play its role.
Implications for Policy, Delivery or Practice:
Income may be a governing effect of the behaviors
of the MSA usage. Polices targeted to different
income groups may be needed to strengthen the
role of MSA. For low income insured employees,
government can give additional subsidies to their
personal MSA. Regarding high income groups, we
suggest to extend the application of available MSA
balances. For example, enrollees can use their own
MSA to pay for family members’ medical expenses,
or purchase complementary private insurances.
Poster Number: 88
Can Family Income Impact Children's Limitations
Due to Chronic Conditions?
Presented by: Lucy Bilaver, M.P.P., M.S., Ph.D.
candidate, School of Social Service Administration,
University of Chicago, 1313 E. 60th Street,
Chicago, IL 60637; Phone: (773) 256-5158; Email:
labilave@uchicago.edu
Research Objective: To identify a causal effect of
income on parent's perceptions of their children's
limitations due to chronic conditions. Analysis of
NHIS and NLSY data has revealed that parents of
children in poverty report more activity limitations
than their higher income counterparts, but it is
unknown whether this is due to a spurious or causal
relationship.
Study Design: A individual fixed effects
instrumental variable design will be applied to
observational data. The instrumental variable for
family income will exploit large expansion of the
Earned Income Tax Credit during the 1990's.
Population Studied: Data on the children of the
NLSY 1979 female respondents will be used.
Eleven waves of data have been collected on over
11,000 children. The study population will include
over 9,000 children with outcome data in at least
one time period between 1988-2006.
Principal Findings: Twelve percent of the children
in the sample had an activity limitation reported at
some point during the study period. In crosssectional models controlling for child, mother, and
other household characteristics, a $10,000 increase
in family income was associated with a -.36%
decrease in the probability of parents reporting a
activity limitation. After controlling for child fixed
effect, the association decreases and becomes
statistically insignificant. Using expansion in the
EITC to create an instrumental variable for reported
income reveals a positive causal effect of income
on the probability of reporting any activity limitation.
Specifically, a $10,000 increase in family income
implies a .8 percent increase in the likelihood of
reporting a activity limitation.
Conclusions: Family income is a causal
determinant of the way in which parents perceive
the limiting effect of their children's chronic
conditions.
Implications for Policy, Delivery or Practice:
There may be a disparity in the availability and
affordability of services to support children with
chronic health conditions in families with high
income versus low income. There needs to be
further description of this causal relationship to
understand the source of this disparity.
Funding Source(s): AHRQ
Poster Number: 95
The Transportation Disadvantage Index: Progress
Towards Developing a New Community Measure of
Child Health Access Barriers
Presented by: Stephen Borders, Ph.D., M.S.H.P.,
Associate Professor, School of Public, Nonprofit
and Health Administration, Grand Valley State
University, 401 West Fulton Street, Grand Rapids,
MI 49504; Phone: (616) 331-6569; Email:
borderss@gvsu.edu
Co-authors: Roy Grant, M.A.; Dennis Johnson,
M.A.
Research Objective: Transportation is a frequently
cited barrier to healthcare access, but researchers
have rarely evaluated the problem in depth. It is
generally recognized that having health insurance
alone does not assure access to the healthcare
system. Particularly in health professional shortage
areas, geographic problems (distance and
availability of transportation) may complicate
healthcare access. We sought to identify areas
where transportation barriers are particularly acute
for children by developing a Transportation
Disadvantaged Index.
Study Design: Using 2004-2006 hospital discharge
data from the Texas Department of State Health
Services (TDSHS), a multiple regression design
was used with ambulatory care sensitive conditions
(ACSC) admissions as a proxy for children’s access
to healthcare services. We used ordinary least
squares regression to identify predictors of nonemergent ACSCs by zip code in Travis County,
Texas and seven contiguous counties. Using
Geographical Information System (GIS), we
mapped the level and intensity of ACSCs by zip
code. Covariates included the ratio of automobiles
to population 16 and older, distance to the nearest
federally qualified health center (FQHC), percent of
working population 16 and older that takes public
transportation to work, urban or rural Census
Bureau designation of zip code and a rich array of
personal socioeconomic characteristics. We used
factor analysis to group socioeconomic factors into
meaningful categories, revealing three underlying
dimensions: 1) zip codes dominated by high levels
of minorities receiving public benefits in the form of
welfare and Medicaid (“TANF Families”), 2) zip
codes dominated by high poverty areas that were
most likely eligible but not enrolled for public
benefits and spend a high proportion of income on
housing (“Working Poor”) and 3) zip codes
dominated by single-female headed households
with high numbers of women in the workforce, less
likely to be minority (“Working Mothers”).
Population Studied: Using hospital discharge data
from TDSHS, we identified 15,862 emergency
department (ED) visits in the eight county area
attributable to children 17 years or younger.
Principal Findings: Of the 15,862 ED visits during
the study period by children ages birth-17, 2,883
(18.1%) were classified as non-emergent ACSCs.
Among children utilizing ED services for ACSCs,
there were strong inverse relationships by zip code
between ratio of automobiles (p<.013) and
percentage of Working Mothers (p<.05). ED use for
ACSCs was also associated with percentage of
working population taking public transportation to
work (p<.004) and percentage of population living in
a rural zip code (p<.028). Travel distances to
FQHCs were greatest in rural areas.
Conclusions: The results suggest that distance to
health care services (rural zip code) and availability
of personal transportation are associated with
greater utilization of ED services for ACSCs. This
suggests that transportation is a strong barrier to
timely access of child healthcare services.
Implications for Policy, Delivery or Practice:
Factors associated with transportation barriers to
child healthcare access can be identified and
mapped using hospital discharge data. Policy
makers can target these areas for interventions to
enhance the availability of non-emergency medical
transportation, deployment of mobile clinics and
additional primary care services (e.g., new FQHCs)
to improve child access to appropriate preventive
and primary healthcare services, potentially
improving outcomes and lowering the cost of care.
Funding Source(s): WKK, The Children's Health
Fund
Poster Number: 96
Clinical Characteristics Influence Parent-Reported
Quality of Life among Children with Attention Deficit
Hyperactivity Disorder
Presented by: Eugenia Chan, M.D., M.P.H.,
Assistant in Medicine, Division of Developmental
Medicine, Children's Hospital Boston, 300
Longwood Avenue, Boston, MA 02115; Phone:
(617) 355-3070; Email:
eugenia.chan@childrens.harvard.edu
Co-authors: Chao-Yu Guo, Ph.D.; Sara Toomey,
M.D., M.P.H.; Janine Molino, M.S.; Stephen Porter,
M.D., M.P.H.
Research Objective: Children with ADHD
experience poorer health-related quality of life
(HRQL) relative to typical peers, but the extent to
which HRQL is associated with key clinical
characteristics is unknown. We investigated the
relationship between parent-reported HRQL and
disease control, ADHD medication side effects, comorbid psychiatric conditions, and complexity of
ADHD care environment.
Study Design: We analyzed data collected during
a clinical trial on how parents shared information
about their child’s ADHD with medical providers.
Parents reported on ADHD symptoms, current
prescribed ADHD medications and side effects, and
HRQL. The normalized Psychosocial Summary
Score (range 0-100) from the 50-item Child Health
Questionnaire-Parent Form was the primary
outcome. Predictors included disease control status
based on NICHQ Vanderbilt symptom counts;
severity of ADHD medication side effects; results of
Vanderbilt screen for anxiety/depression,
oppositional defiant disorder, and conduct disorder;
and whether the primary manager of the child's
ADHD also prescribed the ADHD medications.
Multivariate linear regression analyses adjusted for
child age, gender, and parent race/ethnicity, and
adjusted beta estimates were re-transformed into
scaled HRQL scores. We considered a 5-point
difference in scaled HRQL score to be clinically
meaningful.
Population Studied: Parents of children age 5-12
with ADHD treated with prescription medication,
recruited from diverse communities in greater
Boston. Of 271 parents screened, 194/271 were
eligible and 182/194 were enrolled. We analyzed
177 parents with complete data.
Principal Findings: Overall, mean±SD
Psychosocial Summary Score was 37.5±12.
Parents reported their children to have poor disease
control (63%), more than minimal side effects
(71%), and positive co-morbidity screen (50%). In
both crude and adjusted analyses, higher HRQL
was significantly associated with good disease
control (adjusted beta 0.7, p<0.001), minimal
medication side effects (adjusted beta 0.36,
p<0.006), negative co-morbidity screen (adjusted
beta 0.67, p<0.001), and the same clinician
managing care and prescribing medications
(adjusted beta 0.33, p<0.03). Expected
improvement in scaled HRQL scores were +9.2 for
good (vs. poor) disease control, +8.8 for a negative
(vs. positive) comorbidity screen, +4.7 for minimal
(vs. more than minimal) side effects, and +4.3 for
the same (vs. different) manager as prescriber.
Conclusions: In this community sample, parents
reported significantly higher HRQL for their children
with ADHD when there was better symptom control,
fewer medication side effects, no symptoms of
psychiatric co-morbidities, and a less complex care
team structure to support disease management.
Implications for Policy, Delivery or Practice:
Optimizing care for children with ADHD may
substantially improve their HRQL. Strategies for
optimizing HRQL may include both improvements in
disease-specific management as well as a more
streamlined care team.
Funding Source(s): NLM
Poster Number: 97
Do Provider-focused Interventions Improve the
Prescribing of Daily Asthma Controller Medications
to Children?
Presented by: Janet Coffman, Ph.D., M.A.,
M.P.P., Assistant Professor, Philip R. Lee Institute
for Health Policy Studies, University of California,
San Francisco, 3333 California Street, Suite 265,
San Francisco, CA 94118; Phone: (415) 476-2435;
Email: Janet.Coffman@ucsf.edu
Co-authors: Michael D. Cabana, M.D., M.P.H.; Arpi
Bekmezian, M.D.; Christine Cho, M.D., M.P.H.;
Adam Hersh, M.D., Ph.D.; Megumi Okumura, M.D.
Research Objective: Since 1997 the National
Heart Lung and Blood Institute has recommended
that physicians prescribe daily controller medication
to all persons with persistent asthma to reduce the
frequency and severity of asthma exacerbations.
Previous studies suggest that physicians do not
consistently follow this guideline. Our objective was
to conduct a systematic review of studies of
interventions to increase the prescribing of daily
controller medication regimens to children with
persistent asthma.
Study Design: We conducted a systematic review
randomized controlled trials (RCTs) and cluster
RCTs that assessed the effect of interventions
aimed at improving the prescribing of asthma
controller medications. The primary outcome
assessed was the likelihood of prescribing an
asthma controller medication (e.g., an inhaled
corticosteroid). The secondary outcomes were
utilization of acute care services (e.g., emergency
department visits for asthma), and health outcomes
(e.g., days with asthma symptoms). We searched
MEDLINE, EMBASE, CINAHL, and the Cochrane
Central Register of Controlled Trials. Two teams of
three reviewers independently reviewed abstracts.
Differences among team members and between the
two teams were resolved by consensus.
Population Studied: Selection criteria included
publication in English and analysis of effects on the
prescribing of asthma controller medications to
children aged 0-17 years. We included interventions
delivered to individual clinicians with prescribing
authority, groups of clinicians, or health care
organizations. Because we were interested in
interventions that sought to directly influence
clinicians’ prescribing behavior, interventions
delivered only to patients or caregivers were
excluded.
Principal Findings: We screened 699 abstracts,
selected 98 articles for further review, and identified
nine articles that met the inclusion criteria. All
interventions had two or more components, which
included educational materials (n=8 studies),
conferences (n=6), educational outreach visits
(n=5), individualized treatment recommendations
for clinicians' patients (n=5), audit and feedback
(n=1), and local opinion leaders (n=1). Five of the
nine studies reported that the interventions were
associated with statistically significant increases in
the prescribing of daily controller medications.
Interventions that provided clinicians with treatment
recommendations tailored to the needs of individual
patients were more likely to improve prescribing
practices than interventions that provided general
information about recommended prescribing
practices. Of the six studies that examined the
effects of the intervention on use of acute care
services for asthma, only two reported a statistically
significant reduction in emergency department visits
and none reported a reduction in hospitalizations.
The two studies that evaluated the impact of the
intervention on health outcomes found few
statistically significant differences between the
intervention and control groups.
Conclusions: RCTs of interventions to increase
prescribing of asthma controller medications to
children with persistent asthma have reported
inconsistent effects on prescribing, acute care use,
and health outcomes. Interventions that provide
specific treatment recommendations for individual
patients are more likely to increase prescribing than
interventions that only provide general information.
Implications for Policy, Delivery or Practice:
Provider-focused interventions alone appear to be
insufficient to improve prescribing and health
outcomes for children with asthma. Multi-faceted
interventions which eM.P.H.asize individualized
treatment recommendations and which target
patients and caregivers as well as providers may be
necessary.
Funding Source(s): World Health Organization
Poster Number: 98
Resource Use for Children Undergoing Congenital
Heart Surgery in Pediatric Hospitals
Presented by: Jean Connor, D.N.Sc., R.N.,
Director of Nursing Research Cardiovascular
Program, Cardiovascular, Childrens Hospital
Boston, 300 Longwood Avenue, Boston, MA 02115;
Phone: (617) 355-8890; Email:
jean.connor@cardio.chboston.org
Co-authors: Kimberlee Gauvreau, Sc.D.
Research Objective: To describe overall resource
use and breakdown of charges by type of service
for patients undergoing congenital heart surgery in
free-standing pediatric hospitals.
Study Design: ICD-9-CM codes were used to
identify patient admissions undergoing congenital
heart surgery using the Pediatric Health Information
System (PHIS) database years 2007-2008.
Adjusted billed charges (charges billed in the
medical record adjusted by the wage/price index)
were abstracted for patients using PHIS CTC
codes. Aggregate charges overall and by type of
service were identified. Charges were further
stratified by Risk Adjustment for Congenital Heart
Surgery (RACHS-1) risk categories. Items within
each charge category were abstracted for further
description.
Population Studied: <18 years of age undergoing
congenital heart surgery in a pediatric institution
with an annual volume of 25 cases or more over 2
years. Patient admissions coded as heart
transplants and premature infants/neonates with
PDA closure as the only cardiac procedure were
excluded.
Principal Findings: The total aggregate billed
charges for the 19,565 admissions from 34
hospitals were $3,894,072,623. The median
adjusted billed charges per admission were
$114,721. Of the six sub-charge categories, “Other”
accounted for 45.4% of the aggregated charges
($1,766,633,414, median $52, 545 per admission,
with a maximum of $2,719,266). After “Other”, Lab
14.5%, Clinical 13.0%, Pharmacy 11.5%, Supply
9.8%, and Imaging 5.5% accounted for decreasing
percentages of the total aggregate charges. These
sub-charge proportions were similar within each of
the six RACHS-1 risk categories, with “Other”
accounting for 42.8%-47.2% of the adjusted billed
charges. Items with the highest aggregate charges
identified as “Other” included, “operating room
service”, “pediatric cardiac intensive care unit”,
“pediatric intensive care unit”, “neonatal intensive
care unit”, and “pediatric coronary care unit”.
Conclusions: Hospital care for children undergoing
congenital heart surgery generated significant billed
charges with a wide variation in range. Of note,
“Other” services including operating room and
intensive care unit services contributed to over 40%
of the billed hospital charges. The proportion of
“Other” billed charges remained relatively
unchanged regardless of RACHS-1 risk category.
Implications for Policy, Delivery or Practice:
Currently between 75% and 85% of health care
resources and costs are consumed by 10% of the
population. Children born with congenital heart
disease are included in this complex population of
patients. Care for these children in this study
generated significant billed charges with wide
variation. The magnitude of clinical complexity and
its impact on billed charges remains unclear.
Further examinations require focus on institutional
variation of overall charges and sub-charge
categories as well as examinations of institutional
variation in charge structures for specific area/ unit
services.
Poster Number: 99
A Comparison of Head Circumference Growth
Curves for Children from Birth to 3 Years
Presented by: Carrie Daymont, M.D., Fellow in
General Academic Pediatrics, Department of
Pediatrics, University of Pennsylvania/Children's
Hospital of Philadelphia, 3535 Market Street Suite
1550, Philadelphia, PA 19104; Phone: (267) 4260684; Email: daymont@email.chop.edu
Co-authors: Wei-Ting Hwang, Ph.D.; Chris
Feudtner, M.D., M.P.H., Ph.D.; David Rubin, M.D.
M.S.C.E.
Research Objective: To determine the validity of
the head circumference growth curves produced by
the U.S. Centers for Disease Control and the World
Health Organization in a large U.S. primary care
population.
Study Design: Retrospective cohort study. We
created smoothed head circumference curves from
our population and compared them to published
curves, specifically addressing the proportion of
observations above the 95th and below the 5th
percentile for each set of curves.
Population Studied: We analyzed 415,458 head
circumference observations from the electronic
medical records of 75,412 patients in a primary care
network that covers three mid-Atlantic states.
Principal Findings: Large differences exist
between the primary care network curves and the
CDC and WHO curves, especially at the 95th
percentile.
Conclusions: The overall proportion of subjects
above the 95th/below the 5th percentile was
5.0%/4.5% for primary care network, 6.1%/4.6% for
NCHS, 8.0%/3.1% for CDC, and 13.3%/3.1% for
WHO. The proportions varied by age. Using the
CDC curves, the proportion above the 95th
percentile increased from 0.23% for children under
two weeks of age to 12.0% for children 12 months
old. Using the WHO curves, the proportion above
the 95th percentile was over 5% at all ages with a
maximum of 18.0% for children above 24 months.
Implications for Policy, Delivery or Practice: Use
of the CDC and WHO head circumference curves
could result in clinically significant misclassifications
of children with larger heads. The direction of bias
has the potential to delay diagnosis of intracranial
pathology in children under two months of age
using the CDC curves and to cause excessive
evaluation of healthy children beyond six months of
age using either the CDC or WHO curves.
Funding Source(s): NIH-NRSA
Poster Number: 100
Understanding Fast Food Choices and the
Influence of Calorie Labeling on Children and
Adolescents
Presented by: Brian Elbel, Ph.D., M.P.H.,
Assistant Professor of Medicine and Health Policy,
Division of General Internal Medicine, New York
University, 423 East 23rd Street, 15-120N, New
York, NY 10010; Phone: (212) 263-4283; Email:
Brian.Elbel@nyumc.org
Co-authors: Joyce O. Gyamfi, M.S.
Research Objective: Obesity is an enormous
public health problem, and children have been
particularly highlighted for intervention. Particularly
of concern is children’s fast food consumption.
However, we know very little about how children or
their parents make fast food choices, including how
they react to calorie labeling. The goal of this study
was to determine the influence of mandatory, citywide calorie labeling on food choices of children
and adolescents in New York City and to better
understand their fast food choices.
Study Design: Natural Experiment. Survey and
receipt data were collected from New York City, and
Newark, NJ (as a comparison city). Study
restaurants included four of the largest chains
located in NYC and Newark: McDonald’s, Burger
King, Wendy’s, and Kentucky Fried Chicken (KFC).
Population Studied: Sample included 349 children
and adolescents aged 0-17 years who visited the
restaurants with their parents or alone before or
after labeling was introduced. Children and
adolescents who came to the restaurant with a
parent comprise of 69% of the study sample, with
31% visiting the restaurant alone. By design, 76%
of our sample was surveyed in NYC. Approximately
47% of the participants were males, with 54% in the
13-17 age groups. 65.9% of the participants
identified themselves as Black, 23.5% were Latino,
and the remaining 10.6% were of mixed race or
White.
Principal Findings: Among adolescents, 57% of
the sample in NYC and 18% in Newark saw calorie
labels post-labeling; although only 9% indicated that
the labels influenced their meal choice. The mean
calories purchased by children (via their parents)
and adolescents in the pre-labeling period were 684
for males and 612 for females; this did not
significantly differ in the post-labeling period for
NYC or Newark. Approximately 35% of adolescents
ate fast food six or more time per week.
Additionally, 72% of adolescents reported that taste
was the most important factor in their meal
selection. The majority of adolescents in NYC
underestimated or did not know the amount of
calories in their meals (70.1%); this was marginally
improved after the introduction of labeling (61.0%).
Almost all adolescents indicated they made their
fast food choices without the involvement or input of
the parent. However, about half indicated that at
home, their parents decide (24%) or they decide
jointly (16.8%).
Conclusions: In our relatively small study, while
most adolescent in low-income, urban areas saw
calorie labels, few indicated utilizing the information
in their food choice. Calories purchased were
unchanged for children and adolescents during our
study period. Adolescents tended to underestimate
the calories they purchased. Most adolescents
value taste most of all in their food choice. Finally,
parental involvement tends to play a larger role in
food choice at home than at fast food restaurants.
Implications for Policy, Delivery or Practice: In
our study, food choice of adolescents and parents’
choices for their children did not change as a result
of calorie labeling. This work highlights the need for
greater attention to food and obesity policies that
pay particular attention to children, adolescents,
and families, including how fast food choices are
made and can be influenced.
Funding Source(s): RWJF
Poster Number: 101
2009 NHPS Delaware Child Care Provider Survey:
Exploring the Relationship Between Awareness,
Policies, and Practices and Behaviors
Presented by: Tiho Enev, Ph.D., Senior Scientist,
Policy, Evaluation and Research, Nemours Health
and Prevention Services, 252 Chapamn Road,
Newark, DE 19711; Phone: (302) 444-9246; Email:
tenev@nemours.org
Co-authors: Alex Camacho, Ph.D.
Research Objective: The 2009 Delaware Child
Care Provider Survey is a critical component of a
systems-level evaluation model that studies the
relationship between a) the center director
awareness of nutrition and physical activity
guidelines, b) center policies, and c) staff practices
and behaviors in childcare settings. It provides data
to empirically test and validate the systematic
relationships between these three components.
Study Design: The 2009 Delaware Child Care
Provider Survey has a version for center directors
and for teachers. The Director’s Version consists of
four sections: 1) Center Demographics 2) Nutrition:
a)Awareness of Nutrition Standards b)Center
Policies Regarding: i. Nutrition Standards, ii.
Nutrition Education, Training, and Monitoring, iii.
Staff Behaviors, 3) Physical Activity (PA) and
Screen Time (ST): a) Awareness of PA and ST
Standards, b) Center Policies Regarding: i. PA and
ST Standards, ii. PA Education, Training, and
Monitoring, iii. Staff Behaviors, and 4) Directors’
perceptions on: a) Support from Administration,
Staff, and Parents, b) Available Resources, and c)
Monitoring of Childcare Quality. The Teacher’s
Version is based on a 1-to-1 mapping framework –
whereas questions addressing center policies in the
Director’s Version have corresponding practice or
behavior questions in the Teacher’s version. This
two-level survey design allows comprehensive
analysis of 1) the congruency between directors’
and teachers’ self-reports, 2) Relationships
between a) directors’ awareness, b) center policies,
and c) staff practices and behaviors, and 3) effect of
center and staff demographics, center resources,
training and monitoring on the above domains and
their dynamics. An innovative approach was used
to administer the survey to achieve better results.
Previous experience with data collection in child
care settings has shown that: a) it is difficult to
reach the teachers without going through the center
director, which might contaminate the information,
b) surveying only the center’s director about staff
practices and behaviors might lead to aggregation
and distortion of information, and c) traditional
approaches (mailing surveys) are not rewarding
and have to be combined with multiple telephone
follow-ups in order to achieve the desired response
rate. To overcome these limitations, the surveys
were administered to the center director and two
randomly selected teachers during a scheduled visit
at the center by a survey administrator.
Population Studied: A stratified random sample of
160 licensed childcare centers, representative of
Delaware’s population of 450 childcare centers
spread in all three counties.
Principal Findings: The preliminary findings from
the pilot study suggest a positive relationship
between the level of a) awareness, b) policy
implementation and c) staff practices and
behaviors. Comprehensive analyses will be
performed once the final data is available
(February, 2010).
Conclusions: Targeted multi-level surveys can be
used to study complex environments and provide
insights to inform the programmatic work of public
health organizations.
Implications for Policy, Delivery or Practice: The
results of the analysis provide critical insight into the
mechanisms behind NHPS’ programmatic work
within the child care sector. Specifically, results
assist the organization in understanding where to
concentrate programmatic work in the future by
determining the extent that awareness, policies,
and practices are interrelated.
Funding Source(s): RWJF
Poster Number: 102
Factors Associated with Myopia Treatment
Adherence for School Children in Taiwan
Presented by: Yi-Ting Fang, Student, Institute of
Health & Welfare Policy, School of Medicine,
National Yang-Ming University, 155 Sec. 2 Li-Nong
Street Peitou, Taipei. Taiwan ROC, Taipei, Taiwan
11221; Phone: +886228267017; Email:
yt620601@yahoo.com.tw
Co-authors: Pu Christy, Ph.D.; Huang Nicole,
Ph.D.; Ginger Liu, M.S.; Chou Yiing-Jenq, Ph.D.;
Chou Pesus, Dr.PH.
Research Objective: To determine the factors
associated with myopia treatment adherence by
using a national representative longitudinal
database. To measure the impact of Atropine
concentration on myopia treatment adherence
Study Design: The National Health Insurance
(NHI) claim data (1999-2005) was used to conduct
a historical cohort study for the population in
Taiwan. We choose the newly Atropine users in
2001 for treating their myopia, and followed for 4
years. Treatment effect model was used to correct
the endogeneity between dependent and
independent variables in this analysis. The
dependent variable was Atropine adherence, which
was defined by Medication Possession Ratio
(MPR). Good adherence was defined as MPR
>=80%, bad adherence as MPR<20%. The
independent variables including age, gender,
school grade, residential area, urbanicity, mydriatic
drug and Atropine concentration (1%, 0.5%, 0.3%,
0.1%).
Population Studied: National health insurance
claim data is a representative sample (1/20) of all
medical claims in Taiwan. We choose those
between 4~15 years of age in 2001, which included
14,475 school children.
Principal Findings: Mean MPR of all mydriatic
eyedrop decreased substantially over time, with the
trend being 115.6%, 74.0%, 42.1%, 29.7%, &
22.5% in 0.5 yr, 1yr, 2yr, 3yr & 4yr follow up.
Proportion of good atropine adherence
(MPR>=80%) was 54.3%, 34.8%, 12.0%, 3.6% &
1.1% in 0.5 yr, 1yr, 2yr, 3yr & 4yr. Factors
associated with lower MPR including school grade
(grade 6-9, p<0.05 respectively), residential area
(northern area, p=0.019), urbanicity (combination
area, p<0.0001), most common used mydriatic as
tropicamide (p<0.0001) and higher Atropine
concentration, but not gender (p=0.942). Results
from the treatment effect model revealed that lower
concentration atropine eyedrop such as 0.5%
atropine (p<0.0001)or tropicamide(p<0.0001) as the
most common mydriatics eyedrop predicted better
treatment adherence compared with using 1%
atropine.
Conclusions: The myopia treatment adherence of
school children was dropping rapidly from 115.6%
in 0.5 year to 22.5% in 4 year post first prescription.
The adherence to Atropine was even worse (54.3%
in 0.5 year and 1.1% in 4 years). School grade,
residential area, urbanicity and different mydriatic
drug are key factors associated with MPR. Lower
concentration and short acting mydriatic eyedrop
showed a better treatment adherence, while less
photophobia and near work difficulty was
associated.
Implications for Policy, Delivery or Practice: The
prevalence of myopia was rising rapidly in Taiwan,
Hong Kong and Singapore, its 60-80% in young
adults. The prevalence of high myopia was rising as
well (0.2% in 1983 and 3.4% in 2000). High myopia
was associated with increasing risk of cataract,
glaucoma and retinal diseases. There are evidence
that regular Atropine eyedrop application (usually
use every night before sleep) can halve myopia
progression. Good control of myopia progression
depends heavily on mydriatic eyedrop adherence.
Ophthalmologists may consider changing the
prescription of mydriatic eyedrop to enhance
myopia treatment effect.
Poster Number: 103
Bed Usage in the Pediatric Intensive Care Unit
Presented by: Evan Fieldston, M.D., M.B.A.,
M.S.H.P., Robert Wood Johnson Clinical Scholar,
RWJ Clinical Scholars Program and Pediatrics,
University of Pennsylvania School of Medicine, 423
Guardian Drive, 1303A Blockley Hall, Philadelphia,
PA 19104; Phone: (215) 573-2585; Email:
fieldsto@upenn.edu
Co-authors: Christian Terwiesch, Ph.D.; Joshua
Metlay, M.D., Ph.D.; PICU Research Group
Research Objective: Use real-time observation to
describe utilization, flow through, and bottlenecks of
pediatric intensive care unit (PICU) beds at a large,
urban, academic children's hospital.
Study Design: A recording tool was developed
through clinical and operational expert input and
through iterative pilot use. It was then used to
collect >20,000 bed-hours of real-time observation
of patient flow in the pediatric intensive care unit
(PICU) over 5 non-consecutive weeks.
Population Studied: Twenty-four beds in a PICU
at a children's hospital.
Principal Findings: 82% of bed-hours were used
for critical-care value-added services, 8% on
logistics, 9% on non-value-added uses, and 1%
unclear. 95% of all bed-hours were consumed on
14 activities. There were wide variations in timing
on certain logistical and non-value-added services,
suggesting that bottlenecks can be identified and
improved to maximize functional capacity.
Conclusions: Much time in the PICU was spent
caring for patients at high levels of intensity.
Identifiable periods of logistical and non-valueadded occupancy were identifiable, as well as
bottlenecks. Process improvement work can be
directed at areas of delay uncovered by this
systematic frontline analysis, to maximize functional
capacity, which may improve quality (safety,
efficiency, effectiveness, timeliness, patientcenteredness, equity), access, staff satisfaction,
education, and value.
Implications for Policy, Delivery or Practice:
Pediatric Intensive Care Units (PICU), with limited
number of beds and resource-intensive services,
are a key component of flow and an important place
for optimization. PICUs are crossroads for many
patients and delays in moving them in or out can
negatively impact clinical status and may decrease
overall efficiency. Delays in moving patients out of
the PICU, when no longer in need of critical care
services, may mean that value-added services are
being denied to other patients -- a form of waste.
This study demonstrates a methodology to evaluate
micro-system efficiency, determine utilization of
resources, and identify bottlenecks for
improvement.
Funding Source(s): Leonard Davis Institute of
Health Economics Working Group Grant
Poster Number: 104
Addressing Children's Hospital Crowding by
Smoothing Occupancy
Presented by: Evan Fieldston, M.D., M.B.A.,
M.S.H.P., Robert Wood Johnson Clinical Scholar,
RWJ Clinical Scholars Program and Pediatrics,
University of Pennsylvania School of Medicine, 423
Guardian Drive, 1303A Blockley Hall, Philadelphia,
PA 19104; Phone: (215) 573-2585; Email:
fieldsto@upenn.edu
Co-authors: Matthew Hall, Ph.D.; Samir Shah,
M.D., M.S.C.E.; Marion Sills, M.D., M.P.H.; Anthony
Slonim, M.D., Dr.PH.; Angela Myers, M.D., M.P.H.;
Courtney Cannon; Susmita Pati, M.D., M.P.H.
Research Objective: To quantify the effect of
smoothing on crowding at children's hospitals.
Study Design: Daily inpatient census &
standardized length-of-stay ratios (SLOSR) were
calculated for 39 free-standing, tertiary-care
children's hospitals in the Pediatric Health
Information System (PHIS) for 2007. Midnight
census was used to calculate daily occupancy. A
retrospective smoothing algorithm set each
hospital-day's occupancy to the hospital's mean
occupancy for each week. Post-smoothing
reductions in weekly maximum occupancy were
averaged for the entire year for each hospital. We
determined the number of patients beyond 85%,
90%, 95%, and 100% occupancy levels pre- and
post-smoothing, and the change in the number of
hospitals, patient days, and patients exposed to
these levels of occupancy.
Population Studied: 39 free-standing children's
hospitals.
Principal Findings: Hospitals' mean occupancy
was 70.9%-108.1% on weekdays and 65.7%-94.9%
on weekends . Weekday occupancy exceeded
weekend occupancy (median difference 8.2%points (IQR: 7.2-9.5)). Mean post-smoothing
reduction in weekly maximum occupancy across all
hospitals was 6.6%-points (IQR: 6.2-7.4). By
smoothing, 39,607 patients from the 39 hospitals
were removed from exposure to occupancy levels
>95%.
Conclusions: Smoothing occupancy over a week
can provide hospitals with a useful strategy to
reduce crowding.
Implications for Policy, Delivery or Practice:
High levels of hospital occupancy are associated
with poorer patient outcomes, decreased access,
and increased provider stress. Hospitals respond to
crowding in a number of ways, but often focus on
reducing length of stay (LOS). For hospitals already
functioning at reasonable levels of efficiency,
additional efforts to reduce LOS may not sufficiently
increase functional capacity. In children's hospitals,
median LOS is 2-3 days, so throughput
improvement may be limited in reducing crowding.
Managing scheduled admissions through
smoothing has been proposed as an alternative
strategy to reduce variability in daily occupancy and
thereby reduce weekday crowding. Our study
confirms that smoothing occupancy would reduce
the risk of crowding and the number of patients
exposed to high-occupancy conditions.
Funding Source(s): RWJF
Poster Number: 105
Patterns of Patient Flow at Children's Hospital
Presented by: Evan Fieldston, M.D., M.B.A.,
M.S.H.P., Robert Wood Johnson Clinical Scholar,
RWJ Clinical Scholars Program and Pediatrics,
University of Pennsylvania School of Medicine, 423
Guardian Drive, 1303A Blockley Hall, Philadelphia,
PA 19104; Phone: (215) 573-2585; Email:
fieldsto@upenn.edu
Lead Author: Meera Ragavan
Co-authors: Bhuvaneswari Jayaraman, M.P.H.;
Evan Fieldston, M.D., M.B.A., M.S.H.P.
Research Objective: Determine how
admission/discharge and occupancy patterns vary
over time at a large children's hospital.
Study Design: Census data for a fiscal year were
abstracted: date/time of admission/discharge,
emergent/scheduled, length of stay (LOS). Hourly
census was calculated with a SAS macro; peak
daily census were extracted. For pts with LOS 7
days, pt-hours-generated by admit day = avg LOS
(hrs) for each day x number of admissions.
Coefficient of variation (CV; std dev/mean)
assessed variability.
Population Studied: All pediatric patients
hospitalized in one fiscal year at a large, urban,
academic children's hospital.
Principal Findings: Among 22,310 admits (mean
1859/month, SD 150, CV 8.1%), 22% were
scheduled. CVs on emergent & scheduled admits
per month were equal (9.8% vs. 9.7%). May, July,
August had >1 SD fewer emergent admits than
mean, while June had >1 SD more scheduled
admits. By day of week, emergent admits had CV
12.0%, while scheduled admits had CV 65.3%,
driven by lack of weekend admits (and fewer on
Thurs-Fri). For LOS 7 days (84% pts), each day
generated mean 154,248 pt-hours of business (SD
33,816; CV 21.9%). Emergent pt-hours generated
by admit day CV 9.6%, while scheduled CV 15.7%.
Mondays generated 25.2% of all scheduled pthours; Tues, Wed 20% each. Median LOS for
scheduled patients admitted on Mon was 2.1 days,
while 1.2-1.3 on other weekdays. Wed and Thurs
had highest peak census (1 SD above mean
census).
Conclusions: Scheduled admissions are a major
contributor to variability in occupancy and risk of
mid-week crowding. Predictable patterns lead to
high occupancy on some days and unused capacity
on others. Hospitals interested in optimization,
value, and avoiding crowding should assess their
patterns. Further studies should link variability to
outcomes (quality, access, finance, education, and
staff satisfaction).
Implications for Policy, Delivery or Practice:
Optimization of patient flow is crucial to improving
quality and value in healthcare, but variability
comes from many sources, including artificial
variability on scheduling patients to enter hospitals.
This study confirms that artificial variability can be a
significant challenge (and opportunity) for hospitals
that face days of high occupancy.
Funding Source(s): RWJF
Poster Number: 106
Delayed and Forgone Care Due to Cost for
Families with Chronic Conditions in High-Deductible
Health Plans
Presented by: Alison Galbraith, M.D., M.P.H.,
Assistant Professor, Department of Population
Medicine, Harvard Medical School and Harvard
Pilgrim Health Care Institute, 133 Brookline
Avenue, 6th Floor, Boston, MA 02215; Phone:
(617) 509-9893; Email:
alison_galbraith@hms.harvard.edu
Co-authors: Dennis Ross-Degnan, Sc.D.; Stephen
B. Soumerai, Sc.D.; Meredith B. Rosenthal, Ph.D.;
Charlene Gay, B.A.; Tracy A. Lieu, M.D., M.P.H.
Research Objective: Cost sharing is known to
reduce health care use, leading to concerns about
high-deductible health plans (HDHPs). Few studies
of HDHPs have evaluated families. Our objectives
were to (1) evaluate the probability of delayed or
forgone care due to cost in families with chronic
conditions in HDHPs compared with traditional
health plans, and (2) compare the effects of HDHPs
on health care use by adults and children within
families.
Study Design: We conducted a cross-sectional
phone and mail survey between April and
December 2008. The primary outcome was report
of any delayed or forgone care (including acute
care, emergency department visits, chronic disease
care, check-ups, or tests) due to cost in the prior 12
months. We collected data for adults and children
separately, conducted family-level analyses using
multivariate logistic regression, and calculated
predicted probabilities of delayed or forgone care
due to cost using the mean or modal values for covariates in the model.
Population Studied: We surveyed families with a
child <18 years old insured by a large
Massachusetts health plan. We identified families
enrolled for the prior 12 months with at least one
member with an ICD-9 code for a chronic condition.
Of these, we selected all families in HDHPs (with
annual family deductibles >$1000) and twice as
many families in traditional plans without
deductibles.
Principal Findings: We surveyed 297 families in
HDHPs (median family deductible = $2000) and
523 families in traditional plans (response rate
46%). In adjusted analyses, families in HDHPs
were significantly more likely than those in
traditional plans to report delayed/forgone care due
to cost overall for any family member (24.2% vs.
9.3%, respectively; p< 0.001) and for any adult in
the family (17.5% vs. 7.8%, p< 0.001), but not for
children (2.1% vs. 1.1%; p=0.23).
Conclusions: Delayed or forgone care due to cost
is significantly more likely for families with chronic
conditions in HDHPs compared to traditional plans.
In particular, adult members of families enrolled in
HDHPs appear to be at increased risk compared to
those in traditional plans. For children, the
prevalence of delayed or forgone care due to cost is
low and not significantly different within families in
HDHPs and traditional plans.
Implications for Policy, Delivery or Practice: A
differential response to increased cost-sharing in
HDHPs may exist for adults and children within a
family. The clinical significance of reported delayed
or foregone care merits further investigation.
Funding Source(s): RWJF
Poster Number: 107
Stemming the Increase of Childhood Overweight
and Obesity in Delaware
Presented by: Allison Gertel-Rosenberg, M.S.,
Senior Policy and Program Analyst, Office of VP of
Policy and Prevention, Nemours, 252 Chapman
Road, Christiana Building, Suite 200, Newark, DE
19702; Phone: (302) 444-9171; Email:
agrosenb@nemours.org
Co-authors: Debbie I. Chang, M.P.H.; Vonna
Drayton, Ph.D.
Research Objective: The evaluation assesses the
population-level longitudinal impact of Delaware’s
multi-sector healthy eating and physical activity
interventions on children’s knowledge, attitudes,
behaviors and health outcomes. The evaluation
also explores changes to the system-level
environments where children live, learn and play.
Study Design: A quasi-experimental evaluation
design, documenting and measuring specific
indicators (related to knowledge, attitude and
behavior changes), assesses the aggregate impact
of policy and practice changes in the child’s
environment and on population behavior over time.
The design is efficient for longitudinal research
involving longer time periods, which are necessary
for demonstrating population level health outcome
changes. This evaluation also captures the degree
of policy, practice, and system changes occurring at
the state and local levels that are likely to facilitate,
and sustain, population-level changes in the long
term.
Population Studied: The population studied is a
representative sample of Delaware children and
related systems and environments.
Principal Findings: Results from the 2008
Delaware survey suggest that the prevalence of
overweight and obesity for Delaware children ages
2-17 years has not changed significantly since
measured in 2006. Specifically, no statistically
significant (p=0.09) differences in BMI categories
were observed between 2006 and 2008.
Furthermore, no statistically significant changes in
prevalence of overweight and obesity were
observed within geographic and demographic
categories between 2006 and 2008. For example,
the prevalence in Sussex County (41.6% in 2006
versus 39.4% in 2008) and the prevalence among
non-Hispanic Whites (35.9% in 2006 versus 37.5%
in 2008) were statistically unchanged. There are,
however, clear demographic differences within
years. Compared to non-Hispanic Whites, nonHispanic Blacks had significantly higher prevalence
of overweight and obesity in 2006 (p=0.019), and
also in 2008 (p=0.003). Overall, there was a fourfold increase (from 5% to 19%) between 2006 and
2008 in Delaware households’ awareness of the 52-1-Almost None message. When there was
parental message awareness, significantly more
children age birth through 17 years engaged in one
hour of physical activity per day (10% vs. 26%,
p=0.000) and moderate to vigorous physical activity
for more than 20 minutes (21% vs. 33%, p=0.022)
between 2006 and 2008.
Conclusions: Early results show that the initiative
has stemmed the increase in the prevalence of
childhood overweight and obesity with findings that
prevalence is statistically unchanged for children
since 2006. Nemours’ initiative has also spurred
increased knowledge of healthy eating and
increased physical activity in school and care
settings, resulting in policy and practice changes.
This model, of focusing on population health and
changing policies and practices in multiple settings
through strategic partnerships, knowledge
mobilization and social marketing, can effectively
address obesity and other complex problems facing
children today.
Implications for Policy, Delivery or Practice: The
data indicate that changes are being implemented
in systems – schools, child care, primary care –
with associated changes in healthy eating and
physical activity behaviors, and ultimately health
outcomes. The results of this initial evaluation
provide evidence for the efficacy of a
comprehensive prevention-oriented model.
Investments have proved valuable in many ways,
including establishing community collaboration and
capacity to address other childhood issues and
leveraging resources from multiple sources to affect
children’s health.
Funding Source(s): Nemours
Poster Number: 108
Healthy Kids, Healthy Future: Health Promotion in
Early Care and Education
Presented by: Allison Gertel-Rosenberg, M.S.,
Senior Policy and Program Analyst, Office of the VP
of Policy and Prevention, Nemours, 252 Chapman
Road, Christiana Building, Suite 200, Newark, DE
19702; Phone: (302) 444-9171; Email:
agrosenb@nemours.org
Lead Author: Debbie Chang, M.P.H.
Co-authors: Anne De Biasi
Research Objective: Combating childhood obesity
in early care and education settings is an untapped
opportunity. Current national and state efforts tend
to focus on improving wellness and preventing
obesity in elementary and secondary schools, often
overlooking the need for wellness policies in the
early care and education system. Research,
however, supports the need to provide healthy
nutrition and physical activity environments for
children at young ages. Therefore identification,
promotion and evaluation of policies and practices
that foster healthy environments must become a
high priority.
Study Design: Scattered efforts around the country
aim to improve nutrition and increase physical
activity in early care and education settings, but
greater collaboration and dissemination of these
efforts is needed. Recognizing the need for a multidisciplinary approach to child health promotion and
obesity prevention, a conference to advance
research, policy and practice in the areas of child
care and obesity prevention was initiated. Three
explicit goals of the conference were identified.
Identify promising practices and tools across the
nation that are addressing the current childhood
obesity epidemic; Identify current promising policies
as well as policy opportunities at the state and
federal levels that would support obesity
prevention/health promotion in child care settings;
Identify research gaps in the field.
Population Studied: The conference was used to
elicit multiple stakeholders’ views on the issues and
opportunities pertaining to childhood obesity in early
child care settings.
Principal Findings: Many participants at the
conference have been working to develop, support
and implement children’s health promotion and
obesity prevention polices at the local, state and
federal levels. This conference was one of the first
opportunities for stakeholders across disciplines to
share efforts that have been successfully
implemented in their communities. The nation’s
leaders discussed their expertise in policies,
practices and tools in topics including the
importance of oral health, limited screen time and
breastfeeding, as well as general nutrition and
physical activity wellness strategies.
Conclusions: As a next step to continue advancing
the field, leaders have decided to form the Healthy
Kids, Healthy Future Steering Committee, intended
to support continued collaboration among the field’s
top leaders. The committee will follow up on the
many ideas generated at the conference to move
the field forward; will continue sharing best
practices and policies to encourage dissemination
and adoption in the states; and will partner on other
innovative strategies to improve children’s health in
early care and settings, including policy change and
research.
Implications for Policy, Delivery or Practice:
Bringing together the leaders and experts in the
early care and education, nutrition, physical activity,
oral health, and child health fields, who in the past
have generally worked in silos, with the purpose of
discussing promising practices and tools, policy
opportunities, and gaps in the evidence base
propelled the field forward. Supporting collaboration
of efforts, thereby breaking down these silos, allows
for a larger scale more effective outcome. These
combined efforts, to improve nutrition and physical
activity environments in early care and education
settings will help create a lifetime of healthy habits
for children in the United States.
Funding Source(s): Nemours
Poster Number: 109
Nationwide Trends in Inpatient Pediatric Cardiac
Interventional Procedures from 1997-2007: On-Line
Estimates from the Healthcare Cost and Utilization
Project (HCUP) of the Agency for Healthcare
Research and Quality
Presented by: Darryl Gray, M.D., Sc.D., Medical
Officer, Center for Quality Improvement and Patient
Safety, Agency for Healthcare Research and
Quality, 540 Gaither Road, Rockville, MD 20850;
Phone: (301) 427-1326; Email:
darryl.gray@ahrq.hhs.gov
Co-authors: Kamal Pourmoghadam, M.D.; Alan
Hsu, B.A.; Jennifer Moore, M.P.H.; Claudia Steiner,
M.D., M.P.H.; Marshall Jacobs, M.D.
Research Objective: Pediatric heart disease
(including congenital and acquired lesions of the
heart/great vessels) affects 1-2 million Americans of
all ages. However, little is known about overall
frequencies or outcomes of procedures used to
treat these conditions. We used aggregate
administrative data to examine US care patterns
and outcomes associated with inpatient pediatric
cardiac interventional procedures.
Study Design: This retrospective descriptive study
used estimates from HCUP’s annual Nationwide
Inpatient Sample (NIS) and its triennial Kids’
Inpatient Database (KID). The NIS captures
administrative data for all discharges from a
weighted sample of ~20% of US community
hospitals. The KID reflects weighted samples ~10%
of normal newborn discharges and ~80% of other
pediatric discharges. We identified International
Classification of Diseases-Clinical Modification
(Ninth Revision) codes for therapeutic (rather than
purely diagnostic) cardiac procedures performed via
transcatheter or surgical approaches. We queried
HCUP’s on-line aggregate data
(http://hcupnet.ahrq.gov) to estimate volumes,
mean length of stay (MLOS), billed hospital
charges, payer mix and inpatient mortality for 0-17
year-olds discharged with these principal procedure
codes. Hospital charges, which generally exceed
costs but do not include physician fees, were
expressed in 2007 dollars using Producer Price
Index ratios. We used Census data to generate
population-based rates.
Population Studied: Weighted, nationally
representative estimates of US discharges
Principal Findings: NIS procedure discharge rates
for 0-17 year-olds in 1997 and 2007 were 37.5
/100,000 and 36.4/100,000 respectively. However,
yearly rates fluctuated considerably (e.g., 38.1,
61.0, and 39.3/100,000 in 2004, 2005 and 2006
respectively). KID’s figures were more stable. For
an estimated 30,816 discharges (43.5/100,000) in
1997, inpatient mortality=4.6%, MLOS=12.8 days,
aggregate charges=$3.3 billion. Payers included
private insurance (56.5%; 52.6% of charges) and
Medicaid (33.2%; 37.4% of charges); 3.1% were
uninsured. KID rates for 2000 and 2003 were 41.1
and 42.5/100,000, while NIS rates were 32.8 and
26.5/100,000 respectively. The 2006 KID estimated
that there were 34,434 discharges (46.7/100,000).
Mortality=3.4%, MLOS=22.1 days, aggregate
charges=$5.9 billion. While private insurance
covered 48.0% (44.3% of charges), Medicaid cases
(42.8%) generated 47.2% of charges; 2.3% were
uninsured. The 2006 KID estimated that there were
19,531 discharges (472.9/100,000) of infants.
Mortality=5.1%, MLOS=33.7 days, mean
charges=$230,743/case. In contrast, among 14,903
discharges of 1-17 year-olds (21.4/100,000),
mortality=1.1%, MLOS=6.1 days, mean
charges=$94,907/case. These findings mirror
patterns seen in earlier KID and NIS data.
Conclusions: Sampling considerations and other
factors make the KID a more robust source than the
NIS for inpatient pediatric cardiac procedure data.
KID procedure volumes and rates rose modestly
from 1997 to 2006; unadjusted mortality declined.
However, as measures of resource use (and
indirect measures of treatment intensity), MLOS
and inflation-adjusted charges nearly doubled, and
Medicaid payments increased. Infants had
considerably higher procedure rates, mortality,
MLOS and charges than did 1-17 year-olds.
Implications for Policy, Delivery or Practice: The
considerable discharge volumes, resource use and
mortality associated with pediatric cardiac
procedures (especially for infants) indicate the
societal importance of these interventions. These
findings, along with our observations of growing
Medicaid involvement and of possible underrepresentation of uninsured patients (~11-15% of
US children) reflect HCUPnet’s useful aggregate
data. They argue for more population-based
encounter-level research on patterns and outcomes
of pediatric cardiac procedures.
Poster Number: 110
Oral Health in Rural South Carolina: The
Importance of the Relationship Between School
Nurses and the Dental Community
Presented by: Jeff Hatala, M.M.C., M.B.A.,
Graduate Research Associate, SC Rural Health
Research Center, University of South Carolina, 220
Stoneridge Drive, Suite 204, Columbia, SC 29210;
Phone: (803) 251-6317; Email:
jhatala31@gmail.com
Lead Author: Amy Brock-Martin, Dr.PH.
Co-authors: Kyle Shah; Janice Probst, Ph.D.
Research Objective: To understand the
challenges faced by school nurses in getting their
students dental care, and to see if rural school
nurses have greater challenges than urban school
nurses.
Study Design: Electronic survey was used to
gather the data from school nurses. Electronic list of
nurses is maintained by SC Dept. of Health and
Environmental Control. Urban Influence Code was
matched with the zip code of the school to
determine which schools were rural.
Population Studied: school nurses in South
Carolina (n=900)
Principal Findings: Over 85% of school nurses
responded to the survey. Rural school nurses face
more challenges than urban school nurses. When
examining the challenges faced by school nurses
with rural location, we find that lack of parent
involvement is the only statistically significant
finding (p <.05). Via logistic regression, we further
investigated the relationship between lack of parent
involvement and rural location; we found rural
school nurses, when compared with urban school
nurses, are almost 1.7 times more likely to face the
challenge of lack of parent involvement.
Conclusions: The challenges that school nurses
reported they faced are not driven by the dental
community, but by the parents of the children
needing oral health services. The key to providing
successful oral health care to school-age children
lies in parental involvement and support for their
children.
Implications for Policy, Delivery or Practice: As
many of these parents may be unmarried parents
with hourly jobs, taking time from work may reduce
the family income. One possible recommendation is
for dentists in rural areas to periodically offer
extended hours during non-traditional work time or
some weekend hours to allow those children to
receive dental care. Nurses, schools and the dental
community can leverage their positive partnership
to develop ways that further encourage parents to
support their children. Future research may want to
examine perceptions of school nurses from other
states as well.
Funding Source(s): American Dental Association
and the South Carolina Dental Association (SCDA)
Poster Number: 111
Is Emergency Department Utilization Associated
with Primary Care Utilization among Low-Income
Children?
Presented by: Margaret Holland, Ph.D., M.P.H.,
postdoctoral fellow, Department of Pediatrics &
School of Nursing, University of Rochester, 601
Elmwood Avenue, Box SON, Rochester, NY 14618;
Phone: (585) 273-2054; Email:
margaret_holland@urmc.rochester.edu
Co-authors: Byung-Kwang Yoo, M.D., Ph.D.;
Harriet Kitzman, Ph.D.; Peter Szilagyi, M.D.,
M.P.H.; Linda Chaudron, M.D., M.S.; Helena
Temkin-Greener, Ph.D., M.P.H.
Research Objective: Experts are concerned about
high use of emergency departments (ED), resulting
in elevated healthcare costs and frequent
overcrowding of EDs. Low-income children are
known to use the ED more than higher-income
children. There is some evidence that increased
community-level primary care capacity may
decrease ED use. However, few studies have
examined the relationship between ED and primary
care use among low-income children. We examined
if primary care use (well-child care (WCC) and sick
visits) is associated with overall ED visits and with
potentially avoidable ED visits among low-income
children.
Study Design: Ordered logistic regression was
utilized to determine if primary care use was
associated with each of 5 ED visit categories: all ED
visits, injury/ingestion only, reasons other than
injury/ingestion, ambulatory-care sensitive (ACS)
conditions (may be avoidable with timely primary
care), and non-ACS conditions. All models
controlled for children's chronic conditions,
birthweight, and demographics. Medical records for
children were collected from birth to 24 months.
Interviews at the mother’s enrollment (during
pregnancy) and at 12 and 24 months after birth
provided demographic information.
Population Studied: 433 low-income, urban
mother-child pairs from the Nurse-Family
Partnership MeM.P.H.is trial control group
(observed without intervention). Primiparous
women were recruited from an obstetrical clinic and
met 2 of 3 high-risk criteria: unmarried,
unemployed, or less than 12 years of education.
Principal Findings: On average, children had 2.7
ED visits, 10.1 sick visits, and 5.0 of 9
recommended WCC visits over 2 years. Twelve
percent of ED visits were for injury/ingestion; 49
percent of ED visits were for ACS conditions. More
sick visits to primary care were associated with
more ED visits when all ED visits was the outcome
(OR: 1.18, p<0.001). This association held for all
other subsets of ED visits, although the magnitude
was significantly lower for injury/ingestion visits only
(OR: 1.03, p<0.04) and non-ACS visits (OR: 1.11,
p<0.001). WCC visits were not associated with ED
utilization. These results did not change when WCC
and sick visits were estimated together in the same
models.
Conclusions: Sick visits to primary care were
associated with increased ED use among
individuals in this population of low-income children,
even when controlling for chronic conditions. Wellchild care visits were not associated with ED use.
Unlike in other populations, in which greater access
to primary care visits are associated with fewer ED
visits, in this low-income population more primary
care visits for illness are associated with more ED
visits.
Implications for Policy, Delivery or Practice: If
these findings are replicated, it may be that among
low-income children primary care visits do not deter
ED visits. Efforts to lower healthcare costs may
need to focus on other areas.
Funding Source(s): AHRQ
Poster Number: 112
How Well Does the Children with Special Health
Care Needs Screener Identify Children with
Disabilities?
Presented by: Amy Houtrow, M.D., M.P.H.,
Assistant Professor, Pediatrics, University of
California, San Francisco, 500 Parnassus, San
Francisco, CA 94143; Phone: (415) 476-3899;
Email: houtrowa@peds.ucsf.edu
Co-authors: Joan Hilton, Ph.D.; Megumi Okumura,
M.D.
Research Objective: Children with disabilities are
amongst the most vulnerable members of society. It
is a public health goal to improve the health and
quality of life of children with disabilities. A
screening instrument that identifies children with
disabilities would be of benefit to the health services
research community. The Children with Special
Health Care Needs (CSHCN) Screener is used in
multiple national surveys to identify CSHCN and
could be a valuable tool to identify those with
disabilities. At this time, there are no studies
validating its use as a screening tool for disabilities.
Therefore, our objective is to determine the
sensitivity, specificity, positive predictive value
(PPV) and negative predictive value (NPV) of the
CSHCN Screener for CSHCN with disabilities.
Study Design: We used the National Survey of
Children with Special Health Care Needs 2005-06
(N= 40,723). CSHCN with disabilities could
potentially be identified in the Screener by the set of
questions used to determine which children have
conditions that last at least one year and are
associated with limitations in their ability to do the
things most children of the same age can do. In the
Health and Functional Status of the Survey, seven
of the 14 ‘difficulty’ questions are relevant to
disability – difficulties with mobility/coordination,
self-care, fine motor skills, learning, communicating,
vision despite correction, and hearing despite
hearing aids. We used these questions as our
markers of disability in order to test the sensitivity,
specificity, positive and negative predictive value of
the Screener.
Population Studied: Nationally representative
sample of CSHCN aged 0-17 years in the United
States.
Principal Findings: Using our seven disability
questions, 49.3% of CSHCN were identified as
having at least one disability. Cognitive disabilities
were identified in 43.4% of CSHCN, while physical
disabilities were identified in only 22.8%. In
contrast, 21.4% of respondents identified their child
as having a disability on the Screener. The
Screener’s PPV was 82.1%, the NPV was 59.6%,
the sensitivity was 35.7% and the specificity was
92.4%. While nearly 60% of respondents with a
child who had mobility, self care or fine motor
disabilities identified their child has disabled on the
Screener, only one third of children with cognitive
disabilities were identified by the Screener as
having a disability.
Conclusions: The CSCHN Screener does not
identify all CSHCN with disabilities.
Implications for Policy, Delivery or Practice: If
the Screener is used to estimate the prevalence of
disability in childhood, millions of children with
disabilities will not be identified. This will lead to low
estimates of disability prevalence that could have
significant policy implications. As children with
disabilities require more health and related services
than other children and face significant social
challenges and barriers to their well-being, it is
imperative to accurately identify these children in
order to provide needed services and funding.
Researcher utilizing the CSHCN Screener should
consider using other measures of disability
available in national datasets when studying
children with disabilities.
Funding Source(s): AHRQ, NIH-NICHD
Poster Number: 113
Enhancing Pediatric Psychopharmacology for
ADHD in North Carolina’s Medicaid Program
Presented by: Charles Humble, M.S.P.H., Ph.D.,
Director, Analytic Services, AccessCare, 3500
Gateway Centre Boulevard, Suite 130, Morrisville,
NC 27560; Phone: (919) 380-9962; Email:
chumble@ncaccesscare.org
Co-authors: Peter Scott Jensen, M.D.; Lynn
Wegner, M.D.; Steven Wegner, M.D., J.D.;
Christopher Kratochvil, M.D.; Alan Stiles, M.D.
Research Objective: To examine the
characteristics and baseline treatment practices of
primary care providers (PCP) who care for pediatric
patients with Attention Deficit/Hyperactivity Disorder
(ADHD) in North Carolina’s largest network of
Medicaid providers (AccessCare). These data will
underpin subsequent tests of the relative impact of
two interventions of differing intensity, each
designed to increase the likelihood that providers
will employ expert-recommended “best practices” in
the assessment and treatment of children and
adolescents with ADHD, the most common
psychiatric diagnosis in children. Results of this pilot
program will inform decisions regarding future
directions in provider training in North Carolina’s
Medicaid program and possibly elsewhere.
Study Design: Physicians and other providers who
prescribe medications in AccessCare member
practices were invited to attend an intensive, 3-day
training in pediatric psycho-pharmacology in
November, 2009. One condition for attendance was
willingness for the PCP’s practice to be randomized
to receive complementary training for the practice’s
care coordinator to enhance the coordination of
ADHD care. The program includes baseline and
follow-up assessment of trainees’ knowledge of
current practice guidelines for ADHD and selfassessed competence using them to diagnose and
monitor care of patients. Clinical teaching is
reinforced though hands-on opportunities for
sustaining practice and problem-solving vis-à-vis
obstacles to implementation, the use of key opinion
leaders, access to additional web-based tools that
PCPs can use to obtain information and learn at
their own pace. Support networks are developed by
conducting training in small PCP peer groups and
through bi-weekly conference calls. The
intervention group received additional training for
care coordinators and office staff in specific
strategies designed to identify and remove
obstacles to use of diagnostic scales and symptom
checklists.
Population Studied: A volunteer sample of 55
providers actively caring for pediatric patients with
ADHD covered by NC’s Medicaid program
Principal Findings: Initial analyses show that
providers in the Intervention and Usual Care arms
of the study are balanced in terms of numbers of
patients with ADHD diagnoses and regional
dispersion across the state. Overall they report
seeing an average of 91 pediatric patients with
psychiatric problems per week. At baseline, 95% of
providers rate their competence treating ADHD as
Good or Excellent. Sixty-one percent of providers
report caring for 80-100% of child patients with
ADHD without consulting other providers. Stimulant
therapy is the favored method of treatment. Fiftyeight percent said that prescribers in similar settings
use parent and teacher rating scales to assess
ADHD symptoms at baseline; 29% of similar
providers would test different medication doses
each week during initial titration; and only 19%
would use parent side effects scales during regular
visits.
Conclusions: Although Medicaid providers at
baseline report high levels of competence treating
ADHD in pediatric patients, their descriptions of
community norms in various aspects of treatment
suggest opportunities for improving the use of
clinical practice guidelines for ADHD diagnosis and
treatment with pediatric patients – the principal
outcomes in this pilot study.
Implications for Policy, Delivery or Practice:
Demonstration of successful improvement in ADHD
care in the primary care setting may allow more
cost-effective options for appropriate and patientfriendly care for North Carolina’s pediatric Medicaid
clients.
Funding Source(s): NIMH
Poster Number: 114
Lifetime Medical Cost Attributable to Early
Childhood Obesity and a Simulation of Affordable
Interventions
Presented by: Frick Kevin, Ph.D., M.A.,
Professor, Department of Health Policy and
Management, Johns Hopkins Bloomberg School of
Public Health, 624 N. Broadway, Rm. 606,
Baltimore, MD 21205; Phone: (410) 614-4018;
Email: kfrick@jhsph.edu
Lead Author: Sai Ma, Ph.D.
Co-authors: Kevin D. Frick, Ph.D.
Research Objective: Obesity continues to be a
public health crisis in the United States: over the
last three decades, the obesity rate for preschoolers
has tripled to 12%. Childhood obesity is linked to
numerous adverse health consequences such as
obesity in adulthood and other comorbidity, which
can further result in unfavorable social
consequences such as reduced lifetime academic
achievement, loss of productivity, and a reduced
income earning capacity. Consequently, policymakers and researchers began to pay increasing
attention to early interventions. Yet, no calculation
has been made to assess under what conditions
early interventions could be economically justified.
Therefore, this study seeks to answer: 1) What are
the lifetime medical costs attributable to early
childhood obesity? And 2) At what level of
effectiveness and cost would a population-based
early intervention be affordable?
Study Design: This study relies on three data
sources: the prevalence of obesity at all ages was
estimated from the National Health and Nutrition
Examination Survey, the persistence of obesity
from childhood to adulthood was determined by a
thorough literature review, and medical costs
attributed to obesity was estimated from the 2006
Medical Expenditure Panel Survey, using a
standard two-part econometric model. Simulations
were then conducted to estimate cost-savings of
interventions that take place at different ages with a
range of effectiveness.
Population Studied: The study population includes
all noninstitutionalized Whites, Blacks and
Hispanics living in the United States. Pregnant
women are excluded.
Principal Findings: The literature review suggests
between 50% and 75% obese children at age 5 will
remain obese in adulthood. In 2006, each obese
adult spent $1,548 (41%) more, and each obese
child (6-17) spent $264 (27%) more on annual
medical expenditure than normal-weight
counterparts. Lifetime medical cost attributable to
obesity is calculated by race, gender, and smoking
status. For example, compared to a white nonobese female non-smoker, the additional average
lifetime medical cost of a white obese female nonsmoker is $140,919, among which, $36,279 is
attributable to early childhood obesity. Results of
the simulations reveal that if the goal is to prevent
500,000 obese children becoming obese adults, we
would need early interventions (0-5) at 55%
effectiveness level, or interventions (6-11) at 25%
effectiveness or later interventions (12-17) at an
effectiveness of 20%. As for cost-savings, at 50%
effectiveness, population-based interventions at 0-5
can achieve $41 per capita medical cost-savings,
interventions at 6-11 yield $90 per person, and
interventions at 12-17 can save $119 medical cost
per person.
Conclusions: Our simulation shows that
population-based childhood interventions, even with
moderate effectiveness and only considering
medical cost-savings, would yield positive economic
returns.
Implications for Policy, Delivery or Practice:
Interventions could play a critical role in preventing
childhood obesity, reversing obesity after it occurs,
and avoiding adult obesity among those who are
obese as children. Our study demonstrates the
significant returns of intervening early childhood
obesity and should motivate policy-makers to take
serious actions to encourage the development of
new effective intervention, especially given that
government currently takes the largest share of
medical cost attributable to obesity.
Funding Source(s): The PEW Charitable Trusts &
the Zanvyl and Isabelle Krieger Fund.
Poster Number: 115
Association Between Acanthosis Nigricans and
Risk Factors for Type 2 Diabetes among Hispanic
Children in a Texas-Mexico Border Community
Presented by: Sion Lee, Research Assistant,
South Texas Border Health Disparities Center, The
University of Texas - Pan American, ITT Building,
Room 1.404Q, 1201 West University Drive,
Edinburg, TX 78541-2999; Phone: (956) 316-7937;
Email: sionlee@fas.harvard.edu
Co-authors: Suad Ghaddar, Ph.D.; Doreen Garza,
M.P.H.; David Salazar, M.S.; Patricia Keck, M.S.N.,
R.N.
Research Objective: Acanthosis nigricans (AN) is
a dermatologic condition characterized by velvety
hyper-pigmentation and thickening of skin folds. In
general, AN is associated with obesity and can be
an indication of various conditions related to insulin
resistance. The objective of this study is (1) to
report on AN prevalence rates among Hispanic
children in a Texas-Mexico border community, and
(2) to examine associations of AN with other risk
factors for type 2 diabetes.
Study Design: Data is from the 2005 Texas Risk
Assessment for Type 2 Diabetes in Children, a
state-level program that screens children for the AN
marker in Texas public and private schools.
Children who are positively identified with the
marker undergo additional assessments of body
mass index (BMI) and blood pressure. Our analysis
focuses on a subset which includes children who
were screened for obesity and hypertension
irrespective of their AN diagnosis. We utilize binary
logistic regression to assess factors associated with
the presence of the AN marker.
Population Studied: Students of a school district in
a predominantly Hispanic Texas-Mexico border
community characterized by high rates of obesity
(34% of county population vs. 26% nationwide) and
diabetes prevalence (17% of county population vs.
8% nationwide).
Principal Findings: Of 6,822 students screened,
99.6% were Hispanic and 51% male. Students were
almost evenly distributed among four grade levels
(3rd, 5th, 7th, and 9th). AN was detected among
22.8% of the subjects. Approximately 9% of the
students screened were hypertensive and 11.6%
pre-hypertensive. About a quarter were obese and
17.2% overweight based on age- and genderspecific BMI categorization. Binary logistic
regression results revealed that AN was less
prevalent among males (OR: 0.55, 95% CI: 0.480.64), and more prevalent among 5th (OR: 1.76,
95% CI: 1.44-2.14) and 7th (OR: 1.66, 95% CI:
1.37-2.03) graders relative to the younger group.
Those who were obese (OR: 41.39, 95% CI: 33.8850.57), overweight (OR: 8.02, 95% CI: 6.49-9.91),
hypertensive (OR 1.41, 95% CI: 1.12-1.77), and
pre-hypertensive (OR 1.47, 95% CI: 1.21-1.79)
were also more likely to have the AN marker.
Conclusions: Our findings reflect the increasing
trend in AN prevalence rates among Hispanics.
While earlier studies reported rates of 5-7% within
this group, more recent studies report rates that are
similar to those found in this study (22.8%).
Furthermore, the presence of the AN marker is
associated with obesity and hypertension, both of
which are risk factors for type 2 diabetes.
Implications for Policy, Delivery or Practice:
Screening programs constitute the basis to
identifying at-risk populations and to guiding
prevention efforts for type 2 diabetes. Our results
underscore the importance of intervention programs
that address childhood obesity. Current policies of
the Texas Risk Assessment program include
notifying parents and issuing medical referrals.
Future efforts should include following-up on ANpositive students and the outcomes of the physician
referral process. Findings from this program have
also triggered interventions in the border school
district analyzed (increasing physical activity
duration beyond the state-mandated level, nutritious
lunch menus and snacks, etc.). Preliminary results
suggest positive effects on obesity rates. Future
research will explore the effects of these
interventions on obesity, hypertension, and AN
prevalence rates.
Funding Source(s): CDC
Poster Number: 117
Seven Year Mortality Trend at U.S. Children's
Hospitals for Patients Critically Ill on the First
Hospital Day
Presented by: James Levin, M.D., Ph.D., Chief
Medical Information Officer, Children's Hospital of
Pittsburgh of UPMC, 4401 Penn Avenue,
Pittsburgh, PA 15224; Phone: (412) 692-6756;
Email: jim.levin@chp.edu
Research Objective: Although mortality is
relatively uncommon at United States children's
hospitals, the subset of patients who are critically ill
on the first day of hospitalization are at higher risk.
Successfully treating these patients challenges care
delivery systems. A previous report from The
Children's Hospital of Pittsburgh of UPMC (CHP)
concluded that there had been a significant
increase in risk adjusted mortality in the five months
after the introduction of the computerized provider
order entry (CPOE) phase of an electronic medical
record (EMR). This study extends that analysis in
four ways: 1) All patients on vasoactive medications
on the first day of hospitalization are included, 2)
Post-implementation time frame is extended to 6
years, 3) The population is broadened to include
patients from 20 hospitals, 4) The APR-DRG risk of
mortality (ROM) classification is employed to predict
mortality ratios.
Study Design: A retrospective study was
performed using administrative data. Patient
encounters were included if the patient received
one or more vasoactive medications on the first day
of hospitalization. These medications are commonly
used for a broad range of critically ill patients and
are rarely used outside of an intensive care setting.
The mortality ratio for each ROM group within the
study population was used to calculate
standardized mortality ratios (SMR).
Population Studied: Data were analyzed from the
Pediatric Health Information System (PHIS, Child
Health Corporation of America), 1,544,037
consecutive inpatient discharges from 2002 to 2008
were evaluation. 684,682 patients received one or
more vasocative medications on the first day of
hospitalization.
Principal Findings: From 2002 to 2008 the
unadjusted mortality ratio declined from 8.4% to
6.7% and the SMR declined from 1.16 to 0.91 (22%, p<0.01). At CHP, the SMR declined from 1.20
to 0.91 (-24%, p<0.10). All 20 hospitals showed
improvements. The hospital interquartile range was
-9% to -32%. During the first 6 months of CPOE at
CHP the SMR increased from 1.08 to 1.17 (+8%,
NS) when compared to the last 6 months prior to
CPOE. Increases of this magnitude or larger
occurred during 32% of the quarters at CHP and
33% of all hospital quarters.
Conclusions: Between 2002 and 2008 there was
significant decrease in the proportion of critically ill
patients failing to survive to discharge from 20
children's hospitals. It is common for the SMR to
fluctuate when looking at a single hospital over a
relatively short period of time.
Implications for Policy, Delivery or Practice: The
study demonstrates improvements in pediatric
critical care over 7 years. Comparisons over shorter
time frames or between individual hospitals need to
be interpreted with caution, however. Improvements
in risk adjustment methodology are needed but the
low frequency of deaths will continue to make
comparisons challenging. CHP was one of the first
children's hospitals in the country to begin a mutliyear implementation of an EMR and in 2009
became the first pediatric hospital to achieve a
completely digital care environment. Although a
previous study raised the question of an unintended
consequence of health information technology, the
current analysis does not support an unusual
increase in mortality in the first months after CPOE
introduction.
Poster Number: 119
Measuring Medical Homes: Tools to Evaluate the
Pediatric Patient- and Family-Centered Medical
Home
Presented by: Rebecca Malouin, Ph.D., M.P.H.,
Assistant Professor, Department of Family Medicine
and Department of Pediatrics and Human
Development, Mcihigan State University, B-101
Clinical Center, East Lansing, MI 48824; Phone:
(517) 884-0453; Email:
rebecca.malouin@ht.msu.edu
Co-authors: Sarah Merten, M.P.H.
Research Objective: The pediatric medical home
has historically been defined by care that is
accessible, family-centered, continuous,
comprehensive, coordinated, compassionate and
culturally effective. Differentiating a practice that
meets criteria as a medical home from one that
does not is of great importance in determining (1)
which activities successfully transform practices
and (2) which degree of “medical homeness” is
associated with better outcomes and reduced
healthcare expenditures. Furthermore, identification
of some standard measurement criteria is
recommended for practices participating in pilot
projects to reduce provider and practice reporting
fatigue, and allow for comparison of outcomes
across projects. The objective of this research was
to identify and review instruments for use as an
assessment tool for “medical homeness” using the
attributes within the American Academy of
Pediatric’s (AAP) Desirable Characteristics of a
Medical Home.
Study Design: The study design is that of a
literature review. Instruments were identified
through a literature review using PubMed and
through informal interviews with opinion leaders in
the fields of maternal and child health and
pediatrics. Instruments were rated based on
number of questions pertaining to the following
characteristics: comprehensiveness, coordination,
continuous, accessibility, family-centered,
compassion and cultural effectiveness.
Population Studied: The unit of analysis is an
instrument. Instruments were selected for review if
their purpose is to explicitly assess pediatric
medical homeness or if they include at least two of
the attributes of the AAP’s Desirable Characteristics
of a Medical Home.
Principal Findings: Sixteen instruments were
identified and reviewed. Across all instruments
assessed, coordination and family-centered care
were the most represented attributes of the patient
and family-centered medical home. Accessibility
and compassion were the least represented
attributes, likely due to the level of specificity used
in the examples of accessibility within the Desirable
Characteristics of a Medical Home described by the
AAP and the difficulty in developing a measure of
compassion. The Family-Centered Care SelfAssessment Tool and the Primary Care
Assessment Tools (PCAT) contained the most
elements of the pediatric medical home as defined
by the AAP. None of the tools included more than
one element of all domains as defined by the
Desirable Characteristics of a Medical Home.
Conclusions: The fact that none of the instruments
included more than one question relating to the
domains as defined by the Desirable
Characteristics of a Medical Home may be partly
due to the fact that the elements described within
the AAP practice accessibility domain, such as
acceptance of all insurance and accessibility by
public transportation, are quite specific. Consensus
on the specific elements to be included within each
attribute of the patient and family-centered medical
home is needed.
Implications for Policy, Delivery or Practice:
Identifying instruments with immediate practical
relevance for providers, families and patients is of
critical importance. Having immediately useful
information available from comparable practice
surveys will encourage patients, families and
providers to remain engaged in the evaluation
process as well as provide critical information that
can actually affect the transformational process
through provision of quality improvement reports.
The American Academy of Pediatrics is publishing
this work as a book titled, “Measuring Medical
Homes: Tools to Evaluate the Pediatric- and
Family-Centered Medical Home”.
Funding Source(s): HRSA, American Academy of
Pediatrics
Poster Number: 120
The Role of State Early Intervention (EI) Eligibility
Policy on EI Participation Rates Among Young
Children with Special Health Care Needs
Presented by: Beth McManus, P.T., M.P.H.,
Sc.D., RWJ HSS Scholar, Population Health
Sciences, University of Wisconsin-Madison, 610
Walnut Street, 707 WARF, Madison, WI 53726;
Phone: (617) 529-8138; Email:
mcmanus2@wisc.edu
Co-authors: Marie McCormick, M.D., Sc.D.;
Dolores Acevedo-Garcia, Ph.D.; Micheal Ganz,
Ph.D.; Penny Hauser-Cram, Ed.D.
Research Objective: In this study, we sought to
describe differences in Early Intervention (EI)
participation across US state among a cohort of
young children with parent-reported developmental
delays and disabilities.
Study Design: This was a cross-sectional multilevel analysis. Logistic multi-level models were
used to estimate the relative contributions of child
socio-demographic and developmental
characteristics, and state EI eligibility policy on EI
participation. We first derived a model of access to
EI services based on the socio-demographic and
special needs characteristics of the children. We
then examined the effects of state eligibility policies
using multi-level modeling to account for the
clustering of children nested within states. This
approach also permitted us to examine the crosslevel interaction between state EI eligibility policy
and individual poverty status, controlling for other
factors.
Population Studied: We used the 2005-2006
National Survey of Children with Special Health
Care Needs (CSHCN) to select a cohort of CSHCN
at risk for neurodevelopmental delay. We selected
children less than three years of age who had a
diagnosis or parent report of developmental
difficulty that affected their function at least some of
the time. In addition, children also met one of the
following Maternal and Child Health Bureau
CSHCN criteria: requiring more medical,
educational, or behavioral services when compared
to children of similar age; or receiving or needing
physical, occupational, or speech therapy; or having
functional limitations; or having behavioral or
developmental difficulties.
Principal Findings: We observed an overall rate of
EI participation of 45.7% that ranged 23.1% to
83.3% across the states. EI participants were less
likely to be Hispanic, living in a multi-racial family,
poor, with a developmental delay, or with less
severe impairments. The probability of receiving EI
was lower for children living in states with more
strict EI eligibility criteria than those with liberal
criteria (0.43 vs. 0.52).Poverty influenced this
association with the adjusted probabilities of
receiving EI for poor (<100% Federal Poverty Level,
FPL) and non-poor children (> 185% FPL) being
0.18 and 0.36, respectively. Among non-poor
children, those living in states with strict eligibility
criteria were nearly as likely as poor children living
in states with liberal eligibility criteria to receive EI
(0.33 vs. 0.36).
Conclusions: Our results of this study reveal
marked disparities and unmet needs in EI
participation as a function of both characteristics of
the child and the state program.
Implications for Policy, Delivery or Practice: Our
findings highlight areas of unmet need for EI
services within vulnerable populations and the
contribution of state policies to ameliorating or
exacerbating these disparities. They have important
policy implications to guide improved outreach to
families of children with special health care needs.
Funding Source(s): CDC
Poster Number: 121
Social Inequalities in Preschool Special Education
Participation Among Children with Developmental
Delays and Disabilities
Presented by: Beth McManus, P.T., M.P.H.,
Sc.D., RWJ HSS Scholar, Population Health
Sciences, University of Wisconsin-Madison, 610
Walnut Street, 707 WARF, Madison, WI 53726;
Phone: (617) 529-8138; Email:
mcmanus2@wisc.edu
Co-authors: Adam Carle, Ph.D.; Dolores AcevedoGarcia, Ph.D.; Michael Ganz, Ph.D.; Penny HauserCram, Ed.D.; Marie McCormick, M.D., Sc.D.
Research Objective: In this study, we examined
state variability in preschool Special Education
participation among a cohort of children in the
United States with parent-reported developmental
delays and disabilities. In addition, we sought to
determine to what extent child characteristics and
state socio-economic attributes explained state
variability.
Study Design: We used a cross-sectional multilevel analysis and fit a series of multi-level logistic
models to determine 1) overall state differences in
children’s receipt of preschool Special Education, 2)
the relative contribution of child socio-demographic
characteristics, and 3) the effect of state income
inequality, Special Education funding sources and
state Special Education eligibility policy, controlling
for child and family characteristics.
Population Studied: We used data from the 20052006 National Survey of Children with Special
Health Care Needs (CSHCN) and selected a
sample population of preschool-aged children (ages
3-5) whose parents reported a developmental delay
or disability that affected the child’s function at least
some of the time In addition, children also met one
of the following Maternal and Child Health Bureau
CSHCN criteria: requiring more medical,
educational, or behavioral services when compared
to children of similar age; or receiving or needing
physical, occupational, or speech therapy; or having
functional limitations; or having behavioral or
developmental difficulties.
Principal Findings: The observed an overall rate
of Special Education participation of 42.5%.
However, we found significant variation across
states in rates of Special Education ranging from
14.3%% to 65.4%. Special Education recipients
were more likely to be non-poor [adjOR: 1.3, 95%
CI:(1.02, 1.73)]; male [adjOR:1.67; 95% CI: (1.32,
2.09)], and have severe developmental disabilities
[(OR: 3.68; 95% CI: [2.76, 4.90)]; and least likely to
be Black, non-Hispanic [adjOR: 0.65; 95%CI: (0.47,
0.90)]. Special Education participation rates were
highest (54%) in states with the least income
inequality and lowest (31%) in the most unequal
states (p<0.001). Above and beyond child and
family characteristics, the income disparity between
most and least equal areas was significantly
associated with an increased odds [adjOR: 2.29;
95% CI: (1.5, 5.42)] of receiving Special Education
and this pattern persisted across all states [IOR:1.5,
3.5]).
Conclusions: We found marked social disparities
and unmet need in preschool Special Education
participation that stemmed, in part, from individual
child characteristics, but to an additional extent,
from state socio-economic attributes.
Implications for Policy, Delivery or Practice: Our
findings suggest that state income inequality
translates into disparate opportunities for education
for children with disabilities. Strategies to ameliorate
inequalities might include reducing barriers to
Special Education by 1) promoting community-level
developmental screening and referral, 2) ensuring
seamless transition between Early Intervention and
preschool Special Education, 3) bolstering universal
preschool and inclusion programming for children
with disabilities, and 4) assessing Special
Education implementation to reduce undue financial
burden to communities and psychological distress
to families.
Funding Source(s): CDC
Poster Number: 122
The Effect of Sibling Configuration on Children’s
Access and Use of Medical Care
Presented by: Sean Orzol, M.P.H., M.A.E.,
Doctoral Candidate, Health Management and
Policy, University of Michigan, 1415 Washington
Heights M3141, Ann Arbor, MI 48109; Phone: 734560-3111; Email: orzol@umich.edu
Co-authors: Thomas Buchmueller, Ph.D.
Research Objective: Family structure is
acknowledged as a potentially important factor in
determining a child’s access to medical care.
Previous research has documented an inverse
relationship between the size of the family and
children’s use of medical care. These findings are
often used to lend support to the “resource dilution
hypothesis” – the larger the family the less
resources, including health care resources, can be
allocated to any one child. What is often ignored in
such research is that family size is correlated with
birth order, and that some of the observed
relationship between family size and access to
medical care might therefore be a marker for a birth
order effect on access to care. Whereas the
resource dilution hypothesis implies that children in
larger families may be receiving less than the
optimal amount of medical care, if it is birth order
rather than family size that matters, this normative
conclusion may not be warranted. This study
investigates the effect of sibling configuration on
children’s access and use of medical care.
Specifically, it examines the extent to which birth
order and the number of children in a family is
related to various measures of access to health
care.
Study Design: We use data obtained from the
1998-2006 National Health Interview Survey (NHIS)
and the 1996, 2000, and 2004 panels of the Survey
of Income and Program Participation (SIPP) and
proceed in two steps. First, we run multivariate
regressions on cross sectional data, allowing us to
explore the relationship by looking at “betweenfamily” differences in sibling configuration. Second,
we utilize information on multiple children per family
in the SIPP and run models accounting for
unobserved heterogeneity using family fixed effects.
Population Studied: A nationally representative
sample of children aged 0 to 15 years.
Principal Findings: Findings using the NHIS
suggest that some aspects of children’s health care
may be influenced by family size, giving support to
the dilution of resources hypothesis. We also find
some evidence that birth order may be related to
the type and amount of care children receive,
independent of family size.
Conclusions: Our preliminary results suggest that
both family size and birth order may each be an
important factor influencing some types of children’s
use of medical services. We find suggestive
evidence that the resource most impacted by size
of the family is parental time, not financial.
Implications for Policy, Delivery or Practice:
Sibling configuration is complex and important
information might be missed if the correlation
between birth order and family size is not
considered when modeling its relationship on child
outcomes. Children are dependent upon their
parents to obtain and initiate health care services,
as such, health care practitioners may need to
consider a child’s family context when administering
or planning their health care services. Additionally,
understanding the mechanism through which family
size may affect health care access and utilization is
essential if we are to develop effective tools to
ensure timely and appropriate health utilization of
children at risk.
Poster Number: 123
Antipsychotic Polypharmacy among Michigan
Medicaid Youth
Presented by: Robert Penfold, Ph.D., Instructor,
Department of Population Medicine, Harvard
Medical School, 133 Brookline Avenue, 6th Floor,
Boston, MA 02215; Phone: (617) 509-9826; Email:
robert_penfold@hphc.org
Co-authors: Kelly Kelleher, MD, M.P.H.; Brandon
Strange, M.D.; Kathleen Pajer, M.D.; Wei Wang,
M.Sc.
Research Objective: The prevalence of
antipsychotic polypharmacy has been increasing in
adult populations. Our objective was to measure the
prevalence of pediatric antipsychotic polypharmacy
and identify the the most common diagnoses
associated with this prescribing pattern.
Study Design: We employed a longitudinal cohort
study design. Antipsychotic polypharmacy was
defined as the concurrent use of two or more
antipsychotic medications for a minimum of 60
consecutive days. We calculated age and sex
standardized rates of polypharmacy between 2000
and 2003. We also calculated the prevalence of
mental health diagnoses among children using
multiple antipsychotic medications.
Population Studied: We identified all individuals
aged 6 to 21 years who were continuously enrolled
in Michigan Medicaid between January 2000 and
December 2003.
Principal Findings: The rate of antipsychotic
polypharmacy in the Michigan Medicaid youth
population nearly tripled between March 2000 and
October 2003 from 1.85 to 5.73 per 100,000 person
days of eligibility. Of the 122 individuals using
multiple antipsychotic medications, 44.0% had at
least one diagnosis of mental retardation.
Approximately 12% of individuals had at least one
diagnosis for a psychotic disorder. Concomitant
prescribing of risperidone with another atypical
antipsychotic was the most common (42.5%)
treatment strategy with quetiapine being the most
common addition.
Conclusions: Antipsychotic polypharmacy among
Michigan Medicaid youth followed a similar growth
trend to that observed in adult populations over the
same time period. However, the treatment strategy
is most common among children with moderate to
severe mental retardation. Antipsychotic
polypharmacy in the Michigan Medicaid population
does not appear to be related to treatment
refractory psychotic disorders or externalizing
disorders.
Implications for Policy, Delivery or Practice:
There is very little evidence supporting the use of
multiple antipsychotic medications in children and
adolescents. The results of this study suggest that
much antipsychotic polypharmacy in the Michigan
Medicaid population is inappropriate. Given the
risks that antipsychotic medications pose to the
developing body and brain, policy makers should
consider strategies that strictly limit antipsychotic
polypharmacy.
Funding Source(s): NIMH
Poster Number: 124
Primary Injury Prevention and Mortality from Motor
Vehicle Crashes among Children and Adolescents
in the United States: From Safety Behaviors to
Healthy Environments
Presented by: Luis Pinet-Peralta, M.Sc.,
Research Assistant, Public Policy, University of
Maryland Baltimore County, 8030 Watermill Court,
Elkridge, MD 21075; Phone: (410) 680-6969;
Email: lpinet1@umbc.edu
Research Objective: The objective of this research
is to evaluate the effects that primary prevention
policies, particularly those associated with the
physical and the social environments, have had on
reducing the mortality risk from motor-vehicle
crashes involving young children and adolescents
in the United States.
Study Design: This is a retrospective study based
on 1997-2005 data from the FARS, the IIHS, the
U.S. Census, U.S. Presidential elections and U.S.
labor statistics, and analyzed using negative
binomial regression, maximum likelihood estimation
and fixed-effects.
Population Studied: Children and adolescents
involved in fatal motor vehicle crashes in the United
States between 1997 and 2005.
Principal Findings: I expect to find significant
reductions in the proportion of deaths among states
with primary enforcement of restraint laws,
mandatory driver education programs and
checkpoint s compared with states with secondary
or inexistent laws. I also expect to find mortality
reductions among states with speed reduction and
traffic calming measures compared with states
where such measures are not available.
Conclusions: The research provides an initial
evaluation of environmental modification policies
and fatal injuries from motor vehicle crashes among
children and adolescents at the national level and
using a longitudinal perspective.
Implications for Policy, Delivery or Practice:
Results may be useful to policy-makers and public
health practitioners involved in injury prevention and
control program development and implementation.
Studies that have provided evidence supporting the
positive impacts of environmental modifications on
injury prevention have been executed in Europe
only, where childhood fatal motor vehicle injury
rates have been reduced considerably more than in
the United States.
Poster Number: 125
Do Parents Perceive What Their Children Perceive
about Symptoms and General Health? An Analysis
Using Longitudinal Data
Presented by: Christy Pu, Ph.D., Assistant
Professor, Department of Public Health, National
Yang-Ming University, Li-Nong Street, Sec 2,
Peitou, Taipei, Taiwan 155; Phone:
+886285906914; Email: christypu@hotmail.com
Co-authors: Nicole Huang, Ph.D.; Yiing-Jenq
Chou, Ph.D.
Research Objective: This research examines the
relationship between perceptions of symptoms by
adolescents and their self-rated health (SRH) and
whether these perceptions matter when the parents
rate the general health of these adolescents.
Study Design: We analyzed 1,634 adolescents
(814 boys and 820 girls) from the Youth Project
conducted by the Academia Sinica of Taiwan. Data
from two waves (wav1=2001, wave 3=2003) were
used. Two physical and two mental symptoms were
used :(1) headache (2) feeling dizzy (3) feeling
lonely and (4) feeling depressed. The generalized
estimation equation (GEE) with a time trend was
used to determine the association between these
symptoms and the reporting of reduced health by
adolescents. A separate regression was carried out
using the reports of the parents. The kappa statistic
was used to determine agreement between the
adolescents’ own reports and their health as
reported by their parents and how such agreement
change over time.
Population Studied: 1,634 adolescents form
Northern part of Taiwan
Principal Findings: The results showed that for the
adolescents’ own reports, the four perceived
symptoms (headache, dizziness, loneliness and
depression) were closely associated with reporting
reduced general health by the adolescents, but
these perceptions were not picked up by their
parents in their ratings of the adolescents’ general
health. For depressive symptom, parents of girl
subjects who had reports of “a little” (OR=0.73,
p<0.05), and “sometimes” (OR=0.57, p<0.01) were
significantly associated with lower odds of reporting
reduced health for these adolescents. Time was
significant with OR=0.17 (p<0.01) and OR=0.64
(p<0.01) for the adolescent and parent reports,
respectively. The kappa statistic showed that
agreement between adolescents’ reports and
parents’ reports for general health decreased over
time.
Conclusions: Perceived symptoms are an
important factor in adolescents’ rating of their own
general health, but this is no true for the parents’
reports of the adolescents’ general health;
furthermore, the degree of agreement decreases
over time. Parents may not be good proxies when
assessing the health of adolescents.
Implications for Policy, Delivery or Practice:
When adolescents are the subjects of interest,
parents are often used as proxy raters since
children and adolescents are sometime assumed to
be unreliable respondents. In addition, parents,
instead of physicians, are the adults with whom the
adolescents have most frequent contact, thus they
are responsible for monitoring the health status of
the adolescents. It is thus important to ensure
parents are aware of their children’s health status
for proper treatments to be undertaken.
Funding Source(s): Academia Sinica, Taiwan
Poster Number: 126
School Gardens: A Comprehensive Nutrition
Education Approach to Improve Vegetable
Knowledge, Attitudes and Consumption among
School Aged Children
Presented by: Shelia Roundtree, M.D., M.P.H.,
Physician, Internal Medicine, Department of
Veteran's Affairs, 3510 Augusta Road, Greenville,
SC 29605; Phone: (864) 525-8482; Email:
tuunee01@yahoo.com
Co-authors: James Hardin, Ph.D.; Denise Hodo,
M.P.H.; Khosrow Heidari, M.S.; Joseph Beraho,
M.D.; George Nixon, M.D.
Research Objective: Childhood overweight and
obesity constitute a formidable public health
challenge. The cost in terms of morbidity, mortality
and fiscal burden are virtually incalculable.
Overweight children tend to become overweight
adults. This study sought to determine whether a
school garden based curriculum could favorably
influence childrens' vegetable knowledge, attitude
and consumption.
Study Design: Two groups of school children
participated in a comprehensive, validated school
garden curriculum. The children in the control group
received usual grade-appropriate instruction with no
special eM.P.H.asis on nutrition or vegetable
consumption. Those in the intervention group
planted, prepared and consumed fresh vegetables.
Throughout the intervention classroom teachers
taught nutritional facts as outlined in the curriculum.
Pre and post intervention test data were analyzed
using analysis of variance adjusting for school, sex,
group, time, and the interaction of group and time.
Knowledge was assessed from a 5-item multiple
choice test while satisfaction was assessed using
10 vegetable-specific Likert scale questions.
Population Studied: Our sample consisted of 345
school-aged (8-12 years) children from two schools
in one school district.
Principal Findings: Students in the intervention
group had significantly higher post-study knowledge
scores than those students in the control group
(p<0.001). Compared to the control group, students
in the intervention group showed higher satisfaction
for vegetables (p=0.003). Specifically, the children
in the intervention group showed higher satisfaction
scores for romaine lettuce (p<0.001), peppers
(p<0.001), and spinach (p=0.017). There were no
significant findings for sex or school.
Conclusions: This school garden project favorably
influenced students' knowledge and attitudes
concerning vegetables as well as their vegetable
consumption.
Implications for Policy, Delivery or Practice:
This work underscores the critical role classroom
instruction can play in the armamentaria to combat
obesity and overweight. Nutrition education can be
more effectively integrated into existing science
curricula. Likewise, gardening can occur in the
setting of physical education and even the related
arts. Further, this work implies the need for
development of standards based curricula materials
and teacher training in the areas of gardening and
nutrition. Clearly, adequate funding must be in
place to support these efforts.
Funding Source(s): Local Foundation
Poster Number: 127
Reasons for Critical Care Inter-facility Transports of
Neonatal and Pediatric Patients to a Tertiary Care
Academic Children’s Hospital
Presented by: Ranna Rozenfeld, M.D., Associate
Professor of Pediatrics, Pediatric Critical Care
Medicine, Children's Memorial Hospital/Feinberg
School of Medicine, 2300 Children's Plaza, Box
#73, Chicago, IL 60614; Phone: (773) 880-4891;
Email: rrozenfeld@northwestern.edu
Co-authors: Donna Woods, EdM, Ph.D.; Sally
Reynolds, M.D.; Michael Kelleher, M.D.; Paul
Leonardi, Ph.D.; Borko Jovanovic, Ph.D.
Research Objective: To assess the reasons for
inter-facility transfer of pediatric patients to an urban
tertiary/quaternary care academic pediatric medical
center.
Study Design: All inter-facility transports brought to
the facility, including critical care transports by
dedicated transport teams were reviewed for
medical diagnosis, day, time, month, and location of
accepting unit (ED, PICU, NICU, ward). Categories
of reason for transport diagnoses were
independently assessed by 5 physicians
representing differing pediatric specialties involved
in transports (ED, critical care, neonatology, general
pediatrics). The transport data was then classified
into broad categories of diagnosis. Frequencies
were estimated.
Population Studied: All pediatric transports from
September 2002 through January 2008 were
included - premature infants, neonates, infants,
children and adolescents.
Principal Findings: There were 8346 transports of
which return transports (397) and transports out of
the institution for radiation therapy (177) were
excluded. There were 7,772 transports remaining
for inclusion. The reasons for transfer included 134
diagnoses with at least 3 transports (6933). For the
remainder of transports (839) the reason occurred
in <3 transports. Of the 7,772 patient transfers, the
most frequent diagnoses included: respiratory
distress (861), seizures (652), congenital heart
disease (CHD) (403), prematurity (242), diabetic
ketoacidosis (226), and prematurity with respiratory
distress syndrome (213). Of the diagnoses listed as
CHD, 279 were listed as CHD, 22 coarctation of the
aorta, 20 tetrology of fallot, 17 transposition of the
great vessels, 9 aortic stenosis, 3 hypoplastic right
heart, 24 hypoplastic left heart, 12 pulmonary
stenosis, 8 total anamolous pulmonary venous
return and 9 ventricular septal defect. Reasons for
transport were classified as follows: respiratory
(23.8%), neurologic (12.0%), pediatric surgery
(10.3%), neonatal (10.3%), cardiac surgery (7.4%),
general pediatric conditions (7.0%), trauma (5.8%),
acute life threatening events (4.0%), neurological
surgery (4.0%), otolaryngology (ENT) (3.7%),
endocrine/allergy (3.6%), cardiology (3.2%),
gastroenterology (2.6%), hematology/oncology
(1.4%), and kidney/urology (1%). The month of
January had the highest volume of transfers and
transfers were most frequent on the first day of
each month. Transports also occurred most
frequently on Wednesdays and Fridays. The most
frequent time of day for transport was consistently
between 3PM and 12AM.
Conclusions: This study provides details of the
pattern of pediatric emergency transports to an
urban pediatric center. Over 50% of the transfers
were related to only 4 of the reason for transfer
categories: respiratory (23.8%), neurologic (12.0%),
pediatric surgery (10.3%), neonatal (10.3%).
Implications for Policy, Delivery or Practice: The
results of this study could be useful for institutional
and regional planning and resource allocation.
Understanding the timing of volume and flow of
transports can enable the transport services to be
adequately staffed for the potential transport calls.
Understanding the most frequent types of
transports can assist with development of
appropriate staffing of the transport team, the
appropriate units in the hospital, as well as having
the appropriate consultants available to care for
these patients.
Funding Source(s): AHRQ
Poster Number: 128
Racial/Ethnic Variation in the Relationship Between
Adolescent Health Status and the Intention to
Attend College
Presented by: Tetine Sentell, Ph.D., Assisant
Professor, Office of Public Health Studies,
University of Hawaii at Manoa, 1960 East West
Road, Biomed D104G, Honolulu, HI 96822; Phone:
(808) 956-5781; Email: tsentell@hawaii.edu
Research Objective: Educational attainment is a
key predictor of adult health status, yet decisions to
pursue higher education are typically made in
adolescence. The relationship between adolescent
health status and college intention is not fully
understood. If poorer adolescent health status is
associated with lower expectations for college
education, this will inform future interventions to
improve both adolescent and adult health.
Understanding variation in this relationship by
race/ethnicity will help target efforts to decrease
health disparities.
Study Design: Data were used from the 3,971
adolescents ages 12-17 who completed the 2003
California Health Interview Survey and were
currently attending school. Participants selfreporting plans to attend college/university were
compared to those reporting other plans (e.g., job,
military). Self-reported health status was measured
as (1) poor/fair, (2) good, and (3) very
good/excellent. Control variables in multivariate
logistic models included age, poverty, gender,
race/ethnicity, and parental education. Models were
performed with the entire sample, and for each
race/ethnicity separately to identify variation in the
relationship of adolescent health to education
intention.
Population Studied: Participants were diverse:
34% Hispanic, 9% Black, 10% Asian, 42% White,
and 5% other. Average age was 14; 49% were
female. Nineteen percent did not plan to attend
college; this varied significantly by race/ethnicity
from 22% of Hispanics (highest) to 11% of Asians
(lowest) (Chi2=33.04;p=.01). Ten percent reported
poor/fair health, 30% good health, and 60% very
good/excellent health.
Principal Findings: In unadjusted analyses, 27%
of adolescents with poor/fair health, 23% in good
health, and 17% in poor health did not plan to
attend college. In multivariate models, both
adolescents in poor/fair health (OR 0.50; 95%CI:
0.32-0.80) and those in good health (OR 0.63;
95%CI: 0.48-0.82) were significantly less likely to
plan on college compared to those in very
good/excellent health. When racial/ethnic groups
were considered separately, Asian adolescents in
fair/poor health were dramatically less likely to plan
on college (OR 0.09; 95%CI: 0.14-0.62) compared
to Asians in very good/excellent health. White
adolescents in good health were significantly less
likely to plan on college (OR .53; 95%CI: 0.39-.72)
compared to those in very good/excellent health.
No significant associations between health and
college intention were seen in African American or
Hispanic subpopulations.
Conclusions: Nearly 1 in 5 adolescents did not
plan to attend college, a key predictor of adult
health status. Adolescents who self-reported either
fair/poor or good health were significantly less likely
to plan on college compared to those in very
good/excellent health. Important variation was
found in the relationship of adolescent health to
college intention by race/ethnicity.
Implications for Policy, Delivery or Practice:
Promotion of adolescent health is important for
teens' immediate quality of life, but also represents
a critical contribution to their future health and
health-related opportunities. This study suggests
that investment in adolescent health to promote
higher educational attainment may be a key
pathway by which child health promotion efforts
could lead to better adult health opportunities.
However, this study also suggests that health is not
the critical factor limiting Hispanic and African
American college intentions, which should be
considered in efforts to decrease health disparities.
Poster Number: 129
Recent Trends in Sugar-Sweetened Beverage
Consumption and Childhood Obesity in California
Presented by: Lu Shi, Ph.D., Senior Analyst,
Health Services, University of California, Los
Angeles, 3271 Sawtelle Boulevard, Apartment 106,
Los Angeles, CA 90066; Phone: (315) 395-8025;
Email: lushi.pku@gmail.com
Co-authors: Jeroen Van Meijgaard, M.A.
Research Objective: The prevalence of overweight
among US children and adolescents appears to be
leveling off. Few studies have looked at recent
changes among risk factors that might help explain
these encouraging developments in obesity trends.
We present an analysis of recent data from the
California Health Interview Survey (CHIS) that
examines trends in childhood and adolescent
obesity as well as trends in sugar-sweetened
beverage (SSB) consumption, and discuss the
possible associations between these trends.
Study Design: We compared three separate crosssectional samples (2003, 2005 and 2007) from
biennial California Health Interview Survey (CHIS)
for three age groups, age 2-5, age 6-11, and age
12-17. We estimate the crude prevalence of high
SSB consumption (defined as having more than
one soda or other sweetened drinks during the
previous day) for each age group in each wave of
CHIS. Two measures of obesity were used: weightfor-age (WTA) at or above the 95th percentile on
national growth charts for children aged 2 to 11,
and Body Mass Index (BMI) for age at or above the
95th percentile on national growth charts for
adolescents aged 12 to 17. Logistic regression
analysis is used to estimate adjusted odds ratios of
consuming high levels of sugar-sweetened drinks in
2005 and 2007 compared to the baseline year of
2003.
Population Studied: Children and adolescents
living in California.
Principal Findings: From 2003 to 2007, each age
group experienced a substantial decline in high
sugar-sweetened drink consumption (16.4% to
5.0% for age 2-5, p<.001; 22.5% to 9.9% for age 611, p<.001; 35.7% to 25.7% for age 12-17, p<.001).
Declines in the prevalence of children’s obesity
were significant among children age 2-5 (p<.001)
and age 6-11 (p <.05) but not among adolescents
(p=.42). Children and teenagers surveyed in 2005
and 2007 were significantly less likely than those
surveyed in 2003 to have consumed more than one
SSB in the previous day after adjusting for gender,
age, race/ethnicity, poverty level and parental
education.
Conclusions: This study shows a downward trend
in the consumption of SSB among California
children and adolescents, a trend that is
independent of demographic and economic trends
in the state. While there has been no significant
change in adolescent obesity prevalence, the
substantial decline in both childhood obesity and
adolescent SSB consumption has the potential to
lead to a decline in obesity prevalence among
adolescents in future years. However, there is
insufficient information in the CHIS survey to
determine whether a reduction in SSB consumption
can be attributed to the legislation of California's SB
677, which put restrictions on soda sales on school
properties.
Implications for Policy, Delivery or Practice:
Policy actions on SSB consumption might have had
relatively rapid impacts on the prevalence of health
risk factors in the population. Further research is
needed to examine the contribution of declining
SSB consumption on the leveling off of obesity
trends and the extent to which these declines are
attributable to new policies and programs.
Particularly valuable will be comparisons between
states, comparisons with national data and
continued monitoring of trends in California.
Poster Number: 130
Racial Disparities in the Incidence of Adverse
Events for Hospitalized Pediatric Patients
Presented by: Richard Smith, Ph.D., Associate
Professor, College of Business, University of South
Florida St. Petersburg, 140 Seventh Avenue South,
COB 348, Saint Petersburg, FL 33701; Phone:
(727) 873-4943; Email: smithrb@mail.usf.edu
Co-authors: Lisa Simpson, M.B., B.Ch., M.P.H.,
F.A.A.P.; Gerry Fairbrother, Ph.D.
Research Objective: Racial disparities in health
care are well documented phenomena, although
questions persist about their cause. Increasingly,
the literature cites quality differences in facilities
where white and nonwhite patients receive care as
opposed to differences in treatment by providers or
hospitals. In the area of children’s health, AcevedoGarcia and colleagues (2008) suggest various
pathways in which children’s health is adversely
affected by residential racial segregation, including
segregated health care settings and the associated
disparity in quality of treatment. The objective of this
study is to determine whether hospitals that largely
treat nonwhite patients provide a lower quality of
care to children than other hospitals. Much attention
has been given to the incidence of preventable
medical errors in hospitals in the wake of two
alarming reports from the Institute of Medicine, To
Err is Human and Crossing the Quality Chasm.
Thus, this study uses recently developed, and
pediatric-specific, patient-safety indicators to
examine the incidence of adverse events for all
non-newborn inpatients up to age 17.
Study Design: A multivariate logistic regression, at
the patient-discharge level, is used to estimate the
probability a pediatric patient experiences a
potentially preventable adverse event, controlling
for patient and hospital characteristics. The key
explanatory variable characterizes the hospital
according to the proportion of nonwhite pediatric
discharges.
Population Studied: The sample consists of over 1
million discharges of children, age 0 to 17, but
excluding normal newborns, who met criteria for
being at risk of experiencing at least one of 8
preventable adverse events. Data come from the
2007 Nationwide Inpatient Sample (NIS) from the
Healthcare Cost and Utilization Project (HCUP),
sponsored by the Agency for Healthcare Research
and Quality (AHRQ). The 2007 NIS contains
complete inpatient discharge information for over
1,000 community hospitals in 40 states. Weights
are provided to allow for setting the value of sample
measures relative to national estimates.
Principal Findings: Preliminary results indicate
there is a small but statistically insignificant
increase in the odds of an adverse event for
children in hospitals with a higher proportion of
nonwhite discharges. On the other hand, there is a
large and significant decrease (OR = 0.492) in the
odds of an adverse event for children who are in
hospitals with the highest relative to the lowest
proportion of Medicaid patients.
Conclusions: Based on these results, racial
disparities in health outcomes for children do not
seem to stem from differences in the quality of care
between hospitals that treat a high proportion of
nonwhite patients and other hospitals.
Implications for Policy, Delivery or Practice:
These findings suggest that policymakers need to
look to other pathways, perhaps even outside the
health care setting (e.g. education, environment) to
explain racial disparities in the health outcome of
children.
Funding Source(s): HRSA, Maternal and Child
Health Bureau
Poster Number: 131
Geographic Disparities in Service Utilization by
Medicaid and CHIP-Eligible Children with Autism
Presented by: Angela Snyder, Ph.D., M.P.H.,
Senior Research Associate, Georgia Health Policy
Center, Georgia State University, 14 Marietta
Street, Suite 221, Atlanta, GA 30303; Phone: (404)
413-0285; Email: angiesnyder@gsu.edu
Co-authors: Daniel Crimmins, Ph.D.; Mei Zhou,
M.S.
Research Objective: To compare service
utilization patterns of children diagnosed with
autism living in urban and rural areas of Georgia.
Study Design: Retrospective cross-sectional study
comparing children diagnosed with autism living in
rural and urban areas of Georgia on both the types
of providers accessed and the frequency and costs
of health services utilized. County-level Rural-Urban
Continuum Codes (RUCC) are used to define rural
and urban residence for each child and two
measures of provider supply are used as
independent variables. Control variables are
categorized according to the theoretical framework
for the study of healthcare access first described by
Aday and Andersen (1974). Predisposing factors
include the child’s race/ethnicity and gender,
enabling factors include eligibility type and whether
or not the child was continuously enrolled in public
insurance for the full year. Age group is used as a
measure of need because previous research has
shown that age is associated with the breadth of
services children use. Multivariate methods
including both logistic and ordinary least squares
regression will be used to study specific utilization
patterns that show a geographic disparity in the
descriptive and bivariate analyses.
Population Studied: The study sample includes
5,682 children who were younger than 21 years of
age and received a primary or secondary diagnosis
for autistic disorder associated with a Medicaid or
CHIP reimbursed claim in calendar year 2005 in
Georgia. This sample represents 0.47% of the 1.2
million Medicaid and CHIP enrolled children in that
year.
Principal Findings: Descriptive results suggest
that children living in rural areas are somewhat
similar on the predisposing characteristics, with
black children less likely to live in rural areas. There
is a difference in the enabling characteristics
regarding program eligibility: more disabled children
live in rural areas and more children on Medicaid
waiver programs, with higher incomes, live in urban
areas. Need as measured by age group is similar
between groups. Similar to previous research,
children living in urban areas are more likely to visit
a pediatrician, while children living in rural areas are
more likely to visit a family practitioner. Surprisingly,
children in rural areas were slightly more likely to
visit a physician specialist. More predictably,
children in urban areas are more likely to visit a
therapist. In regards to expenditures, children
residing in rural areas utilize more pharmacy
services while children residing in urban areas
utilize more outpatient services. Multivariate results
are in process and will be completed by June 2010.
Implications for Policy, Delivery or Practice:
Because this study examines community and
delivery system factors that may create geographic
disparities in access, findings may suggest policy
interventions to decrease barriers and improve the
systems of care for children with special health care
needs including autism.
Funding Source(s): Pending funding from HRSA
Poster Number: 132
Is Parent Health Literacy Associated with ParentReported Quality of Life among Children with
Attention Deficit Hyperactivity Disorder?
Presented by: Sara Toomey, M.D., M.Phil.,
M.P.H., Instructor, Division of General Pediatrics,
Children's Hospital Boston, 21 Autumn Street, 2nd
Floor, Boston, MA 02215; Phone: (857) 218-4076;
Email: sara.toomey@childrens.harvard.edu
Co-authors: Janine Molino, M.S.; Eugenia Chan,
M.D., M.P.H.; Chao-Yu Gao, Ph.D.; Stephen Porter,
M.D., M.P.H.
Research Objective: The aim of our study was to
investigate the relationship between parent health
literacy and health-related quality of life (HRQL) in a
community-based sample of children with ADHD.
Study Design: We conducted a cross-sectional
analysis of parents of children with ADHD. Parents
reported on ADHD symptoms, current prescribed
ADHD medications and side effects, and HRQL.
The normalized Psychosocial Summary Score
(range 0-100) from the 50-item Child Health
Questionnaire-Parent Form (CHQ50) was the
primary HRQL outcome. The primary independent
variable was parent score on the Test of Functional
Health Literacy in Adults (TOHFLA). Multivariate
linear regression analysis adjusted for disease
control status, side effects, parent race/ethnicity,
parent gender, and parent level of education.
Scaled HRQL effect sizes were calculated using the
adjusted beta-coefficients. A 5-point difference was
considered to be a clinically meaningful change in
HRQL for the CHQ50.
Population Studied: We analyzed data collected
from parents of children with ADHD from diverse
communities in greater Boston who participated in a
clinical trial on information-sharing in ADHD. Of 271
parents screened, 194/271 (72%) were eligible and,
of those, 182/194 (94%) were enrolled. Two parents
were excluded post-hoc. We analyzed 177 parents
with complete data. 94% of parents were female
and 50% were non-White. 27% had a high school
education or less.
Principal Findings: The parents were largely
literate with only 6% scoring inadequate/marginal
on the TOFHLA. Mean HRQL was 33.9 (SD 12.5)
for parents with inadequate/marginal literacy
compared to 37.7 (SD 12.0) for parents with
adequate literacy. In multivariate analysis, parents
with inadequate/marginal literacy had a clinically
relevant lower HRQL although it did not reach
statistical significance (expected difference in
scaled HRQL 5.9, p=.12). Additional significant covariates included disease control (expected
difference in scaled HRQL 14.4, p<.001) and
presence of more than minimal side effects
(expected difference in scaled HRQL 7.6, p<.001).
No additional parent traits were associated with
HRQL.
Conclusions: Parent health literacy may be
associated with HRQL for children with ADHD and
demonstrates the strongest association with child
HRQL among the parent traits studied.
Implications for Policy, Delivery or Practice: Low
health literacy is known to be associated with poor
health outcomes and HRQL. Children with ADHD
experience poorer HRQL relative to typical peers. If,
as our study suggests, parent health literacy is
associated with HRQL for children with ADHD,
interventions targeting improvement in HRQL
should consider addressing parent literacy in
proposed strategies.
Funding Source(s): National Library of Medicine
Poster Number: 133
Adolescent Body Mass Index is Associated with an
Increase in Medical Encounters for Musculoskeletal
Injuries
Presented by: Carolyn Valdez, Quality
Improvement Associate, Patient Safety and Quality,
Denver Health, 645 Acoma Street, Denver, CO
80204; Phone: (303) 436-5491; Email:
Carolyn.Jagunich@dhha.org
Lead Author: Greg P. Gutierrez, M.D.
Co-authors: Simon J. Hambidge, M.D., Ph.D.
Research Objective: There is limited data on the
association between body mass index (BMI) and
musculoskeletal (MSK) injury in children and
adolescents. Most information is limited to
questionnaires and chart reviews of cohorts of
about 200 children. Our large urban vertically
integrated health care system allows us to analyze
a large cohort of individuals over the course of
many years. We hypothesized that an increase in
BMI would be associated with an increased risk of
MSK injuries in our pediatric population.
Study Design: Our study consisted of a
retrospective chart review of pediatric patients with
access to Denver Health’s community health care
centers. We examined patients in terms of
demographics, documentation of BMI data, history
of injury, and comorbid conditions in order to
examine this association.
Population Studied: The cohort was narrowed
down to 30,509 children ages 2-17 years old with 2
or more primary care visits from 2005 to 2008 in the
Denver Health system and limited to patients with
documented BMI data (or height and weight) in our
records. Patients with a BMI under 10 or over 60
were excluded along with patients that had no
recorded BMI data during the study period. Patients
were grouped into BMI categories according to the
national percentiles according to their age and
gender. Patients with percentiles greater than or
equal to 95% were classified as obese, and
between 85-94% were classified as overweight. All
others fell into the non-obese category.
Principal Findings: Of the 30,509 children, 18%
were obese; 15% were overweight; 67% were non
obese; and 39% of the cohort had an Injury Related
Medically Attended Event (IRMAE). The number of
injuries per child ranged from 0 (70%) to 5+ (1%).
18% of the cohort had comorbid conditions. In
bivariate analysis, higher BMI was associated with
a greater population percentage of IRMAEs: nonobese 22%, overweight 25% and obese 27%
(p<.0001). There were increased IRMAEs in
patients with a comorbidity (P<.0001), with
increasing age (P<.0001), and in males versus
females (P<.0001). Multivariate analysis
demonstrated an increasing trend for IRMAE with
increasing weight: overweight children were more
likely to be injured than non-obese (OR 1.090, 95%
CI 1.019-1.161, p<0.01) and obese children were
more likely to be injured than non-obese (OR 1.102,
95% CI 1.025-1.185, p< 0.01).
Conclusions: This is the first study in American
children to demonstrate an increased risk for
musculoskeletal injury requiring a medical visit with
increased BMI. Furthermore, the risk appears to
increase from non-obese to overweight to obese.
Given that obese children have been shown to be
less physically active than their non-obese peers,
our results suggest that children with higher BMI
are at significantly increased injury risk when they
are active. Possible reasons include relative muscle
weakness, poor proprioception and deconditioning.
Implications for Policy, Delivery or Practice: The
findings from this study will be taken into
consideration by our institution while planning
interventions related to obesity in the pediatric
setting. Our findings suggest that, in addition to
counseling families about the importance of
physical activity, providers and parents should be
aware of the increased risk for injury in obese
children.
Poster Number: 134
Rural/Urban, Racial/Ethnic, and Socioeconomic
Disparities in Lifestyle Counseling for Overweight
and Obese Adolescents
Presented by: Amy Wallace, M.D., M.P.H., Senior
Scholar, Veterans Rural Health Resource Center-Eastern Region, Veterans Affairs Medical Center,
215 N. Main Street, WRJ, VT 05009; Phone: (802)
295-9363 ext. 6072; Email: aew@dartmouth.edu
Research Objective: Overweight and obesity are
associated with serious health and social
consequences, and are steadily increasing among
rural, minority, and socioeconomically
disadvantaged populations. Encouraging lifestyle
changes: diet, exercise, and weight loss, should be
the initial approach for providers treating
adolescents with these issues. We examined the
likelihood of lifestyle counseling among patients age
13 to 21 years across gender, rural/urban locales,
racial/ethnic groups, median ZIP Code income,
educational and area poverty levels during
preventive and primary care appointments.
Study Design: We used the 2006 and 2007
National Ambulatory Medical Care Survey database
in a cross-sectional study design to evaluate the
likelihood of lifestyle counseling for overweight and
obese adolescents during preventive and primary
care visits. We examined prevalence of overweight
and obesity and calculated odds ratios to determine
the relative likelihood of lifestyle counseling across
residence locale, demographic, and socioeconomic
characteristics.
Population Studied: From the 2006 and 2007
National Ambulatory Medical Care Survey
database, we extracted 7,714 records representing
262 million preventive and primary care visits for
non-pregnant patients age 13 to 21 years with
documented BMI>25 or diagnosis of obesity.
Among these, 3,248 records representing
approximately 109 million patients received lifestyle
counseling.
Principal Findings: The overall prevalence of
overweight or obesity among 13 to 21 year olds
seen in outpatient visits was not statistically
different across gender (21% of females and 20%
of males) or residence locale (21% for urban and
suburban patients, 20% for rural patients).
Providers offered lifestyle counseling during only
33% of preventive and primary care visits for
overweight and obese adolescents. Across all
demographic groups, providers were 91% more
likely to counsel overweight urban (p<.0001) and
31% more likely to counsel suburban adolescents
compared to rural adolescents (p=.009). Odds ratio
for counseling for urban compared to suburban and
rural adolescents remained highly significant after
adjustment for other demographic and
socioeconomic variables. Other factors associated
with greater likelihood of lifestyle counseling were
male gender (OR 1.45, p<.0001), Asian race
(OR=1.52 , p=.003 compared to non-Hispanic
whites, all other race/ethnic comparisons nonsignificant), private insurance payer (OR=1.28,
p<.0001 compared to other payer types). Percent
community poverty, median community income and
community educational level did not significant
predict for lifestyle counseling.
Conclusions: Overall, providers are neglecting
lifestyle counseling for overweight and obese
adolescents during preventive and primary care
visits. When provided, demographic groups with
high obesity prevalence: rural patients, females,
Blacks and Hispanics are less likely to receive
counseling those groups for whom obesity is less
prevalent (urban and suburban adolescents, males,
and Asians). Community level economic variables
that describe relative disadvantage: lower median
community income and education level, and higher
percent poverty, did not predict lower likelihood of
counseling; however, lack of private insurance was
significantly associated with lower likelihood of
counseling.
Implications for Policy, Delivery or Practice:
While providers should provide counseling aimed at
nutrition, exercise, and weight loss to all overweight
adolescents during routine preventive care visits,
they should be especially attentive to population
segments with higher obesity prevalence. Early
adoption of healthy lifestyle habits may prevent a
lifetime of obesity-related health problems.
Funding Source(s): VA
Poster Number: 135
Raising Awareness of Youth Healthy Eating and
Physical Activity Behaviors: Short-Term
Effectiveness of a Social Marketing Campaign
Presented by: Hongmei Wang, Ph.D., Assistant
Professor, Health Services Research and
Administration, College of Public Health, University
of Nebraska Medical Center, 984350 Nebraska
Medical Center, Omaha, NE 68198-4350; Phone:
(402) 559-9413; Email: hongmeiwang@unmc.edu
Co-authors: Liyan Xu, M.S.; Anne Skinner, B.S.;
Molly O'Dell, M.D.
Research Objective: Changing the eating and
physical activity behaviors is considered the very
first step to tackle the youth obesity epidemic.
Behavior change in turn is based on the improved
knowledge of youth on the important health
messages. Across the nation, mass media
campaign is one of the mostly commonly used tools
to covey the health messages to youth. It is
important for the policy makers to understand the
effectiveness of a social marketing campaign in
delivering health messages and changing
behaviors. This study aims to measure the
effectiveness of a local social marketing campaign
promoting five recommendations on dietary and
physical activity behaviors to youth.
Study Design: The social marketing campaign is
funded locally and carried out by an advertising
company using mass media including TV, radio,
billboards, and website. The five principle
recommendations for youth are: 5 servings of fruits
and vegetables; 4 glasses of water; 3 servings of
low-fat dairy; 2 or less hours of screen time, and 1
or more hours of physical activity a day. The
researchers utilize a quasi-experiment design to
measure the campaign effectiveness. A phone
survey was conducted after the campaign is
completed in September 2009 to collect awareness,
knowledge, and behavior information of the
targeted audience. Knowledge and behavior
changes of respondents exposed to the campaign
were then compared to those of respondents not
aware of the campaign. Demographic information
was also compared between these two groups to
ensure the group similarity. T-test and Chi-squared
test were used to examine the significance of the
results.
Population Studied: A random sample was
selected from all households in local community
with children aged 5 to 18 years old and with a fixed
phone number listed. A total of 1,030 households
were surveyed. For households with youth aged 12
to 18 years, one youth was randomly selected to
complete the survey. The total sample size of the
youth was 504, with a response rate of 59.6%. For
households with children aged 5 to 11 years, a
parent or guardian was asked to answer knowledge
questions and behavioral questions of a randomly
chosen child in the household. 526 parents were
included and the response rate of this group was
78.1%.
Principal Findings: For both parents and youth
surveyed, those who were aware of the campaign
are more likely to reported correct responses to the
knowledge questions regarding the five
recommendations than those not aware of the
campaign. Significant differences in behaviors were
also found between those who were aware of the
campaign and those not aware of the campaign.
However, the significance in behavior changes was
not consistent across the five recommendations.
Parents who were aware of the campaign were
even found less likely to report that their child ate
five servings of fruits and vegetables and less than
two hours of screen time a day.
Conclusions: Social marketing campaign targeting
youth can be effective in delivering health
messages to increase knowledge of youth in a short
term. However, additional in-depth education and
health promotion efforts may be necessary for
children to understand the messages and follow the
recommendations to change their behaviors in the
long run.
Poster Number: 136
Prescription Antihypertensive Drug Use in Children
Presented by: W. Pete Welch, Ph.D., Senior
Consultant, Federal Health, The Lewin Group, 3130
Fairview Park Drive, Sutie 800, Falls Church, VA
22042; Phone: (703) 269-5605; Email:
Pete.Welch@Lewin.com
Co-authors: Wenya Yang, M.A., M.P.A.; Joseph
Flynn, M.D., M.S.; Perdita Taylor-Zapata, M.D.
Research Objective: Legislative initiatives
intended to improve pediatric labeling of
prescription medications have had significant
impact on pediatric drug development. In 1997, the
Food and Drug Administration Modernization Act
(FDAMA) provided incentives for companies to
voluntarily test drugs in pediatric populations in
exchange for 6 additional months of market
exclusivity. In 2002, the Best Pharmaceuticals for
Children Act (BPCA) created a mechanism for
studies of drugs without patent protection. In 2003,
the Pediatric Research Equity Act (PREA) required
that pharmaceutical companies conduct studies in
pediatric populations of any medication likely to be
used in children. These developments have
resulted in a proliferation of pediatric clinical trials
and updating of pediatric labeling. In this study we
sought to determine whether anti-hypertensive
drugs prescribed for use in children are actually
indicated for pediatric use and have FDA-approved
pediatric labeling.
Study Design: Using the database of a large
commercial insurer with approximately 20 million
covered lives, we defined our study population as
children who were continuously eligible for medical
and pharmacy benefits in the calendar year 2008.
We identified pediatric patients (age=18) with
hypertension using Episode Treatment Groups
(ETGs), a commercial software package that
organizes claims into episodes of care by
combining information from diagnosis, procedure
codes, and drug NDC codes from patient claims.
Pharmacy claims of the patients receiving antihypertensive drugs were then extracted to calculate
drug user rate and days of supply by therapeutic
class and drug agent. The prescribed drugs were
categorized based on the availability of pediatric
labeling (as indicated either in the FDA-approved
label or in the 2004 Fourth Report by the National
High Blood Pressure in Children and Adolescents
Working Group and an indication for pediatric use.
Population Studied: 2,912 pediatric patients with
primary (n=2,379) and secondary (n=533)
hypertension.
Principal Findings: 685 (22%) of primary
hypertensive patients were treated with any antihypertensive drugs, as were 384 (72%) secondary
hypertensive patients. Of the 1,069 patients treated,
87% were treated with labeled drugs that were
indicated (73% of supply days), 32% were treated
with non-labeled drugs that were indicated (23% of
supply days), and 3% were treated with a labeled
drug that was not indicated (2% of supply days).
Because some patients were treated with multiple
drugs, these percentages sum to more than 100%.
5% of patients (2% of supply days) could not be
categorized. Additionally, at least 12% of pediatric
patients with hypertension received treatment with
agents whose label information predated FDAMA.
Conclusions: Among hypertensive pediatric
patients receiving any type of antihypertensive
medications, the vast majority were being treated
with drugs that are indicated for pediatric use.
However, a substantial proportion of children with
hypertension are being treated with drugs that lack
FDA-approved pediatric labeling, or whose labeling
predates the recent legislative initiatives.
Implications for Policy, Delivery or Practice:
While recent legislative efforts have greatly
improved the availability of pediatric labeling for
antihypertensive medications, our finding that many
non-labeled medications are still widely used
suggests that greater efforts are needed to fulfill the
mandate of FDAMA, BPCA and other legislative
initiatives to provide safe and effective medications
for children.
Funding Source(s): National Institute for Child
Health and Human Development
Poster Number: 137
Economic Perspectives on Child Obesity
Presented by: Frederick Zimmerman, Ph.D.,
Chair, Department of Health Services, University of
California, Los Angeles, Box 951772, Los Angeles,
CA 90095-1772; Phone: (310) 825-1971; Email:
fredzimmerman@ucla.edu
Research Objective: Child Obesity is high and
rising, and represents one of the most significant
public health threats of our era. Considerable
research has explored the sources of socioeconomic disparities in childhood obesity and diet.
Economists have made a modest contribution to
this discussion to date, eM.P.H.asizing the role of
food subsidies in lowering prices for calorie-dense
foods, and the role of labor-saving devices in
reducing the time costs of food preparation. This
analysis critically examines three theoretical
traditions within economics—neo-classical
economics, behavioral economics, and institutional
economics—for their value in explaining and
contributing solutions to the childhood obesity
epidemic.
Study Design: For each of the three strands of
economics, a formal model of diet choice is
developed, and testable implications of the model
are derived. In the neo-classical model, diet is an
individual, rational choice by fully-informed parents.
Preferences are assumed to be parent-specific and
exogenous, and prices are assumed to be
exogenous, but subject to change by tax policy. In
the behavioral economics model, diet choices are
individual, but are strongly influenced by availability,
by hyperbolic discounting, and by physiological
cues such as portion size. Prices, availability, and
portion size are assumed to be modifiable by policy.
In the institutional model, neo-classical and
behavioral models are embedded in a particular
cultural-economic structure that shapes both
preferences and choice sets. Prices, availability,
and portion size are modifiable by policy, but so are
demand shifters, such as advertising, industrial
structure, and the ways in which public discussions
over obesity are framed.
Principal Findings: Little empirical support exists
for the neo-classical model of rational, informed
decision-making in its strict representative agent
form. Considerable evidence exists to support a
behavioral economics model of diet choice.
However, behavioral factors are useful in explaining
the aggregate trend of diet choice, they are not very
good at explaining socio-economic disparities in
diet outcomes. The institutional model predicts
increasing obesity when the effective price of
advertising falls, as it has over the past 3 decades,
and that children of parents with the lowest wages
are the most likely to be exposed to the obesogenic
effects of television advertising.
Implications for Policy, Delivery or Practice: The
neo-classical model suggests that a tax on
obesogenic foods, such as sugar-sweetened
beverages, will disincentivize high-caloric diets.
This policy is likely to fail because the price
elasticity of sugar-sweetened beverages is low,
implying that a very large tax—in excess of 50% of
the purchase price—will be required to have any
meaningful effect on demand. The behavioral
economics model suggests solutions such as
making fruits and vegetables more prominently
available in the school lunch room. Such policies
hold considerable promise, but may fail if they are
not broadened to include the whole food
environment. The institutional model suggests that
poor diet choices are largely the result of heavy
advertising. While this problem will not be easy to
redress, it is the only way to achieve meaningful
impact on the epidemic in child obesity.
Poster Number: 138
Specialty Referral Communication and Completion
in the Community Health Center Setting
Presented by: Katharine Zuckerman, M.D.,
M.P.H., Assistant Professor, Pediatrics, Division of
General Pediatrics and Child and Adolescent
Health Measurement Initiative, Oregon Health and
Science University, 707 SW Gaines Road, Mail
Code CDRC-P, Portland, OR 97239; Phone: (503)
494-6726; Email: zuckerma@ohsu.edu
Co-authors: Kerrie Nelson, Ph.D.; Tara Bryant,
B.A.; James Perrin, M.D.; Karen Donelan, Sc.D.
Research Objective: Parent and provider
disagreement about a child’s care at the time of
specialty referral may lead to incomplete referral
(not attending a specialty visit when referred by a
primary care provider). This project’s objectives
were (1) to assess the overall rate of specialty
referral completion among children referred by
pediatric community health center providers for
specialty care at an affiliated tertiary care center (2)
to assess parent-provider concordance in views of
seriousness of the referred child’s health problem,
necessity of specialty referral, and parental
understanding of the referral, and (3) to assess
whether parent and provider ratings of seriousness
of problem, necessity of referral, and parental
understanding were associated with referral
completion.
Study Design: Two months after referral to
specialty care, parents and primary care providers
were contacted to complete a survey which rated
seriousness of problem, necessity of referral, and
parental understanding on a 4-part rating scale
(“definitely yes”-“definitely no”). Parents were
surveyed by telephone, and providers completed
self-administered surveys attached to a copy of
their clinic note from the referral visit. Using
electronic health records, we assessed whether
specialist visits were completed or missed. We
used Pearson correlation coefficients [PCC] to
summarize parent-provider concordance in
seriousness of problem, necessity of referral, and
parent understanding. We applied logistic
regression to test the associations of parent and
provider ratings for each variable with referral
completion.
Population Studied: Parents of 501 children
seeking care from 28 pediatric primary care
providers in two community health centers located
in disadvantaged Massachusetts communities. All
children were referred for pediatric specialty
consultation at an affiliated tertiary care center.
Principal Findings: 27% of specialty referrals were
not completed. Parents and providers agreed in
ratings of seriousness of the problem 33.3% of the
time, in ratings of the necessity of referral 49.2% of
the time, and in ratings of parental understanding of
referral 62.1% of the time. Parent-provider
concordance on rankings was low for all measures
(PCC 0.10 for seriousness of problem, -0.09 for
necessity of referral, and -0.02 for parent
understanding of referral). On multivariate analysis,
significant factors associated with incomplete
referral included lower parent rating of necessity of
referral (Odds Ratio [OR] of referral completion 0.14
for “somewhat no” vs. “definitely yes”; confidence
interval [CI] 0.12-0.66) and lower provider rating of
seriousness of problem (OR 0.41 for “somewhat no”
vs. “definitely yes”; CI 0.17-0.97). The survey
response rate was 70% for parent surveys for 98%
provider surveys.
Conclusions: Nearly 3 in 10 pediatric specialty
referrals were not completed. Parents and providers
frequently disagreed about the necessity of
specialty referral and the seriousness of referred
children’s health problems. When parents thought a
problem was less serious, and when providers
thought a referral was less necessary, the referral
was less likely to be completed.
Implications for Policy, Delivery or Practice:
These data suggest that there is significant room for
improvement in parent-provider specialty referral
communications, and that low-quality
communication may be associated with incomplete
referral. To decrease incomplete referral rates,
providers should take a family-centered approach to
referral decision-making and provide additional
support to families that feel referral is less
necessary.
Funding Source(s): CRICO Risk Management
Foundation
Poster Number: 139
Redeveloping Closed Military Installations with a
Health in All Policies Approach
Presented by: Holly Avey, Ph.D., M.P.H., Senior
Research Associate, Georgia Health Policy Center,
Georgia State University, 14 Marietta Street, Suite
221, Atlanta, GA 30303; Phone: (404) 413-0291;
Email: havey@gsu.edu
Co-authors: Heather Devlin, M.A.; Naima Wong,
Ph.D., M.P.H.; Jane Branscomb, B.E.; Karen
Minyard, Ph.D.; Karen Cheung
Research Objective: Health in All Policies (HiAP)
is a strategy that strengthens the link between
health and other policies, creating an environment
that supports people’s ability to lead healthy lives.
HiAP addresses health across sectors, including
housing, transportation, education, employment,
land use, environment and agriculture.
Incorporating health-supporting policies in all areas
of community development can address the key
determinants of health equity. Fort McPherson is an
Army installation in Georgia that set to close in
2011 as part of the Department of Defense’s Base
Realignment and Closure (BRAC) program. The
vacated military property offers Metro Atlanta its
largest single redevelopment opportunity in many
years, with profound economic and health
implications for new residents, neighboring
communities and the region. The Georgia Health
Policy Center and Centers for Disease Control and
Prevention (CDC) are collaborating with McPherson
Local Redevelopment Authority (LRA) and other
stakeholders to bring a Health in All Policies
perspective to BRAC redevelopment activities at
Fort McPherson. The research objective is to use
Fort McPherson as case study and work with
national, state and local stakeholders to
operationalize the concept of Health in All Policies
and create a framework for incorporating health into
future base closures and other large-scale
redevelopments.
Study Design: Up to 20 semi-structured key
informant interviews will be conducted with federal,
state, and community stakeholders to understand
their diverse perspectives and needs. In keeping
with the exploratory and qualitative nature of this
study, key informants will be identified using a
snowball sampling technique in which individuals
refer those they know, who in turn refer those they
know, and so on. Initial informants will be identified
with the assistance of colleagues. Interview
categories will include federal stakeholders to
understand the BRAC process and authority
structure, and community and state leaders to
understand the policy environment. Based on the
knowledge gained from key informant interviews,
stakeholders will be mapped with respect to power
and influence. Structured opportunities will be
designed, using principles from whole scale system
change, for community, state, and federal
stakeholders to mutually explore health-promoting
development policies. A Health in All Policies
framework will be developed using results and
insights from these exploratory meetings.
Population Studied: BRAC stakeholders and
decision-makers.
Principal Findings: Base closure is a complex and
highly prescribed process that takes years to
complete and redevelopment may span decades
and yet there are a limited number of opportunities
to interject a Health in All Policies approach. These
include the design guidelines, zoning stipulations
and the developer’s agreement.
Conclusions: By incorporating health into the
policy development process across sectors and at
multiple levels, we can begin to address the key
determinants of health and health equity with a
more systematic and holistic approach.
Implications for Policy, Delivery or Practice: The
redevelopment of Fort McPherson presents a
unique opportunity to demonstrate the impact on
the health and well-being of residents when a
community is designed with a Health in all Policies
approach. Findings will inform a model that could
be used to incorporate health-promoting policies in
other base closures and other large-scale
redevelopments such as automobile manufacturing
plants.
Funding Source(s): CDC, through a cooperative
agreement with the National Network for Public
Health Institutes
Poster Number: 143
Does Loss of Safety Net Hospitals in a Community
Disproportionately Affect Racial and Ethnic Minority
Groups?
Presented by: Gloria Bazzoli, Ph.D., Bon Secours
Professor, Health Administration, Virginia
Commonwealth University, P.O. Box 980203,
Richmond, VA 23298-0203; Phone: (804) 8285223; Email: gbazzoli@vcu.edu
Co-authors: Woolton Lee, Ph.D.; Hui-Min Hsieh,
M.B.A.; Lee Rivers Mobley, Ph.D.
Research Objective: Safety net hospitals (SNHs)
are thought to play a critical role in the U.S. health
system by providing health care services for
vulnerable populations. SNHs also frequently offer
a variety of social and outreach services, which
could potentially reduce disparities among ethnic
and racial minority groups. This may be especially
true for minority individuals living near an SNH
because they will have ready access to these
facilitating services. Our study examined whether
closure or ownership conversions of nearby SNHs
had a disproportionate effect on uninsured and
Medicaid patients, especially for non-Hispanic black
and Hispanic individuals.
Study Design: Hospital discharge data for the
states of Arizona, California, Florida, and Wisconsin
were examined from the early 1990s to 2003. Our
primary access measure was distance traveled by
patients to hospitals for inpatient care. Both closure
and for-profit conversion represent events that
reduce safety net resources available to nearby
communities. Our analysis looked separately at five
types of hospital admissions based on primary
diagnosis: ambulatory care sensitive conditions,
referral sensitive conditions, marker conditions,
births, and mental health and substance abuse
admissions. The first three types are commonly
examined in studies of access to care, and the
remaining two represent types of hospitalizations
common among the uninsured and Medicaid
populations.
Population Studied: We examined patients treated
in urban hospitals in the four study states. Patients
were classified based on race/ethnicity and payer
status. Based on our analysis of travel patterns to
urban hospitals in each study state, we classified
patients based on whether or not they lived near to
SNHs that ultimately closed or converted to forprofit status by the ending year of the study. In
addition, other patient characteristics that might
affect patient travel patterns were controlled in
multivariate models, including comorbidities, the
emergent nature of the admission, hospital
characteristics, health system characteristics, and
neighborhood characteristics.
Principal Findings: Generally, we found that
uninsured, Medicaid, and Medicare patients
experienced added travel for many types of hospital
care in the ending year of our study relative to what
was observed for privately insured patients. The
uninsured and Medicaid effects were more common
among non-Hispanic blacks and Hispanics than in
white populations. Longer travel for Medicare
patients was predominantly present among whites
and Hispanics. However, these effects appear to be
related to general health system changes and not
associated with SNH closure or conversion per se.
Conclusions: Our results suggest that access to
inpatient care, as measured by distance traveled to
obtain care, may have become more problematic
for certain insurance types and for minority patients
over time. However, these access problems appear
to be related to general health system changes
rather than proximity to changes in SNH structure
and resources.
Implications for Policy, Delivery or Practice: Our
results suggest that SNHs are not a panacea to
eliminating disparities across insurance or
race/ethnicity groups. Policymakers likely need to
look more broadly at systemic health system issues
and changes over time to identify factors that have
created and sustained health disparities.
Funding Source(s): National Heart, Lung and
Blood Institute
Poster Number: 144
Variations in Specialty Care in Patients with Chronic
Diseases: A Case of Primary-Specialty Disconnect
or Patient Preference?
Presented by: Jessica Bellinger, Ph.D., M.P.H.,
Postdoctoral Fellow, Institute for the Partnerships to
Eliminate Health Disparities, University of South
Carolina, 220 Stoneridge Drive, Suite 204,
Columbia, SC 29210; Phone: (803) 251-6317;
Email: jbellinger@sc.edu
Co-authors: Saundra Glover, Ph.D.; Edith
Williams, Ph.D.; Rahnuma Hassan, M.H.A.
Research Objective: Existing health disparities
research typically explores patient levelcharacteristics and system-level characteristics,
such as provider availability and characteristics.
However, a model of patient-centered care includes
aspects of shared care. Primary care serves as a
complement for specialty services by providing a
medical home with continuity of care while the
expertise of specialists improve outcomes. The
study purpose is to determine racial and ethnic
variation in specialty care utilization by perceived
health and chronic disease status.
Study Design: The Commonwealth Fund’s 2006
Health Care Quality Survey is a nationallyrepresentative telephone survey of adults living in
the United States measuring health care utilization
and quality of care. Data include self-reported
health status, patient preference and behaviors,
and other measures. The survey has an adult
sample of 3,535 respondents. A stratified minority
sample design was employed to ensure a
representative sample of minorities. Logistic
regression and chi-square analysis were used in
the analysis of the primary outcome, self-report of
specialty care. SAS 9.2 was used to perform all
analyses appropriately weighted for survey
estimates.
Population Studied: To reduce loss of information
and account for missing data, respondents with
missing values from relevant items were not
included in our final sample (n=2475), whites
(43%), African Americans (32%), and Others (25%).
More African Americans reported poor health (23%)
than whites (20%). African Americans reported
more chronic diseases (hypertension, diabetes,
heart disease) than other respondents (p<.0001).
Principal Findings: African Americans were less
likely to report specialty care than whites when
controlling for age, income, and health insurance
status (AOR=0.81; 95% CI 0.67, 0.99). Compared
to the youngest respondents (18-44 years old), the
oldest respondents were more likely to report
specialty care (AOR=1.76; 1.35, 2.31).
Respondents lacking health insurance were less
likely to report specialty care use (AOR=0.42; 0.28,
0.61). Other negative predictors of specialty care
were lack of prescriptive coverage with health
insurance plan (AOR=0.59; 0.42, 0.85) and poor
perceived health (AOR=1.74; 1.33, 2.23). African
Americans were more likely to report at least one
chronic disease than whites (AOR=1.66; 1.33,
2.08).
Conclusions: Overall, perceived need (as
measured by self-report of health), race/ethnicity,
access, and age were predictors of both specialty
care use and chronic disease status. Low income
respondents were more likely to report specialty
care use, which suggests that public health
insurance programs increase appropriate access to
care.
Implications for Policy, Delivery or Practice:
African Americans reported higher rates of chronic
disease, indicating risk of poor health outcomes and
a greater need for specialty care; however, African
Americans were less likely to report specialty care
use. Such findings represent an opportunity to work
towards the elimination of health disparities.
System-level barriers, such as the availability of
providers (both primary and specialty), could
positively impact racial and ethnic disparities given
that lack of a usual source of care was a negative
predictor of specialty care and chronic disease.
However, this is not the complete solution in
isolation. Over 90% of the African Americans in the
sample reported a regular health care provider. This
suggests a need for interventions targeting personlevel barriers, such as self-efficacy and chronic
disease knowledge.
Funding Source(s): National Center on Minority
Health and Health Disparities (NCMHD) at the
National Institutes of Health (NIH)
Poster Number: 145
Disparities in Cervical Cancer Prevention and
Control Practices, Knowledge, and Preferences
among Women Living in the Southeastern United
States
Presented by: Jessica Bellinger, Ph.D., M.P.H.,
Postdoctoral Fellow, SC Rural Health Research
Center, University of South Carolina, 220
Stoneridge Drive, Suite 204, Columbia, SC 29210;
Phone: (803) 251-6317; Email: jbellinger@sc.edu
Co-authors: Heather Brandt, Ph.D.; Jan Probst,
Ph.D.; Saundra Glover, Ph.D.
Research Objective: The introduction of the Pap
test contributed to significant reductions in cervical
cancer mortality. The introduction of the HPV
vaccine may contribute to reductions. Effects of
appropriate screening have not been distributed
equally across groups, contributing to excess
mortality and disparities. In South Carolina, AfricanAmerican women are more likely to die of cervical
cancer than white women. Access, social
determinants, and women’s knowledge and
preferences for appropriate prevention and
screening contribute to disparities. The objective of
this study was to examine the effects of access,
social determinants, and women’s knowledge and
preferences on cervical cancer prevention and
control practices.
Study Design: Random-digit dial telephone
interviews were used to collect data in a populationbased HPV survey in South Carolina. Access
measures included health insurance and a regular
source of care. Screening was measured by a Pap
test within three years of interview. HPV vaccine
acceptability was measured by receipt of the
vaccine or willingness to vaccinate a child under 18
years of age. HPV and Pap test knowledge were
measured separately using indices. Social
determinants of interest were lifetime experiences
of discrimination and distrust in the health care
system. Multivariate logistic regression determined
outcome predictors (Pap test or HPV vaccine
acceptability) using appropriate survey design
procedures in SAS 9.2.
Population Studied: Potential subjects (18-70
years) were randomly selected. The interview
response rates for the primary sample and the
African American oversample were 36.7% and
34.0% with 1,029 respondents. Incomplete
interviews and “other” minority groups were
excluded from analysis (n=986), African-American
(30.5%) and white respondents (69.5%).
Principal Findings: Most respondents (87%)
reported Pap tests. More African Americans
(70.3%) preferred annual Pap tests even if advised
by a provider about longer intervals compared to
white women (55.4%, p<.0001). Twenty percent
reported the HPV vaccine; however, 77% were
willing to vaccinate themselves or their children
under 18 years of age. More African Americans
reported high experiences of perceived
discrimination (34.2%) than whites (4.0%)
(p<.0001). Nearly half of the African-American
respondents reported provider counseling during
their last Pap compared to a third of white women;
however, African Americans reported lower
knowledge (p<.0001). Access was a greater
predictor of cervical cancer screening than social
determinants or preferences. Women with a regular
provider (AOR=2.819; 95% CI 1.449, 5.483) and
private health insurance (AOR=2.970 CI 1.364,
6.469) were more likely to report Pap tests,
controlling for race, and knowledge. Provider
counseling had an impact on Pap tests
(AOR=2.007 CI 1.27, 3.19) not vaccine
acceptability.
Conclusions: There was high acceptability for
innovation in cervical cancer prevention and control
practices. However, African-American women
viewed the health care system more negatively and
reported more discrimination, which may contribute
to a lack of trust in provider recommendations.
Access was the strongest predictor of a recent Pap
test.
Implications for Policy, Delivery or Practice: The
findings indicate barriers to cervical cancer
prevention and control, which may contribute to
mortality disparities. Further research is needed to
identify preferences for HPV innovation to inform
the development of HPV interventions. Improved
access, especially to safety net health centers,
would increase cervical cancer screening and may
lead to reduced disparities.
Funding Source(s): NCI
Poster Number: 146
Through Our Eyes: Perspectives of Young Black
Men on Factors Affecting Transition to Manhood
Presented by: Nazleen Bharmal, M.D., M.P.P.,
NRSA Fellow, Department of GIM/HSR, University
of California, Los Angeles, 911 Broxton Avenue, 3rd
Floor, Los Angeles, CA 90024; Phone: (310) 7940150; Email: nbharmal@mednet.ucla.edu
Co-authors: Arleen Brown, M.D., Ph.D.; David
Kennedy, Ph.D.; Loretta Jones, M.A.; Charles LeeJohnson, M.S.W.; D'Ann Morris
Research Objective: To have young, Black men
from Los Angeles County identify and discuss
factors affecting their transition to manhood using
photovoice
Study Design: We used CBPR/CPPR principles to
form a 12-person collaborative of academic and
community members from various organizations. To
date, the collaborative has defined goals for the
partnered project to address premature mortality in
Black men, developed a project protocol,
participated in data collection and analysis, and
developed a preliminary dissemination plan. The
protocol was approved by the RAND Corporation
IRB. We used photovoice for data collection. Each
participant was given a digital camera to take a
minimum of 50 and maximum of 300 photographs
over four months. Participants individually
discussed photographs with an advisory board
member using the SHOWeD method. Each
participant used pile-sorting methods to discuss
major themes with an advisory board member.
Participants then selected 6-10 representative
pictures to put into a group collection of
photographs. In a separate session, each
participant, pile-sorted the group photographs and
discussed themes with an advisory board member.
All pile-sorting sessions were audio recorded and
transcribed. Pile-sorts of group pictures were
analyzed using multidimensional scaling and
hierarchical cluster analysis to systematically
compare participants’ themes and identify patterns
of associations between sorted photographs. Major
themes and related quotes were also elicited from
the pile-sorting transcripts.
Population Studied: Eligible participants were 1626 year old Black men who had lived most of their
lives in Los Angeles. Candidates were recruited by
advisory board members.
Principal Findings: Twelve recruited participants
completed all phases of the project. The
participants had a mean age of 20 years, 25% were
enrolled in higher education, 33% employed full
time, 75% grew up in single-parent/fostercare
homes, and 50% had been incarcerated and/or
involved in gangs. Four clusters emerged from the
group collection of photographs. The participants
interpreted the clusters as: 1) Struggles faced in
manhood. Pictures and quotes of police
harassement, incarceration, institutional racism,
lack of stable finances or job opportunities,
homelessness, neighborhood effects, lack of
direction/despair, family challenges (including poor
health of caretakers), and substance
abuse/dependence. 2) Inspiration and/or social
support. Pictures and quotes depicted family,
friends, coaches, religion and church, music,
nature, and people from car clubs & communitybased organizations. 3) Role of Sports. Pictures
and quotes showing sports as teamwork, an
outlet/escape, fitness, entertainment, and personal
accomplishment pivotal during their transition to
adulthood. 4) Los Angeles lifestyle. Commentary
and pictures about the opulence and excess of
living in LA, including expensive homes, rich
neighborhoods, high-end fashion, entertainment
industryluxury/sports cars, beautiful/picturesque
views, and abundance of fast food restaurants.
Most men discussed this lifestyle as unattainable
(except the fast food), but recognized the influence
of LA life on their personal and professional goals.
The theme analysis of experiences, which is
ongoing, will generate additional themes that
participants identify from their individual photos as
important in their transition to manhood. Based on
the project findings, priority areas will be identified
by the advisory board and participants. Community
dissemination of the results may include sharing
findings with community, school-based, and policy
organizations that affect healthy outcomes for Black
men through workshops, town hall meetings,
interactive gallery exhibit, and popular press
articles.
Conclusions: CBPR and photovoice are effective
methods to engage young, Black men to identify
and discuss factors affecting their transition to
manhood, contextualize research findings, and
participate in dissemination.
Implications for Policy, Delivery or Practice:
These methods not only allow traditionally
disenfranchised individuals and communities to
identify and document pertinent issues of concern,
but also to create the design and implementation of
strategies that foster positive change. Our small
project has led to the formation of a community
service/advocacy young men's group for the
participants, as well as a broader coalition among
the advisory board members to pursue a charter
school focused on young boys of color.
Funding Source(s): RWJF, Clinical Scholars
Program
Poster Number: 147
Massachusetts Leads the Way: Pay for
Performance to Reduce Racial/Ethnic Disparities
Presented by: Jan Blustein, M.D., Ph.D.,
Professor of Health Policy and Medicine, Health,
New York Unversity/Wagner Graduate School, 295
Lafayette Street, New York, NY 10012; Phone:
(212) 998-7427; Email:
jan.blustein@wagner.nyu.edu
Research Objective: Massachusetts is the first
state to provide financial incentives to hospitals to
adopt practices and processes that reduce
racial/ethnic disparities in health care. Hospitals
recently began to receive payments though the
MassHealth (Medicaid) program in a phased
program that began with public reporting and is
moving toward performance-based payments. This
study sought to understand the terms of
MassHealth's P4P/disparities program, assess
implementation challenges, and identify factors
associated with variations in hospital performance.
Study Design: We used a mixed methods
approach to understand the program’s workings,
implementation, and early impact. We reviewed
public documents and held open-ended in-depth
interviews with state employees. Additionally, we
analyzed the first year of performance and payment
data to assess the extent to which hospitals met
reporting requirements, and to determine the
degree to which performance on disparities
measures was associated with hospital
characteristics.
Population Studied: Our quantitative analysis
examined the performance of the 66 acute care
hospitals that participated in MassHealth's Hospital
Pay-for-Performance initiative in RY2008.
Principal Findings: During RY2008 hospitals were
eligible for payments for reporting and/or
performance on two dimensions: (1) organizational
cultural competence, as measured by a selfassessment tool, and (2) disparities in clinical care
for 4 conditions. Implementation challenges
included: (a) rolling out a complex program during
the relatively tight timeline dictated by statute; (b)
selecting and developing measures that were
appropriate for a Medicaid population; (c) conveying
reporting requirements to hospitals; and (d)
identifying valid and reliable disparities indices.
Hospital success in meeting reporting and
performance requirements varied. For the cultural
competence self-assessment, 21 of the hospitals
(32%) were ineligible for incentive payment due to
data deficiencies (failed timely reporting or
inadequate documentation). Of the 45 that reported
successfully, most received scores below 50% on
the self assessment instrument. Hospitals were
more successful in reporting clinical performance by
racial/ethnic subgroup, with only 4 (6%) failing to
provide acceptable data. However, there was little
measurable within-hospital disparity in care on the
clinical measures. In part, this was due to small
sample sizes for each clinical condition, and
relatively homogeneous patient populations in most
hospitals.
Conclusions: The MassHealth P4P/disparities
program is in the early stages of implementation
and refinement. Hospitals had difficulties with
reporting requirements around the cultural
competence self-assessment, but were able to
provide patient-level race/ethnicity data for clinical
services. Realities such as small sample sizes and
limited within-hospital racial diversity have proven to
be significant barriers to executing a credible
program to enhance racial equity in health care.
Implications for Policy, Delivery or Practice:
Pay-for-performance may be a useful tool to reduce
disparities in health care. Massachusetts'
experience provides a model for other states and
entities hoping to adopt this approach. However,
early experience suggests that the task is
logistically and technically challenging. National
well-validated measures are needed that apply to
low income populations served by Medicaid.
Successful implementation of P4P/disparities will
require flexibility and sustained engagement by
policymakers and stakeholders.
Poster Number: 148
Geographic Variation in Service Connection
Disability Ratings in the American Community
Survey
Presented by: Michel Boudreaux, M.S., Research
Assistant, School of Public Health, Division of
Health Policy and Managment, University of
Minnesota, 2221 University Avenue SE, Suite 345,
Minneapolis, MN 55414; Phone: (651) 503-4079;
Email: boudr019@umn.edu
Co-authors: Yvonne Jonk, Ph.D.; Kathleen Thiede
Call, Ph.D.
Research Objective: In 2008, the American
Community Survey (ACS) added a question
addressing veterans’ service connection (SC)
disability ratings. The primary objective of this paper
is to examine geographic variation in veterans’ SC
ratings by state and to identify explanatory factors.
We also assess the quality of this new variable for
Department of Veteran Affairs (VA) policy related
questions.
Study Design: Descriptive analysis of SC ratings
by state is followed by regression analyses
accounting for potential explanatory variables. To
asses the quality of the SC variable, the
consistency of SC responses with other information
obtained in the survey is described. Next, ACS
results are compared to counts published in the
Veterans Benefits Administration (VBA) annual
Benefits Report. Since the ACS sample frame does
not include all the populations included in VBA data
and covers a slightly different time period (CY 2008
versus FY 2008), VBA reports are not a perfect
standard. Thus, we examine the relative difference
of the ACS and other general population surveys to
VBA data.
Population Studied: U.S. veterans. Data comes
from the ACS, an ongoing survey of the U.S.
population that samples roughly 3 million
addresses. Related survey topics include veteran
status, period of service, socio-economic
characteristics, and disability measures that are
related, but not equivalent to SC disability.
Principal Findings: Initial examination of SC rating
by state found that nationally, 21% of veterans had
ratings of 70% or more, after excluding veterans
reporting a 0% rating and those refusing to report
their rating. West Virginia had the highest
percentage of such veterans (29.8%) and North
Dakota had the lowest (13.7%). Since some
variation is to be expected, further adjusted
analyses will address whether state-level variation
in SC ratings can be explained by differences in
veterans’ socio-economic and servicecharacteristics. Anomalies were found for veterans
reporting a 0% rating, but overall results
approximated administrative counts. Discrepancies
were similar to other general population surveys.
Self-reported income was generally consistent with
self-reported SC rating. However, the average
‘other income’ (which includes all “VA payments”)
for veterans reporting a 50-60% SC rating was less
than VBA compensation amounts published by the
VA. The percentage of those with an ACS defined
disability by rating category trended in the right
(higher rated veterans were more likely disabled).
Conclusions: Preliminary results found geographic
variation in SC ratings by state. These results
indicate that the geographic variation found by
previous researchers persists. While data
anomalies were found, the ACS is a useful source
of information for policy makers and analysts
interested in the well-being of veterans.
Implications for Policy, Delivery or Practice: SC
rating is a primary determinant of the level of
financial support and health care access granted to
disabled veterans. Previous studies have found
ratings disparities by gender, race, and geography
that cannot be explained by differences in
impairment. Thus, quality data and careful
monitoring of disabled veterans is of critical concern
to policy makers seeking to improve the equity of
the VA disability system and to those interested in
assessing the needs of disabled veterans.
Poster Number: 149
Perceived Discrimination and Patient Experience
with Care
Presented by: Thomas Bryant III, M.S.W.,
Doctoral Student, Health Services Research,
Management & Policy, University of Florida, Health
Science Center, PO Box 100195, Gainesville, FL
32610-0195; Phone: (904) 253-2263; Email:
bryantt3@ufl.edu
Co-authors: Robert Weech-Maldonado, M.B.A.,
Ph.D.; Allyson Hall, Ph.D.
Research Objective: Examine the relationship
between patients' perceptions of discrimination, and
their experiences with Medicaid as shown by the
Consumer Assessments of Healthcare Providers
and Systems (CAHPS) reports and ratings of care.
Study Design: Data consist of randomly surveyed
Florida Medicaid beneficiaries from September
2007 to December 2007. The survey included the
Health Plan CAHPS 4.0 instrument and a set of
items assessing patient experiences with
discrimination. There was a 40% response rate,
1877 completed surveys. The data is limited to
respondents with a personal doctor, resulting in
1509 completed surveys. The dependent variables
are Health Plan CAHPS 4.0 reports (getting needed
care, timeliness of care, doctor communication, and
health plan customer service) and ratings of care
(personal doctor, specialists, health care, and
health plan). Independent variables include
measures of patient experiences with racial/ethnic
discrimination. Variables related to systematic
differences in survey responses are used as casemix adjustors: gender, age, education, and health
status. Data are analyzed using analysis of
variance (ANOVA) and ordinary least squares
regression.
Population Studied: Non-HMO Florida Medicaid
beneficiaries from September 2007 to December
2007.
Principal Findings: Study results show that a
significant proportion (9%) of the sample perceive
discrimination as a result of race/ethnicity.
Respondents who experienced discrimination
sometimes significantly report more difficulties
getting needed care (B=-21.7; p<.0001), getting
care quickly (B=-11.5; p=.005) and communicating
well with their doctors (B=-13.8; p<.0001) than
those who never experienced discrimination.
Moreover, patients who perceived discrimination
sometimes significantly gave lower ratings for care
received (â=-13.2; p<.0001), their personal doctor
(B=-7.5; p=.005), specialist (B=-7.8; p=.017) and
health plan's customer service (B=-13.83; p<.0001)
compared to respondents who never experienced
discrimination. These results suggest that
perceptions of race/ethnic discrimination are
associated with lower CAHPS reports and ratings of
care.
Conclusions: The results suggest that perceptions
of discrimination based on race/ethnicity and
Medicaid insurance are associated with lower
CAHPS reports and rating of care.
Implications for Policy, Delivery or Practice:
Findings imply that Medicaid health plans should
use quality improvement efforts to address
perceptions of discrimination as a barrier to care for
their enrolled patient population. Findings suggest
that reducing perceptions of discrimination can
result in improved CAHPS reports and ratings of
care. The study also suggests the importance of
assessing cultural competency from the patients'
perspective, and including these measures in
patient health care surveys.
Poster Number: 150
Rise of Tobacco Use among the Hmong in
Minnesota as Perceived by Community Leaders
Presented by: Diana Burgess, Ph.D., Assistant
Professor/Core Investigator, Department of
Medicine and Center for Chronic Disease
Outcomes Research (CCDOR), University of
Minnesota and Minneapolis VA Medical Center,
One Veterans Drive, Minneapolis, MN 55417;
Phone: (612) 467-1591; Email:
diana.burgess@va.gov
Co-authors: Jeremiah Mock, M.Sc., Ph.D.; Nina
Alesci, M.P.H.; Barbara Schillo, Ph.D.; Steven
Foldes, Ph.D.; Jessie Saul, Ph.D.
Research Objective: Research suggests that
smoking is becoming an increasing problem among
the Hmong in the United States. Traditionally, in
Laos, very few Hmong adults smoked tobacco and
it was almost unheard of for Hmong youth to
smoke. However, studies of tobacco use among
this population are sparse. The purpose of this
study was to gain a deeper understanding of
tobacco-related attitudes, beliefs, and behaviors in
the Hmong community and to gain insights into
strategies for reducing tobacco use in this
population.
Study Design: Data were derived from a larger
community-based participatory research study
examining tobacco use within four Southeast Asian
communities in Minnesota through key informant
interviews with formal and informal community
leaders. Southeast Asian members of the research
team, who were fluent in English and their native
language, conducted semi-structured interviews
lasting 60-90 minutes. Interviews were taperecorded, transcribed, and translated into English
when necessary and reviewed by another bilingual
researcher to assure the accuracy of the
translation. Data were analyzed using grounded
theory. Initial results were presented these findings
to members of Hmong community who served as
informal advisors to the research project, and were
modified based on their feedback.
Population Studied: Key informants consisted of 7
female and 7 male Hmong key informants, ages 24
to 65, living in Minnesota, who had been in the US
from 12-24 years. The key informants volunteered
with civic organizations. Nine key informants spoke
English and Hmong equally and 5 spoke only
Hmong. All female key informants stated that they
had never smoked cigarettes. Of the men, 2 were
current smokers, 2 were former smokers, and 3 had
never smoked.
Principal Findings: Informants felt that living in the
U.S. had contributed to the rise of cigarette smoking
within their ethnic community. Whereas in Laos,
very few Hmong male youths smoked and were
admonished for smoking, and female youths were
never seen smoking, in Minnesota it had become
commonplace to see both genders of Hmong
youths smoking. The key informants’ construction of
this problem was centered primarily on social
concerns rather than health concerns, in which
Hmong youths’ public displays of smoking were
seen as acts of disregard for, and rejection of
traditional Hmong cultural norms, due to adoption of
“American” values. Informants believed that middleaged Hmong men and elders took up smoking while
living in Minnesota because of their traumatic and
difficult life experiences rather than because they
were trying to preserve “traditional” roles.
Informants also believed that Hmong people in
Minnesota had learned very little about the negative
health effects of smoking.
Conclusions: Key informants believed that traumaand relocation-induced stress, anxiety, and
depression (among elders) and the adoption of
American values (by youth), had a substantial
negative influence on tobacco use and norms in
their ethnic community.
Implications for Policy, Delivery or Practice:
There is a need for additional research to validate
the findings that emerged from this qualitative
study, as well as research that focuses on smoking
among youth, which was a major concern among
key informants. Nonetheless, this study points to
the need for policy makers and advocates to
address the specific root causes underlying
smoking among different segments of the Hmong
community. It is also important that health care
professionals who care for Hmong refugees and
immigrants to be aware of and provide culturally
meaningful resources to address their smoking as
well as other psychiatric disorders and psychosocial problems.
Funding Source(s): Clearway Minnesota and Blue
Cross Blue Shield of Minnesota
Poster Number: 151
Self-stigma, Stress, and Smoking among African
American and American Indian Women Smokers:
An Exploratory Qualitative Study
Presented by: Diana Burgess, Ph.D., Assistant
Professor/Core Investigator, Department of
Medicine and Center for Chronic Disease
Outcomes Research (CCDOR), University of
Minnesota and Minneapolis VA Medical Center,
One Veterans Drive, Minneapolis, MN 55419;
Phone: (612) 467-1591; Email:
diana.burgess@va.gov
Co-authors: Rachel Widome, Ph.D.; Sean Phelan,
M.P.H.; Michelle van Ryn, Ph.D., M.P.H.; Steven S.
Fu, M.D., M.S.C.E.
Research Objective: Research suggests that
negative emotions and difficulty coping with
stressful events might impede women’s ability to
quit smoking. The following qualitative study
explores the interrelationship between smoking
behavior and negative emotions among a sample of
racial/ethnic minority female smokers.
Study Design: Data were derived from a larger
study, the Perspectives of Ethnic Minority Smokers
(POEMS) project, involving sixteen 90-minute focus
groups with current and former smokers from ethnic
minority communities who were recruited by
community organizations. In the first stage of
analysis, focus group transcripts were coded and
analyzed by three reviewers to identify emergent
themes. In this second stage, grounded theory was
used to analyze a subset of the findings involving
smoking both a way to cope with and also a source
of negative emotions and stress.
Population Studied: The current analyses include
16 female African American and American Indian
smokers from 3 focus groups. The focus groups
were conducted in a large metropolitan area in the
Midwest.
Principal Findings: Consistent with prior research,
participants reported using smoking as a tool to
cope with stress and negative emotions.
Deprivation from smoking—either due to a
deliberate quit attempt or to the inability to obtain
cigarettes—was associated with negative states
such as anger, irritability, and distress. In addition,
women expressed negative emotions (e.g. guilt,
anxiety, disgust) and negative images of
themselves (e.g., weak, “not very smart”) due to
their inability to quit, despite the negative
consequences they experienced from smoking.
These negative self-perceptions are consistent with
stigmatized views of smokers held by the public.
Women also expressed feelings of defiance about
their smoking despite pressure to quit and identified
external factors, including those linked to race and
social disadvantage, which contributed to their
inability to quit. These findings are discussed in the
context of research on social stigma.
Conclusions: The negative emotions, self-stigma
and shame experienced by low income American
Indian and African American women smokers may
contribute to continued smoking and disrupt
attempts to quit smoking. Future research is
needed to explicitly examine the role of negative
emotions, stress and stigma among low-income
minority women smokers in order to develop
effective tobacco cessation intervention for this
vulnerable group of smokers.
Implications for Policy, Delivery or Practice:
Results point to the need for tobacco control
initiatives aimed at socially disadvantaged women
to address the negative emotions that may hinder
cessation. Such interventions might include helping
women find alternative ways to cope with life
stressors and negative emotions. It is also
important to ensure that tobacco control initiatives
do not contribute to the stigmatization felt by
women who smoke. In particular, it is suggested
that messages avoid depicting socially
disadvantaged mothers who smoke as a source of
harm to their children and instead depict them as
individuals who are important in their own right, and
worthy of the health benefits that quitting smoking
would bring.
Funding Source(s): Clearway Minnesota
Poster Number: 152
Racial and Ethnic Differences in Diabetes among
People with Psychiatric Disorders: Results from the
National Epidemiologic Survey on Alcohol and
Related Conditions
Presented by: Leopoldo Cabassa, Ph.D.,
Assistant Professor, Psychiatry, Columbia
University/New York State Psychiatric Institute,
1051 Riverside Drive Room 1715, Unit 11, New
York, NY 10032; Phone: (212) 543-5311; Email:
cabassa@pi.cpmc.columbia.edu
Co-authors: Roberto Lewis-Fernández, M.D.;
Jorge López Castroman, M.D.; Carlos Blanco,
M.D., Ph.D.
Research Objective: The aim of this study was to
examine racial and ethnic differences in diabetes in
a nationally representative sample of adults with
psychiatric disorders.
Study Design: Data were drawn from Wave 2 of
the National Epidemiologic Survey on Alcohol and
Related Conditions (NESARC), a nationally
representative study conducted in the United States
(U. S.). The AUDADIS-IV, a structured diagnostic
interview, was used to determine psychiatric
diagnostic assessments of past-year mood, anxiety,
and alcohol/ substance abuse disorders, and
lifetime personality disorders. The presence of
diabetes and other common medical conditions
(e.g., cardiovascular disease, hypertension, liver
disease) was determined by two questions.
Participants were asked whether they had a
particular medical condition (e.g., diabetes) in the
last 12 months. If they answered “yes”, then they
were asked whether this diagnosis was confirmed
by a doctor or other health professional. Those
responding “yes” to both questions were
categorized as having the medical condition in
question. Multivariate logistic regression models
adjusting for sociodemographic variables (e.g., age,
gender, education) and common diabetes risk
factors (i.e., body mass index, comorbid medical
conditions and use of psychotropic medications)
were used to examine racial and ethnic differences
in diabetes across a variety of psychiatric conditions
(e.g., mood, anxiety, alcohol/substance abuse).
Diabetes rates were compared between nonHispanics whites (reference group) and several
racial and ethnic minority groups including African
Americans, Hispanics, Native Americans, and
Asians/Pacific Islanders.
Population Studied: This study used a nationally
representative sample of 34,653 noninstitutionalized adults 18 years of age or older.
Principal Findings: African Americans, Hispanics,
and Native Americans with a variety of psychiatric
conditions reported significantly higher rates of
diabetes than non-Hispanics whites and
Asian/Pacific Islanders with similar mental health
needs. The higher prevalence of diabetes among
African Americans and Hispanics with psychiatric
disorders when compared to their non-Hispanic
whites counterparts persisted even after adjusting
for sociodemographic variables and common
diabetes risk factors. Odd Ratios (OR’s) for African
Americans ranged from 1.75 (95% CI:1.31-2.32) for
diabetes and any personality disorders to 1.96 (
95% CI:1.24-3.11) for diabetes and any mood
disorders. OR’s for Hispanics ranged from 1.87
(95% CI: 1.23-2.86) for diabetes and any mood
disorders to 3.09 (95% CI 1.50-6.36) for diabetes
and any alcohol/drug abuse disorders. Across each
of the psychiatric disorders examined, no significant
differences were reported between non-Hispanic
whites and Native Americans and Asian/Pacific
Islanders in the rates of diabetes once we
controlled for sociodemographic variables and
diabetes risk factors.
Conclusions: This study presents racial and ethnic
disparities in diabetes among a nationally
representative sample of people with common
psychiatric conditions.
Implications for Policy, Delivery or Practice: The
higher prevalence of diabetes among African
Americans, Hispanics, and Native Americans with
psychiatric conditions suggests that targeted efforts
are needed to screen and monitor the physical
health of these groups in both mental health and
primary care settings. Culturally appropriate
prevention and treatment strategies are needed to
reduce the negative impacts of diabetes and its
complications among racial and ethnic minority
communities in the U. S.
Funding Source(s): NIDA, New York State Office
of Mental Health
Poster Number: 153
Are Black Patients More Likely to Receive Total
Knee Arthroplasty in Lower-Quality Hospitals?
Presented by: Xueya Cai, Ph.D., Research
Assistant Professor, Internal Medicine, University of
Iowa Carver College of Medicine, SE 615 GH, 200
Hawkins Drive, Iowa City, IA 52242; Phone: (319)
384-8512; Email: xueya-cai@uiowa.edu
Co-authors: Peter Cram, M.D.; Mary VaughanSarrazin, Ph.D.
Research Objective: Total knee arthroplasty (TKA)
is performed to relieve pain and improve function
for patients with advanced knee osteoarthritis.
Substantial evidence suggests that despite the
dramatic expansion in indications of TKA, blacks
are less likely to receive this procedure compared
to whites. It is less clear, however, whether blacks
are more likely to undergo procedures in lower
quality hospitals than whites when they do receive
the procedure. Two previous studies found that
nonwhite patients tend to receive TKA in lower-
volume (likely lower-quality) hospitals. In this study,
we estimated hospital quality using a composite
outcome defined as complication or mortality within
90 days of surgery, adjusted for patient risk. We
then examined the association between race and
admissions to low-quality and high- quality
hospitals.
Study Design: We analyzed 100% Medicare
Provider Analysis and Review (MedPAR) data to
identify patients undergoing primary TKA (ICD-9CM code 81.54) between July 2002 and June 2005
(n=635,439). A random-intercept logistic regression
model of 90-day postoperative complications
(sepsis, hemorrhage, pulmonary embolism, deep
vein thrombosis, wound infection requiring
readmission and death) was estimated, adjusting
for patient demographics, admission source,
admission year, and co-morbidities using a
previously published algorithm. The ratios of
observed-to-expected outcomes (O/E ratio) were
calculated for each hospital. Hospitals were then
classified into high-quality (O/E in the bottom
quintile of all hospitals), low-quality (O/E in the
highest quintile), and intermediate-quality (middle
three quintiles) groups. We further identified white
and black patients undergoing primary TKA
between July and December of 2005. A multinomial
logit model was estimated to test the independent
impact of race on admissions to high-quality and
low-quality hospitals defined above. The models
controlled for patient age, gender, admission
source, comorbidities, education level and
household income (based on zipcode-level US
census data), travel distance to hospital (based on
miles between patient residence and hospital
zipcode), and geographic location of patient
residence (urban or rural).
Population Studied: Over 727,411 Medicare
admissions to 3611 hospitals for primary TKAs
between 2002 and 2005.
Principal Findings: Between July and December
of 2005, 87,019 whites and 4,953 blacks underwent
TKA. Blacks were less likely than whites to receive
TKA in high-quality hospitals (6.1% vs 10.2%,
p<0.0001), and more likely to receive TKA in lowquality hospitals (21.4 vs 11.9%, p<0.0001). Black
patients had lower household income and traveled
shorter distance for admission. Multivariable
analyses showed that blacks were less likely to use
high-quality hospitals (OR=0.84, 95% CI 0.72-0.92),
and more likely to use low-quality hospitals
(OR=1.76, 95% CI 1.61-2.10) than whites.
Sensitivity analysis confirmed our findings when
high-quality and low-quality hospitals were defined
using alternative cut-off points of the risk-adjusted
complication rate.
Conclusions: Black Medicare patients tend to
receive TKA in lower-quality hospitals.
Implications for Policy, Delivery or Practice:
This study finds that in addition to the welldocumented under-utilization of arthroplasty among
blacks, black patients who do receive TKA tend to
be admitted to lower quality hospitals. Further
studies are necessary to uncover the mechanisms
at the patient, referring physician and system-ofcare levels that lead to the racial disparity in access
to high-quality hospitals for TKA.
Funding Source(s): RWJF
Poster Number: 154
Systematic Measurement Error’s Influence on
Disparities in National Rates of Children with
Special Health Care Needs Across Spanish and
English Speaking Households
Presented by: Adam Carle, Ph.D., M.A., Assistant
Professor, Department of Pediatrics: Division of
Health Policy and Clinical Effectiveness, Cincinnati
Children's Hospital and Medical Center; University
of Cincinnati, 3333 Burnet Avenue, Cincinnati, OH
45226; Phone: (513) 803-1650; Email:
adam.carle@cchmc.org
Research Objective: Non-English speaking
children often experience poorer health and greater
care obstacles than English speakers. This includes
English speaking children with non-English
speaking parents. Yet, research suggests large
differences in the rates of US children with special
health care needs (CSHCN) across English (14%)
and Spanish speaking families (4.6%). A result
opposite much disparities research. And, a finding
that suggests that children from Spanish speaking
homes may require fewer resources than children
from English speaking families. However, culture
and translation may affect the way parents answer
questions about their children, resulting in
differences that reflect systematic measurement
error (i.e., measurement bias) not true differences.
Before concluding that disparities across Spanish
and English speaking children reflect true
differences and allocating resources, research must
establish that the instrument used to identify
CSHCN, the CSHCN Screener, provides equivalent
measurement in English and Spanish. This study
investigates this.
Study Design: I used multiple group (MG) multiple
indicator-multiple cause (MIMIC) item response
theory (IRT) to investigate whether the CSHCN
Screener provides equivalent measurement across
English and Spanish versions. MG-MIMIC models,
based in structural equation modeling and IRT,
powerfully investigate measurement bias: the
possibility that parents of children with identical
health respond dissimilarly to questions about their
children’s health as a function of the language in
which they answer the questions. MG-MIMIC do
this by developing equations that describe how
parents respond to questions about their children.
Subsequently, one tests the equivalence of these
equations across Spanish and English speaking
parents.
Population Studied: Data came from the 20052006 National Survey (NS) of CSHCN, a large,
representative survey of US children (n = 350,345)
aged 0-17 years old.
Principal Findings: MG-MIMIC that included
cross-group differences in education and income as
influences on measurement uncovered no
statistically significant systematic measurement
error across English and Spanish versions of the
CSHCN Screener. However, a “naïve” model
excluding education and income’s influences, found
statistically significant measurement bias.
Conclusions: Results demonstrated that
responses provided equivalent measurement
across English and Spanish versions among
children with similar socio-demographics,
increasing confidence in research showing
differences in the prevalence of CSHCN across
English and Spanish speaking children examined
within similar sociodemographics. However, results
showed that comparisons across Spanish and
English speaking families without regard to other
sociodemographics will lead to biased results,
calling into question work comparing CSHCN
across Spanish and English families without taking
into account systematic differences in the
distribution of education and income across these
groups.
Implications for Policy, Delivery or Practice:
These findings eM.P.H.asize the value of culturally
sensitive research and clinical practice. Individuals’
sociodemographic backgrounds can potently
influence how they respond to questions, causing
responses in one language to mean something
different than similar responses in another
language. Using model-based estimates can
mitigate systematic measurement error and lead to
more accurate disparities estimates and research.
When conducting research across distinct racial,
ethnic, and/or cultural groups, investigators should
establish the equivalence of the measures used to
insure the validity of results before developing
policy and allocating resources based on the
results.
Poster Number: 155
Conducting Propensity Score Analyses in Complex
Survey Data with Design Weights to Evaluate
Policies Aimed at Reducing Health Disparities
Presented by: Adam Carle, Ph.D., M.A., Assistant
Professor, Department of Pediatrics: Division of
Health Policy and Clinical Effectiveness, Cincinnati
Children's Hospital and Medical Center; University
of Cincinnati, 3333 Burnet Avenue, Cincinnati, OH
45229; Phone: (513) 803-1650; Email:
adam.carle@cchmc.org
Research Objective: Policy makers and
investigators frequently wish to evaluate the
effectiveness of policies aimed at reducing health
disparities. However, an inherent tension exists in
research of this type: one cannot randomly assign
individuals to policy exposure conditions (policy
exposure vs. no exposure). This limits causal
conclusions, as confounding variables likely
influence the likelihood that individuals are exposed
to a policy. For example, states differ in their
policies (or lack thereof) aimed at eliminating health
disparities. Simultaneously, sociodemographic
variables predict individuals’ states of residence. As
a result, the propensity to benefit (or not) from a
policy is correlated with characteristics of
individuals exposed to the policy. Thus, to make
stronger causal statements about a policy’s
effectiveness, analysts need to acknowledge that
the propensity for exposure to a policy differs
across individuals. Propensity score methods have
gained prominence as an effective method for
handing this bias. In essence, propensity score
methods “balance” the groups (exposed vs.
unexposed) in terms of a set of covariates in an
effort to equate the groups and make stronger
causal conclusions. Survey data often offer the best
datasets within which to conduct this research.
However, large scale surveys frequently employ
complex designs and include design weights.
Unfortunately, little to no published literature guides
investigators on how (or whether) to include weights
in propensity score methods, including score
development and utilization of the scores in
analyses. In this paper, I address this. I offer
practical summary advice for researchers seeking
to incorporate propensity score methods in the
presence of design weights.
Study Design: Using data from the 2005-2006
National Survey of Children with Special Health
Care Needs (NS-CSHCN), I compare the
performance of unweighted and weighted
propensity score methods across four popular
propensity score matching methods: nearestneighbor, radius, kernel, and stratification matching.
I also demonstrate the effect of failure to include
design weights when estimating the propensity
score.
Population Studied: The NS-CSHCN provides
data on the prevalence of special health care needs
and their impact on children and their families.
Children (n = 40,723) ranged in age from 0 to 18
years. Design weights make the data representative
of children nationally and within states.
Principal Findings: Failure to include weights in
the propensity score’s development resulted in
appreciably different propensity scores compared to
developing the scores by including the weights.
Likewise, across matching methods, failure to
include weights in propensity score analyses
resulted in significantly different, biased results.
Conclusions: When evaluating policy’s effects in
complex survey data, analysts should include the
weights in the propensity score development and
subsequent analyses. Researchers should not
ignore weights. Failure to do so results in biased
estimates.
Implications for Policy, Delivery or Practice:
Propensity score methods let health disparities
research make stronger causal statements
regarding the effects of policies. Large scale
surveys offer health services research excellent
opportunities to advance health disparities research
and evaluate effective policy and practice,
especially when incorporating propensity score
methods. However, in order to properly use large
scale survey data with design weights, researchers
must adopt the practices described in this research.
Otherwise, they will draw inappropriate conclusions.
Poster Number: 156
Patient and Clinician Discussions of Physical
Activity: Findings from an Underserved Population
Presented by: Jennifer Carroll, M.D., M.P.H.,
Assistant Professor, Family Medicine, University of
Rochester Medical Center, 1381 South Avenue,
Rochester, NY 14620; Phone: (585) 506-9484;
Email: jennifer_carroll@urmc.rochester.edu
Co-authors: Paul Winters, M.S.; Ronald Epstein,
M.D.; Kevin Fiscella, M.D., M.P.H.
Research Objective: The 5As guidelines are an
evidence-based framework for behavioral health
counseling. The 5As, in which a clinician Asks
about, Advises, Agrees upon, Assists and Arranges
a follow-up plan, is promising as a counseling
strategy in primary care to promote physical activity
in underserved populations. The research objective
is to report findings of patient and clinician 5As use
in physical activity discussions in an urban primary
care setting.
Study Design: The study design is a two-group
pilot RCT with family medicine clinicians as the unit
of randomization. The RCT is a clinician training
intervention currently underway to promote 5As
discussions about physical activity in an
underserved primary care population in Rochester,
NY.The intervention includes novel electronic health
records tools to prompt 5As discussion and tailor
referral options for accessible community resources
for physical activity. Here, we report on 70 baseline
audiorecorded patient-clinician visits followed by a
patient survey and interview asking about their
recall of physical activity discussions with their
primary care clinician. Clinicians also completed
surveys asking about use of the 5As in physical
activity discussions.
Population Studied: Twelve family medicine
clinicians from two federally qualified community
health centers and 70 of their patients. Clinicians
were family physicians (64%), nurse practitioners
(18%) and physician assistants (18%), averaging
15 years work experience (range 2-33 years).
Patients had a mean of 42.6 years, and were 72%
African American, 16 % Hispanic, and 14%
Caucasian. Most(71%) had Medicaid insurance.
Principal Findings: Patients’ average BMI was
32.6; weight-related comorbidities included diabetes
(18%), hypertension (34%), depression (24%), and
osteoarthritis or chronic pain (62%). Clinicians and
patients reported similar communication gaps in
5As discussions. Clinicians reported difficulty with
Assist and Arrange steps such as low confidence
about negotiating a treatment plan (79%) and
limited knowledge of community resources (75%).
Patients also reported that their clinicians used
Assist and Arrange skills less frequently (41%, and
18% respectively) compared the other As. Ninetysix percent of patients were highly interested getting
a physical activity guided plan or referral from their
clinician to an accessible, low cost community
program for physical activity. Clinicians indicated a
strong desire to learn more about referral options
incorporated via the electronic health records
system (100%).
Conclusions: Clinicians and patients reported that
Assist and Arrange skills were least frequent in
physical activity discussions. Though clinicians
reported Assist and Arrange to be challenging, both
clinicians and patients were highly interested in
incorporating Assist and Arrange into physical
activity counseling through enhanced knowledge of
accessible community resources for physical
activity.
Implications for Policy, Delivery or Practice: To
improve clinician-patient communication about
physical activity for underserved populations,
accessible, low cost community resources for
physical activity referrals are needed. We also need
simple and effective referral procedures for
clinicians and their patients. Electronic health
record tools may be a promising strategy to
facilitate communication and generate referrals to
community programs.
Funding Source(s): NCI
Poster Number: 157
Investment in Nursing Key to Reducing Disparities
in Outcomes in Minority Serving Hospitals
Presented by: Margo Brooks Carthon, Ph.D.,
R.N., Post Doctoral Fellow, Center for Health
Outcomes and Policy Research, University of
Pennsylvania School of Nursing, 418 Curie
Boulevard, Philadelphia, PA 19104; Phone: (215)
898-8050; Email: jmbrooks@nursing.upenn.edu
Co-authors: Ann Kutney Lee, Ph.D., R.N.; Jeannie
Cimmiotti, Ph.D., R.N.; Doug Sloane, Ph.D.; Linda
Aiken, Ph.D., R.N.
Research Objective: Few studies have examined
the relationship between minority patient health
outcomes and nursing care, despite nurses’
centrally important role in acute care. This research
study’s objective was to determine whether
hospitals with higher proportions of black patients
differ in nurse and patient satisfaction compared to
hospitals with fewer black patients, and if so why.
Study Design: This study was a cross sectional
analysis of three data sources collected in 20062007, including a large survey of hospital nurses’
perceptions of working conditions in four large
states (California, New Jersey, Pennsylvania, and
Florida), hospital patient discharge data obtained
from these four states, and the Hospital Consumer
Assessment of Healthcare Providers and Systems
(HCAHPS) survey. Linear regression models were
used to evaluate the relationship between the
proportion of Black patients, patient satisfaction,
and nurse-assessed quality measures.
Population Studied: Hospital patient discharge
data were used to obtain the percentage of Black
patients in each hospital. The percentage of Blacks
in hospitals was divided into three equal categories.
A total of 429 hospitals in the 4 states met the data
requirements for inclusion.
Principal Findings: The percentage of patients
who would definitely recommend the hospital to
friends and family was 3% lower in hospitals with
the highest proportion of Black patients, as
compared to the hospitals with the fewest Blacks. In
the context of HCAHPS data, this difference is
sizable and statistically significant. Fewer patients
in the hospitals with the highest numbers of Black
patients also reported that nurses always
communicated well with them, and always received
help when they needed it. Nurses working in
hospitals with the highest proportion of Black
patients reported poorer quality of care and were
less confident that patients could manage their care
upon discharge, as compared to nurses working in
hospitals with the fewest Black patients. These
negative effects persisted, but were moderated,
when controlling for nurse staffing levels and the
nurse work environment.
Conclusions: Patients in hospitals with higher
proportions of Blacks are more dissatisfied than
patients in hospitals with the fewest number of
Blacks. Nurses’ perceptions of quality follow a
similar pattern. Fewer Black patients received care
in hospitals with good nurse work environments
including adequate nurse staffing, explaining in part
their higher levels of dissatisfaction.
Implications for Policy, Delivery or Practice: The
results suggest that a promising strategy for
reducing racial disparities in hospital outcomes is to
invest in improving nurse staffing and the quality of
the nurse work environment in minority serving
hospitals.
Funding Source(s): NINR/NIH T32 NR 0714
Poster Number: 158
Looking Forward – Looking Back: The Use of
Historical Analysis as a Qualitative Method to
Address Contemporary Health Disparities
Presented by: Margo Brooks Carthon, Ph.D.,
R.N., Post Doctoral Fellow, Center for Health
Outcomes and Policy Research, University of
Pennsylvania School of Nursing, 418 Curie
Boulevard, Philadlephia, PA 09104; Phone: (215)
898-8050; Email: jmbrooks@nursing.upenn.edu
Research Objective: Despite the recognition that
historical and cultural factors significantly influence
patterns of health and illness among minorities,
studies which employ rigorous historical analysis of
community health activism in minority communities
are limited. This research study’s objective was to
examine the pre and post health outcomes of an
anti- tuberculosis campaign taking place between
1913 and 1925, in a Black South Philadelphia
community. Results are then used to determine the
utility of historical analysis as a useful qualitative
method to evaluate the effectiveness of past
policies and programs to reduce disparities.
Study Design: Social historical methodology was
employed for the completion of this study. Main
outcomes of interest were tuberculosis death rates
pre and post anti-tuberculosis campaign and the
influence of collaborative networks in advancing
public health interventions.
Population Studied: Primary archives were used
to describe and define the collaborative antituberculosis efforts of three Philadelphia-based
civic associations: the Starr Centre, the Whittier
Centre and the Phipps Institute. Sources were
reviewed from the Barbara Bates Center for the
Study of the History of Nursing, Temple Urban
Archives, and the University of Pennsylvania
General Archives. Secondary source analysis of
vital statistics, newspapers, major historical work on
nurses and physicians and sociological and
historical monographs detailing the lived experience
of Black Philadelphians was performed.
Principal Findings: In 1900, Blacks living in
Philadelphia suffered from TB mortality rates which
were two - three times higher than native white
residents. After decades of coping with excessive
deaths, Black community members, along with a
number of civic associations and health
professionals organized to wage a war on the
infectious disease. Subsequent to these efforts, a
precipitous decline in TB mortality in the Black
community occurred. Prior to the TB campaign of
1914, Philadelphia’s Black community was largely
invisible in the war on TB. The inclusion of Black
nurses and physicians, in conjunction with collective
efforts of public-private partnerships to address
both the health and material needs of Black
community members, led to an increase in TB
prevention efforts among members of the Black
community.
Conclusions: Findings from this study reinforce the
importance of social cohesion and collective
efficacy to improve health outcomes in minority
communities. Results further suggest that the use
of historical analysis may serve as a valuable
qualitative measure to evaluate evidence and
identify diverse strategies to address enduring
public health dilemmas.
Implications for Policy, Delivery or Practice:
Given the persistence of disparities among racial
minorities over the last century, the purposeful
integration of history, policy, and practice offers
compelling insights and direction for health equity
initiatives today.
Funding Source(s): AHRQ, NINR/NIH T-32
(NR0714)
Poster Number: 159
Barriers to Accessing Care among South Asian
Breast Cancer Patients
Presented by: Neetu Chawla, M.P.H., Research
Associate; Doctoral Candidate, Health Services;
Division of Cancer Prevention and Control
Research, University of California, Los Angeles,
650 Charles Young Drive South, Los Angeles, CA
90095-6900; Phone: (310) 825-7430; Email:
nchawla@ucla.edu
Research Objective: To understand the barriers
faced by South Asian breast cancer survivors in
utilization of cancer care.
Study Design: Semi-structured, in-depth interviews
were conducted among South Asian breast cancer
survivors in English and four South Asian
languages (Bengali, Hindi, Gujarati, and Urdu). All
interviews lasted between 60 and 90 minutes, were
audiotaped, and later transcribed. Interviewers were
trained according to the protocol of the UCLA
Institutional Review Board. The data collection
instrument included both quantitative and qualitative
questions. Open-ended questions covered a range
of topics, including the utilization of care, spiritual
needs, psychological well-being, and social support.
Data from the utilization of care domain were
analyzed for emergent themes and identification of
barriers to cancer care.
Population Studied: Forty South Asian women
living in Southern California ages 18 and older with
a previous diagnosis of breast cancer.
Principal Findings: The ages of women ranged
from 28 to 80 years, with the average age of
women being 55 years. The ethnic breakdown of
the sample was as follows: 53% Indian; 19%
Pakistani; 19% Sri Lankan; and 8% Bangladeshi.
Approximately 19% of the participants were in
active treatment for their breast cancer whereas
78% of participants reported that their cancer was
in remission. The time of diagnosis varied widely in
the sample, ranging from March of 1980 to August
of 2007. The majority of the sample (75%) reported
that cancer was diagnosed at an early stage.
Approximately 75% of women completed the
interview in English and the remainder in a South
Asian language. The sample included notable
proportions of South Asians with lower
socioeconomic status, with approximately 28%
reporting high school education or less and 23%
reporting an annual household income of less than
$32,000 per year. Analysis of open-ended
responses within the utilization of care domain
revealed a number of barriers to care faced by
South Asian women. Women who interviewed in a
South Asian language were more likely to describe
problems in communicating with providers and
obtaining information. Across insurance status,
women expressed difficulties in obtaining
diagnoses, particularly those who were younger
ages and who found their breast cancer
themselves. Women without a family history of
breast cancer suggested that physicians seemed
less concerned about the disease and less likely to
order additional tests during the diagnostic phase.
Other barriers described distance to treatment
facilities during chemotherapy and radiation, lack of
information about treatment options, and difficulties
obtaining second opinions. Several women also
described receiving help from another breast
cancer survivor in navigating their treatment and felt
this positively shaped their treatment experience.
Conclusions: South Asian breast cancer survivors
described several barriers to obtaining their care
and treatment. Communication with providers,
particularly among non-English speaking South
Asians was raised. In addition, women described
delays in diagnosis and difficulties in navigating
treatment (e.g. obtaining second options, need for
additional information).
Implications for Policy, Delivery or Practice:
Health care practitioners should make efforts to
reduce delays in breast cancer diagnosis, improve
communication, and provide navigational support to
South Asians with breast cancer. Cancer facilities
should provide language assistance and consider
peer navigators when delivering services to this
population.
Funding Source(s): California Breast Cancer
Research Program
Poster Number: 160
Health Disparities among Elderly Medicare
Beneficiaries with Diabetes and Multiple Co-Morbid
Chronic Conditions
Presented by: Lichun Chia, Ph.D., Postdoctoral
Fellow, Medicine, University of Pittsburgh, 230
Mckee Place, Pittsburgh, PA 15213; Phone: (412)
692-2011; Email: lic22@pitt.edu
Co-authors: Howard Degenholtz, Ph.D.
Research Objective: Racial and ethnic disparities
in health care have important implications for health
care providers, administrators and policy makers.
Health care providers are expected to provide high
quality of care to all patients based on need. In
patients with diabetes, inadequate treatment and
management of the conditions can result in
complications. Racial and ethnic minorities have
disparities in access to and use of primary care.
Evidence-based studies have shown that
complications of diabetes can be prevented with
appropriate monitoring and management; however,
there are disparities in quality for DM care. Few
studies have examined the contribution of
race/ethnicity to health disparities among people
with diabetes and multiple chronic conditions. The
purpose of this study was to compare access to
care and diabetes related care management and
outcomes between racial and ethnic minorities and
Caucasians with diabetes and additional comorbid
conditions.
Study Design: This was a secondary data analysis
using the nationally representative 2002 Medicare
Current Beneficiary Survey data. Self reported
information related to functional status, diabetes
conditions, care management and access to care
are included in the analysis. Univariate and
bivariate, and regression analyses were used to
examine the racial and ethnic differences in access
to care and utilization of health care service.
Population Studied: Medicare Beneficiaries 65
and older who have diabetes were included in the
study. People who have end-stage renal diseases
and live in nursing homes or assisted living facilities
were excluded from the analysis. The final analytic
sample had 2372 observations. Beneficiaries were
divided into four racial/ethnic groups:
Caucasian/Non Hispanic, African American/Non
Hispanic, Hispanic, and others.
Principal Findings: Approximately 64% of
Caucasian beneficiaries reported that their general
health was good to excellent compared to 50-54%
for African Americans, Hispanic and other
racial/ethnic groups respectively. While the number
of comobidities was similar among the groups, the
prevalence of specific comorbid conditions varied.
Caucasians had higher rates of coronary heart
diseases and osteoporosis, Hispanics reported
higher rates of Chronic Obstructive Pulmonary
Disorders, and African Americans had higher rates
of hypertension. African Americans were also more
likely to be prescribed insulin for controlling their
diabetes and had worst blood glucose control.
Hispanic and African Americans were more likely to
report using an emergency room for primary care
and had more difficulty of getting needed care.
Finally, racial and ethnic minorities were more
prone to diabetes related complication such as eye
problems.
Conclusions: These findings indicate that
racial/ethnic disparities among Medicare
beneficiaries with diabetes are evident. Minority
groups are less likely to have usual source of care,
poor control of their blood sugar, having more
trouble of getting needed care and have more
diabetic complications.
Implications for Policy, Delivery or Practice:
There is significant potential for better providing
needed care to improve health outcomes and
reduce complications among vulnerable minority
populations. Longitudinal studies are necessary to
assist policy makers in tracking system
improvements.
Funding Source(s): AHRQ
Poster Number: 161
Reducing Racial and Ethnic Disparities in Health
Care: Partnerships Between Employers and Health
Plans
Presented by: Patricia Collins Higgins, Ph.D.,
M.P.H., Researcher, Health, Mathematica Policy
Research, 1549 Wild Rose Court, Golden, CO
80403; Phone: (303) 997-7216; Email:
thiggins@mathematica-mpr.com
Co-authors: Melanie Au, M.P.P.; Erin Fries Taylor,
Ph.D., M.P.P.
Research Objective: To assess current health
plan/employer partnerships addressing disparities,
provide an overview of barriers that prevent more
partnerships from forming, and discuss strategies to
encourage increased involvement of employers in
the future.
Study Design: Qualitative case study including
document review and in-depth interviews with over
a dozen key informants.
Population Studied: We interviewed large
employers, health plan representatives, government
officials, and national experts.
Principal Findings: Despite an increased focus on
value-based purchasing among employers, it
remains rare for employers to be working directly on
disparities, either internally or in partnership with
health plans. A few notable partnerships have
recently emerged between large employers and
health plans and between the Office of Minority
Health and the National Business Group on Health.
Additionally, some employers are beginning to
assess health plans’ disparities work through
evaluation tools such as eValue8. Barriers to
greater employer involvement in disparities
reduction include lack of employer awareness of the
importance and cost of disparities, concerns about
the legality of sharing employees’ race and ethnicity
data, and the perceived administrative burden and
cost of disparities assessments. Several strategies
may increase employer involvement in this work,
including educating employers and employees on
the causes and consequences of disparities,
promoting the business case for disparities
reduction, eM.P.H.asizing the legality of r/e data
sharing, and creating more visible and cohesive
national leadership around these issues.
Conclusions: In the current economic
environment, disparities assessment and reduction
is a secondary priority for employers and health
plans alike according to several purchasers and
plans. However, some employers, particularly those
with a diverse workforce, realize that addressing
disparities ultimately has the potential to reduce
health care costs and create a more productive and
healthy workforce. Leading health plans continue to
work towards more fruitful collaborations with
employers and development of better methods of
disparities assessment.
Implications for Policy, Delivery or Practice: As
disparities continue to be discussed by health
plans, purchaser coalitions, and political figures,
more employers may recognize disparities as a
salient issue that is closely tied to value-based
purchasing. Moreover, demographic shifts in U.S.
workforce may eventually spur more employers and
health plans to work together to reduce health care
disparities, according to several experts. As the
U.S. population continues to diversify and
employers become more attuned to the direct and
indirect costs of health care disparities,
collaboration between purchasers and health plans
on reducing disparities may increase.
Funding Source(s): AHRQ
Poster Number: 162
Impact of Health Literacy and Socioeconomic
Status on Ethnic Disparities in Asthma
Presented by: Laura Curtis, M.S., Statistical
Analyst, Institute for Healthcare Studies,
Northwestern University, 750 N Lake Shore Drive,
10th Floor, Chicago, IL 60611; Phone: (312) 5035538; Email: l-curtis@northwestern.edu
Co-authors: Michael S. Wolf, Ph.D., M.P.H.; Kevin
B. Weiss, M.D., M.P.H.; Leslie C. Grammer, M.D.
Research Objective: Racial/ethnic disparities are
well documented in asthma. While socioeconomic
status has long been implicated as an explanatory
factor, limited literacy has more recently been
viewed as a potentially modifiable trait driving
asthma inequities. We sought to determine whether
literacy mediated the association between race and
asthma outcomes.
Study Design: Participants completed a face-toface interview and were followed longitudinally by
phone every 3 months for 2 years as part the
Chicago Initiative to Raise Asthma Health Equity
(CHIRAH) study. Health literacy was measured
using the Rapid Estimate of Adult Literacy in
Medicine (REALM) and data regarding participant
demographics, socioeconomics, and asthma
(symptoms, limitation to activities, medication and
urgent care use) were collected by self report at
baseline. Asthma knowledge was assessed using
an 11-item assessment developed for this study.
Asthma control (well controlled vs. not well/poorly
controlled) was classified based on NAEPP 2007
guidelines and the mini-Asthma Quality of Life
Questionnaire (AQLQ) was used to measure
asthma quality of life (AQoL). We used a form of
exploratory regression-based mediational analysis
to examine the explanatory nature of health literacy
and socioeconomic status in the association
between race and longitudinal asthma related
outcomes (knowledge, quality of life, and control)
across 7 time points.
Population Studied: 345 community dwelling
adults (99 Latino, 198 African American, and 48
White) in Chicago with persistent asthma and
complete baseline data were included in this
analysis.
Principal Findings: Rates of limited literacy
differed significantly across the three racial groups:
31% Latinos, 38% African Americans, compared to
13% Whites (p=0.003). Controlling for age, gender,
and duration of asthma, African American
participants had less asthma knowledge (ß=-.66,
95% CI -1.02 to -0.29, p<.001) and lower asthma
quality of life (ß=-.48, 95% CI -0.80 to -0.16,
p=.003) than White participants. Literacy alone
reduced these disparities by more than 30% for
both outcomes (Knowledge ß=-.45, 95% CI -0.78 to
-0.11, p=.01; AQoL ß=-.32, 95% CI -0.62 to -0.02,
p=.04). The addition of SES reduced these
differences further, to a point of non-significance.
Differences in asthma knowledge and quality of life
between Latino and White participants were
partially explained by literacy and SES, but
remained significant (Knowledge ß=-.61, 95% CI 0.99 to -0.23, p=.002; AQoL ß=-.31, 95% CI -0.59
to -0.03, p=.03). Ethnic disparities in asthma control
were fully accounted for by literacy and SES across
all racial groups (African American vs. White
OR=1.08, 95% CI 0.61 to 1.92, p=.79; Latino vs.
White OR=0.79, 95% CI 0.44 to 1.41, p=.42).
Conclusions: Limited literacy was identified as a
significant mediator to the relationship between
race/ethnicity and all examined asthma outcomes.
Socioeconomic indicators also explained these
associations.
Implications for Policy, Delivery or Practice:
Efforts to reduce asthma disparities should consider
health literacy promotion strategies as well as
improving access and quality of care. Specifically,
evidenced based patient education and counseling
interventions are available that could mitigate the
impact of literacy and thereby reduce inequities.
Funding Source(s): NHLBI
Poster Number: 163
The Effect of Patient-Physician Racial/Ethnic and
Language Concordance on Quality of Diabetes
Care
Presented by: Lisa Diamond, M.D., M.P.H.,
Research Physician, Health Policy Research, Palo
Alto Medical Foundation Research Institute, 795 El
Camino Real, Ames Building, Palo Alto, CA 94301;
Phone: (650) 853-3383; Email:
diamondl@pamfri.org
Co-authors: Sukyung Chung, Ph.D.; Latha
Palaniappan, M.D., M.S.; Harold Luft, Ph.D.
Research Objective: Diabetes mellitus (DM) and
its complications are disproportionately higher in
racial/ethnic (R/E) minority groups. Emerging
evidence suggests that patient-physician R/E and
language concordance may affect patient treatment
and medication adherence. We examine whether
R/E and language concordance affects DM quality
measures in a diverse ambulatory health care
system where all patients have similar access to
health care, high quality providers, and interpreter
services. We hypothesized that patient-physician
R/E and language concordance has more effect on
outcomes, rather than process measures, as the
latter requires patient education and adherence.
Study Design: Cross-sectional
Population Studied: We linked data on patient and
physician characteristics and DM quality measures
from the Palo Alto Division of the Palo Alto Medical
Foundation (PAMF), a large multi-specialty group
practice in northern California. PAMF collects selfreported R/E and preferred language information at
registration. R/E and language concordance was
based on the primary care physician (PCP)
assigned to the patient. All active patients in the DM
registry (n=5,994) as of August 2009, matched with
168 PCPs were included. Multilevel random effects
models took account of clustering within physicians
while including patient (demographics,
comorbidities, likelihood of seeing the same PCP,
and insurance type) and physician (experience and
DM panel size) factors. The quality indicators
included four process measures (glycosylated
hemoglobin (HbA1c) within 6 months, blood
pressure (BP), LDL cholesterol, and foot
examinations within one year) and four outcome
measures (HbA1c<7% or <8%, target BP<130/80,
and target LDL cholesterol = 100mg/dL).
Principal Findings: Patients represent diverse R/E
(45% non-Hispanic white (NHW), 27% Asian/Pacific
Islander (API), 11% Hispanic/Latino, 5%
Black/African American, and 12% other race) and
preferred language (English 73%, Spanish 3%,
Chinese 3%, 11% other). Physicians were also of
diverse R/E (60% NHW, 36% API, and 4%
Hispanic/Latino), with some non-English language
proficiency (Spanish 23%, Chinese 14%), 22%
other). Hispanic/Latino patients were least likely to
have a R/E concordant PCP (2%), compared to API
(14%) and NHW (32%) patients. Most (98%) of
non-English speaking patients had a language
discordant PCP. In multivariate analyses, there
were significant differences in DM care by patient
R/E: API patients were more likely to have had a
foot exam (p<.001) and to achieve target BP
(p<.01) and LDL (p<.01) compared to NHW
patients. Black/African American patients were less
likely to achieve target BP compared to NHW
(p<.01), and Hispanic/Latino patients were less
likely to have HbA1c<8% (p<.0001). Hispanic/
Latino patients with a R/E concordant PCP were
more likely to have had their HbA1c checked
(p<.01) and API patients with R/E concordant PCP
were more likely to achieve a HbA1c<7% target
(p=.02) compared to R/E discordant patients.
Preferred language and language concordance
were not significantly associated with any of the DM
quality measures.
Conclusions: R/E and patient-physician R/E
concordance were associated with better
performance on some DM quality measures.
Neither language preference nor language
concordance was associated with DM process or
outcome measure achievement.
Implications for Policy, Delivery or Practice:
Disparities in DM care across R/E groups, even
among insured patients with access to the same
PCPs within an organization, suggest that we need
a better understanding of how R/E and R/E
concordance affect DM quality of care. Future
interventions to address quality of DM care may
seek to target R/E discordant patient-physician
groups.
Funding Source(s): The California Endowment
Poster Number: 164
Use of Different Treatment Options by Rural Early
Stage Prostate Cancer Patients
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, Family Medicine, WWAMI Rural Health
Research Center, University of Washington, 4311
11th Avenue NE, Suite 210, Seattle, WA 98105;
Phone: (206) 685-0402; Email:
mdoesche@uw.edu
Lead Author: Laura-Mae Baldwin, M.D., M.P.H.
Co-authors: Roger Rosenblatt, M.D., M.P.H.;
Shilpen Patel, M.D.; Michael Porter, M.D., M.S.;
Holly Andrilla, M.S.
Research Objective: Maintaining the highest
quality cancer care is inherently difficult in rural
areas due to the paucity of cancer care specialists,
diagnostic services, and therapeutic services. This
study compares use of different treatment options
between rural and urban prostate cancer patients to
identify potential inequities in service access and
treatment choice.
Study Design: Retrospective cohort study using
2004-2006 data from rural and urban in 10 states
with Surveillance, Epidemiology, and End Results
(SEER) cancer registries. We compared unadjusted
rates of different treatments between rural and
urban cancer patients, and conducted logistic
regression analysis to examine the odds of
definitive treatment with recommended surgery or
radiation therapy versus non-definitive treatment
(active surveillance or non-standard surgical
treatment) among rural and urban patients adjusting
for sociodemographic (age, race/ethnicity, marital
status), cancer (stage), and contextual
characteristics (state).
Population Studied: 78,874 urban and 10,638
rural prostate cancer patients ages 18 and older
with clinical T1c and T2 stage disease.
Principal Findings: 70.3% rural and 73.4% urban
patients (p=.000) received definitive treatment;
25.6% rural and 24.4% urban patients were in
active surveillance (p=.005); 4.1% rural and 2.3%
(p=.000) urban patients received non-standard
surgical treatment only (local tumor
destruction/excision, simple prostatectomy). Of
patients with definitive treatment (n=62,397), 47.9%
rural and 50.0% urban patients (p=.001) were
treated with radical prostatectomy; 33.6% rural and
34.0% urban (p=.45) with external beam radiation;
and 18.5% rural and 16.0% urban (p=.000) with
radioactive pellet implantation. The adjusted odds
of urban patients receiving definitive therapy was
1.31 (CI 1.23-1.39) times that of rural patients.
Conclusions: Rural patients are significantly less
likely to receive definitive treatment for early stage
prostate cancer and are significantly more likely to
receive non-standard, less comprehensive surgical
treatments and to be in active surveillance. When
they do receive definitive treatment, it is more likely
to be a one-time radioactive pellet treatment. Taken
together, these findings suggest that rural prostate
cancer patients may make treatment decisions
based on logistical factors such as geographic
barriers or limited capacity.
Implications for Policy, Delivery or Practice:
These findings, combined with results from further
studies designed to understand how rural patients
learn about and choose their prostate cancer
treatments, will elucidate whether these patients
have difficulties accessing their preferred cancer
services, and why. With this information,
interventions at the level of physician practice and
service delivery can be developed to alleviate ruralurban disparities in cancer service.
Funding Source(s): HRSA
Poster Number: 165
Does Residential Sprawl Contribute to Women's
Risk of CHD?
Presented by: Christine Eibner, Ph.D.,
economist, Economics and Statistics, RAND
Corporation, 1200 S. Hayes Street, Arlington, VA
22202; Phone: (703) 413-1100 ext. 5913; Email:
eibner@rand.org
Co-authors: Beth Ann Griffin, Ph.D.; Chloe E. Bird,
Ph.D.; Regina A. Shih, Ph.D.; Mary Ellen Slaughter,
M.S.; Jose J. Escarce, M.D., Ph.D.
Research Objective: We assessed the relationship
between metropolitan statistical area (MSA) level
urban sprawl and incident coronary heart disease
(CHD) among women, adjusting for individual
sociodemographic characteristics, baseline health
status and health behaviors.
Study Design: We used 2-level hierarchical Cox
proportional hazard models (e.g. shared frailty
models) to analyze the Women’s Health Initiative
Clinical Trial (WHI CT) data, merged with a
continuous measure of MSA-level sprawl developed
by Ewing and colleagues. Participants were ages
50-79 at baseline, recruited at 40 clinical centers
and 36 satellite locations, enrolled between 1993
and 1998, and followed until March 2005. We
examined 3 outcomes: time until first CHD event
(myocardial infarction (MI), revascularization, and
hospitalized angina), time until CHD death or first
MI, and time until CHD death. The urban sprawl
index captured residential density, mixed land use,
the concentration of development within the central
city, and street connectivity for 83 MSAs.
Population Studied: The sample (n=44,418, after
limiting to WHI CT women living in MSAs with
sprawl data) was 79.5% non-Hispanic white, 11.3%
non-Hispanic black, 4.5% Hispanic, and 4.7% other;
60.0% were married at baseline, 94.6% had at least
a high school education, and 65.9% had household
incomes between $20,000 and $75,000 (categories
ranged from <$10,000 to =$150,000). Although
restricting the WHI CT (n=68,132) to women in
MSAs with sprawl data reduced sample size,
demographic characteristics were similar in the full
and restricted samples.
Principal Findings: After controlling for individuallevel sociodemographic characteristics including
age, race, education, income, martial status, region,
family history of MI, and study arm as well as
baseline health measures (BMI, waist hip ratio,
physical activity, self-reported history of diabetes,
hyperlipidemic medication use and/or self-reported
high cholesterol, hypertension) and health
behaviors (smoking, alcohol use, hormone use), we
found that women residing in sprawling MSAs
experienced shorter time to first CHD event or time
to CHD death or MI. Living in an area with less
urban sprawl was associated with decreased risk of
a first CHD event and CHD death or MI (HR = 0.95
per standard deviation increase in the Ewing index,
95% CI = 0.91-0.99; HR = 0.91, 95% CI = 0.860.97, respectively). When we examined the four
components of sprawl individually, residential
density had the most pronouced effect. Living in a
higher residential density MSA was associated with
a decreased risk of CHD event (HR = 0.94, 95% CI
= 0.91-0.97) and CHD death or MI (HR = 0.90, 95%
CI = 0.86-0.95).
Conclusions: Urban sprawl was independently
associated with greater CHD risk, above and
beyond individual-level baseline characteristics.
The index’s residential density component had the
most noteworthy independent effect, suggesting
that a beneficial effect of population may explain the
findings, which held even after adjusting for
neighborhood SES.
Implications for Policy, Delivery or Practice: Our
findings suggest that land use and development
patterns in urban areas may influence
cardiovascular disease risk. Due to their increased
risk of CHD events, screening and prevention
efforts may need to be more aggressive for women
living in areas with greater sprawl.
Funding Source(s): National Heart, Lung, and
Blood Institute
Poster Number: 166
Rural – Urban Disparity in Colon Cancer Screening
among Elder Medicare Beneficiaries: Does Rurality
Matter?
Presented by: Lin Fan, M.S., Student, Community
and Preventive Medicine, University of Rochester,
601 elmwood Avenue, Rochester, NY 14642;
Phone: (585) 705-7534; Email:
lin_fan@urmc.rochester.edu
Co-authors: Katia, Noyes, Ph.D.; Supriya, Mohile,
M.D., M.S.
Research Objective: To examine urban-rural
disparity of Colorectal cancer (CRC) screening
among the elder Medicare beneficiaries and assess
rurality’s independent impact on receipt of CRC
screening.
Study Design: Cross sectional study. We included
community dwelling Medicare beneficiaries 65
years of age and older from 2005 Medicare Current
Beneficiary Survey (N=11,488). Rural-Urban
Commuting Area Codes (RUCAs) were used to
characterize 4 types of rural and urban status. Selfreported time-appropriate CRC screening practice
(fecal occult blood test [FOBT] last year or
colonoscopy / sigmiodoscopy in past 5 years),
FOBT in past 2 years, and colonoscopy /
sigmiodoscopy in past 5 years were compared
among the 4 types of urban and rural areas;
independent impact of rurality on receipt of
screening was evaluated using weighted logistic
regression.
Population Studied: Community dwelling
Medicare beneficiaries 65 years of age and older
Principal Findings: Residents of urban areas were
significantly more likely to receive time-appropriate
screening (55.5% [95% CI: 53.7% - 57.2%] of urban
beneficiaries, 50.1% [95% CI: 45.2% - 54.9%] of
those from large rural / town, 47.5% [95% CI:
43.0% - 52.0%] from small rural / town, and 42.7%
[95% CI: 38.3% - 47.0%] from isolated rural / town ,
P<0.0001). Similar patterns were observed for
FOBT in past 2 years and colonoscopy /
sigmiodoscopy in past 5 years. After controlling for
age, gender, race, income, education, health
insurance, HMO, health status, and health
conditions, rurality was still significantly associated
with time appropriate screening (P<.0001), FOBT in
past 2 years, and colonoscopy / sigmiodoscopy in
past 5 years.
Conclusions: CRC screening is less than
satisfactory, and about half of the elder Medicare
beneficiaries do not receive time-appropriate CRC
screening. There is evidence of urban-rural
disparity in CRC screening among elderly Medicare
beneficiaries.
Implications for Policy, Delivery or Practice:
Special attention should be paid to rural areas,
particular small and isolated rural / towns to
increase CRC screening practice. Further research
is needed to understand the reasons for the urbanrural disparity to better inform policy intervention.
Poster Number: 167
Residential Segregation and the Availability of
Primary Care Physicians
Presented by: Darrell Gaskin, Ph.D., Associate
Professor, African American Studies, University of
Maryland, 2169 Lefrak Hall, College Park, MD
20742; Phone: (301) 405-1162; Email:
dgaskin@aasp.umd.edu
Co-authors: Gniesha Dinwiddie, Ph.D.; Kitty Chan,
Ph.D.; Rachael McCleary, B.A.
Research Objective: To examine the association
between residential segregation and geographic
access to primary care physicians.
Study Design: Using data from the 2006 American
Medical Association master file and the 2000
Census, we computed the population to primary
care physician (PCP) ratio for each zip code in a
metropolitan statistical area (MSA). We defined
PCP shortage areas as those zip codes with no
PCP or a population to PCP ratio of greater than
3500. We excluded zip codes with populations of
less than 200 persons. Using logistic regression
analysis we estimated the association between
whether a zip code was a PCP shortage area and
the minority composition of the zip code and degree
of segregated in the zip code’s MSA. Minority
composition was measured by a set of indicator
variables that denoted whether the zip code was
predominately (50% or more) African American,
predominately Hispanic, or predominately Asian.
We used the Census Bureau’s MSA measures of
segregation. Five segregation measures for each
minority group were considered: dissimilarity index,
isolation index, spatial/clustering index,
centralization index and concentration index. In the
logistic regression, we controlled for distribution of
the population by age, gender, poverty status,
education attainment, and home ownership in the
zip code. We also controlled for clustering within
county.
Population Studied: Our sample consists of
15,900 zip codes located partially or completely in
an MSA or which 15801 have segregation data for
each minority group. Of these zip codes, 4.9% were
predominately African American, 4.4% were
predominately Hispanic, and 0.2% were
predominately Asian.
Principal Findings: Overall we found that
predominately African American zip codes were
more likely to be a PCP shortage area (OR = 1.65 p
= 0.001). However, this relationship varies with the
degree of segregation in the MSA. As segregation
measured by the dissimilarity index, centralization
index and concentration index increased the odds
of being a PCP shortage area increased for African
American zip codes. The odd ratios for low
segregation area ranged from 0.96 to 1.23
compared to high segregation areas which ranged
from 2.16 to 2.70. We found that when the
clustering index for Asian was included in the
model, the predominately Asian zip codes in low
segregation areas were more likely to be PCP
shortage areas (OR=3.43) compared to high
segregation areas (OR = 0.93). We did not find an
association between predominately Hispanic zip
codes and PCP shortage areas.
Conclusions: African American communities in
highly segregated MSAs and Asian communities in
low segregated MSAs are at risk of having too few
PCPs.
Implications for Policy, Delivery or Practice:
Efforts should be made to improve geographic
access to PCPs in African American communities in
segregated MSAs and Asian communities in low
segregated MSAs. Future studies should assess
whether limited geographic access to PCPs in
these predominately African American and Asian
communities contribute to race disparities in the use
of PCP services.
Funding Source(s): NCMHD
Poster Number: 168
Residential Segregation, Physician Shortage and
the Use of Primary Care Services
Presented by: Darrell Gaskin, Ph.D., Associate
Professor, African American Studies, University of
Maryland, 2169 Lefrak Hall, College Park, MD
20742; Phone: (301) 405-1162; Email:
dgaskin@aasp.umd.edu
Co-authors: Gniesha Dinwiddie, Ph.D.; Kitty Chan,
Ph.D.; Rachael McCleary, B.A.; Thomas LaVeist,
Ph.D.
Research Objective: Do residential segregation
and physician availability explain disparities in the
use of primary care services.
Study Design: Using data from the 2006 Medical
Expenditure Panel Survey, the 2000 Census and
the 2006 American Medical Association master file,
we determined whether race/ethnic disparities in
primary care service use were associated with
residential segregation and the availability primary
care physician (PCP). We used 5 measures of
primary care service use: 1) office based physician
visits, outpatient department physician visits, office
based non-physician visits, outpatient department
non-physician visits, and having a usual source of
care (USC). At the individual level we controlled for
age, gender, marital status, insurance status,
educational attainment, employment status, region,
and health status. We linked to the MEPS data, 6
measures of residential segregations: the minority
composition of the zip code and 5 MSA measures
of segregation for each minority group. Minority
composition was measured by a set of indicator
variables that denoted whether the zip code was
predominately (50% or more) African American,
predominately Hispanic, or predominately Asian.
The five segregation measures for each minority
group were the dissimilarity index, isolation index,
spatial/clustering index, centralization index and
concentration index. To measure PCP availability,
we computed the population to primary care
physician (PCP) ratio for the zip code. We defined
PCP shortage areas as those zip codes with no
PCP or a population to PCP ratio of greater than
3500. We estimated logistic and negative binomial
regression analyses assuming primary care use
has two parts: initiation and subsequent use. We
estimated a base model using on the MEPS data,
and then models using each of the segregation and
physician availability variables separately.
Population Studied: We restricted are our sample
to 20775 adults living in MSAs.
Principal Findings: Preliminary findings suggest
that disparities in primary care use were associated
with residential segregation and PCP availability
especially for non-physician services. The impact
varies by segregation measure and by race/ethnic
minority group.
Conclusions: African Americans and Hispanics
who are exposed to segregation are less likely to
use primary care services. Asians in segregation
areas are more likely to use primary care services.
Implications for Policy, Delivery or Practice:
Efforts should be made in improve access to
primary care physician services in African American
and Hispanic communities in segregated areas.
Funding Source(s): NCMHD
Poster Number: 169
Patient Navigation Utilization among Native
American Cancer Patients in the Northern Plains
Presented by: B. Ashleigh Guadagnolo, M.D.,
M.P.H., Assistant Professor, Radiation Oncology,
The University of Texas M.D. Anderson Cancer
Center, 1515 Holcombe Boulevard, Unit 97,
Houston, TX US 77030; Phone: (713) 563-2341;
Email: aguadagnolo@gmail.com
Co-authors: Michelle Sargent, R.N.; Amy Boylan,
R.N.; David Koop, R.N.; Kevin Molloy, R.N.; Daniel
Petereit, M.D.
Research Objective: Native Americans (NA) in the
Northern Plains suffer disparately high cancer
mortality rates compared with the general US
population and NA in other regions. We developed
a culturally-specific patient navigation (PN) program
as a component of a multi-faceted effort to reduce
cancer-related health disparities in this region.
While studies of PN for assisting patients through
the cancer screening/diagnosis process are
available, relatively little published data exists
regarding PN for patients undergoing cancer
therapy. Our goal was assess relative intensity of
PN utilization for different courses of therapy.
Study Design: This is a prospective cohort study to
evaluate PN utilization among NA cancer patients.
The PN program (Walking Forward Program)
provides culturally-competent navigators to assist
patients with navigating cancer therapy, obtaining
medications, insurance issues, communicating with
medical providers, and travel and lodging logistics.
Navigators included trained lay health workers (both
NA and non-NA) as well as individuals with formal
health care training (e.g., R.N.). Lakota language
services and educational materials were available.
All patients received PN throughout cancer
treatment. Additionally, community navigators live
and work on the reservations providing an
embedded resource for cancer patients returning to
their reservation communities following cancer
treatment. The non-parametric Mann-Whitney test
was used to test for differences between mean
number of hospital-based navigator encounters for
subgroups of patients receiving various cancer
treatment regimens.
Population Studied: Hospital-based recruitment of
332 NA cancer patients in Rapid City, SD who
received PN services throughout cancer treatment
between 2/2004 and 9/2009.
Principal Findings: The median participant age
was 60 years (range, 16-89 years). Median
distance lived from the cancer center was 121 miles
(range, 10-444 miles). The majority of patients were
referred for PN for financial assistance (54%) or
logistical support (35%). Other reasons included (>
1 reason possible): psycho-social support (35%),
care coordination (22%), advocacy (20%), or
education (20%). The median number of contacts
with a navigator was 12 (range, 1-119). Median
time spent with the navigator at first contact was 40
minutes (range 10-250 min.) and 20 min for
subsequent contacts. Patients undergoing radiation
therapy (RT) had more contacts with a navigator
than those not receiving RT (mean, 25 vs. 13
contacts, respectively; p=0.0001), as did those
receiving chemotherapy vs. those who did not
(mean, 24 vs. 15 contacts, respectively; p=0.001).
Patients who received both chemotherapy and RT
also had more encounters with a navigator than
those not receiving both modalities (mean, 30 vs.
16 contacts, respectively; p=0.0001). Patients
undergoing curative treatment had an average of 21
contacts with a navigator compared to 17 contacts
for patients receiving palliative treatment (p=0.02).
Conclusions: NA patients who received more
intensive cancer care sought more PN services.
The primary reasons for accessing PN services
included financial assistance and travel/lodging
support.
Implications for Policy, Delivery or Practice:
These data may have implications for resource
allocation for PN in vulnerable populations receiving
intense or complicated treatment regimens.
Research is needed as to whether PN improves
treatment adherence, outcomes, and quality of life.
Funding Source(s): NCI
Poster Number: 170
Disparities in Receipt of Supportive/Palliative Care
Services among Individuals with Lung Cancer
Presented by: Michael Halpern, M.D., Ph.D.,
Senior Fellow, Health Services and Social Policy
Research, RTI International, 701 13th Street NW, #
750, Washington, DC 20005; Phone: (202) 9747813; Email: mhalpern@rti.org
Co-authors: Debra Holden, Ph.D.; Anne Larsen,
M.S.
Research Objective: Increased eM.P.H.asis is
being placed on patient-centered care, particularly
for individuals with serious medical conditions. For
individuals diagnosed with cancer, patient-centered
care involves addressing all symptoms, including
those caused by the underlying disease or by
cancer treatments. However, reports suggest
disparities in receipt of care for symptoms such as
pain, decreased levels of functioning, and
psychological distress among individuals with
cancer. We conducted this study to explore
disparities in receipt of supportive/palliative care
services among an insured elderly population with
lung cancer.
Study Design: We used data from NCI’s
Surveillance, Epidemiology and End Results
(SEER) cancer registry linked to Medicare claims
data (SEER-Medicare) to assess disparities in
receipt of pain management, physical/occupational
therapy (PT), psychological/psychiatric testing or
treatment services during the first 12 months after
diagnosis. Separate multivariate logistic regressions
were performed to assess disparities based on
patient sociodemographic and clinical factors on the
receipt of each category of supportive/palliative
care services. Regressions controlled for cancer
stage; age at diagnosis; sex; race/ethnicity;
rural/urban residence; dual Medicare-Medicaid
enrollment (a proxy for low income status);
comorbidities; and area-based socioeconomic
status indicators.
Population Studied: All individuals in the SEER
registry with a lung cancer diagnosis continuously
enrolled in Medicare parts A and B for at least 12
month following diagnosis were included.
Principal Findings: Among 33,186 individuals with
lung cancer, Black patients were significantly less
likely to receive pain management services or
psychological treatment compared with White
patients. Women were more likely than men to
receive PT and psychological testing or treatment.
Older patients were more likely to receive PT or
psychological testing, but less likely to receive pain
management or psychological treatment compared
with younger patients. Patients with dual MedicareMedicaid coverage were less likely to receive pain
management than were non-duals, but more likely
to receive PT and psychological testing and
treatment. Individuals with multiple comorbidities
were more likely to receive all of the
supportive/palliative care services examined.
Patients from areas with higher rates of adults
having only a high school education were
significantly less likely to receive pain management
or PT. Those from areas with high rates of nonEnglish speakers were also less likely to receive
pain management services, but were more likely to
receive psychological testing.
Conclusions: Significant disparities were observed
in receipt of supportive and palliative care services
among an elderly, insured population with lung
cancer. Disparities based on patient
sociodemographic characteristics were observed
despite controlling for clinical characteristics such
as cancer stage and level of comorbidities.
Implications for Policy, Delivery or Practice:
From these data, we cannot determine whether
these observed disparities are related to differences
in patient preference, clinician practice patterns,
type of health care facility where treatment is
received, and/or other factors. Further research is
needed to understand the basis for these
disparities. In particular, studies need to examine
the effects of patient preference and
recommendations or referrals for
supportive/palliative care services by clinicians on
receipt of these services.
Poster Number: 171
Massachusetts Health Reform and Use of ReferralSensitive Inpatient Procedures
Presented by: Amresh Hanchate, Ph.D.,
Research Assistant Professor, General Internal
Medicine, Boston University School of Medicine,
801 Massachusetts Avenue, #2077, Boston, MA
02118; Phone: (617) 638-8889; Email:
hanchate@bu.edu
Co-authors: Karen Lasser, M.D., M.P.H.; Alok
Kapoor, M.D., M.P.H.; Danny McMormick, M.D.,
M.P.H.; Meredith Manze, M.P.H.; Nancy Kressin,
Ph.D.
Research Objective: Massachusetts (MA) is the
setting of a landmark experiment in health care
reform that has sharply expanded health coverage
since its inception in 2006. In the absence of any
previous research, our objective is to examine
whether expanded insurance coverage has
improved access to selected major inpatient
procedures (total knee replacement (TKR),
percutaneous transluminal coronary angioplasty
(PTCA), coronary artery bypass graft (CABG) and
total/partial hip replacement (THR)) that are
typically underutilized by minority patients.
Study Design: We analyzed MA Hospital Inpatient
Discharge Data on all non-federal MA hospital
discharges from 2004-2008 (fiscal years) for
2,656,554 white patients, 195,481 black patients,
and 172,852 Hispanic patients. State-level annual
counts of individual procedures were stratified by
county, sex, race/ethnicity and age (21-44, 45-54
and 55-64), resulting in n=1,320 observational units
after combining the five years of longitudinal data.
To convert procedure counts into incidence rates
(number of procedures/10,000 population), we
obtained Census population data to create an
‘exposed population measure’ for each
observational unit. We used direct standardization
and hierarchical negative binomial regression
models to estimate changes in the use of
procedures over two years following the onset of
reform. We used the Wald test to estimate
differences in utilization rates among black and
Hispanic non-elderly adults (age 21-64) compared
to their white counterparts. To separate the impact
of health reform from secular temporal effects that
may also effect untargeted population, we use a
difference-in-difference approach contrasting the
pre/post change with those among patients >= age
65, assuming they have continuous Medicare
coverage.
Population Studied: MA-residing patients
discharged from MA hospitals during 2004-2008 for
the following surgical procedures: TKR (n=53,633),
PTCA (n=60,475), CABG (n=16,701) and THR
(n=42,991).
Principal Findings: Between 2006 and 2008, total
state-level volumes increased for TKR (+22%) and
THR (+7%), and decreased for PTCA (-15%) and
CABG (-8%). While there was a secular increase in
TKR and THR incidence rates between 2004-2006
(pre-reform) and 2008 (post-reform), the increases
were larger for blacks (TKR=53% THR=13%; all pvalues<0.05) and Hispanics (TKR=44%; THA=8%
p-values>0.05) than for whites (TKR=24%;
THR=8%). While there was a decrease in PTCA
incidence rates, the decreases for blacks (17%)
were smaller than that for whites (22%; p values
<0.05). For CABG, the decrease in incidence rate
was of similar magnitude across the three cohorts.
While similar changes in rates for minorities were
found among the elderly, their magnitude was
generally smaller; for instance, the increase in
incidence of joint replacement procedures for nonelderly minorities was larger than that for elderly.
Conclusions: Following health reform, there has
been a larger increase (or smaller decrease) in the
use of some common inpatient procedures among
minority non-elderly adults in MA. We will further
examine if utilization differences are associated with
geographical location and access to safety-net
hospitals.
Implications for Policy, Delivery or Practice:
These preliminary findings indicate that insurance
coverage expansion from the MA health reform may
have increased utilization of procedures among
racial and ethnic minorities due to pent-up demand.
Funding Source(s): NIH/NINDS
Poster Number: 172
Racial & Ethnic Disparities in Inpatient Outcomes
for Ischemic Stroke Admissions: A Stratified
Socioeconomic Status Approach
Presented by: Amresh Hanchate, Ph.D.,
Research Assistant Professor, General Internal
Medicine, Boston University School of Medicine,
801, Massachusetts Avenue, #2077, Boston, MA
02118; Email: hanchate@bu.edu
Co-authors: Elaine Hylek, M.D.; Griffith Bell, M.Sc.
Research Objective: Even though there are over
half a million inpatient admissions for acute stroke
in the US each year, little attention has been given
to examining racial and ethnic disparities in their
outcomes, especially adverse outcomes such as
inpatient mortality and paralysis. This may partly be
due to the fact that unadjusted inpatient mortality
rates are lower for minorities compared to whites;
given that minority stroke patients are relatively
younger, it appears that these rates obscure large
differences in risk factor profile by race and
ethnicity. Adjusting for these differences, our
objective is to quantify the extent of differences in
adverse outcome rates by race and ethnicity. In
particular, our focus is on these differences
associated with poorer socioeconomic status
among minorities. Such an investigation requires
stratifying each racial and ethnic cohort by
socioeconomic status and examining the immense
diversity within each (racial/ethnic) cohort.
Stratifying each cohort by median zip-code income,
we estimate rates of three adverse outcomes for
patients admitted for acute stroke -- inpatient
mortality, paralysis and coma, and compare these
rates across income strata within same racial/ethnic
cohort.
Study Design: Pooling the universe of inpatient
discharge records from four states (Arizona,
Massachusetts, Pennsylvania and Texas), we
stratified all admissions for ischemic stroke for
adults >= 45 years of age by sex, age (44-64, 6574, 75-84 and 85+), race/ethnicity (white, black,
Hispanic and other) and zip-code median income
(first quartile, second quartile, top two quartiles
combined). Paralysis and coma were defined based
on secondary diagnosis codes. Secondary
diagnosis codes (ICD-9) were used to identify
paralysis and coma, and to identify patient risk
factors (including hypertension, heart failure and
diabetes). Pooled multilevel logistic regression
models were estimated to measure mean
differences in outcomes across strata.
Population Studied: All patients admitted with the
principal diagnosis of ischemic stroke during 20052008 in all civilian hospitals in AZ (n=15,831), MA
(n=16,438), PA (n=17, 958) and TX (n=20,066).
Principal Findings: Unadjusted mean outcomes
rates were 5.1% (inpatient mortality), 22.6 percent
(paralysis) and 1.3 percent (coma). Confirming
previous findings, unadjusted inpatient mortality
was lowest for blacks (3.8%), followed by Hispanics
(4.1%) and whites (5.4%). However, comparing the
income gradients for inpatient mortality for the three
racial and ethnic cohorts, the gradient for blacks is
the steepest, followed by that for Hispanics. Among
blacks, risk-adjusted inpatient mortality ranged from
2.7% among the highest income cohort to 4.5%
among the poorest cohort (66 percent higher; p
<0.01); the corresponding excess mortality was
52% among Hispanics and 31% among whites.
Similar comparison for paralysis outcome indicated
that the gradient for Hispanics was the steepest. No
significant gradient was found for coma.
Conclusions: Among minority patients admitted for
ischemic stroke, those residing in poorer
neighborhoods experienced worse outcomes
(inpatient death and paralysis) than their
counterparts living in higher-income neighborhoods;
in comparison, this gap is narrower among white
patients.
Implications for Policy, Delivery or Practice:
While mean rates of adverse inpatient outcomes for
patients admitted for ischemic stroke are lower for
minorities in general, compared to those for whites,
they obscure substantial disparities associated with
poorer socioeconomic status. One implication is
that barriers to care may be potentially higher for
poorer minorities when compared to those for
poorer whites.
Funding Source(s): NINDS
Poster Number: 173
Factors Associated with Early Postpartum
Depression Among Black and Latino Mothers
Presented by: Elizabeth Howell, M.D., M.P.P.,
Assistant Professor, Health Evidence & Policy,
Mount Sinai School of Medicine, One Gustave L.
Levy Place, Box 1077, New York, NY 10029;
Phone: (212) 659-9567; Email:
elizabeth.howell@msnyuhealth.org
Co-authors: Jason Wang, Ph.D.; Howard
Leventhal, Ph.D.
Research Objective: To assess the relationship
between race, demographic factors, and clinical
factors with early reported postpartum depression
among Black and Latino mothers.
Study Design: As part of a randomized controlled
trial, enrolled mothers are interviewed at baseline
during their postpartum hospital stay to assess
depression, anxiety, and other factors. Depression
is assessed using a validated depression screening
instrument, the Edinburgh Postnatal Depression
Scale. The associations of early postpartum
depression with race, other demographic
characteristics, past history of depression, and
clinical factors were examined using bivariate and
multivariable analyses.
Population Studied: 460 self-identified Black and
Latino postpartum women who had uncomplicated
deliveries at an urban hospital. Eligible subjects are
women >18 years of age, English or Spanish
Speaking, had working telephones, and had infants
whose birthweights were > 2500 grams and 5minute Apgar scores >6. The recruitment target for
this randomized controlled trial is to enroll 530
Black and Latino mothers.
Principal Findings: Mean age was 28 (range 1846); 66% of the sample were Latino and 34% were
Black. Thirty-six percent were foreign born, 21%
spoke Spanish, 62% had Medicaid insurance and
56% earned < $30,000 annually. Overall 15%
screened positive for depression (EPDS score >10)
and there were no racial/ethnic differences in rates
of depression. Depression was more common in
mothers born in the US than mothers born outside
of the US (10% vs. 18%, p=.01), English speakers
than Spanish speakers (17% vs. 8%), p=.02), single
mothers as compared with married mothers (19%
vs. 13%, p=.056), mothers with comorbid illness vs.
mothers without comorbid illness (22% vs. 13%,
p=.02), and mothers with a past history of
depression as compared with mothers without a
past history of depression (47% vs. 9%, p<.0001).
Rates of depression did not differ by insurance
status, delivery type, or presence or absence of
antepartum complications. After adjusting for race,
other demographic factors, past history of
depression, delivery type, comorbid illness, and
complication status, depression was more common
in mothers with a past history of depression (OR of
8.21, 95% CI: 4.47, 15.08).
Conclusions: Racial/ethnic differences in rates of
reported postpartum depression did not exist
among Black and Latino early postpartum mothers.
Past history of depression in this low income
population of mothers was the strongest correlate of
early postpartum depression.
Implications for Policy, Delivery or Practice:
Further research is needed to assess the impact of
early depression screening on the mental health
outcomes of high risk women, particularly in Black
and Latino communities.
Funding Source(s): National Center for Minority
Health and Health Disparities
Poster Number: 174
How are Iraq/Afghanistan Veterans with Health
Problems Faring in the Labor Market?
Presented by: Jennifer Humensky, Ph.D., PostDoctoral Fellow, Center for Management of
Complex Chronic Care, Edward Hines Jr VA
Hospital, 5000 S. Fifth Avenue (151H), Building 1,
B251, Hines, IL 60141; Phone: (708) 202-5863;
Email: Jennifer.Humensky@va.gov
Co-authors: Denise M. Hynes, Ph.D.; Neil Jordan,
Ph.D.; Kevin Stroupe, Ph.D.
Research Objective: Veterans returning from
World War II and the Korean War generally fared
well in the labor market following their service.
However, this pattern was different for Vietnam
veterans, who encountered more difficulties in the
labor market. Previous research has shown that
veterans in better health and those who learned
skills more translatable to the private sector fared
better in the job market. The popularity of specific
wars has also been shown to affect re-entry into the
job market. However, the labor market experience
for Iraq/Afghanistan veterans is not wellunderstood. While the Iraq/Afghanistan wars are
largely unpopular with the American public, support
for the troops is high, thus the impact of any stigma
is uncertain. Our research objective was to examine
the labor market experience of Iraq/Afghanistan
veterans, and then to examine whether labor
market outcomes varied by health status, and
whether this relationship has varied over time.
Study Design: We used a multilevel, repeated
observations model to identify a trajectory of labor
market outcomes (participation in the labor force,
probability of any earnings, and enrollment in
school) for veterans in good and poor health. The
empirical model was based on the Ownsworth and
McKenna (2004) model of labor market participation
of persons with traumatic injuries, and controlled for
a variety of individual and contextual characteristics
associated with labor market outcomes, including
demographics, education level, previous military
service, locality, and state unemployment rate.
Population Studied: We analyzed the most recent
three panels from the March Current Population
Survey (CPS): 2006-2007, 2007-2008 and 20082009. Information on service in Iraq/Afghanistan
was not available in CPS prior to 2006. The 20082009 cohort included 125,623 persons age 18-65;
about 925 were Iraq/Afghanistan veterans. Of the
Iraq/Afghanistan veterans, about 86% were in the
labor force, about 89% had earnings and about 5%
were enrolled in school.
Principal Findings: Iraq/Afghanistan veterans in
good health had similar employment outcomes to
non-veterans. No statistically significant differences
were found between veterans and non-veterans in
good health in 2009 in labor force participation and
probability of any earnings. Veterans were less
likely to be enrolled in school (ß=-0.06, p<0.01)
than non-veterans. However, veterans in fair or
poor health in 2009 had lower odds of labor force
participation (OR=0.11, p<0.01), and any earnings
(OR=0.74, p<0.01), compared to veterans in good
health. Results were generally consistent from
2006-2009.
Conclusions: Although Iraq/Afghanistan veterans
in good health have been doing well in the labor
market, veterans in fair or poor health have been
encountering more difficulties. Persons in poor
health generally perform less well in the labor
market, and this holds true for veterans. More
research is needed to determine which health
problems are associated with employment
problems and whether participation in specific
treatment programs results in better outcomes.
Implications for Policy, Delivery or Practice:
Returning veterans with health problems may be
especially vulnerable to employment difficulties.
The Veterans Administration and other veteran
service organizations should consider targeted
approaches to help veterans succeed in the labor
market.
Poster Number: 175
The Impact of Race and Hospital Racial Makeup on
Readmissions for Heart Failure
Presented by: Karen Joynt, Fellow, Health Policy
and Management, Harvard School of Public Health,
677 Huntington Avenue, Boston, MA 02115;
Phone: (617) 432-5551; Email:
kjoynt@partners.org
Co-authors: Ashish K. Jha, M.D., M.P.H.
Research Objective: Reducing readmissions for
patients admitted with heart failure (HF) is a major
policy initiative, partly driven by the fact that HF is
the most common cause of hospitalizations and
readmissions in the Medicare program. Prior
studies have suggested that blacks may have
higher readmission rates than white patients, yet
the reason for this disparity is not well understood.
We sought to determine whether location of care
plays an important role in racial disparities.
Study Design: We used Medicare claims data from
2006-2007 to categorize hospitals based on the
proportion of black patients they cared for,
designating those hospitals in the top decile as
“minority-serving.” We created risk-adjusted
generalized linear regression models, accounting
for clustering within hospitals, with patient race and
hospital racial makeup as our primary predictors,
and 30-day readmission as our primary outcome.
Population Studied: 699,450 Medicare patients
with a primary discharge diagnosis of HF.
Principal Findings: In analyses fully adjusted for
patient characteristics, black patients had higher
30-day readmission rates (24.4% versus 23.7%,
p<0.0001) than white patients, but this difference
disappeared after accounting for where black
patients received their care: adjusting for whether
blacks were admitted to minority-serving hospitals
or not, we found no racial difference in readmission
rates (25.0% for black patients versus 24.7% for
white patients, p=0.15). Risk-adjusted 30-day
readmission rates were higher at minority-serving
hospitals overall (26.0%) than at non-minorityserving hospitals (23.6%, p<0.0001), and this
difference was consistent after controlling for race
(26.0% versus 23.7%, p<0.0001). At minorityserving hospitals, black patients had a slightly
higher 30-day readmission rate than white patients
(26.6% versus 25.6%,p=0.004). However, there
were no disparities in readmission rates at nonminority-serving hospitals: both black and white
patients had identical 30-day admission rates,
at23.5% (p=0.96).
Conclusions: We found that elderly black patients
in the U.S. overall have higher 30-day readmission
rates than white patients. However, this effect
disappeared when taking into account the hospitals
where black patients receive a large proportion of
their care: minority-serving hospitals had
significantly higher readmission rates, and
accounting for the location in which patients sought
care explained most of the national disparity in
readmission rates. In fact, black patients at nonminority-serving hospitals had significantly lower
admission rates than white patients at minorityserving hospitals.
Implications for Policy, Delivery or Practice:
Racial disparities in HF outcomes seem to be
primarily a systems problem – the hospital at which
a patient seeks care, and not patient race, is the
primary driver of racial disparities in readmission.
These findings suggest that focusing on quality
improvement at poor-performing, minority-serving
hospitals may have the potential to improve quality
of care for all HF patients while simultaneously
reducing disparities in HF care.
Funding Source(s): NIH Training Grant
Poster Number: 176
Exploring Cultural Sensitivity in the National Drug
Abuse Treatment System Survey
Presented by: Mimi Kim, Ph.D., Investigator, Cecil
G. Sheps Center for Health Services Research,
University of North Carolina at Chapel Hill, 725
Martin Luther King Jr. Boulevard, CB#7590, Chapel
Hill, NC 27599-7590; Phone: (919) 843-6883;
Email: mimi@unc.edu
Co-authors: Daniel L. Howard, Ph.D.; Paul Godley,
M.D., Ph.D., M.P.P.
Research Objective: Evidence suggests that one
of the most effective strategies for addressing the
racial disparities in substance abuse treatment is to
provide services in a culturally sensitive manner yet
defining cultural sensitivity remains complicated. A
limited amount of research has suggested the
importance of client-provider ethnic matching as
one dimension of cultural sensitivity that could be
effective in the delivery of health services. Cultural
sensitivity is one way for organizations to promote
better outcomes for the repeat clients, many of
whom are minorities, they treat. Despite the
apparent outcome advantage to providing culturally
sensitive services in treatment, little research has
gone further to examine other service
characteristics that may define culturally sensitive
treatment. This analysis assesses the extent to
which service and organizational level variables
predict cultural sensitivity, as defined by a clientprovider ethnic matching in addition to cultural
sensitivity, over time in outpatient substance abuse
treatment (OSAT) units.
Study Design: The data were collected for the
1995 and 2005 National Drug Abuse Treatment
System Survey (NDATSS) data set. Repeated
measure models examined the relationship of
cultural sensitivity over time to several service,
organizational, and client characteristics variables.
Population Studied: The unit of analysis was a
physical facility with resources dedicated primarily
to treating individuals with substance abuse
problems on an ambulatory basis.
Principal Findings: In the multivariable analysis
final models, results indicate that OSAT units with
smaller client/staff ratios, affiliations with hospitals,
clients who are younger than 50 years of age, and a
higher percentage of dually diagnosed clients are
more likely to be culturally sensitive.
Conclusions: Focusing on service provision and
organizational characteristics that positively predict
cultural sensitivity can inform various ways to foster
and develop culturally sensitive treatment practices
and longer, more positive client outcomes.
Implications for Policy, Delivery or Practice:
These findings are significant for several reasons.
First, this analysis uses a nationally-representative
sample and thus the data are particularly useful for
understanding the OSAT system as a whole.
Second, because the study uses a split-panel
design, this analysis examines changes in the
nation’s outpatient drug abuse treatment system
over time as well as providing an accurate picture of
the system (i.e., as a result of adding treatment
units to the sample at each wave of data collection).
Third, this analysis is examining treatment practices
that have a potentially significant empirical
relationship to specific treatment outcomes and the
long-term well-being of an underserved population.
Therefore, examining specific treatment
characteristics as they relate to specific treatment
outcomes can aid researchers, policymakers, and
clinical managers to examine the extent to which
their treatment system is meeting clients’ needs and
where the strengths and weaknesses of the system
may lie. Finally, this analysis examines
organizational characteristics that may be
associated with changes and variation in treatment
practices that significantly impact long-terms
outcomes in the client population. By looking at
these associations, this analysis identifies specific
characteristics that contribute to a treatment units’
ability to establish longer, more positive client
outcomes or areas of improvement related to
positive outcomes.
Funding Source(s): NCMHD
Poster Number: 177
Predicting Glaucoma Diagnosis in an Elderly
Sample: Revisiting the Established Populations for
Epidemiologic Studies of the Elderly Study
Presented by: Mimi Kim, Ph.D., Investigator, Cecil
G. Sheps Center for Health Services Research,
University of North Carolina at Chapel Hill, 725
Martin Luther King Jr. Boulevard, CB#7590, Chapel
Hill, NC 27599-7590; Phone: (919) 843-6883;
Email: mimi@unc.edu
Lead Author: Daniel Howard, Ph.D.
Research Objective: The prevalence of vision
impairment and blindness are higher among older
individuals in the United States. Furthermore,
African-Americans are twice as likely to be blind or
visually impaired compared to their Caucasian
counterparts. There is also evidence of several
potential risk factors associated with vision loss in
addition to or independent of natural aging.
Research has found a wide range in the prevalence
of vision impairment among different racial groups.
Though research has examined race as an
important factor in defining the scope of glaucoma
in the United States, further exploration into other
specific factors may help inform the design of
preventive services to decrease rates of blindness
due to glaucoma among minority groups. There
remains an underreporting of individual and service
factors that may predict glaucoma diagnosis by
race. This paper intends to further analyze the
association between glaucoma diagnosis and
several other covariates, including service level
factors, with particular attention to race among the
EPESE participants.
Study Design: This analysis employed generalized
estimating equations (GEE) to simultaneously test
for trends across time for each of the demographic,
health characteristics, health services, and
healthcare provider characteristic domains
according to each racial group using a logistic link
function in the GEE analysis.
Population Studied: This observational,
longitudinal study is based on participants from the
North Carolina Established Populations for the
Epidemiologic Studies of the Elderly (NC EPESE).
Principal Findings: When controlling for patient
health characteristics for both racial groups, having
diabetes is a significant predictor for Caucasians
and a marginally significant predictor for AfricanAmericans but in opposition between the two
groups. Specifically, Caucasians without diabetes
are more likely to have a glaucoma diagnosis. In
contrast, having diabetes was a marginally
significant predictor of a glaucoma diagnosis among
African-American study participants. Also, the
slightly higher number of women in the sample in
relation to the prevalence of glaucoma suggests a
possible association between gender and glaucoma
as suggested by other research.
Conclusions: These findings suggest that specific
demographic, patient health characteristics, health
provider characteristics, and the burden of disease
may be important predictors of glaucoma diagnosis
over time for African-Americans and should be
further investigated. Furthermore, the finding for
diabetes in the multivariable models suggests a
potentially important physiological difference
between the racial groups in relation to glaucoma.
Implications for Policy, Delivery or Practice: The
findings suggest that working Caucasians who have
cataracts, but not diabetes, and are hesitant to see
a doctor when necessary are more likely to have a
glaucoma diagnosis. The diagnosis of cataract is
common and this finding may represent an
association with having a condition to seek eye care
more than a true association between cataracts and
glaucoma. Furthermore, glaucoma may be more
likely in African-Americans who have cataracts with
co-occurring diabetes and have been dissatisfied
with the medical care they have received. In sum,
this analysis provides individual-level characteristics
and different service-level issues within different
ethnic groups. These data may help to design
studies to develop and test more precisely targeted
interventions to improve a full range of healthcare
services that are often needed by elderly patients.
Funding Source(s): NCMHD
Poster Number: 178
Examining Race and Depression among a Sample
of Aging Medicaid Patients: A Longitudinal Analysis
Presented by: Mimi Kim, Ph.D., Investigator, Cecil
G. Sheps Center for Health Services Research,
University of North Carolina at Chapel Hill, 725
Martin Luther King Jr. Boulevard., CB#7590,
Chapel Hill, NC 27599-7590; Phone: (919) 8436883; Email: mimi@unc.edu
Lead Author: Daniel Howard, Ph.D.
Co-authors: Mimi Kim, Ph.D.; Paul Godley, M.D.,
Ph.D., M.P.P.
Research Objective: Research has shown that
depression is associated with morbidity, mortality,
decreased quality of life, increased medical costs,
and the co-occurrence with a myriad of other
diagnoses. Rates of depression for older,
community dwelling adults have ranged between
one-fourth to one-fifth percent. Service barriers in
addition to low detection, misdiagnosis, and lack of
treatment are particular concerns for older adults
with depresion. Furthermore, due to the high needs
of aging populations, depression is also often
placed as a lower treatment priority for older adults,
especially when coupled with such a high rate of
physical health comorbidities. Yet, despite the
prevalence of depression, its documented
association to other chronic medical conditions, and
the inarguable impact of socioeconomic conditions
(e.g. race, income) on both physical and mental
health status, there is a paucity of research that has
examined depression among the aging and
differences between races. This study explores
what differences there are between AfricanAmericans and Caucasians in terms of major, mild,
and other depression when controlling for other
demographic and treatment variables.
Study Design: Analyses for this study employed
generalized estimating equations (GEE) to
simultaneously test for trends across time for each
of the demographic, geographic areas, types of
mental health treatment, antidepressant drug class,
and principal diagnosis domains across the three
different depression categories using a logistic link
function in the GEE analysis.
Population Studied: This study used 2000 and
2004 Medicaid claims data from six states
(Michigan, New Jersey, Florida, Georgia, Colorado
and Washington). Participants were included
according to the following criteria: 55 years or older;
African-American or Caucasian; and at least one
long term care Medicaid claim.
Principal Findings: The three individual,
multivariate logistic models look at predicting major
depression, minor depression, and other
depression. When looking at the results of the
multivariable model for major depression, according
to domains for the ease of discussion here, older
African-American men who are not Medicare
recipients are less likely to have a diagnosis of
major depression. Similarly, older African-American
men who are not Medicare recipients are also less
likely to have a diagnosis of minor depression when
controlling for all other variables in the model over
time. But older African-American men who are not
Medicare recipients are more likely to have a
diagnosis of other depression.
Conclusions: With such variability in severity of
diagnosis comes the challenge of service providers
to provide appropriate and patient-specific care for
aging adults who face a myriad of psychosocial and
physical challenges.
Implications for Policy, Delivery or Practice:
This analysis strives to provide a stepping stone
from which future research can examine depression
in order to contribute to treatment development for
this high-need population. Considering the unique
needs of individuals with varying levels of
depression can help to tailor interventions to more
effectively target the needs of aging individuals who
are challenged in coping with their mental health
needs.
Funding Source(s): NCMHD
Poster Number: 179
Perinatal Depression among Hispanic Women:
Defining Depression and Treatment Preferences
Presented by: Sandraluz Lara-Cinisomo, Ph.D.,
Assistant Professor, Child and Family
Development, University of North Carolina,
Charlotte, 9201 University City Boulevard,
Charlotte, NC 28223; Phone: (704) 687-8179;
Email: slaracin@uncc.edu
Co-authors: Katherine Wisner, M.D.; Rachel
Burns, M.P.H.; Diego Chavez-Gnecco, M.D.
Research Objective: Perinatal depression is a
significant mental health condition that may lead to
serious health and mental health outcomes for
mothers and their children (Weissman et al., 2004).
There has been a growth in the number of studies
examining treatment options for perinatal
depression including the use of antidepressants
and psychotherapy. However, most studies have
focused primarily on white women (Gaynes et al.,
2005). Given the expected rapid growth of the
Hispanic community and the public health
importance of treating depression, it is vital that we
have a better understanding of how Hispanic
women define depression and determine their
depression treatment preferences if we are to
effectively and appropriately treat Hispanic perinatal
women suffering with depression. To address this
need, we conducted a qualitative study with
Hispanic perinatal women to gather information
about their definition of depression, determine their
treatment preferences and establish guidelines for
implementing their preferred mental health
treatments.
Study Design: Focus group interviews were
conducted with Hispanic perinatal women. A
standardized interview protocol available in Spanish
and English was used to guide the interviews. Five
areas of inquiry guided the interviews: Defining
depression and postpartum depression, Treatment
Preferences, Barriers to Treatment, and
Suggestions for treatment delivery. Qualitative data
analyses were conducted with the focus group
interview as the unit of analysis.
Population Studied: Twenty-one prenatal and
postpartum Hispanic women participated in focus
group discussions.
Principal Findings: Our analyses revealed that
Hispanic women in our study define depression as
a set of symptoms: crying, irritability and
withdrawal. With regard to coping with these
symptoms, the data revealed three primary
approaches, including seeking social support from
family, friends, distracting oneself (i.e., focusing on
a specific task) and withdrawing from others.
We also found a hierarchy of treatment
preferences, with use of their own coping as the
preferred approach followed by the use of social
support, and third, but less desirable, was seeking
help from a mental health profession. However,
there were a lot of misconceptions about what a
mental health provider can do to help. There were
also a number of fears and noted risks associated
with seeking professional mental health care, such
as feeling pressured to take antidepressants or
losing custody of their child. The use of
antidepressants as a treatment option was not
endorsed. However, participants said they would be
more willing to take antidepressants after the birth
of their child than during pregnancy and only if no
other treatments were effective.
Conclusions: Given the treatment preferences
reported, it is important that health care providers
and researchers ensure that perinatal depression
interventions address the treatment preferences of
the target population.
Implications for Policy, Delivery or Practice: If
we are to address the mental health needs of
Hispanic perinatal women suffering with
depression, it will be vital that we take great care in
providing culturally appropriate health care services
that meet the social and mental health needs of this
population.
Poster Number: 180
Racial/Ethnic Disparities in the Health and
Healthcare of U.S. Adolescents: Has Anything
Changed Over the Years?
Presented by: Hua Lin, Ph.D., Biostatistical
Consultant III, Department of Pediatrics, University
of Texas Southwestern Medical Center, 5323 Harry
Hines Boulevard, Dallas, TX 75390; Phone: (214)
648-5034; Email: hua.lin@utsouthwestern.edu
Lead Author: May Lau, M.D., M.P.H.
Co-authors: Rosa Avila, M.S.P.H.; Hua Lin, Ph.D.;
Glenn Flores, M.D.
Research Objective: Little is known about
racial/ethnic disparities in health and healthcare of
adolescents, and whether these disparities have
changed over time. The goal of this study was to
examine adolescent racial/ethnic disparities in
medical and oral health, access to care, and use of
services, and to determine whether these
disparities changed from 2003-2007.
Study Design: The National Survey of Children's
Health (NSCH), sponsored by the Maternal and
Child Health Bureau and conducted by the National
Center for Health Statistics using the State and
Local Area Integrated Telephone Survey
mechanism, was a random-digit-dial telephone
survey conducted in 2003 and 2007 on a nationally
representative sample of households with children
0-17 years old. Disparities were examined for 40
measures of medical and oral health, access, and
use of services for youth 10-17 years old. Bivariable
and multivariable analyses were used to identify
disparities in African-American, Latino,
Asian/Pacific Islander, Native American, and
multiracial youth, compared with white youth. Zscores were used to examine time trends from 2003
to 2007.
Population Studied: A nationally representative
sample from the 2003 and 2007 NSCH (n=48,741
and n=45,897, respectively) of US households of
adolescents 10-17 years old at the time of the
interview.
Principal Findings: After adjusting for eight
socioeconomic factors in multiple logistic regression
analyses of the 2007 wave, persistent disparities
were noted over time for one or more racial/ethnic
minority groups. For example, minority adolescents
were more likely to have the following: suboptimal
medical and oral health, overweight/obesity,
asthma, no health insurance, no personal doctor or
nurse (PDN), insufficient time spent with the PDN,
problems obtaining specialty care, and no specialty
or mental health care in the past year. Compared
with white youth in 2007, Latino, African-American,
and multiracial youth had about double the odds of
suboptimal medical and oral health, and Latinos,
African-Americans, Native Americans, and
multiracial youth had significantly greater odds of
being uninsured and having no PDN. Certain
disparities were especially prominent for specific
racial/ethnic groups: for Latinos, having suboptimal
medical health; for African-Americans, having
suboptimal oral health, being overweight/obese,
and having sporadic insurance; for Asian/Pacific
Islanders, having no specialty and mental health
care in the past year; and for Native Americans,
having behavior problems, no medical insurance,
no PDN, insufficient time spent with PDN, problems
getting specialty care, and needing more medical
care than others. Multiracial youth also had many
disparities compared with white adolescents. Most
racial/ethnic health and health care disparities
persisted between 2003 to 2007, and two
significantly worsened: problems obtaining specialty
care in African-Americans, and ADHD in multiracial
youth.
Conclusions: US adolescents experience many
racial/ethnic disparities in physical and oral health,
access to care, and use of services. Almost all
disparities persisted from 2003 to 2007, while some
disparities worsened over time.
Implications for Policy, Delivery or Practice:
Results from this study may inform future research
and policy on reducing adolescent racial/ethnic
disparities in health and healthcare.
Poster Number: 181
The Primary Language Spoken at Home and Health
and Healthcare Disparities in U.S. Adolescents:
Has the Tincture of Time Reduced Disparities?
Presented by: Hua Lin, Ph.D., Biostatistical
Consultant III, Department of Pediatrics, University
of Texas Southwestern Medical Center, 5323 Harry
Hines Boulevard, Dallas, TX 75390; Phone: (214)
648-5034; Email: hua.lin@utsouthwestern.edu
Lead Author: May Lau, M.D., M.P.H.
Co-authors: Rosa Avila, M.S.P.H.; Hua Lin, Ph.D.;
Glenn Flores, M.D.
Research Objective: Little is know about
disparities and their trends over time in youth from
non-English-primary-language (NEPL) homes
compared with youth from English primarylanguage (EPL) homes, and even less is known
about disparities among NEPL youth. The goal of
this study was to examine disparities and their
trends over time in medical and oral health, access
to care, and use of health services in NEPL youth.
Study Design: The National Survey of Children's
Health, sponsored by the Maternal and Child Health
Bureau and conducted by the National Center for
Health Statistics using the State and Local Area
Integrated Telephone Survey mechanism, was a
random-digit-dial telephone survey conducted in
2003 and 2007 of households with children 0-17
years old. Bivariable and multivariable analyses
were performed for all youth 10-17 years old to
identify disparities in 40 measures of medical and
oral health, access to care, and use of services.
Among NEPL youth, sub-analyses compared
Latinos, Asian/Pacific Islanders (APIs), and whites.
Z-scores were used to analyze time trends in
disparities from 2003-2007 (significance tested at
a<.05).
Population Studied: A nationally representative
sample from the 2003 and 2007 NSCH (n=48,741
and n=45,897, respectively) of US households with
adolescents 10-17 years old at the time of the
interview.
Principal Findings: In multivariable analyses
adjusting for 8 sociodemographic factors, NEPL
youth had significantly greater adjusted odds than
EPL youth of having suboptimal medical and oral
health status, no medical insurance, no personal
doctor/nurse, and no specialist or routine preventive
dental visit in the past year. In the sub-analysis for
the 2007 wave, compared with NEPL white youth,
NEPL Latinos had greater odds of having
suboptimal medical (odds ratio [OR], 4.8; 95%
confidence interval [CI], 2.5-9.9) and oral (OR, 2.4;
95% CI, 1.1-5.4) health, no medical insurance (OR,
2.7; 95% CI, 1.1-6.5,), and unmet dental care needs
(OR, 16.2; 95% CI, 1.7-154.1). NEPL APIs had
greater odds of no mental health visit in the past
year (OR, 7.2; 95% CI, 1.4-36.3), and NEPL Latinos
and APIs had greater odds of having no personal
doctor/nurse (OR, 3.9; 95% CI, 1.5-10.1; OR, 5.7;
95% CI, 1.4-22.8) and no preventive dental care in
the past year (OR, 4.7; 95% CI, 1.4-15.4; OR, 4.4;
95% CI, 1.2-16.1). Almost all disparities for NEPL
youth (vs. EPL youth) persisted between 2003 and
2007. Two disparities significantly worsened: NEPL
vs. EPL youth needing more medical care than
others, and for NEPL Latinos vs. whites, having
ADHD.
Conclusions: NEPL youth experience multiple
disparities in medical and oral health, access to
care, and use of health services, and almost all
disparities persisted over time. Compared with
NEPL white youth, NEPL Latinos and APIs also
face many disparities, most of which have
continued over time.
Implications for Policy, Delivery or Practice:
Results of this study may help target policy to
improve health and healthcare disparities in NEPL
youth, including reducing language barriers for
NEPL youth.
Poster Number: 182
Promoting Access and Reducing Out-of-Pocket
Prescription Drug Costs for Vulnerable Medicare
Part D Patients
Presented by: Helene Lipton, Ph.D., Professor,
Institute for Health Policy Studies, University of
California, San Francisco, 3333 California Street,
Suite #265, San Francisco, CA 94118; Phone:
(415) 476-2964; Email:
smitha@pharmacy.ucsf.edu
Co-authors: Amanda Smith, M.P.H.; Timothy
Cutler, Pharm.D.; Marilyn Stebbins, Pharm.D.
Research Objective: Given the complexity of the
Medicare Part D prescription drug benefit and
annual plan changes, many Medicare beneficiaries
often lack knowledge and skills needed to select an
optimal Part D plan. This problem is especially
prevalent in vulnerable populations, where lack of
resources and limited English proficiency are
barriers to navigating the benefit effectively. Our
objective was to reach these vulnerable populations
and lower their annual out-of-pocket (OOP)
prescription drug costs through one-on-one
prescription drug plan counseling.
Study Design: Students from seven California
schools of pharmacy were trained in Part D to
provide one-on-one counseling in multiple
languages, under faculty supervision. Outreach
events were conducted primarily during the Part D
open enrollment period, which occurs each year
between November 15 and December 31. Annual
estimated OOP costs for each participant’s current
prescription drug plan and the lowest-cost plan (if
different) were obtained by entering their current
medication regimen online, using the Medicare
Prescription Drug Plan Finder tool. Prescription
drug cost-related non-adherence behaviors also
were identified through a questionnaire.
Population Studied: Outreach was targeted
toward vulnerable Medicare beneficiaries, defined
as low-income individuals whose annual income is
less than 300% of the Federal Poverty Level.
Events were conducted in low-income clinics,
housing complexes, community centers and other
settings serving vulnerable Medicare patients. From
October 2008 through December 2009, 90 outreach
events were conducted and 1,552 beneficiaries
received individualized Part D counseling. The
majority of those counseled (80%, n=1,245) were
identified as low-income, using the criterion
specified above. Our study population of 1,029 is a
subset of these low-income beneficiaries who were
enrolled in a stand-alone prescription drug plan (as
opposed to a Medicare Health Plan, which provides
both medical care and prescription drug coverage).
Principal Findings: Only 26% of the study
population (n=243) were enrolled in the lowest-cost
prescription drug plan. Based on counseling
recommendations, 400 beneficiaries enrolled in a
lower-cost Part D plan, going from a median plan
cost of $354.75 to $124.50, for an average OOP
cost-savings of 69%. Of the 1,029 beneficiaries,
949 (92%) were receiving additional governmental
assistance with their prescription medication costs
(i.e., Medicaid or the low-income subsidy (LIS));
867 (84%) were from ethnic/racial minority groups;
624 (61%) had limited or no English proficiency;
and 490 (48%) did not complete high school. In
addition, 50 beneficiaries were identified as LISeligible and received assistance with their LIS
applications during counseling. Of note, 304 (30%)
beneficiaries reported prescription drug cost-related
non-adherence behaviors during the last year,
including not filling prescriptions due to cost and/or
skipping doses.
Conclusions: Evidence suggests that increases in
out-of-pocket prescription drug costs lead to nonadherence behaviors with adverse health
consequences. Our findings suggest that targeted
outreach by trained pharmacy advocates can locate
vulnerable patient populations in need of Part D
counseling and successfully reduce their out-ofpocket prescription drug costs. Our data further
show that the majority of vulnerable beneficiaries
identified in this study is still struggling with
medication costs and can maximize their benefit
with targeted outreach by trained experts.
Implications for Policy, Delivery or Practice:
Vulnerable populations of Medicare beneficiaries
are ill-equipped to navigate the complex Part D
program alone. That so many are unable to realize
the intended benefits of Part D suggests a gap in
policy objectives and implementation. Our research
suggests that policymakers can use targeted
outreach programs to ensure that vulnerable
populations have access to benefits offered and are
not adversely affected by policy changes.
Funding Source(s): The Amgen Foundation
Poster Number: 183
Use of Outpatient Care in VA and Medicare among
Disability-Eligible and Age-Eligible Veterans
Presented by: Chuan-Fen Liu, Ph.D., M.P.H.,
Investigator, Health Services Research &
Development, VA Puget Sound Health Care
System, 1100 Olive Way, Suite 1400, Seattle, WA
98101; Phone: (206) 764-2587; Email: chuanfen.liu@va.gov
Co-authors: Christopher Bryson, M.D., M.S.; Mark
Perkins, Pharm.D.; Nancy Sharp, Ph.D.; Matthew
Maciejewski, Ph.D.
Research Objective: Prior research on Medicareeligible veterans has focused almost exclusively on
age-eligible veterans, but veterans who are
disability-eligible for Medicare represent a large and
growing proportion of VA users. This study
compares use of VA and Medicare outpatient
services by veterans who are disability-eligible or
age-eligible for Medicare.
Study Design: This is a retrospective cohort study.
We obtained information on primary care, specialty
care, and mental health visits from 2001-2004 VA
administrative datasets and Medicare claims. We
examined differences between disability- and ageeligible veterans in primary care, specialty care, and
mental health outpatient visits using generalized
estimating equations.
Population Studied: The study includes a cohort
of 4,704 disability-eligible veterans and 10,816 ageeligible veterans who were VA primary care users
and were also eligible for Medicare in FY2000.
Principal Findings: A significant proportion of
Medicare eligible veterans used Medicare for
primary care and specialty care, but not mental
health care. Disability-eligible veterans were less
likely than age-eligible veterans to use Medicare
services for primary care (30% vs. 40% in 2001,
p<0.001) and specialty care (62% vs.73% in 2001,
p<0.001), but not mental health care (6% vs. 2% in
2001, p<0.001). Similar patterns were observed in
each year of the study period. Multivariate analysis
results show that disability-eligible veterans had
more VA visits for primary care (adjusted incidence
rate ratio [aIRR]=1.11, p<0.001) and specialty care
(aIRR=1.19, p<0.001), and more total
(VA+Medicare) visits for primary care (aIRR=1.07,
p<0.01) and specialty care (aIRR=1.07, p<0.01)
than age-eligible veterans. Disability-eligible
veterans were more likely than age-eligible
veterans to have any VA mental health visit (odds
ratio (OR)=1.71, p<0.001), any Medicare mental
health visit (OR=1.46, p<0.01), and any mental
health visit in either system (OR=1.73, p<0.001).
There were no significant differences in numbers of
VA mental health visits, Medicare mental health
visits or total mental health visits among mental
health care users.
Conclusions: This study shows that both disabilityand age-eligible VA primary care patients use
significant primary care and specialty care services
outside of VA. Further, this study documents
greater outpatient care needs for primary care,
specialty care, and mental health care among
disability-eligible veterans than age-eligible
veterans, especially for VA care.
Implications for Policy, Delivery or Practice:
Current disability-eligible veterans have greater
health care needs than age-eligible veterans. VA
and Medicare will need to be prepared for the future
needs of disability-eligible veterans, particularly
younger veterans returning from Iraq and
Afghanistan and Iraq wars. Establishing care
coordination protocols between VA and Medicare
systems will ensure efficient use of taxpayer
resources and smoother transitions to VA care for
disabled veterans.
Funding Source(s): VA
Poster Number: 184
Successful Recruitment of Black and Latino Women
in a Behavioral Educational Intervention to Reduce
Postpartum Depression
Presented by: Anika Martin, Assistant Professor,
Health Evidence & Policy, Mount Sinai School of
Medicine, One Gustave L. Levy Place, Box 1077,
New York, NY 10029; Phone: (212) 659-9567;
Email: elizabeth.howell@msnyuhealth.org
Co-authors: Amy Balbierz, M.P.H.; Edna Bisson;
Howard Leventhal, Ph.D.; Elizabeth Howell, M.D.,
M.P.P.
Research Objective: To determine reasons for
refusals among self-identified Black and Latino
postpartum mothers approached to participate in a
postpartum depression intervention trial; and to
revise recruitment strategies based on patient
feedback.
Study Design: This is an ongoing randomized
controlled trial of 530 Black and Latino postpartum
mothers testing whether a behavioral educational
intervention reduces postpartum depression.
Participants randomized to the intervention arm
receive a 2-step intervention that prepares and
educates mothers about modifiable factors
associated with postpartum depression (e.g.
distress from physical symptoms, deficits in social
support, low self-efficacy, and distress from infant
colic), bolsters social support, enhances
management skills, and increases participants’
access to resources. Participants are surveyed at 3weeks, 3-months, and 6-months to assess
depressive symptoms and other factors. All women
who refuse to participate in the trial are asked openended questions about their reasons for refusal.
Responses were transcribed, reviewed for similar
themes, arranged into distinct categories, and
organized as a taxonomy of reasons for refusal.
Based on the taxonomy, we made targeted
revisions to our recruitment methods.
Population Studied: Eligible participants are selfidentified Black and Latino postpartum women who
had uncomplicated deliveries at an urban hospital.
Eligible subjects are women >18 years of age,
English or Spanish Speaking, had working
telephones, and had infants whose birthweights
were > 2500 grams and 5-minute Apgar scores >6.
The study has enrolled 460 self-identified Black and
Latino mothers and this study presents results for
the114 Black and Latino mothers who have refused
to participate in this randomized controlled trial.
Principal Findings: Of the 574 Black and Latino
postpartum mothers approached during their
hospital stay, 114 women refused to participate
(Refusal rate of 19.9%). Seventy-two percent
(n=82) of the patients were Latino and 38% (n=32)
were Black. The taxonomy of refusals derived from
patient responses has six main categories related
to: 1) lack of interest, 2) having already been
through childbirth before, 3) time commitment, 4)
feeling overwhelmed, 5) patients moving from the
area, and 6) confidentiality. Based on patient
responses we revised our recruitment strategy to
alleviate mothers’ concerns. We eM.P.H.asized the
following elements: 1) a focus on women’s health,
2) postpartum mother’s health is understudied and
participants could add to the knowledge base, 3)
flexible schedule for follow-up, and 4) explained the
process of confidentiality. The refusal rate has
declined over 25 weeks of enrollment from 40% to
19.9%.
Conclusions: Tracking of patient refusals and
modifying recruitment methods based on patient
feedback improved recruitment of Black and Latino
mothers for a postpartum depression intervention
trial.
Implications for Policy, Delivery or Practice:
Tracking patient refusals and tailoring recruitment
methods based on patient responses may improve
recruitment of Black and Latino participants in
randomized controlled trials.
Funding Source(s): National Center for Minority
Health and Health Disparities (NIH)
Poster Number: 185
Health Care Screening and Preventive Practices
among Hispanic and Non-Hispanic Adults in the
United States
Presented by: Margaret McDonald, Ph.D.,
Director, Health Economics and Outcomes
Research, Global Market Access, Pfizer Inc, 235
East 42nd Street, New York, NY 10017; Phone:
(212) 733-3093; Email:
margaret.m.mcdonald@pfizer.com
Co-authors: Michael B. Lustik, M.S.
Research Objective: Ethnic disparities in health
care quality have been documented previously, but
current knowledge of screening and preventive
measures among Hispanic adults remains
incomplete. This study contributes to our knowledge
of preventative care by examining current national
estimates of flu shots, sigmoidoscopy or
colonoscopy, prostate-specific antigen tests,
mammograms and Pap tests among Hispanic and
non-Hispanic white adults aged 20 years and older.
Study Design: Cross-sectional observational study
design.
Population Studied: Analysis of nationally
representative data collected from Hispanic and
non-Hispanic white adults aged 20 years and older
participating in the Behavioral Risk Factor
Surveillance System (BRFSS) 2008 (N= 23,325
Hispanic; 320,417 non-Hispanic whites).
Principal Findings: Hispanic adults aged 50 years
and older have significantly lower rates of having
received a flu shot in the past year compared with
same aged non-Hispanic white adults (45% vs.
54% for men, respectively; 50% vs. 59% for
women, respectively; age-adjusted). Significant
differences in annual flu shot rates persist among
Hispanic and white adults 65 years and older (64%
and 72%, respectively). Hispanic men aged 50
years and older are significantly less likely than
same-aged white men to have had a prostatespecific antigen test within the past 2 years (58%
vs. 71%, respectively) or of ever having had a
colonoscopy/sigmoidoscopy (47% vs. 66%).
Hispanic women 50 years and older have
significantly lower screening rates for colon cancer
than white women (52% vs. 65%, respectively; ageadjusted). Mammogram rates for Hispanic and
white women aged 40 to 64 years are 73% and
76% (p = 0.02); mammogram rates are 80% and
79%, respectively, among women aged 65 years
and older. Rates for having a Pap test within the
past 3 years are similar for Hispanic and white
women: 88% and 89% (age 20-39 years); 86% and
87% (age 40-64 years); and 71% and 67% (age
65+ years).
Conclusions: Hispanic adults are less likely than
non-Hispanic white adults to have received flu shots
and sigmoidoscopies or colonoscopies. Hispanic
men have lower rates of having had a prostate
specific antigen test.
Implications for Policy, Delivery or Practice:
Effective approaches, including physician education
and community outreach are needed to reduce
these preventative medicine disparities.
Poster Number: 186
The Social Construction of Racial/Ethnic Disparities
in Diabetes – A Case of Misplaced Concreteness?
Presented by: John McKinlay, Ph.D., Senior Vice
President and Chief Scientist, Health Services and
Disparities Research, New England Research
Institutes, Inc, 9 Galen Street, Watertown, MA
02472; Phone: (617) 972-3012; Email:
jmckinlay@neriscience.com
Co-authors: Carol Link, Ph.D.; Rebecca
Shackelton, Sc.M.; James Meigs, M.D.; Lisa
Marceau, M.P.H.
Research Objective: Racial/ethnic disparities in
diabetes are widely considered an incontrovertible
epidemiologic fact. This paper questions whether
the reported racial/ethnic variation in diagnosed
diabetes accurately reflects the actual distribution of
(undiagnosed) diabetes in the population.
Study Design: We employ data from two different
but complementary studies—a community-based
epidemiologic random sample survey and a
balanced factorial experiment concerning primary
care physician decision making for diabetes.
Population Studied: Study 1 – The Epidemiologic
Survey: 5503 adults (1767 black, 1877 Hispanic,
1859 white; 2301 men and 3202 women) from
throughout the greater Boston, MA area. Study 2 –
The Factorial Experiment: 192 licensed internists,
family physicians, or general practitioners practicing
in New Jersey, New York, or Pennsylvania.
Principal Findings: No significant race/ethnicity
differences were found in the prevalence of
diabetes symptoms within SES level (?2 p>.35).
Significant SES differences are found (?2 p<.0001),
with higher prevalence among lower SES. Our
experimental data show physicians pattern their
diagnosis of diabetes by race rather than by SES
(blacks (73.4%), Hispanics (60.9%), whites (48.4%)
(p=.009) despite the fact that all patients presented
with exactly the same signs and symptoms of
diabetes. Diagnosed diabetes in the epidemiologic
survey and NHANES data followed the same
pattern.
Conclusions: Our results suggest a racial/ethnic
patterning of clinical decisions by primary care
providers. While the actual prevalence of
undiagnosed diabetes symptoms is patterned far
more strongly by SES than race/ethnicity, when
they are presented for diagnosis they are patterned
more by race/ethnicity (than by SES). Physicians
attending to (more visible) race/ethnicity, ignoring
(less obvious) SES, may produce prevalence rates
similar to the SES distribution of diabetes.
Misplaced concreteness results in searching for
genetic and bio-physiologic explanations which
distract from the more important SES differences
which shape the actual prevalence of diabetes.
Implications for Policy, Delivery or Practice:
Searching for causes and eventually interventions
associated with race/ethnicity (like genetics, family
history and bio-physiology) may distract from the
search for factors associated with the real culprit –
the significant effect of SES on the prevalence of
diabetes in society.
Funding Source(s): NIDDK and NCMHD
Poster Number: 187
State Policies and Language Access in California's
Commercial HMOs
Presented by: Christopher Moreland, M.D.,
M.P.H., Fellow, Internal Medicine, University of
California, Davis, 4150 V Street, Suite 2400,
Sacramento, CA 95834; Phone: (916) 734-7005;
Email: cmoreland@ucdavis.edu
Co-authors: Dominique Ritley, M.P.H.; Julie
Rainwater, Ph.D.; Patrick Romano, M.D., M.P.H.
Research Objective: Report the status of language
access services for California’s limited English
proficient (LEP) health maintenance organization
(HMO) members within the context of public
reporting (on a state-sponsored website) and
impending state regulation.
Study Design: We conducted a retrospective,
descriptive analysis of HMO responses to the
California Office of the Patient Advocate’s annual
survey, administered from 2003 through 2008.
Responses were stratified by product line and/or
language.
Population Studied: California’s nine largest fullservice HMOs and other HMOs which serve the
state’s Medicaid (Medi-Cal) population. Combined,
these groups represent over 90% of Californian
HMO members.
Principal Findings: From 2003 to 2008, during
California’s public reporting of HMO language
services and before implementation of state
language mandatory reporting requirements in
2009, the percentage of HMOs providing in-person
interpretation at medical points of contact free of
charge to consumers increased to over 80% across
all product lines. From 2006 to 2008, the
percentage of commercial product lines providing
in-person medical interpretation in threshold
languages (for which such services became
mandatory in 2009) increased from 64% to 94%.
From 2004 to 2008, the percentage of commercial
HMO product lines matching new LEP members to
providers by language increased from 50% to 71%.
In 2005, 64% of commercial product lines and 56%
of Medicare product lines provided telephone
interpretation services in all seven surveyed
languages, compared with 79% and 83%,
respectively, in 2008. Throughout the reporting
period, Medi-Cal and Healthy Families product lines
provided nearly all language services due to state
contract requirements.
Conclusions: HMOs increasingly provided
language access services in varying forms for LEP
Californians from 2003 to 2008 during a period of
public reporting but before implementation of state
regulations. Such changes within commercial
product lines appear more pronounced than those
within state-contracted product lines.
Implications for Policy, Delivery or Practice:
Public reporting of HMO language access services
may have been effective in encouraging such
services, particularly within commercial product
lines. It is too early to assess the additional effects
of the regulatory mandate implemented in 2009.
Funding Source(s): Health Resources and
Services Administration
Poster Number: 188
The Impact of Residential Segregation on Racial
Disparities in Infant Health
Presented by: Kwame Nyarko, B.A., Graduate
Research Assistant, Health Management and
Policy, University of Iowa, 5206 Westlawn, Iowa
City, IA 52242; Phone: (319) 335-7180; Email:
kwame-nyarko@uiowa.edu
Co-authors: George Wehby, Ph.D.
Research Objective: African-American infants
have significantly poorer health outcomes
compared to White infants. We examine the effects
of racial residential segregation on the birth
outcomes of African-American infants and the
extent to which population and area-level
characteristics explain these effects. Furthermore,
we evaluate the role of unobserved heterogeneity in
the relationship between segregation and infant
health and assess the heterogeneity in the effects
of residential segregation by unobserved area-level
endowments that impact infant health.
Study Design: The study uses county-level data
from the US census and Natality surveys for 1990
and 2000 and from the area resource files. We
measure residential segregation by the dissimilarity
and the isolation indices. The study outcomes
include low birth weight (LBW) and preterm birth
rates of African-American infants. We estimate the
effects of segregation on the means of the infant
health outcomes using random and fixed effect
models. We also apply quantile regression to
estimate the effects of segregation on the quantiles
of the infant health outcomes. Quantile regression
evaluates the heterogeneity in effects of
segregation by unobserved factors that determine
the county’s location on the distribution of these
outcomes.
Population Studied: The study sample includes
235 counties of 100,000 or more residents and at
least 50 annual African-American live births. Health,
behavioral, demographic and socioeconomic
county-level factors are measured for the AfricanAmerican population.
Principal Findings: Ignoring unobserved factors,
both measures of residential segregation have
significant negative effects on infant health that is
explained to a large extent by the observed
population and area-level characteristics. The
remaining effects of residential segregation are
entirely explained in most segregation and infant
health specifications by unobserved fixed effects
that may result in overestimation of the effects of
residential segregation if they are not causally
related to segregation. The study also finds that the
negative impacts of residential segregation are
primarily at the low quantiles of adverse birth
outcomes and not at the higher quantiles. In other
words, segregation impacts infant health in areas
with low rates of adverse birth outcomes and high
area-level endowments that improve infant health,
but segregation has insignificant effects in areas
with high rates of adverse birth outcomes and poor
endowments for infant health.
Conclusions: The study results suggest that
residential policies alone may not have large effects
on reducing racial disparities in infant health,
especially in areas with very high rates of adverse
birth outcomes among African-American infants,
and that they should be combined with policies that
improve area-level endowments for infant health.
Implications for Policy, Delivery or Practice:
Policies that enhance area-level endowments that
are relevant for infant health are needed to increase
the returns of residential policies in reducing racial
disparities in infant health.
Funding Source(s): CDC, NIH
Poster Number: 189
Racial Differences in Patient Satisfaction with a
Blood and Marrow Transplant Case Management
Program
Presented by: Nancy Omondi, M.B.A., M.S.,
Manager, Health Services Research, Health
Services Research, The National Marrow Donor
Program, 30001 Broadway St. N.E Suite 100 OPA,
Minneapolis, MN 55413; Phone: (612) 884-8518;
Email: nomondi@nmdp.org
Co-authors: Ellen Denzen, MS; Debbie Jacobson;
Elizabeth Murphy, Ed.D, R.N.
Research Objective: Blood and marrow
transplantation (BMT) is a complex and resourceintense treatment for patients with life-threatening
hematologic disorders. The National Marrow Donor
Program operates the national registry of volunteer
donors and facilitates BMT. Its Office of Patient
Advocacy (OPA) includes a case management
team that provides one-to-one telephone
counseling to patients and their caregivers to
address transplant-related questions. Previous
research has addressed the impact of race on
patient satisfaction; it has been demonstrated that
satisfaction increases access to health services.
We surveyed recipients of OPA case management
services to evaluate patient satisfaction by race.
Secondarily, survey responses were used to
improve OPA’s case management services.
Study Design: This cross-sectional study utilized
an eight item questionnaire administered via mail
without follow-up. Key evaluation areas were:
helpfulness of the information and services
provided; overall satisfaction with case
management; and demographic information.
Subjects were asked to describe follow-up actions
taken as a result of the information/services
provided and to recommend ways to improve these
services. Helpfulness of the information and
services provided was assessed using a three-point
Likert scale. Satisfaction was assessed through the
response to: “Would you recommend the OPA to
someone else in your situation?” (yes, maybe, no,
don’t know), which was used to create a Patient
Satisfaction Index (PSI). Race was categorized as
White and Other because of limited respondents in
race categories other than White.
Recommendations for improvement and follow-up
actions were analyzed qualitatively.
Population Studied: Our study population included
all BMT patients and their caregivers living in the
U.S. who directly contacted OPA case managers
between October 1, 2007 and September 30, 2008.
Of the 989 self-administered surveys that were
mailed, 241 (24%) evaluations were returned.
Principal Findings: Whites comprised 86% of the
respondents. More than 65% of respondents found
medical information provided by the Case
Managers helpful. Overall, 54% of respondents took
follow up action as a result of information or
services provided by case management. There
were no significant differences between Whites and
Other Races recommending OPA to others or
taking a follow up action. The overall PSI was 89%.
Respondents who reported satisfaction with OPA
case management services were more likely to take
follow-up action compared to those not satisfied
(p<.05).
Conclusions: OPA case managers successfully
addressed the transplant-related needs of patients
from all racial groups. Most patients were satisfied
with transplant case management and patient
satisfaction did not differ by race. Transplant case
management impacted informed decision making
as shown by follow-up actions taken by
respondents. In addition, respondents provided
feedback regarding other transplant
information/services that would be helpful to them.
Implications for Policy, Delivery or Practice:
One-to-one case management and provision of
transplant information and services have a major
impact on patient transplant decision-making,
ultimately improving access to health services.
Addressing satisfaction disparities could improve
access to health services.
Funding Source(s): The National Marrow Donor
Program
Poster Number: 190
Race Disparities in Hospital Discharge Disposition:
The Role of Model Misspecification
Presented by: Eberechukwu Onukwugha, Ph.D.,
Assistant Professor, Pharmaceutical Health
Services Research, University of Maryland School
of Pharmacy, 220 Arch Street, Baltimore, MD
21201; Phone: (410) 706-8981; Email:
eonukwug@rx.umaryland.edu
Lead Author: Jason Bergtold, Ph.D.
Co-authors: Eberechukwu Onukwugha, Ph.D.
Research Objective: Evidence for race disparities
in hospital discharge disposition have been studied
previously and results may depend on the statistical
model that is used to calculate the risk ratios. In
particular, a misspecified model can provide
misleading evidence regarding the direction and
magnitude of effect. The objective of this study is to
compare the evidence for race disparities in poststroke hospital discharge disposition under
alternative specifications of the multinomial logistic
regression model.
Study Design: The Probabilistic Reduction (PR)
approach provides an alternative approach for
specifying discrete choice models using
observational data. Within the PR approach, the
functional form of the choice probability for an
individual includes an index function that depends
explicitly on the inverse conditional distribution. An
empirical application used cross-sectional data from
a state-wide confidential inpatient hospital
discharge data including admissions occurring
between January 2000 and December 2005.
Population Studied: The sample was limited to
live discharges among adult (i.e. aged greater than
or equal to 18 years) patients admitted for a stroke
in Maryland. A traditional multinomial logistic
regression model specification with limited
interactions and a model specified following the PR
approach were estimated and compared. Marginal
effects for both the individual covariate of interest
(‘Non-Caucasian race’) and the interactions of
interest: (i) ‘patient age’ x ‘race’; (ii) ‘insured status’
x ‘race’; and (iii) ‘hemorrhagic stroke’ x ‘race’, were
estimated. Individual marginal effects were
estimated as the mean marginal effect across
respondents. Interaction effects were calculated as
the mean across respondents. Standard errors for
all marginal effects and interaction effects were
estimated using a delete-d jackknife estimator.
Statistical significance of marginal effects was set at
the 10% level.
Principal Findings: Application of
inclusion/exclusion criteria resulted in 69,921 live
discharges following a stroke-related admission.
The distribution of discharge disposition outcomes
was as follows: home (N = 7,730), home health
care (N = 525), rehabilitation (N = 9,997), nursing
home (including intermediate care) (N =7,323),
discharges against medical advice (N = 4,755), and
all other (N = 39,591). Misspecification tests
indicate that the PR approach provides a better fit
and that the traditional specification of the model is
misspecified. The marginal effect (standard error)
results indicate differences between the traditional
and PR specifications as concerns the likelihood
that non-Caucasians will be discharged home
(32.7% [0.005] less likely vs. 7.6% [0.001] less
likely) or will require additional nursing or
intermediate care (22.5% [0.009] more likely vs.
3.2% [0.001] more likely). Interaction effects of
interest, compared across model specifications,
were either both statistically significant and opposite
in sign (‘age’ x ‘race’) or both statistically significant,
of the same sign, but of different magnitudes
(‘hemorrhagic stroke’ x ‘race’; ‘insured status’ x
‘race’).
Conclusions: We found that the evidence for the
existence and extent of race disparities in poststroke hospital discharge outcomes varied across
statistical model specification approaches.
Implications for Policy, Delivery or Practice:
Efficient public investment in programs that address
race disparities depends on the availability of
reliable information regarding the existence and
extent of disparities. We show that the statistical
model specification approach is a key consideration
when identifying potential sources of bias in
disparities research.
Funding Source(s): NIH Roadmap K12 program
Poster Number: 191
Health Literacy and Communication Quality in
Organizations
Presented by: Chandra Osborn, Ph.D., Assistant
Professor of Medecine, General Internal Medicine
and Public Health, Vanderbilt University Medical
Center, 1215 Twenty-First Ave South, Suite 6000,
MCE North Tower, Nashville, TN 37232; Phone:
(615) 936-8468; Email:
chandra.osborn@Vanderbilt.Edu
Co-authors: Matthew K. Wynia, MD, M.P.H.
Research Objective: Effective health
communication is fundamental to quality care for all
patients. Communication challenges are also
powerful contributors to disparities since patients
with limited health literacy are more often minorities.
Inadequate health literacy leads to
misunderstanding health information, difficulty
following medical instructions and worse health
outcomes. Employing effective communication
practices may be one of the most important
interventions to address the widespread burden of
limited health literacy. We used patient surveys to
explore the relationship between health literacy
status and receiving patient-centered
communication in thirteen clinics and hospitals
serving communication-vulnerable patient
populations.
Study Design: An assessment of organizational
patient-centeredness was developed through a
multistakeholder consensus process. The
assessment was validated in a field test with
thirteen widely-varying health care organizations,
including six hospitals and seven clinics nationwide.
During field testing, patient surveys were distributed
that included seven items to assess perceptions of
patient-centered communication. One third of
patients surveyed also completed items measuring
health literacy.
Population Studied: Surveys were collected from
5,929 patients, of whom 2,116 also completed
items about health literacy.
Principal Findings: Most patients were between
age 25-44, female and a member of a racial/ethnicminority group. One third had limited English
proficiency, and most were receiving care from a
federally qualified health center or large/urban
medical center. Nearly half reported visiting this
facility two-five times in the last six months. The
majority of patients had health literacy challenges,
reporting problems learning about their medical
condition because of difficulty understanding written
information (53%), a lack of confidence in
completing medical forms by themselves (61%) and
needing someone to help them read hospital/clinic
materials (57%). Logistic regression models
showed that, after adjustment for patient
demographic characteristics and organization type,
patients with limited health literacy were 21-78%
less likely than those with adequate health literacy
to report that their organization “always” provides
patient-centered communication across the seven
communication items (20 of 21 p-values <0.05). In
additional multivariable models using a scaled
composite of the seven items, limited health literacy
remained associated with lower communication
quality.
Conclusions: Many of the health consequences
associated with limited health literacy may be
caused or exacerbated by ineffective, non-patientcentered communication. This study demonstrates
that patients with inadequate literacy are
considerably less likely to report receiving patientcentered communication from the organizations
where they seek care, across an array of
communication quality indicators.
Implications for Policy, Delivery or Practice:
Organizational change to alleviate disparities by
addressing health literacy requires leadershipdriven modifications of internal operations,
increasing staff awareness and competency,
improving provider access to training on effective
health communication, and continued monitoring of
organizational performance to identify changes in
quality improvement opportunities.
Funding Source(s): The California Endowment
Poster Number: 192
Stigma among Colorectal Cancer Patients
Presented by: Sean Phelan, M.P.H., Research
Fellow, Family Medicine and Community Health,
University of Minnesota, 925 Delaware Street SE,
Minneapolis, MN 55414; Phone: (612) 625-9106;
Email: phel0085@umn.edu
Co-authors: Michelle van Ryn, Ph.D.; Joan Griffin,
Ph.D.; Diana Burgess, Ph.D.
Research Objective: People with stigmatized
medical conditions experience adverse
psychological and physiological health outcomes
and may delay medical care and if they perceive
that they are devalued by their health care
providers or other people. Furthermore, self-blame
for conditions that are perceived to have behavioral
causes may have negative effects as well. The goal
of the project was to measure experiences of
stigma, self-blame, and discrimination among
patients with colon or rectal cancers, and assess
whether these experiences are related to stressrelated mental health outcomes.
Study Design: A patient questionnaire assessing a
variety of domains of patient characteristics and
experiences was developed by a team of NCI, VA,
DOD and University of Minnesota scientists. A
single item measure of perceived discrimination in
attaining health care was used, as well as items
measuring perceived stigma associated with
colorectal cancer, and self-blame. Outcome
measures consisted of validated measures of
depression, anxiety, overall health, and pain, as
well as measures of patient-centered quality of
care. The questionnaire was pilot-tested in the fall
of 2009 among all VA patients who received a
diagnosis of colorectal cancer in any VA facility in
2008 and were alive at the time of the survey.
Patients were ascertained from the VA Cancer
Registry and vital status confirmed through
computerized medical records. Associations
between stigma related experiences and patient
health outcomes and reported quality of care will be
assessed using multivariate regression techniques.
Population Studied: There were 2018 living
patients with colorectal cancers diagnosed at Stage
1 or higher during 2008 as ascertained from the
registry. Of the 2018 patients in the sample, 67%
(1421) returned completed questionnaires. Most
(94%) were male, reflecting the VA population. The
median age was 67 with a range of 26 to 94.
Principal Findings: There was considerable
variation in patient reports of stigma, discrimination,
and perceived blame. For example, 12% of patient
reported experiencing discrimination by health care
providers, 26% felt that they were at least partially
to blame for their illness, and 25% felt that there
was a stigma that goes with having their condition.
Conclusions: Some patients with colon or rectal
cancers experience stigma and discrimination
related to their condition, and may blame
themselves for acquiring the disease. These
occurrences may have implications for patient
mental health and other outcomes.
Implications for Policy, Delivery or Practice:
Stigma may an important stressor for patients with
colon or rectal cancers that may contribute to their
health outcomes. Physical and mental health care
providers should consider this when working with
this patient population.
Funding Source(s): NCI
Poster Number: 193
What Characterizes Counties Successful in Breast
Cancer Screening?
Presented by: Maria Pisu, Ph.D., Assistant
Professor, Medicine/Preventive Medicine,
University of Alabama at Birmingham, 1530 3rd
Avenue S., MT 628, Birmingham, AL 35298-4410;
Phone: (205) 975-7366; Email: mpisu@uab.edu
Co-authors: Robert S. Levine, M.D.; Robert Oster,
Ph.D.; Alana Cantor, Ph.D.; Michelle Martin, Ph.D.
Research Objective: In the US, there is a wide
geographic and racial variation in breast cancer
screening. The objective of this study is to better
understand how communities successful at
promoting breast cancer screening in black and
white women compare with less successful ones in
socioeconomic status and medical resources
availability and type.
Study Design: We obtained county-level breast
cancer screening rates (CO-BCSR) for white and
black women from Medicare administrative claims
for a sample of 1,000,000 women older than 64
continuously enrolled in Medicare fee-for-service
plans. Women resided in 172 counties with the
highest numbers of breast cancer deaths in black
women. Counties were classified in 4 types: Failing,
i.e., black and white CO-BCSR lower than the
respective race average CO-BCSR; With Disparity,
i.e., black CO-BCSR lower and white CO-BCSR
higher than the respective race average CO-BCSR;
With Reverse Disparity, i.e., black CO-BCSR higher
and white CO-BCSR lower than the respective race
average CO-BCSR; and Successful, i.e., with black
and white CO-BCSR higher than the respective
race average CO-BCSR. Using data from the Area
Resource File, we identified economic factors and
medical resource availability and type that may be
associated with county type. We used discriminant
analysis to identify variables that would characterize
each county type, and multinomial logistic
regression to compare county types.
Population Studied: US Counties and women
enrolled in Medicare
Principal Findings: The average CO-BCSR was
42.3 per 100,000 (range: 20.8-58.2) for black, and
48.7 (range: 29.7-65.2) for white women. Counties
were: 56 Failing, 23 With Disparity, 26 With
Reverse Disparities, and 67 Successful. The
socioeconomic variables that best predicted the
county classification were: percentage of county
population living below the federal poverty level,
county average household income. The medical
resources variables that best predicted county type
were: proportion of hospitals with associated
medical schools and HMO penetration. Other
variables included were: median age of women in
the county, percent of black females, number of
MDs, primary care physicians, and female MDs per
100,000 females, number of hospitals per 100,000
females, percentages of hospitals with
mammography centers. Results of the multinomial
logistic regression showed that, compared to Failing
counties, counties were more likely to be
Successful if they had smaller proportions of foreign
born and black female populations, lower incomes,
fewer hospitals and lower HMO penetration.
Compared with counties with Reverse Disparities,
counties were more likely to be Successful if they
had fewer hospitals, lower proportion of hospitals
associated with medical schools, and lower HMO
penetration. Successful counties did not differ from
counties With Disparities in all variables examined.
Conclusions: Successful counties are not
necessarily richer than failing counties or counties
with disparities, and do not differ significantly in the
availability of most medical resources. Existing type
of care (e.g., medical schools or HMOs) may
differentiate county types.
Implications for Policy, Delivery or Practice:
More research is needed to understand what
makes counties successful and how to transfer
successful models to failing counties or counties
with pronounced disparities.
Funding Source(s): NCI
Poster Number: 194
The Role of Dentists in Alleviating Racial/Ethnic
Disparities in Dental Care: Does Availability of
Discounted Care Increase Affordability?
Presented by: Nadereh Pourat, Ph.D., Associate
Professor, Univeristy of California, Los Angeles
Center for Health Policy Research, 10960 Wilshire
Boulevard, Suite 1550, Los Angeles, CA 90024;
Phone: (310) 794-2201; Email: pourat@ucla.edu
Co-authors: Ron Andersen, Ph.D.; Marvin Marcus,
D.D.S., M.P.H.; Dylan Roby, Ph.D.; Gina Nicholson,
M.P.H.
Research Objective: Research indicates that
racial/ethnic disparities in oral health care may be
attributable to access. Lack of insurance is linked
with fewer dental visits. Also, dental insurance
policies with higher out-of-pocket costs which may
lead to foregoing dental care among insured who
can not afford these costs, including many low
income and non-white groups. The uninsured
seeking lower cost services can obtain care in
community clinics with dental care capacity or
identify private dentists who offer sliding scale fees.
Since the majority (96%) of the dentists are in
private practice, availability of dentists offering
discounts may be crucial in insuring affordability
and subsequent access. This study examines if
higher rates of private dentists who offer discounted
care in an individual’s Medical Service Study Area
(MSSA) has an impact on foregoing needed dental
care due to costs. It is hypothesized that a higher
percentage of dentists in an MSSA who offer
discounted dental care reduces the likelihood of
racial/ethnic groups to forego dental care due to
costs.
Study Design: Data from the 2003 California
Health Interview Survey (CHIS) and a 2003
representative survey of California dentists were
merged at the MSSA level. The percentage of
dentists in an area that reported offering discounted
dental care in their primary practice per MSSA was
merged with CHIS data. SAS v. 9.3 GLIMMIX
models were used and included individual level
demographic and socioeconomic data as well as
MSSA level data. All analyses were adjusted for
CHIS and dentist survey design.
Population Studied: The sample included 40,000
adults, with 25,400 whites, 2,500 African
Americans, 6,700 Latinos, and 3,800 Asian
Americans.
Principal Findings: Latinos (31%) and African
Americans (22%) were more likely to forego needed
dental care due to costs that Asian Americans
(16%) and whites (14%). Preliminary findings
indicate that the higher percentage of dentists who
offer discounted care is associated with the lower
likelihood (OR=0.68) of foregoing dental care due to
costs in an MSSA, in an unadjusted model.
However, the models controlling for individual level
variables indicate that that this relationship is no
longer significant. Rather, lack of dental insurance,
low income, age, and gender are among individual
variables that determine the likelihood of foregoing
dental care due to costs.
Conclusions: Our findings indicate that availability
of discounted dental care may be negatively
associated with foregoing care, however, factors
such as dental insurance are more important in
insuring affordability of dental care. Data is lacking
on the extent of the discount provided by dentists.
Insubstantial discounts are unlikely to change
affordability. If most dentists do not provide
sufficient discounts no impact on affordability is
observed.
Implications for Policy, Delivery or Practice:
Recessions and budget cuts compound the
financial difficulties of populations with the lowest
socioeconomic status, many of whom are members
of non-white racial/ethnic groups and reduce their
ability to obtain dental care. The absence of dental
insurance coverage can not be remedied by
provision of discounted dental care and health care
policy should focus on improving dental coverage
and comprehensiveness of such coverage.
Funding Source(s): NIDCR
Poster Number: 195
Black-White Disparities in Mortality among Middle
Aged Adults: Can Health Insurance Be the
Solution?
Presented by: Janice Probst, Ph.D., Associate
Professor, Health Services Policy and
Management, University of South Carolina, 800
Sumter Street, Columbia, SC 29208; Phone: (803)
777 7426; Email: jprobst@sc.edu
Co-authors: Katrina M. Walsemann, Ph.D.; Jessica
D. Bellinger, Ph.D.; James Hardin, Ph.D.
Research Objective: Disparities in life expectancy
between white and non-white populations in the US
are marked. While infant mortality contributes to this
difference, the majority of excess deaths occur in
middle age. Our research seeks to ascertain
whether health insurance helps to ameliorate racial
disparities in mortality among adults aged 45 - 64.
Study Design: Data were drawn from the 2004
National Health Interview Survey (NHIS) Linked
Mortality File. This file links respondent information
from the nationally representative NHIS to death
records through 2002.
Population Studied: Our analysis examines
respondents between the ages of 45 and 64 during
interview years 1986, 1989-2000. We excluded
1987-1988 because no insurance information was
collected. Records with missing data on any
variable were also excluded, for a sample size of
217,748. We used Cox proportional hazard models
to calculate mortality hazard ratios (HR) by
race/ethnicity and insurance status at time of
interview (privately insured, publicly insured,
uninsured). Respondent demographic, resource
and health status characteristics were held constant
in multivariable analysis. All analyses were stratified
by sex and weighted to account for the complex
sample design.
Principal Findings: Overall, 77.1% of respondents
reported private insurance, 11.1% public insurance,
and 11.8%, uninsured. Minorities were more likely
to be uninsured (18.2% black, 27.2% Hispanic, and
17.5% other adults, versus 9.5% of white
respondents). In analysis holding age and interview
year constant, black adults had an elevated risk of
death (Female: HR 1.62, CI 1.52-1.72. Male: HR
1.57, CI 1.48-1.66) compared to white adults.
Hispanics, regardless of sex, did not differ in risk
from whites, whereas men from other racial origins
were at lower risk of death than whites (HR: 0.83,
CI 0.73-0.95). Adjustment for insurance alone
reduced the hazard ratio for blacks, regardless of
gender, but did not fully eliminate disparities
(Female: HR 1.31, CI 1.23- 1.40. Male: HR 1.32, CI
1.25-1.41), which remained after further adjustment
for socio-economic status, region, rurality, and
health status (Female: HR 1.11, CI 1.04-1.19; Male
HR 1.08 CI 1.02-1.16). Adjustment for insurance
status resulted in significantly lower risk of death for
Hispanic women compared to white women.
Further adjustment for socio-economic status,
region, rurality, and health status resulted in
significantly lower risk of death for Hispanics of
either gender compared to white adults. Of note,
uninsured individuals were at higher risk of death in
all analyses.
Conclusions: Health insurance appears to
ameliorate some of the black-white disparity in
mortality, but does not fully eliminate it.
Implications for Policy, Delivery or Practice:
Expansion of insurance coverage across the US
population through health care reform may not be
sufficient to eliminate race-based disparities in risk
of death. Additional research is needed into
contextual factors, such as neighborhood income or
distance to providers, which may influence health
and survival. Research into the lowered risk of
death experienced by middle-aged Hispanic adults
may provide insight for other minority populations,
as well.
Funding Source(s): HRSA
Poster Number: 196
The Impact of Subjective and Objective
Socioeconomic Status on Health: Do the Results
Differ Between Subjective and Objective Health
Measures?
Presented by: Christy Pu, Ph.D., Assistant
Professor, Department of Public Health, National
Yang-Ming University, Li-Nong Street, Sec 2,
Peitou, Taipei, Taiwan 155; Phone:
+886285906914; Email: christypu@hotmail.com
Co-authors: Nicole Huang, Ph.D.; Yiing-Jenq
Chou, Ph.D.
Research Objective: This research aims to
determine whether the relationship between
subjective socioeconomic status (SES) and health
depends on types of health measures.
Study Design: This research used the data from
the Survey of Health and Living Status of the Middle
Age and Elderly in Taiwan (SHLS) conducted by
the Bureau of Health Promotion, Department of
Health in Taiwan. The survey consists of 5 waves
(1989, 1993, 1996, 1999 and 2003). Educational
attainment and the respondent’s major occupation
were used as measures for objective SES, and selfrated satisfaction on financial position was used to
measure subjective SES. Self-rated health (SRH)
was used to measure subjective health status. SRH
is dichotomized into those who reported reduced
health (SRH not equal to “Excellent” or “Very good”)
and those who did not report reduced health, at the
survey wave before death. Mortality was used to
measure objective health outcome. For both
objective and subjective measures, Cox
proportional hazard models were estimated to allow
for comparison.
Population Studied: 4,049 nationally
representative elders (65 years and above) in
Taiwan
Principal Findings: Subjective SES was positively
and significantly associated with subjective health
measure (SRH). Compared with those who
reported “satisfied” for financial satisfaction, those
who reported “average” (HR= 1.20, p< 0.01) and
“unsatisfied” (HR=1.29, <0.01) were more likely to
report reduced health. However, subjective SES
was not significantly associated with objective
health (mortality). On the other hand, objective SES
(education) was more significantly associated with
mortality than with SRH, with those who had higher
educational attainment had lower HR for mortality.
Conclusions: The correlation found between
subjective SES and subjective health measures
may be simply caused by personal traits, such as,
whether the person is pessimistic. A pessimistic
person may perceive himself to be in poor financial
condition and the same pessimism also lead him to
perceive himself to be in poor health. Thus the
association found between subjective SES and selfreported health may simply reflect such individual
characteristics.
Implications for Policy, Delivery or Practice:
Research that investigates the relationship between
subjective SES and health do not compare results
when subjective and objective health measures are
used. We argue that the association between
subjective SES and health status depends on
whether the health measure is also subjective, and
this should be taken into consideration in related
research.
Poster Number: 197
Availability of Translation Services and Translated
Documents in High Performing Hospitals
Presented by: Marsha Regenstein, Ph.D., M.C.P.,
Research Professor, Dept of Health Policy, Center
for Health Care Quality, George Washington
University, 2121 K Street, NW, Suite 200,
Washington, DC 20037; Phone: (202) 994-8662;
Email: marsha.regenstein@gwumc.edu
Co-authors: Lea Nolan, M.A.
Research Objective: There is little guidance on the
“how” to deliver translation, despite recent activities
by three major accrediting bodies, the National
Committee on Quality Assurance, the Joint
Commission and the National Quality Forum, to
include requirements related to translating vital
documents into languages spoken by patients in a
community. Research has been conducted on the
best way to translate documents, such as the idea
of “back-translating” to ensure quality and leaving a
trail of documentation about how the material was
translated. This study assesses the current
translation practices of hospitals that are
considered high performing in language services.
Study Design: We developed a survey for hospital
translation services directors. The datasource was
completed surveys from telephone interviews,
fielded November 2008 through January 2009. We
generated descriptive statistics on hospital
background, who provides translation services,
policies and procedures regarding translation
services, quality control for translations, storage
and accessibility of translated documents, and
costs of translation services. We also collected
hospitals’ written policies pertaining to translation
services.
Population Studied: The population in this study
are 35 high-performing hospitals, defined as
hospitals who have extensive experience with or
have been innovative in their approach to providing
languages services. We identified 53 potential
subjects, using prior research projects, hospital
association membership listings, and data on
distribution of ethnic diversity and immigrant
population.
Principal Findings: A majority of respondents use
contract translators (29) but almost as many use
staff interpreters to translate (27). Few hospitals
have staff dedicated to translation. The majority of
hospitals (24) translate any document that is
requested, although some make a distinction and
translate any vital document. Hospitals without
standardized policy leave it to the individual
departments to determine what should be
translated. The majority of respondents (27) have
an in-house quality control process in place, but the
remaining 8 do not. Hospitals report a variety of
storage options, including an intranet for all staff,
storage drive in the language services office, and
central repository for all translated documents.
Nearly half of hospitals have centralized costs and
do not charge requesting departments for
translations, four provide free translation services to
hospital staff, and others charge depending on the
type of translated document. Although less than half
could estimate the costs of providing translations,
our data suggests that the annual cost of in-house
translations ranges from $8,000-$900,000. External
costs range from $1,000-$117,000 per year.
Conclusions: Translation services are provided
through a variety of methods, depending on
patients’ specific language needs and availability of
qualified translators. Even with varying resources
available for translating and storing documents, all
hospitals should implement quality control
procedures to ensure the quality of their translated
documents.
Implications for Policy, Delivery or Practice:
This study uses guidance from the field to formulate
trends and determine best practices for the delivery
of translation services in health care settings. These
results also serve as a baseline for accrediting
bodies and policymakers to develop systematic
requirements and processes for institutions on how
to develop effective policies and practices related to
the delivery of translation services and translation of
health care materials.
Funding Source(s): RWJF
Poster Number: 198
Disparities in Meeting the 12 LyM.P.H. Node
Minimum Standard Examination after Resection for
Colon Cancer
Presented by: Kim Rhoads, M.D., M.S., M.P.H.,
Assistant Professor, Surgery, Stanford University,
300 Pasteur Drive H3680F, Stanford, CA 94305;
Phone: (650) 736-8406; Email:
kim.rhoads@stanford.edu
Co-authors: Jennifer Cullen, Ph.D.,M.P.H.; R.
Adams Dudley, M.D., M.B.A.; Sherry Wren, M.D.
Research Objective: Disparities in colon cancer
outcomes are well documented. Improved 5-year
survival has been shown to correlate with having 12
lyM.P.H. nodes (LN) examined after colon cancer
resection, since ensuring that LNs are negative
determines stage and the need for chemotherapy.
The National Quality Forum has recently endorsed
this practice standard as a measure of the quality of
colon cancer surgery. However, there is significant
variation in compliance with this metric. The
purpose of the current study is to determine if there
are racial and ethnic disparities in meeting the12
lyM.P.H. node minimum standard examination and
secondarily to determine the effect of the standard
on disparities in survival after colon cancer
resection.
Study Design: The current study is a retrospective
analysis of a large administrative database from the
California Cancer Registry (CCR) and the California
Office of Statewide Health Planning and
Development (OSHPD). Chi-square analyses and
logistic regression models were used to assess for
differences in the percent of patients having 12 LN
examined and the percent being under staged
(stopping at <12 LN examined when none have
been positive). Models were adjusted for
demographics, tumor location, year of treatment
and hospitals’ annual volume, percent Medicaid,
and teaching status. Cox proportional hazard
models for mortality were adjusted for these factors
and urgency of surgical intervention.
Population Studied: All cases of colon cancer
contained in the CCR database during the years
1996-2006, inclusive, were linked to patient
discharge abstracts from OSHPD for the same
years. Patients with ICD9 codes for the primary
diagnosis of colon cancer, who had concomitant
codes for major colon resection, were included.
Patients with stage IV disease and those with
missing data on lyM.P.H. node counts were
excluded. The final cohort included 62,928 records.
Whites represented 73%; Blacks 7%; Hispanic
patients 11% and Asian/Pacific Islander 9%.
Principal Findings: Blacks, Hispanics and Asians
were 18%, 22% and 6% less likely to have 12 LN
examined. Hispanics had 11% higher odds of being
under staged. Use of high volume, teaching and
high compliance rate hospitals was associated with
a 20%, 21% and 3%, respectively, increased odds
of meeting the standard. Percent Medicaid patients
within a hospital was not significantly correlated
with LN exam variation. Mortality disparities
persisted despite adjustment for receipt of the
standard LN examination. Analysis of mortality
stratified by race showed a significant effect of
having 12 lyM.P.H. nodes examined on outcomes
for whites, but not for other races.
Conclusions: Variation in receipt of the 12 LN
minimum standard after colon cancer resection
does not explain disparities in survival, but is
associated with increased risk for under staging of
Hispanics. Our study suggests that there are other
factors contributing to the persistence of disparities
in colon cancer outcomes.
Implications for Policy, Delivery or Practice:
Increased efforts should be made to ensure that the
12 LN minimum standard is met in all patients to
avoid under staging and potentially, under
treatment of colon cancer.
Funding Source(s): RWJF
Poster Number: 199
Does Socioeconomic Status Moderate the
Association of Race with Morbidities During
Pregnancy Episodes?
Presented by: Douglas Roblin, Ph.D., Senior
Research Scientist, The Center for Health
Research / Southeast, Kaiser Permanente Georgia,
3495 Piedmont Road NE, Building 9, Atlanta, GA
30305; Phone: (404) 364-4805; Email:
Douglas.Roblin@KP.Org
Co-authors: Peter Joski, M.S.P.H.; Donald
Bachman, M.S.; Mark Hornbrook, Ph.D.
Research Objective: African American adults
generally have higher prevalence of morbidities
than whites in the United States. Living in lower
socioeconomic status (SES) neighborhoods
increases the likelihood of poor health. The extent
to which lower SES contributes to a higher rate of
pregnancy complications for African Americans
requires investigation. We explored racial
differences in likelihood of maternal morbidities and
whether these differences were moderated by
neighborhood SES.
Study Design: Pregnancies were defined from
linked birth certificates and hospital records.
Morbidities during pregnancy were ascertained from
all outpatient and inpatient diagnoses and
procedures classified by a previously validated
maternal morbidity grouper. SES was measured
from a factor analysis of 7 US Census measures
associated with tract or block group of residence.
Factor scores were divided into quartiles. Likelihood
of selected morbidities was estimated using logistic
regression adjusted for race, SES quartile, and
mother’s age. Parameter estimates were adjusted
for repeated observations per woman during the
study period.
Population Studied: 9,726 pregnancies among
8,517 non-Hispanic African American women and
10,272 pregnancies among 8,652 non-Hispanic
white women enrolled with Kaiser Permanente
Georgia (KPGA) from 2000-2006.
Principal Findings: 14% of pregnancies were
characterized by non-hereditary anemia, 13% by
hypertension, and 7% by mental disorders. African
American women were more likely than white
women to live in lowest SES quartile neighborhoods
(32% vs. 16%; p<0.05). Adjusted for age and SES,
pregnancies characterized by non-hereditary
anemia and hypertension were significantly
(p<0.05) more likely among African Americans than
whites (OR = 2.24 [95% CI: 2.45, 2.04] and 1.30
[95%: 1.42, 1.18], respectively). Mental disorders
were significantly less likely among African
Americans (OR = 0.50 [95% CI: 0.57, 0.44]).
Stratified by race and adjusted for age, low SES
neighborhoods had a significantly greater
moderating effect among whites than among
African Americans. Among whites, the adjusted
ORs for the lowest SES quartile (vs. highest) were
1.21 (95% CI: 1.40, 1.01) for hypertension, 1.97
(95% CI: 2.40, 1.61) for non-hereditary anemia, and
1.27 (95% CI: 1.57, 1.03) for mental disorders.
Conclusions: Neighborhood SES had a significant
moderating effect on selected morbidities during
live birth pregnancy episodes in this HMO.
Compared to African American women, living in low
SES neighborhoods significantly increased the risk
of major morbidities during pregnancy episodes
among white women.
Implications for Policy, Delivery or Practice: In
an insured and generally middle class population,
both race and SES were associated with likelihood
of hypertension, non-hereditary anemia, and mental
disorders among women during pregnancy. While
coverage, access and affordability have been
suggested as reasons for differences in health risks
by race and SES, our study’s findings suggest other
factors – such as biologic, behavioral, or social
differences – must be considered in the debate
about racial and SES disparities.
Funding Source(s): CDC
Poster Number: 200
Emergency Room Use and Delays in Prescription
Drugs are Higher among African-Americans When
Compared to Other Commercial HMO Enrollees
Presented by: Dylan Roby, Ph.D., Assistant
Professor, Health Services / Center for Health
Policy Research, University of California, Los
Angeles School of Public Health, 10960 Wilshire
Boulevard, Suite 1550, Los Angeles, CA 90024;
Phone: (310) 794-3953; Email: droby@ucla.edu
Co-authors: Gina Nicholson, M.P.H.; Gerald
Kominski, Ph.D.
Research Objective: This study examines delays
in fulfilling prescribed medications, delays in
obtaining needed medical care, visits to emergency
rooms, and the presence of a usual source of care
among insured African Americans in public and
commercial HMOs.
Study Design: Using 2007 California Health
Interview Survey (CHIS) data, we conducted
descriptive analyses of HMO enrollment, payer
source, and race/ethnicity. To determine if these
differences persist after adjusting for
sociodemographic and health differences between
ethnic groups, we conducted separate logistic
regression analyses for each insurance type to
control for selection bias.
Population Studied: These analyses were
conducted for the adult, non-elderly population (1864) enrolled in HMOs based on 2007 CHIS
respondents.
Principal Findings: African-American HMO
enrollees were more likely to use the ER than other
ethnic groups. Although ER use is similarly high for
both whites and African Americans enrolled in
public HMO plans (predominantly Medi-Cal), there
is a distinct difference between African Americans
and whites who are enrolled in commercial HMO
plans. African Americans enrolled in commercial
HMOs were more likely to use the ER (25.4% in
Kaiser and 24.9% in other commercial plans)
compared to other ethnic groups in commercial
HMOs. Rates of prescription drug delays are about
5% higher for commercially insured African
Americans in Kaiser compared to whites. However,
a much larger disparity exists for African Americans
enrolled in other commercial HMO plans compared
to whites. African Americans have an almost 10%
higher rate of delaying needed prescription
medications than whites. Separate regression
models for each insurance type were used to
isolate the potential effects of demographics, health
status and race/ethnicity among the separate
insurance status groups. African Americans were
more likely to use the ER (OR = 1.51, p<0.05) when
enrolled in commercial Kaiser and were more likely
to delay prescriptions (OR = 1.66, p<0.05) when
enrolled in other commercial HMO plans. Predictors
of ER use in the commercial HMO model were
fair/poor health status and chronic illness. In the
commercial Kaiser HMO model, ER use was
predicted by African-American race, fair/poor health
status and chronic illness. Predictors of prescription
delays in commercial HMO enrollees were AfricanAmerican race, female gender, fair/poor health
status and chronic illness.
Conclusions: This disparity suggests other barriers
still exist for HMO-enrolled African Americans in
California in terms of avoiding ER visits and
delaying prescriptions.
Implications for Policy, Delivery or Practice: In
California, the largest proportion of African
Americans in HMO plans are enrolled in Kaiser
Permanente, which operates on an integrated staff
model that could include ER, urgent care, on-site
pharmacy and various other services on one
hospital campus. This type of convenience and
increased access to ER services could potentially
impact patterns of use among Kaiser enrollees.
Greater delays in obtaining prescription drugs for
African-American commercial HMO enrollees could
be due to high copayments, limited pharmacy
benefits, lack of access to in-network pharmacy
services covered by the HMO, or perception of
need by the patient. We do not see similar
prescription delays for African-American
commercial Kaiser HMO enrollees, who may have
better access to on-site pharmacy services
compared to other commercial HMO enrollees.
Funding Source(s): California Office of the Patient
Advocate
Poster Number: 201
Access to Health Care for Latino Mixed-Status
Families
Presented by: Kara Ryan, M.P.P., Health Policy
Research Analyst, Office of Research, Advocacy,
and Legislation, National Council of La Raza, 1126
16th Street, NW, Washington, DC 20036; Phone:
(202) 776-1703; Email: kryan@nclr.org
Co-authors: Patricia Foxen, Ph.D.
Research Objective: Research has shown that
children in mixed-status families are much more
likely to be uninsured than children born to citizen
parents, largely due to barriers to employer-based
and public coverage programs for which they are
eligible. However, there has been limited research
on the experiences of mixed-status families within
the health care system. To assess the potential
implications of the structure of the health care
system for Latino mixed-status families, NCLR
conducted a series of focus groups in October
2009. This research focused on the influence of
parent uninsurance on insured children’s health
care access and quality, with eM.P.H.asis on the
experience of mixed-status families.
Study Design: The seven focus groups took place
in four sites—Charlotte, North Carolina; Chicago,
Illinois; Portland, Oregon; and Fairfax, Virginia. The
sites were chosen based on several factors,
including characteristics of the Latino population
(e.g., size, nascent or established) and the policies
of selected states’ public coverage programs. Five
panels included Latino parents who were uninsured
but whose children had health insurance, with one
panel made up of parents in fully-insured families,
and one panel composed of parents whose family
members were all uninsured. Four of the seven
panels were composed entirely of immigrant
parents with citizen children.
Population Studied: The participants were Latino
parents of U.S. citizen children under age 18. More
than half of the participants were noncitizens,
including both legal and undocumented immigrants,
and the remainder were citizens.
Principal Findings: The following five themes
emerged from this research: 1) Overall, state public
coverage programs are critical to health care
access for citizen children in mixed-status families.
2) However, parents in these families have
dramatically lower access to health coverage and
care and report poorer health. 3) Families’
experiences navigating the public coverage system
were inconsistent from state to state. 4) Affordability
is a significant problem, even when children are
insured. 5) Parents recognize the comprehensive
nature of their children’s health insurance, but
perceived some differential treatment based on
ethnic identity as well as their status as public
coverage recipients.
Conclusions: Access to health coverage for
children in mixed-status families is critical to their
health care access, but varies by community. Even
when children are insured, parents have difficulty
paying for medical services, and they often have no
access to coverage or care themselves. More
research is necessary to understand how Latino
parents’ poor access to health care impacts the
well-being of the family unit.
Implications for Policy, Delivery or Practice: At
this point, final health care reform legislation may
improve access and quality for Latinos in mixedstatus families, or it may actually impose new
barriers, therefore widening health care disparities.
In either case, these findings provide insight into the
issues that must be addressed in order to close
gaps in Latinos’ health care access and quality.
Funding Source(s): Atlantic Philanthropies
Poster Number: 202
Does Patient-Navigator Racial/Ethnic Concordance
Improve Patient Barrier Identification?
Presented by: M. Christina Santana, M.P.H.,
Researcher, Section of Internal and General
Medicine, Women's Health Unit, 801
Massachusettes Avenue, Crosstown Building, Suite
470, Boston, MA 02118; Phone: (617) 638-8036;
Email: christina.santana@bmc.org
Co-authors: Sharon Bak, M.P.H.; Kristine Beaver,
M.P.H.; Stephen Evans; Karen Freund, M.D.,
M.P.H.; Tracy Battaglia, M.D., M.P.H.
Research Objective: Preliminary data suggests
that Patient Navigation, which targets individual
barriers to care, is an effective strategy to reduce
cancer health disparities. While most navigation
programs are culturally tailored, there are no
studies examining the role of navigator
race/ethnicity on outcomes. Boston is one of nine
sites participating in the multi-site, NCI Patient
Navigation Research Program (PNRP) which aims
to reduce delays in cancer care among
underserved with cancer screening abnormalities.
Our objective is to examine the role of racial/ethnic
concordance of patient and navigator on the
identification of barriers to care, as an intermediate
outcome.
Study Design: Three of the six Boston PNRP
Community Health Center sites (CHC’s), with a
breast navigator were included. Racial/Ethnic
concordance was determined by first categorizing
patient and navigators into White or Non-White
groups, and then they were matched for
concordance (yes/no) based on those categories.
The outcome of interest was the presence or
absence of barriers to care, as documented by the
Navigator. Due to racial differences in the identified
barriers outcome analyses, we stratified by patient
racial/ethnic groups of White and Non-White using
chi-squares.
Population Studied: Eligible women were 18 years
or older, and had an abnormal breast cancer
screening test at one of the 3 breast navigated CHC
sites. Women were navigated until a diagnostic
resolution was reached during January 2007-June
2008 study recruitment period.
Principal Findings: Preliminary findings for 503
women showed a mean age of 49 ±12; 58% are
non-White, 63% speak English as their primary
language; 78% had an abnormal imaging test and
35% had private health insurance. Of the 10 breast
Navigators, 70% are White and 20 % speak a
language other than English. Overall 40% of the
patient-navigator pairs were concordant by
race/ethnicity. Overall, Non-White women were
more likely to have barriers present compared with
White women (79% vs. 50%, p <0.0001). NonWhite concordant pairs were as likely as Non-White
discordant pairs to have barriers present (81% vs.
78%, p =0.838), however, White concordant pairs
where less likely to have barriers present compared
to White discordant pairs (41% vs. 72%, p
<0.0001).
Conclusions: Non-White women have more
barriers documented by a Navigator compared to
White women. The lack of association between
racial/ethnic concordance and identification of
barriers may reflect differences in burden of barriers
found in racial groups rather than racial/ethnic
concordance. Our future analyses will include a
larger sample size and thus address the limitation
of our conclusion for racial/ethnic concordance
pairs.
Implications for Policy, Delivery or Practice: Our
findings suggest that Non-White women may have
more barriers to care compared to White women,
supporting cancer care intervention efforts for
racially/ethnic groups. Future studies, with a larger
sample size, can help inform how racially/ethnic
concordant patient and navigator pairs are
associated to barriers to care.
Funding Source(s): NCI
Poster Number: 203
National Scan of Efforts to Reduce Health Care
Disparities
Presented by: Sarah Scholle, M.P.H., Dr.P.H.,
Assistant Vice President, Research, National
Center for Quality Assurance, 1100 13th St. NW,
Suite 1000, Washington, DC 20005; Phone: (202)
955-1726; Email: scholle@ncqa.org
Lead Author: Jessica French
Co-authors: Esther Han, M.P.H.; Ignatius Bau,
J.D.; Sarah Hudson Scholle, M.P.H., Dr.PH.
Research Objective: Despite continuing evidence
that health care disparities persist, little progress
has been made to close the gaps in the quality of
health care delivered to racial, ethnic and linguistic
minorities. Our objective is to catalogue current
national-level efforts to reduce health care
disparities, and to identify gaps and opportunities to
accelerate progress in closing the quality chasm.
Study Design: We conducted an internet search of
disparities reduction efforts and classified each
effort we found by the nature of the project, target
audience(s), and topic(s) addressed. Iterative
reviews by experts in disparities identified missing
efforts and helped classify them. We tabulated the
efforts in each category and conducted cross-tabs
to understand the full range of activities. Finally, we
conducted interviews with experts on their
interpretation of our results and obtained their
recommendations for actions to accelerate
progress.
Population Studied: We sought national-level
efforts using the following search terms: cultural
competency; language access; health care
disparities; health care equity; culturally and
linguistically appropriate services (CLAS);
multicultural health care; minority health; health
care for underserved; health care for vulnerable
populations. We excluded initiatives with diseasespecific, population-specific or organization-specific
foci and efforts related to the social determinants of
health disparities and efforts that were solely
focused on research.
Principal Findings: We found activity in all
categories, with more activity focused on planning,
evaluation, training or education, and fewer efforts
focused on implementing practice changes. More
efforts targeted health care professionals and
policy-makers, and few targeted consumers. We
identified some efforts to develop standards of
performance. Some of these findings reflect
limitations of our methods. Our selection of national
efforts excluded local activity that targeted
consumers or focused on implementing change at a
local practice level. Despite the many efforts
underway, their impact on disparities has been
negligible.
Conclusions: Our experts identified severa l
opportunities to accelerate improvement: The
Department of Health and Human Services Office
of Civil Rights could more proactively enforce Title
VI anti-discrimination provisions. Health care
executives need evidence of a return on the
financial investment required to act. Employers, the
primary purchasers of health care, need evidence
that health care disparities affect their workplaces,
and they need to know of potential interventions.
Medicare and Medicaid have made small, periodic
investments in disparities, but a continued priority
focus is needed to sustain performance
improvements. Consumers need to know that
disparities exist and what they can do about them.
The HITECH Act and a recent Institute of Medicine
report recommending standardized collection and
reporting of race, ethnicity and language data by
health care organizations are potential catalysts for
change through broader implementation of health
information technology. Moreover, both the House
and Senate versions of national health reform
legislation include multiple provisions directed at
reducing health care disparities.
Implications for Policy, Delivery or Practice:
There are many potential sources of leverage, but
ultimately , sustained investment in quality
improvement and practice transformation, together
with increased engagement of purchasers and
consumers, will be necessary to effect significant
progress in reducing health care disparities. Both
investment and engagement depend on
demonstrating the link between disparities and each
stakeholder’s priorities.
Funding Source(s): The California Endowment
Poster Number: 204
Using Search Engine Query Data to Track
Pharmaceutical Utilization
Presented by: Nathaniel Schuster, B.S., Medical
Student, University of Michigan Medical School,
507 Glen Court, Ann Arbor, MI 48104; Phone:
(952) 237-8648; Email: nschu@med.umich.edu
Co-authors: Mary Rogers, Ph.D., M.S.; Laurence
McMahon, Jr, M.D., M.P.H.
Research Objective: Since patients consult the
internet for health information, Google search
behavior might correlate with healthcare utilization.
As a test case, search engine query data for search
terms “Lipitor” and “simvastatin” were evaluated for
temporal and geographic correlation with
benchmarks of resource utilization.
Study Design: Using 2004-2008 Google Trends
and Google Insightsfor Search data, search terms
"Lipitor" and "simvastatin" were evaluated for
change over time. Then, query data was tested
forcorrelation with Lipitor revenues. Finally, query
data of 26 US metropolitan areas was tested for
correlation with service use per Medicare
beneficiary as percent of the national average, a
benchmark of geographic variation.
Population Studied: Individuals using Google
search engine 2004-2008.
Principal Findings: Regression of the slopes of
search engine query datafor search terms “Lipitor”
and “simvastatin” were different pre-June 23,
2006(the day that simvastatin came off-patent)
versus post-June 23, 2006(p<0.001). Lipitor's
annual percent change in revenues for 2004-2008
correlated positively with Lipitor's average query
index (p<0.01) and negatively with simvastatin’s
average query index (p<0.05). Both geographic
variation in the “Lipitor” query index and the “Lipitor”
to “simvastatin” query index ratio for 26
metropolitan areas correlated with service use per
Medicare beneficiary (both p<0.05).
Conclusions: Query data shows search behavior
correlates withpharmaceutical revenues and
metropolitan area healthcare utilization.
Implications for Policy, Delivery or Practice:
This study suggests that search engine query data
has the potential to provide researchers and
policymakers with a novel data source for studying
geographic variations in health care utilization
which can highlight important characteristics of a
community's culture in the use of healthcare.
Search engine query data has great potential for
research since it is available for free, in real-time, at
researchers' fingertips.
Poster Number: 205
The Intended and Unintended Consequences of
Clinical Guidelines in Reducing Health Care
Disparities
Presented by: Rebecca Shackelton, Sc.M.,
Associate Research Scientist, Health Services and
Disparities Research, New England Research
Institutes, Inc, 9 Galen Street, Watertown, MA
02472; Phone: (617) 972-3119; Email:
rshackelton@neriscience.com
Co-authors: John McKinlay, Ph.D.; Lisa Marceau,
M.P.H.; Carol Link, Ph.D.
Research Objective: People of lower
socioeconomic status (SES) carry a greater burden
of disease in the United States than people of
higher SES. Lower SES is associated with
increased coronary heart disease, diabetes and
overall mortality. Physician decision-making and
physicians’ perceptions of their patients are affected
by a patient's SES. Consequently, lower SES
patients can experience inequality at two levels.
Having thoroughly documented variations in clinical
practice by SES, quality assurance experts and
professional organizations developed lists of items
that a physician ought to take when presented with
a likely case of some disease. It was presumed that
close adherence to these lists (guidelines) by
individual practitioners encountering individual
patients would eventually produce desirable
standardization of medical care and eventually a
reduction in the health care variations. The intended
consequence of guidelines, indeed the raison d’être
presented to gain provider acceptance, was the
improvement in the quality of care and ultimately
the reduction or elimination of healthcare variations.
In this study we first examine whether clinical
guidelines are achieving their intended
consequences (reduction of health disparities).
Second, we contemplate potential unintended
consequences of clinical guidelines highlighting an
important distinction between the stated purpose of
some social action and their generally unrecognized
but objective functional consequences.
Study Design: As part of a factorial experiment we
presented primary care doctors (n=192) with
clinically authentic vignettes of a “patient” with
already diagnosed diabetes with an emerging foot
neuropathy systematically varied by patient SES.
Their proposed clinical actions were compared with
established practice guidelines for this clinical
situation.
Population Studied: We recruited a total of 192
primary care physicians, practicing with New York,
New Jersey, and Pennsylvania, stratified by gender
and level of clinical experience.
Principal Findings: Physicians participating in our
experiment were less likely to ask questions, give
advice, prescribe medications, conduct
examinations and testing, and refer the lower
socioeconomic patients. After establishing the
existence of these consistent socioeconomic
disparities in the proposed management of the
diabetes scenario, we found that reported use of
practice guidelines had no measurable effect
towards their reduction (one intended
consequence). However, the reported use of
practice guidelines appeared to precipitate more
clinical actions without eliminating documented
disparities.
Conclusions: Principal reasons for the introduction
of clinical practice guidelines were the desire to
improve the quality of medical care and the
reduction of well-documented healthcare
disparities. From our experimental data and with
respect to the management of diabetes, guidelines
do not appear to be accomplishing their primary
stated purpose. However, they do appear to ensure
that more of everything is done (without reducing or
eliminating healthcare disparities), and potentially
contribute to the ever increasing costs of medical
care.
Implications for Policy, Delivery or Practice:
Consistent with other research we find clinical
practice guidelines are not producing a principal
intended result, and may even produce unintended
consequences.
Funding Source(s): NIDDK
Poster Number: 206
URBAN40: A Quasi-Experimental Study on the
Population Health Impact of a Neighbourhood
Programme in the Netherlands
Presented by: Karien Stronks, Ph.D., Prof.,
Department of Public Health, Academic Medical
Center, Meibergdreef 15, Amsterdam, Netherlands;
Phone: +31205664892; Email:
k.stronks@amc.uva.nl
Co-authors: Anton Kunst, Ph.D.; Mariel Droomers,
Ph.D.; Birthe Jongeneel, M.Sc.; Cees Hoefnagels,
Ph.D.
Research Objective: There is a widespread
agreement that a policy aimed at tackling health
disparities should focus on the social determinants
that are considered the ‘root causes’ of health
disparities, including education, income and the
neighbourhood environment. Little is known about
the extent to which policies aimed to change these
determinants, so-called upstream policies, are
effective in preventing health problems. In view of
this, a window of opportunity was opened by the
decision taken by the Dutch government in 2007, to
spend several billion Euros to increase the
liveability of 40 deprived districts, focusing on
problems of housing, employment, education,
integration and safety. From a public health
perspective, this programme provides a ‘natural
experiment’ to study the health effects of changing
physical and social conditions at the neighbourhood
level.
Study Design: URBAN40 (acronym for Upstream
policy Research on health and Behaviour Across
disadvantaged Neighbourhoods (40)) will study the
impact of the ‘40 districts plan’ in terms of
prevention of health problems among the
inhabitants. It uses a quasi-experimental, repeated
cross-sectional design, based on routinely collected
data. All 40 districts will be compared with 40
control districts, with regard to changes in living
environment (neighbourhood conditions,
unemployment rates, individual socio-economic
position etc.), more specific health determinants
(smoking, housing conditions, physical activity etc.)
and the occurrence of health problems (quality of
life, chronic diseases etc.).
Population Studied: Inhabitants of 40 ‘intervention’
and 40 control neighbourhoods, aged 18 years and
older.
Principal Findings: A large number of health
relevant measures within the 40 districts have been
identified, including creation of green belts, sports
facilities and employment opportunities, and
housing improvements. In addition, the
mechanisms through which the measures might
have an impact on (determinants of) health have
been specified, based on current knowledge on the
link between social determinants and more specific
health determinants on the one hand, and health
problems on the other. These analyses show that
the measures taken within the programme might
affect the health status of the inhabitants through a
material, social and psychosocial mechanism. The
logic models specifying these mechanisms will be
used to derive specific hypotheses as to the effect
of these measures, to be tested in the quasiexperimental design.
Conclusions: The measures taken within a policy
programme aimed at the improvement of the
liveability of deprived neighbourhoods in the
Netherlands, potentially have an impact on the
health status of the inhabitants in these
neighbourhoods, even though health issues were
not part of the policy programme. The actual health
impact will be assessed in future analyses.
Implications for Policy, Delivery or Practice: The
relevance of this study is rooted in the opportunity
to obtain more insight into the effectiveness of
upstream interventions to prevent health problems.
Funding Source(s): Dutch council for health
research and development
Poster Number: 207
Members’ Perceptions of Race and Ethnicity Data
Collection by Health Plans in Massachusetts
Presented by: Kalahn Taylor-Clark, M.P.H.,
Ph.D., Research Director, Engelberg Center for
Health Care Reform, HVHC Initiative, The
Brookings Institution, 1775 Massachusetts Avenue,
Suite 600, NW, Washington, DC, DC 20036-2103;
Phone: (202) 797-6072; Email:
ktclark@brookings.edu
Lead Author: Romana Hasnain-Wynia, Ph.D.
Co-authors: Kalahn Taylor-Clark, M.P.H., Ph.D.;
Ayodola Anise, M.H.S.
Research Objective: To examine perceptions and
concerns of health plan members about race and
ethnicity data collection by health plans and assess
members’ preferences for racial and ethnic
categories.
Study Design: Qualitative study using a moderator
guide with specific questions about attitudes and
preferences for race and ethnicity data collection.
Focus groups were stratified by race and ethnicity
and Spanish language. Themes were derived
based on the series of questions in the guide and
the participants’ discussions and comments.
Analysis was guided by a grounded theory
perspective.
Population Studied: Six focus groups that
included 54 total participants, all of whom belonged
to either a private or public health plan.
Backgrounds of individuals included the following
groups: Caucasian/White, African American/Black,
Chinese, Multiracial (Portuguese, Brazilian, Cape
Verdean), Hispanic/Latino Spanish-speaking, and
Hispanic/Latino English-speaking.
Principal Findings: Emergent themes included: 1)
Skepticism and concern about collecting data; 2)
Utility of collecting data; 3) Members’ need to
understand why health plans would collect these
data; 4) Importance that health plans provide a
clear explanation to members about data collection;
5) Concerns about privacy and confidentiality; and
6) Methods for how health plans should inform
members about the collection of race and ethnicity
data. Participant concerns about health plans
asking their race and ethnicity focused on lack of
transparency about how the information would and
would not be used. Participants expected their
health plans to explain, preferably in writing, why
they were asking for this information. Participants
reacted positively to health plans’ intent to collect
data to improve care. Preferences for racial/ethnic
categories (broad categories vs. more specific
granular ethnicities) differed among the participant
groups.
Conclusions: Members have concerns about
health plans asking for information on race and
ethnicity, but they also understand the utility of
using this information to improve care for diverse
populations. Health plans can improve the process
of and members’ confidence in collection of race
and ethnicity data by: 1) Providing a clear rationale
to members about why plans are collecting these
data; 2) Describing how it would and would not be
used; and 3) Adhering to a list of categories that
strikes a balance between simplicity and enough
granular detail to capture the demographic make-up
of the community. Additional qualitative quotes from
focus group participants provide context to the
emergent themes.
Implications for Policy, Delivery or Practice:
Health plans play a central role in strategies to
address racial and ethnic disparities in health care,
which remain a national concern. The lessons
learned in Massachusetts – where the
Massachusetts Health Care Quality and Cost
Council now requires all health plans to report race
and ethnicity data – may be applied in other states
and nationally as health plans design programs to
systematically collect such data through direct
reporting by their members.
Funding Source(s): RWJF
Poster Number: 208
A Qualitative Case Study of How Organizations Use
Data on Race and Ethnicity to Reduce Disparities
and Improve the Quality of Care
Presented by: Ruth Thorlby, M.Sc., Senior Fellow
in Health Policy, Policy, King's Fund, 11-13
Cavendish Square, London, England UK W1G
0AN; Phone: +442073072646; Email:
r.thorlby@kingsfund.org.uk
Co-authors: Selena Jorgensen, B.A.; Bruce Siegel,
M.D., M.P.H.; John Ayanian, M.D., M.P.P.
Research Objective: Healthcare organizations
have been urged to reduce racial and ethnic
disparities in the quality of care. A first step in this
process is the collection and analysis of healthcare
data by patients’ race and ethnicity. This study
explored the experience of healthcare organizations
that had collected, analyzed and used data to
understand the challenges and benefits that these
efforts foster.
Study Design: In depth interviews were conducted
with participants at various healthcare organizations
across the United States. The organizations were a
selected from a sample identified as being
advanced in the collection and use of race/ethnicity
data by a literature review and contact with expert
informants. Up to five interviews were conducted at
each organization and participants were asked
about the process of collecting race and ethnicity
data and how the data had been analyzed and
used. Interviews were recorded, transcribed and
analyzed for emerging themes using Nvivo 8
qualitative analysis software.
Population Studied: 39 staff members at 8
organizations: a multi-site hospital network, two
teaching hospitals, three health plans and two
community health centers
Principal Findings: Although the organizations
were at varying stages of development in their use
of race and ethnicity data, all reported that staff
were comfortable collecting these data after
appropriate training, and most patients were willing
to report their race or ethnicity. All three health
plans reported that direct data collection from all
their members was difficult because of sensitivities
on the part of some members about the potential
uses of data and had supplemented the patient
data with data based on proxy measures such as
geo-coding. All organizations had attempted to
analyze these data, usually by race stratification of
quality indicators already in use. Some
organizations had attempted to adjust for
confounding factors while others had not. Some
interpreted a disparity as any statistically significant
difference between groups while others had set a
threshold of a certain absolute percentage
difference between groups. Some used ‘white’
patients as the comparator; others interpreted a
disparity as any difference between the best
performing group and other groups. Where
disparities were revealed, data were often used to
design and assess improvement initiatives. Where
disparities were absent, some organizations had
begun to develop new, more sensitive indicators
based on the disparities literature. The community
health centers reported challenges in finding
resources for data analysis. All organizations
believed the data could provide a much more robust
underpinning for action on disparities.
Conclusions: These healthcare organizations had
all successfully collected data from patients but for
health plans the process was more challenging. All
organizations attempted to analyze their data but
with very different methods and understanding of
what constituted a disparity. Analyses of these data
are perceived to provide a robust basis for action to
reduce disparities within healthcare organizations.
Implications for Policy, Delivery or Practice:
Collection and analysis of patient-level data by race
and ethnicity is possible at the level of the
healthcare organization but community health
centers may need extra resources while health
plans may require new strategies, such as state
mandates, to encourage members to share their
data. The heterogeneity of approaches to
understanding and measuring disparities poses a
challenge and may require clearer guidance from
accrediting or governmental agencies if disparities
are to be addressed in a consistent and robust
manner in healthcare organizations across the
United States.
Funding Source(s): CWF
Poster Number: 209
Clinicians’ Views of an Intervention to Reduce
Racial Disparities in Diabetes Outcomes.
Presented by: Ruth Thorlby, M.Sc., Senior Fellow
in Health Policy, Policy, King's Fund, 11-13
Cavendish Square, l, United Kingdom W1G 0AN;
Phone: +442073072646; Email:
r.thorlby@kingsfund.org.uk
Co-authors: Selena Jorgensen, B.A.; John
Ayanian, M.D., M.P.P.; Thomas Sequist, M.D.,
M.P.H.
Research Objective: Interventions that improve
clinicians’ awareness of racial and ethnic disparities
and improve their communication skills are
considered promising strategies for reducing
disparities in health care. This qualitative study
assessed clinicians’ views of an intervention
involving cultural competency training and racestratified performance reports designed to reduce
racial disparities in diabetes outcomes, as well as
their perceptions of the causes and potential
solutions to these disparities.
Study Design: Interviews were conducted with
clinicians who had recently participated in a
randomized intervention to reduce racial disparities
in diabetes outcomes. The clinicians worked with
Harvard Vanguard Medical Associates (HVMA), an
integrated multispecialty group practice based in
eastern Massachusetts. Earlier studies with HVMA
had shown that disparities in intermediate outcomes
of diabetes (achieving adequate glycemic, blood
pressure, and cholesterol control) had persisted
despite the adoption of a quality improvement
program (electronic medical records and chronic
care model). To address these disparities, 62
clinicians were randomized to receive cultural
competency training and monthly race stratified
performance reports for one year. After the
randomized study was concluded, we conducted
semi-structured interviews with clinicians.
Interviews were transcribed and analyzed for
emerging themes using Nvivo 8 software.
Population Studied: 17 clinicians from the original
intervention group, of whom 12 were physicians, 4
were Nurse Practitioners, and 1 was a Physician
Assistant.
Principal Findings: Most clinicians acknowledged
the presence of racial disparities in diabetes control
among their patients, but a small number rejected
this premise or believed that race was not the root
cause of any perceived differences between
groups. Those who did accept that there were
disparities described a complex set of causes,
including socioeconomic factors such as low
incomes, living in areas with poor access to fresh
food and limited opportunities to exercise. Other
factors mentioned included low levels of trust in the
medical profession, poor eating habits,
communication barriers due to language and low
literacy. The clinicians perceived only some of
these causes to be within their power to change,
such as switching patients to less expensive
generic drugs. Clinicians were able to identify
additional services or programs that would help
them address disparities, for example culturally
tailored nutrition advice, outreach to patients within
their communities, and group visits to allow patients
to learn from each other. Some clinicians
challenged the premise of the intervention, focusing
on socioeconomic factors as the primary cause of
disparities, rather than patients’ race. The
performance reports and training were generally
well received but some clinicians did not feel
empowered to act on the information presented.
Conclusions: Cultural competency training and
performance reports may raise clinicians’
awareness of racial and ethnic disparities but may
not be sufficient to empower them to make a
measurable impact on reducing disparities in health
care outcomes.
Implications for Policy, Delivery or Practice:
Organizations committed to reducing disparities
may need to employ strategies that offer concrete
initiatives to improve patient access along with
clinician-focused interventions such as enhanced
information and training for their clinicians.
Clinicians should be seen as a source of ideas for
action based on their insights about why disparities
may be occurring among their own patients.
Interventions that are built on clinician action also
need to respond to those clinicians who have
doubts about the role of race as a driver of poor
outcomes or access.
Funding Source(s): CWF
Poster Number: 210
Disparities in Labor Analgesia Utilization
Presented by: Paloma Toledo, M.D., Postdoctoral Fellow, Institute for Healthcare Studies,
Northwestern University, 251 E. Huron Street F5704, Chicago, IL 60611; Phone: (312) 472-3585;
Email: p-toledo@md.northwestern.edu
Co-authors: Cynthia Wong, M.D.; William
Grobman, M.D., M.B.A.; Romana Hasnain-Wynia,
Ph.D.
Research Objective: There is evidence that there
are racial/ethnic differences in the use of epidural
analgesia for labor pain relief. However, the
reasons for this disparity have not been described.
The purpose of this study was to evaluate the
factors that influence decisions regarding
management of labor pain in patients of different
racial/ethnic backgrounds.
Study Design: A survey was developed based on
existing validated surveys that included domains for
socioeconomic status, assimilation, acculturation,
trust, and family cohesiveness. Questions regarding
knowledge of labor analgesia were developed. The
survey underwent expert panel review. Cognitive
interviews were conducted with members of the
survey population for content validity. The final
survey was administered via face-to-face interviews
with women admitted for labor and delivery who
had not yet requested labor pain relief. The method
of labor analgesia utilized was recorded after
delivery. Groups were compared using the KruskalWallis statistic. After univariable selection,
significant variables were entered into a
multivariable logistic regression model to determine
predictors of epidural analgesia utilization.
Population Studied: English-speaking laboring
women admitted to Northwestern Memorial Hospital
in Chicago: 224 Caucasians, 71 Hispanics, and 53
African-Americans were interviewed.
Principal Findings: Caucasian patients were more
likely to be married, have a higher level of
education, higher income, and have private health
insurance than the other groups (P<0.001). The
rate of epidural analgesia utilization was 92%, 73%,
and 77%, respectively, in the Caucasian, Hispanic
and African-American patients (P<0.01). Despite
expecting more pain during labor than Caucasian
women, Hispanic and African-American women
were more likely to report that they would not use
epidural analgesia (P<0.001). Patients reported
with equal frequency that providers discussed labor
analgesia, however Caucasian patients were more
likely to use of prenatal classes and books as
sources of information, and were more likely to
consult with their family members in their decision
making (medical and non-medical) (P<0.01).
African-American patients trusted their physicians
less than Hispanic and Caucasian patients
(P<0.01). In contrast, Caucasians trusted their
insurance plans less than Hispanic or AfricanAmerican patients (P<0.001).The multivariable
analysis indicated that the method of labor
analgesia the patient expected to use was a
significant predictor of epidural analgesia utilization
(OR 42.4, 95% CI 11.2, 159.7), as was consulting
with family when making decisions (OR 20.6, 95%
CI 1.42, 297.9), and higher income (OR 1.4, 95% CI
1.1, 1.8). Women with less trust in physicians were
less likely to receive epidural analgesia (OR 0.22,
95% CI 0.08, 0.61).
Conclusions: A racial disparity in labor epidural
utilization exists and appears related to pre-existing
expectations, sources of information, and physicianpatient relationships.
Implications for Policy, Delivery or Practice:
Further research should explore the promising
methods and means for educating women about
labor analgesic options in pregnancy.
Funding Source(s): Evergreen Grant
Poster Number: 211
Underutilization of Cancer Screening among
Individuals with a Family History of Breast and
Colorectal Cancer
Presented by: Jennifer Tsui, M.P.H., Doctoral
Student Researcher, School of Public Health,
Department of Health Services, University of
California, Los Angeles, 650 Charles Young Drive
South, Room A2-125 CHS Box 956900, Los
Angeles, CA 90095-6900; Phone: (310) 206-2165;
Email: jentsui@ucla.edu
Co-authors: Ninez Ponce, Ph.D., M.P.P.; Sara
Knight, Ph.D.; Aimee Afable-Munsuz, Ph.D.; Uri
Ladabaum, M.D., M.S., Robert Hiatt, M.D., Ph.D.;
Jennifer Haas, M.D. M.Sc.
Research Objective: Disparities in cancer
screening among average risk populations have
been well-established in the literature. Less is
known about whether and to what extent
sociodemographic and healthcare market factors,
such as having insurance or a recent doctor visit,
explain variations in screening among adults who
report a family history of cancer and may be at
higher risk. This study aims to identify segments of
this population who are vulnerable to breast and
colorectal cancer screening underuse, and who
could benefit from targeted cancer prevention
efforts.
Study Design: Retrospective, cross-sectional
analysis of self-reported data from the 2005
California Health Interview Survey (CHIS). Breast
(BC) and colorectal cancer (CRC) screening
underuse were evaluated separately using
weighted multivariate logistic regression models
accounting for individual factors among adults
reporting moderate or high risk family history. We
use the CDC Risk Stratification Tool to categorize
average versus moderate or high risk individuals.
Screening underuse was defined for BC as women
with a family history of breast and/or ovarian cancer
who did not receive a mammogram within the past
2 years; for CRC as adults with a family history of
colon cancer who did not receive a sigmoidoscopy
within the past 5 years or a colonoscopy within the
past 10 years.
Population Studied: Adult participants from the
CHIS 2005 cancer control module. BC screening
analysis included women, ages 40-64 years old,
with a family history of breast and/or ovarian cancer
and no personal history of breast cancer. CRC
screening analysis included adults, ages 50-64
years old, with a family history of colon cancer and
no personal history of colon cancer.
Principal Findings: Low income (0-99% of the
Federal Poverty Level (FPL)) women with a family
history of breast and/or ovarian cancer had higher
odds (OR 2.40; 95%CI:1.07-5.35) of mammography
underuse compared to women with higher income
(FPL >300%). Compared to women in urban areas,
women in rural areas had increased odds of
underuse (OR 1.68; 95%CI:1.03-2.75). No
significant differences were found in mammography
underuse by race/ethnicity. Among adults at-risk for
CRC, Latinos have increased odds of screening
underuse (OR 3.44; 95%CI:1.51-7.84) compared to
non-Latino whites. Similar to mammography
underuse, low income adults at-risk for colon
cancer had higher odds of underuse (OR 3.20;
95%CI:1.24-8.24) compared to those in the highest
income level. In both models, healthcare market
factors were associated with higher odds of
screening underuse.
Conclusions: We found both healthcare market
and sociodemographic disparities in screening
among individuals with a family history of breast
and colon cancer. Healthcare market factors have
been found in a national population-based study on
breast cancer, but the sociodemographic disparities
we found in the diverse California sample of higher
risk adults is a new finding.
Implications for Policy, Delivery or Practice:
With the advent of new technologies targeting
diagnostics and therapies for familial cancers,
disparities in uptake of these new technologies may
widen by sociodemographic factors. It is important
to continue to target low-income, underinsured and
Latino (for CRC) populations with higher risk of
cancer. Healthcare reform that incorporates
coverage gains would help to ensure cancer
screening disparities do not widen.
Funding Source(s): NCI
Poster Number: 212
and patients for their input, to lead to targeted
interventions.
Population Studied: The population included 1274
patients, 47% are White, and 42% are Black. All
patients recieve care at the UNC Family Medicine
Center.
Principal Findings: Black and White patients
differed (p<.001) in all measures of disease control:
proportions with HbA1c>9 (21.7% vs. 10.1%,
respectively), HbA1c<7 (47.3% vs. 60.2%), diastolic
BP>90 (21.5% vs. 12.0%), LDL>130 (21.9% vs.
13.8%), LDL<100 (60.8% vs. 51.1%) but also
HDL>45 (66.8% vs. 54.4%). Stratified analyses
showed significant differences in the magnitude of
black-white differences across the clinic’s four
provider teams.
Conclusions: These findings of racial group
differences in diabetic outcomes were presented to
the clinic’s providers and staff and focus groups are
now being held with Black patients of the clinic.
Beyond staff education, interventions will be
devised based on recommendations by staff,
providers and patients, and future outcomes
monitored for narrowing of race-group differences.
Implications for Policy, Delivery or Practice:
This unique quality improvement effort could be
used as a model for improving health care
disparities within individual medical systems.
Funding Source(s): UNC Internal Grant
Poster Number: 213
Identifying and Reducing Differences in Disease
Control Between Black and White Diabetic Patients
of the UNC Family Medicine Center: A UNC
ERACE Disparities Initiative
Presented by: Amy Wallace, M.D., M.P.H., Senior
Scholar, Veterans Rural Health Resource Center-Eastern Region, Veterans Affairs Medical Ccenter,
215 N. Main Street, WRJ, VT 05009; Phone: (802)
295-9363 ext. 6072; Email: aew@dartmouth.edu
Co-authors: Anselm Beach, M.S.
Presented by: Eddie Turner, M.D., Fellow, Family
Medicine, University of North Carolina, 590
Manning Drive, Chapel Hill, NC 27599; Phone:
(919) 966-4363; Email: ejturner@med.unc.edu
Co-authors: Donald Pathman, M.D., M.P.H.; Bonzo
Reddick, M.D., M.P.H.; Randall Teal
Research Objective: The methods of quality
improvement have been proposed as one approach
for identifying and then addressing health care
disparities within a given clinical organization. This
project uses quality improvement approaches to
identify and address disparities for Black and White
diabetics who receive care at UNC’s Family
Medicine Center, through analyses of its diabetes
registry.
Study Design: Black and White patients were
compared on group proportions meeting NCQA
target thresholds for glucose, hypertension control
and lipid control. Additional statistical analyses
explored possible reasons for group differences,
and then findings were presented to practitioners
Rural-Urban, Demographic, and Socioeconomic
Disparities in Lifestyle Counseling for Overweight
and Obesity During Preventive Care Visits
Research Objective: Overweight and obesity have
reached epidemic proportions and are associated
with serious health and social consequences.
Obesity is especially prevalent among rural,
minority, and socioeconomically disadvantaged
populations. Achieving and maintaining a healthy
weight by adopting healthy lifestyle habits may
prevent later obesity-related problems and reverse
existing ones. Encouraging lifestyle changes
through proper nutrition, exercise, and weight loss,
is the recommended initial approach for providers
treating overweight patients. We examined the
likelihood of lifestyle counseling among patients age
18 and older across rural/urban locales, gender,
racial/ethnic groups, median ZIP Code income,
educational, and poverty levels during preventive or
primary care appointments.
Study Design: We used a cross-sectional study
design to examine likelihood of lifestyle counseling
during preventive and primary care visits for non-
pregnant overweight and obese adults from the
2006 and 2007 National Ambulatory Medical Care
Survey database. We calculated odds ratios to
describe providers’ likelihood of lifestyle counseling
during preventive/primary care visits across ruralurban residence locale, demographic, and
socioeconomic characteristics described above.
Population Studied: We extracted 11,203 records
representing 832 million preventive and primary
care visits for non-pregnant patients age 18 and
older with BMI>25 or diagnosis of obesity. Of these,
2,939 records representing approximately 103
million visits included lifestyle counseling.
Principal Findings: Providers offered lifestyle
counseling in only 26% of preventive or primary
care visits with overweight and obese patients.
Across all demographic and socioeconomic
categories, providers were 44% more likely to
counsel suburban-dwelling overweight patients
(p<.0001) and 63% more likely to counsel urban
patients (p<.0001) compared to rural patients.
Adjusting for multiple demographic and
socioeconomic variables, odds ratios for lifestyle
counseling was 1.31 (p=.0004) for suburban and
1.42 (p<.0001) for urban patients compared to rural
patients who were overweight or obese. Providers
were significantly more likely to counsel overweight
Hispanic compared to Whites (OR=0.61, p<.0001 or
Asians (OR=0.51, p=.006), and equally likely to
counsel Blacks compared to other races. Patients
ages 40 to 59 were more likely to receive
counseling compared to patients ages 18 to 40
(OR=1.24, p=.009) but those ages 60 and older
were equally likely to receive counseling. Providers
were equally likely to counsel overweight men and
women. Among socioeconomic variables examined:
percent community poverty, median community
income, median community educational level and
insurance type, none were significant predictors for
lifestyle counseling.
Conclusions: Overall, providers are neglecting
lifestyle counseling for overweight and obese
patients during preventive care visits. Providers
may be stretched for time, or doubt the
effectiveness of their efforts; however, we question
whether these justifications explain why providers
offer counseling to some overweight patients and
not others. When provided, only Hispanic ethnicity,
middle age, and urban residence locale predict for
higher likelihood of receiving counseling. With and
without adjustment for other demographic and
economic variables, rural populations are
significantly less likely to receive lifestyle counseling
directed at nutrition, exercise, and weight loss.
Implications for Policy, Delivery or Practice:
Given the high prevalence of overweight and
obesity in rural America, providers treating
overweight rural patients should encourage healthy
lifestyle habits through counseling during preventive
and primary care visits.
Funding Source(s): VA
Poster Number: 214
The Distribution of Medicare Beneficiaries Across
Plans by Race and Ethnicity
Presented by: Robin Weinick, Ph.D., Senior
Social Scientist and Director of Public Sector
Initiatives, 1200 South Hayes Street, Arlington, VA
22202; Phone: (703) 413-1100 ext. 5151; Email:
rweinick@rand.org
Co-authors: Marc Elliott, Ph.D.; Amelia Haviland,
Ph.D.; Katrin Hambarsoomian, M.S.; Elizabeth
Goldstein, Ph.D.
Research Objective: We assess the extent to
which the racial/ethnic composition of Medicare
Advantage (MA) and Part D Prescription Drug
(PDP) plans reflect that of their service areas,
particularly the the extent to which plans
systematically underrepresent or overrepresent
minority beneficiaries-Blacks, Latinos, or
Asians/Pacific Islanders (API).
Study Design: We estimate three-way crosstabulations of self-reported race/ethnicity, Medicare
contract (“plan”) and county of residence using
451,902 responses to the 2009 Medicare CAHPS,
poststratified to represent all 46 million Medicare
beneficiaries using CMS administrative
race/ethnicity and other administrative variables.
This permitted estimates of the self-reported
racial/ethnic composition of 541 MA and 82 PDP
plans, and estimation of the racial/ethnic distribution
of all Medicare beneficiaries in the geographic area
in which each plan operates. For each plan, we
defined its geographic service area as the set of
counties in which it had enrollees. Within this area,
we calculated the racial/ethnic distribution by
weighting all Medicare beneficiaries at the county
level by the geographic distribution of the reference
plan, giving greater weight to the counties in which
the plan was most active. We used scatterplots and
correlations to compare area and plan racial/ethnic
proportions.
Population Studied: Medicare beneficiaries
Principal Findings: Within MA and PDP plans, the
proportion of Medicare beneficiaries from various
racial/ethnic groups is highly correlated with the
racial/ethnic composition of their geographic service
areas; the correlation coefficient for Latinos is 0.91
(0.98) for MA plans (PDPs); for African Americans
0.85 (0.61); and for Asians 0.83 (0.99), p<0.0001 in
all instances. No plan underrepresents Blacks,
Latinos, or API by 20 percentage points relative to
their service areas, but of 541 MA plans, 40 enroll a
minority group at proportions 30 percentage points
or more higher than the corresponding proportion in
the service area. For example, a plan with 61% API
members operates in a service area in which 11%
of Medicare beneficiaries are API. A few plans’
enrollees consist almost solely of Latinos or API
(e.g. 96% Latino enrollment in a 23% Latino service
area); these plans overtly specialize in these
groups’ needs. No PDPs enrolled a minority group
in a proportion that exceeded the overall rate in its
service area by 30%.
Conclusions: Our findings indicate little cause for
concern that some MA and PDP substantially
underserve particular racial/ethnic groups.
However, we do find that certain MA plans
disproportionately draw specific racial/ethnic
minorities. This is particularly true of Latinos and
Asians, for whom plans with certain services – such
as customer service representatives who speak
languages other than English may be more
attractive. Plans that concentrate Blacks may do so
via marketing efforts. The relatively large service
areas for PDPs may limit the ability to concentrate
by race/ethnicity.
Implications for Policy, Delivery or Practice: For
Medicare to contribute to reducing racial and ethnic
disparities in the U.S., beneficiaries must have
equal access to high-quality plan options. Our
results reassuringly indicate that MA and PDP plans
do not underrepresent minority enrollees. Future
research should address the extent to which plans
that concentrate and specialize in the care of
minority beneficiaries provide better experiences for
such beneficiaries than other plans.
Funding Source(s): CMS
Poster Number: 215
Breast Cancer Screening among Medicaid
Managed Care Members with a Work-Limiting
Disability
Presented by: Sharada Weir, D.Phil., Assistant
Professor, Center for Health Policy and Research,
University of Massachusetts Medical School, 333
South Street, Shrewsbury, MA 01545; Phone:
(508) 856-7852; Email:
sharada.weir@umassmed.edu
Co-authors: Heather Posner, M.S.P.H.; Jianying
Zhang, M.P.H.; Georgia Willis, Ph.D.; Jeffrey
Baxter, M.D.; Robin Clark, Ph.D.
Research Objective: Examine evidence of
disparities in breast cancer screening for an
insured, but vulnerable, population of women
enrolled in Medicaid because of a work-limiting
disability.
Study Design: Individual-level data were obtained
for women who met the National Committee for
Quality Assurance’s 2007 Healthcare Effectiveness
Data and Information Set (HEDIS) criteria for the
breast cancer screening measure (i.e., female,
ages 42-69 at the end of 2006, and continuously
Medicaid-insured during 2005-2006). Compliance
with screening guidelines was assessed for each
individual using HEDIS criteria (i.e., at least one
routine mammography screening in 2005 or 2006).
Disability status was determined from Medicaid
eligibility records. We modeled mammography
screening as a binomial logistic function with
member characteristics, comorbidities, health
services utilization, primary care provider type, and
health plan as explanatory variables. Separate
models for women with and without a work-limiting
disability were also estimated. Adjusted odds ratios
were computed from the logistic regression results.
Population Studied: Massachusetts Medicaid
Managed Care members meeting HEDIS criteria for
breast cancer screening in the 2005/2006
measurement period (n=35,878).
Principal Findings: Two-thirds of women in our
data qualified for Medicaid because of disability.
They were older (54 vs. 48 years of age, on
average), and had higher rates of substance
abuse/dependence (17% vs. 8%), smoking (16%
vs. 10%), domestic violence (1% vs. 0.7%) and
major depression (34% vs. 13%) than other women.
Although unadjusted breast cancer screening rates
were roughly equal for women with and without
disability in Massachusetts, after adjusting for
confounders disability status had a significant
negative association with screening mammography
(OR=0.76; p<0.0001). Several variables were
important predictors regardless of disability status.
Women with alcohol or drug use disorders had
substantially lower odds of screening, as did those
who smoked or had at least one emergency
department visit. Having a higher overall illness
burden or a diagnosis of domestic violence reduced
the odds of screening for women with disabilities
but not for non-disabled women. Both groups
benefited from more ambulatory care visits, but the
impact on screening was much larger among
women with disabilities. Living more than 5 miles
from a mammography facility was negatively
associated with screening among women with a
disability but not among those without. Living in an
area with more college educated women was
positively associated with screening in the nondisabled group but did not increase screening
among women with disabilities.
Conclusions: Our analysis provides evidence of
disparities in routine mammography screening for
women with disabilities and suggests several
contributory factors. For instance, our finding on
distance to screening facilities suggests that
inadequate transportation may be one barrier. On
the other hand, our finding on ambulatory visits
suggests that provider efforts to promote screening
are particularly effective for women with disabilities.
Implications for Policy, Delivery or Practice:
Nationwide, as of 2008, rates of routine
mammography for Medicaid managed care plans
averaged only 50-55%. Given that a majority of
eligible women served by Medicaid have
disabilities, and studies have shown that women
with disabilities are more likely to be diagnosed with
late stage disease, a focus on improving rates of
screening for women with disabilities is overdue.
Funding Source(s): MassHealth (Massachusetts
Medicaid Agency)
Poster Number: 216
Racial Disparity in the Quality of Diabetes Care: Do
Physicians Treat Minority Patients Differently or Do
Minority Patients Receive Care from Different
Types of Physicians?
Presented by: Weifeng Weng, Ph.D., Health
Services
Researcher, American Board of Internal Medicine,
510 Walnut Street Suite 1700, Philadelphia, PA
19106-3699; Phone: (215) 606-4134; Email:
wweng@abim.org
Co-authors: Bradley Gray, Ph.D.; Rebecca Lipner,
Ph.D.
Research Objective: Using patient-level chart
audits, we examined disparities in quality of
diabetes care received by minority (black nonHispanic, Hispanic) versus white patients from
board-certified internists. We parsed these
differences between physicians and within
physicians’ patient panel.
Study Design: We defined disparities as
differences in quality of care measures after
controlling for patients’ age and gender. We
measured within versus between physician disparity
by comparing results from linear-probability versus
fixed-effects models, adjusting standard error for
clustering of patients within physicians. We then
utilized Blinder-Oaxaca decomposition to identify
proportion of between physician disparity explained
by physicians’ cognitive skills (equated score on
first attempt of internal medicine certification
examination; certification in endocrinology
subspecialty), a practice-infrastructure measure
(NCQA's Physician Practice Connections (PPC)),
and insurance revenue source (% of revenues from
Medicare and Medicaid patients). Quality measures
included: intermediate-outcomes (A1C, blood
pressure and LDL good and poor controls),
diabetes care process (A1C test, lipid profile, eye
and foot exam, nephropathy assessment) and
preventive care (influenza and pneumococcal
vaccination, nutrition and physical activity plan).
Population Studied: In years 2008-2009, 774
internists completed American Board of Internal
Medicine’s practice characteristics survey and
Diabetes Practice Improvement Module with 17,896
chart audits and 774 physician surveys.
Principal Findings: Overall, disparities in quality of
care across racial/ethnicity groups as measured by
process of care were small (<1%) except for black
patients’ pneumococcal vaccination (difference=3.3%; SE=2.1%). However, a much different pattern
emerged among intermediate-outcome measures
for black patients in that they had worse
intermediate-outcome measures comparing to
whites with the largest disparity in blood pressure
control (<130/80) (difference=-10.4%; SE=1.3%).
Hispanic patients had worse outcomes in only the
A1C measures relative to whites (e.g. A1C>9%:
difference=7.1%; SE=1.3%). Most of the disparities
in intermediate-outcome measures were due to
variation within physicians’ panel of patients (mean
within variation/total variation=75.8%) rather than
between physicians. For example, black patients’
LDL <100: within difference=-6.4% (SE=1.2%)
versus between difference=-2.3% (SE=0.8%). On
average, our physician level factors explained
74.4% of between variation; insurance revenue
source explained 55.9%, cognitive skills explained
24.4% while practice-infrastructure explained only
0.2% of this between variation.
Conclusions: Do physicians treat minority patient
differently? Do minority patients receive care from
different types of physicians? There were no
significant disparities observed in process
measures between or within physician patient
panels. However, blacks had significantly worse
intermediate-outcome measures than whites while
Hispanics had worse outcomes in A1C measures.
Disparities in intermediate-outcome measures were
largely due to variation within physicians’ patient
panel rather than between. For intermediateoutcomes, the dominant physician level factor was
insurance revenue source; practice-infrastructure
explained little between-physician variation.
Implications for Policy, Delivery or Practice: Our
findings suggest that interventions targeted at
improving intermediate-outcome measures rather
than processes of care are more likely to address
racial/ethnic disparity. The fact that between
physician variation was small indicates that
addressing disparity may require interventions
targeted at meeting different needs of minority
patients. Improving cognitive skills for boardcertified internists treating minority patients might
somewhat reduce the between physician variation
that do exist. Conversely, our results indicate that
policies designed to improve practice infrastructure
as measured by the PPC would not materially affect
racial/ethnic disparities.
Poster Number: 217
Infant Mortality in Appalachian States: The
Influence of Poverty, Race/Ethnicity, and
Geography
Presented by: Nengliang Yao, M.S., Ph.D.
candidate,
Health Policy and Administration, The Pennsylvania
State University, 604 Ford Building, University Park,
PA 16802; Phone: (814) 380-1251; Email:
ayao@psu.edu
Co-authors: Marianne Hillemeier, Ph.D.; Stephen
Matthews, Ph.D.
Research Objective: Recent studies have found
that health outcomes in Appalachia are
disproportionately poor compared with those in the
rest of the nation. While historic impoverishment is
thought to be an important factor, the mechanisms
underlying health disadvantage in Appalachia
remain unclear. This study explores spatial
inequalities in infant mortality rates over time
among geographic sub-regions within the 13
Appalachian states, examining the influences of
geographical location, county poverty level, and
race/ethnic composition.
Study Design: This study uses longitudinal countylevel data extracted from the Area Resource File,
which contains data on variable of interest for the
years 1976 through 2004. The dependent variable
is the 5-year average infant mortality rate.
Descriptive statistics are provided for the sampled
counties, and hierarchical multiple regression
models are fit incorporating interaction effects and
spatial lags. The models test the impact on infant
mortalities of sub-regional differences in geography,
percent of the population in poverty, and the
percent of the population who are African American,
and examine interaction effects between location
and poverty at four time periods: 1976-1980, 19871991, 1996-2000, and 2000-2004.
Population Studied: The population of interest
resides in 1064 counties of 13 Appalachian states:
southern mountain region (N=184 counties),
southern non-mountain region (N=197), central
mountain region (N=89), central non-mountain
region (N=66), northern mountain region (N=147),
and northern non-mountain region (N=381).
Principal Findings: Infant mortality rates
decreased substantially in all sub-regions over the
last three decades. The lowest and highest rates of
infant mortality occurred respectively in the northern
mountain region and southern non-mountain region.
In general, mountain counties had lower infant
mortality rates than non-mountain counties, except
in the central region. Regional inequality of infant
mortality persisted over time but decreased in more
recent decades. In the multivariate regression
analyses, higher poverty rate and percent African
American were each predictive of higher rates of
infant mortality in all regions at all time periods. No
interaction effect of poverty and geographic location
on infant mortality was found. Poverty played the
most important part in predicting life chances for
infants born in the regions studied. The percentage
African American also had a significant effect on
infant mortality rates however this effect declined
over time. Living in the southern mountain region
was associated with comparatively higher infant
mortality during three out of the four time periods
studied.
Conclusions: Regional differences in infant
mortality were mostly explained by the inequality in
poverty rates and race/ethnic composition, but
some regional differences remained significant.
Controlling for other factors, the southern mountain
region experienced excess infant death rates.
Implications for Policy, Delivery or Practice:
Understanding spatial inequality and changes in
infant mortality at the county level provides useful
information for health planning and policy
development. In addition to poverty and
disadvantage associated with race/ethnicity,
policymakers in Appalachian states should also pay
attention to health disadvantages associated with
geographic region of residence.
Poster Number: 218
Rural-Urban Differences in Access to Preventive
Care Services
Presented by: Jangho Yoon, Ph.D., M.S.P.H.,
Assistant Professor, Jiann-Ping Hsu College of
Public Health, Georgia Southern University, Cone
Hall, Room 1043, PO Box 8015, Statesboro, GA
30460; Phone: (912) 478-2416; Email:
jyoon@georgiasouthern.edu
Co-authors: Stuart Tedders, Ph.D.
Research Objective: Previous research indicates
that residents of rural areas are at greater risk of
various health problems compared to their urban
counterparts. Lack of access to preventive health
care services has been suggested as one of the
fundamental factors associated with rural-urban
health disparities. The purpose of this study is to
investigate rural-urban disparities in evidencebased utilization of preventive care services.
Study Design: Data were derived from the 2008
Behavioral Risk Factor Surveillance System
(BRFSS) which provides detailed information on
utilization of preventive health care services for
residents in all 50 states and the District of
Columbia. The BRFSS is a random telephone
survey of civilian, non-institutionalized adults, and
uses a complex multi-stage cluster sample design.
The sample used for this study includes 368,401
adults aged 18 to 64. Outcome variables include
dichotomous indicators for a routine checkup within
2 years and clinically meaningful utilization of
preventive care services for diabetes (influenza
vaccine, pneumococcal vaccine, glycosylated
hemoglobin check, and foot exam by a health
professional) and cancer screening (mammogram,
clinical breast exam, Pap test, blood stool test, and
sigmoidoscopy). We estimated multivariate logit
models, accounting for the complex survey design.
Our key independent variables are residence
categories which include rural and non-metropolitan
urban residence. Residence in a metropolitan area
served as a reference category. The residence
variables were constructed using a 9-level ruralurban continuum classification scheme developed
by the U.S. Department of Agriculture. To isolate
the independent effect of rural residence, we
controlled for a comprehensive set of demographic
and socio-economic factors as well as health
status.
Population Studied: U.S. adults aged 18–64.
Principal Findings: The likelihood of obtaining
regular checkups decreased with rurality; lowest for
rural residents (p < 0.01). For persons with
diabetes, there was no significant association
between rural residence and influenza and
pneumococcal vaccination. However, rural
residence was associated in a linear fashion with a
greater likelihood of having glycosylated
hemoglobin check and foot examination (p < 0.01).
Regarding cancer screening, increased rurality was
negatively associated with the likelihood of having
mammogram (p < 0.05), clinical breast examination
(p < 0.05), blood stool test (p < 0.01), and
sigmoidoscopy (p < 0.01). Our results were
consistent to different definitions of rural residence.
Conclusions: Rural residents are less likely than
non-rural residents to obtain clinically meaning
levels of preventive care services, particularly
regular checkups and cancer screening.
Implications for Policy, Delivery or Practice:
Continued efforts are needed to eliminate the
current rural-urban gap in access to preventive care
services. Improving regular checkups and cancer
screening among rural residents, rather than
targeting manageable chronic health conditions
such as diabetes, may be more efficient use of
limited public resources to address rural-urban
health disparities.
Poster Number: 219
Reasons for Refusals among Women Approached
to Participate in a Postpartum Depression
Intervention Trial
Presented by: Amy Balbierz, Assistant Professor,
Health Evidence & Policy, Mount Sinai School of
Medicine, One Gustave L. Levy Place, Box 1077,
New York, NY 10029; Phone: (212) 659-9567;
Email: elizabeth.howell@msnyuhealth.org
Co-authors: Jessica Block; Rebecca Tyrrell; Caron
Zlotnick, Ph.D.; Elizabeth Howell, M.D., M.P.P.
Research Objective: To determine reasons for
refusals among mothers approached to participate
in a postpartum depression intervention trial; and to
examine whether racial/ethnic differences in
reasons for refusal exist.
Study Design: This is an ongoing randomized
controlled trial of 530 Black and Latino, and 530
White and other minority postpartum mothers
testing whether a behavioral educational
intervention reduces postpartum depression.
Participants randomized to the intervention arm
receive a 2-step intervention that prepares and
educates mothers about modifiable factors
associated with postpartum depression (e.g.
distress from physical symptoms, deficits in social
support, low self-efficacy, and distress from infant
colic), bolsters social support, enhances
management skills, and increases participants’
access to resources. Participants are surveyed at 3weeks, 3-months, and 6-months to assess
depressive symptoms and other factors. All women
who refuse to participate in the trial are asked openended questions about their reasons for refusal.
Responses were transcribed, reviewed for similar
themes, arranged into distinct categories, and
organized as a taxonomy of reasons for refusal.
Racial and ethnic differences in reasons for refusal
were then explored.
Population Studied: Eligible participants are
postpartum women who had uncomplicated
deliveries at an urban hospital. Eligible subjects are
women >18 years of age, English or Spanish
Speaking, had working telephones, and had infants
whose birthweights were > 2500 grams and 5minute Apgar scores >6. Bilingual research
assistants recruit and consent patients during their
postpartum hospital stay.
Principal Findings: Of the 1061 women
approached to participate in this study, 245 refused
(23.1%), and 816 mothers enrolled in the trial.
Refusal rates were higher for White and non-Black
or non-Latino mothers (26.9%, 131/487
approached) versus Black and Latino mothers
(19.9%, 114/574 approached). The taxonomy of
reasons for refusals derived from patient responses
has six main categories related to: 1) lack of
interest, 2) having already been through childbirth
before, 3) time commitment, 4) feeling
overwhelmed, 5) patients moving from the area,
and 6) confidentiality. The reasons for refusals did
not differ by race/ethnicity; however, the rank order
of reasons for refusal differed by race/ethnicity.
Black and Latino mothers vs. White and other
minority mothers were more likely to report that they
were not interested in research (44% vs. 27%,
p=.005) and less likely to report that the study was
too much of a time commitments (33% vs. 51%,
p=.005). Black and Latino mothers vs. White and
other minority women were also less likely to report
they felt too overwhelmed with their current
situation to participate and more likely to report
concerns about confidentiality.
Conclusions: Although women from all
racial/ethnic backgrounds refused to participate in
this trial for the same sets of reasons, Black and
Latino versus White mothers rank ordered reasons
for refusals differently.
Implications for Policy, Delivery or Practice:
Tracking patient refusals and tailoring recruitment
methods based on patient responses may improve
recruitment of women from different racial/ethnic
backgrounds for randomized controlled trials.
Funding Source(s): NIMH
Poster Number: 229
A Longitudinal Analysis of the Relationships
Between Neighborhood SES, Neighborhood
Residential Stability, and Mortality among Women
in the U.S.
Presented by: Chloe Bird, Ph.D., Senior
Sociologist, RAND, 1776 Main Street, PO Box
2138, Santa Monica, CA 90407; Phone: (310) 3930411; Email: chloe@rand.org
Co-authors: Regina Shih, Dr.PH.; Christine Eibner,
Ph.D.; Beth Ann Griffin, Ph.D.; Mary Ellen
Slaughter, M.S.; Tamara Dubowitz, Sc.D., Jose
Escarce, Eric Whitsel
Research Objective: A substantial literature finds
that, even after controlling for individual-level
factors such as income and education,
neighborhood-level social conditions are associated
with individual health. Yet, it is unclear whether key
aspects of neighborhood environments such as
neighborhood socioeconomic status (NSES) and
residential stability each influence risk for specific
causes of death. Therefore, we assessed the risk
for different causes of death associated with both
NSES and residential stability, adjusting for
individual-level characteristics, including health
status. We examined both all-cause and causespecific mortality to determine whether NSES and
residential stability independently influence the risk
of cardiovascular (CVD), cancer and accidental
mortality. Such effects may operate through
different social, behavioral and physiologic
pathways.
Study Design: Using 2-level hierarchical Cox
proportional hazard regression models, we analyze
the Women’s Health Initiative Clinical Trials (WHI
CT) data, merged with neighborhood (Census tractlevel) data. We examined all-cause mortality using
age at death, as well as age at death for each of the
three leading causes of death in the WHI cohort:
CVD, cancer, and accidents. The NSES index
includes census tract-level socioeconomic
measures. Residential stability is assessed as the
percentage of census tract residents who lived in
the same home 5 years earlier. We assessed the
sensitivity of our findings to selection effects of both
observed and unobserved covariates.
Population Studied: Participants age 50–79 years
were recruited (1993–1998) at 40 clinical centers
and 36 satellite or remote sites, and followed
through March 2005. We are examined data from
57,113 women (81.6% non-Hispanic white; 9.9%
non-Hispanic black; 4.5% Hispanic; 4.0% other). At
baseline, 60.8% were married, 94.4% had at least a
high school education, and 63.6% had household
incomes under $50,000.
Principal Findings: Women in lower NSES
neighborhoods experienced significantly higher risk
of all-cause death, controlling for individual-level
characteristics. Compared to residing in a
neighborhood at the 75th percentile of NSES, living
in one at the 25th percentile is associated with 1.13
times greater hazard for death (95% CI = 1.06,
1.19). In addition, living in a neighborhood with a
higher residential stability was associated with
decreased hazard for death (0.96; 95% CI = 0.93,
0.99), controlling for NSES. Additional analyses will
examine the relationships between NSES,
residential stability and cause-specific death.
Preliminary analyses suggest that residential
stability is not equally protective across the different
causes of death.
Conclusions: Living in a lower NSES
neighborhood was independently associated with
greater risk of death, suggesting policies that
improve NSES may also yield health returns. In
addition, the protective effect of living in a
neighborhood with higher residential stability
suggests that social aspects of stable communities
are salutary and independent of NSES and
individual-level socioeconomic characteristics and
health status.
Implications for Policy, Delivery or Practice: Our
findings have implications for disease prevention
and treatment, and suggest a need to pay greater
attention to neighborhood contextual effects on
women’s morbidity and mortality both clinically and
in policy development. Further work is needed to
assess what may be different mechanisms for
increased risk of CVD-, cancer-, and accidentaldeath.
Funding Source(s): NHLBI
Poster Number: 230
Individual and Neighborhood Effects Associated
with Stigma and Quality of Life among HIV-infected
Urban African American Men
Presented by: Aaron Buseh, Ph.D., M.P.H.,
M.S.N., Associate Professor, College of Nursing,
University of Wisconsin-Milwaukee, 1921 East
Hartford Avenue, Room 569, Milwaukee, WI 53211;
Phone: (414) 229-5462; Email:
aaronbg@uwm.edu
Co-authors: Sheryl Kelber, M.Sc.; Patricia
Stevens, Ph.D.; Chang Gi Park, Ph.D.
Research Objective: Neighborhood characteristics
are increasingly recognized as contextual factors
necessary for understanding health, risk behaviors,
disease patterns and treatment and prevention
trajectories. This research uses a multidimensional
approach to the understanding of quality of life
among African American men living with HIV/AIDS
by: (a) exploring and describing individual and
neighborhood constructs influencing stigma and
quality of life (QOL), and (b) developing and testing
a model about the pathways of influence between
individual and neighborhood factors in the
experience of HIV stigma relative to QOL.
Study Design: Cross-sectional descriptive survey
design was used to examine selected sociodemographic variables, their association and
influence on the quality of life among a sample of
urban African American males. Questionnaires
were administered assessing: individual factors,
neighborhood social disorder; self efficacy; social
stigma and QOL. Data were analyzed using
descriptive and inferential statistics.
Population Studied: Participants were recruited
through a community-based organization in an
urban setting a Midwestern state. A snowball
sampling technique was used to obtain a sample of
(n=117) African American men (18 years or older)
who self-reported as being HIV-infected.
Principal Findings: Participants ranged in age
from 21 to 60 years (M=42.62, SD=8.25). The
average length of time since diagnosis =13.17
years (SD=7.63). Over 75% of the men had at least
a high school diploma. Over half identified their
sexual orientation as bisexual or homosexual.
Neighborhood effects, self efficacy, perceived
stigma and current drug use were significantly
correlated to QOL. Current drug users perceived
more stigmas, less self efficacy and reported poorer
QOL than non-drug users. Forty-one percent of the
variance in QOL was explained by individual
factors, neighborhood social disorder, self efficacy
and stigma. Stigma mediated the effects of
neighborhood social disorder on QOL and the
domains of physical health and environment.
Conclusions: Results from this study provide
support for the need to recognize and address self
efficacy, as well as stigma and neighborhood social
disorder, including illicit drug use, to improve the
quality of African American HIV-infected individuals’
lives. While focusing on clinical indicators of HIV
disease and associated co-morbidities is
undeniably necessary for improving quality of life for
persons living with HIV/AIDS, it is not sufficient. We
contend that a more multilevel conception of QOL is
necessary, one that fully recognizes the large
effects psychosocial and environmental variables
such as self-efficacy, HIV stigma, and
neighborhood social disorder can have in reducing
quality of life for urban populations.
Implications for Policy, Delivery or Practice: As
AIDS becomes a manageable chronic condition it
has become necessary to develop culturally
appropriate, empirically validated models that will
assist health practitioners in working to improve
QOL among HIV infected African American men.
Our findings demonstrate that neighborhood
characteristics and stigma had negative effects on
QOL while self efficacy contributed positively to
QOL. Interventions are necessary that incorporates
both individual and neighborhood factors.
Funding Source(s): University of Wisconsin
Graduate School Research Growth Initiative (RGI)
Award, Proposal No: 01-075.
Poster Number: 231
Prenatal Health Behaviors and Postpartum
Depression: Is There an Association?
Presented by: Rada Dagher, Ph.D., M.P.H.,
Assistant Professor, Health Services
Administration, University of Maryland, College
Park, 3460 14th Street NW, Apartment 233,
Washington, DC 20010; Phone: (301) 405-1210;
Email: radadagher@yahoo.com
Co-authors: Edmond Shenassa, Sc.D., M.A.
Research Objective: To investigate the
associations of cigarette smoking, caffeine intake,
and vitamin intake during pregnancy with
postpartum depression at 7 weeks after childbirth.
Postpartum depression is a common and
debilitating mental disorder with a complex etiology;
however, scarce research has examined its
association with prenatal health behaviors.
Study Design: Study participants were recruited
from the postpartum floor at the Women and
Infant's Hospital in Providence, Rhode Island, from
2005 through 2009. Eligible women were
interviewed in-person while hospitalized for
childbirth. A follow-up interview was conducted 7
weeks after delivery at each of the participants’
homes with a 79% response rate (N=622).
Postpartum depressive symptoms were measured
using the Edinburgh Postnatal Depression Scale.
Ordinary Least Square (OLS) analyses were
conducted to examine the associations between
prenatal health behaviors and postpartum
depressive symptoms.
Population Studied: The study population
consisted of all women delivering at the selected
hospital. To participate in the study the mothers had
to be at least 18 years of age, speak English, and
agree to sign the consent form approved by the
IRB.
Principal Findings: The women averaged 28
years old, 46% were married, 31% had a college
degree or higher, 76% were white, and 82% were
born in the US. In regards to prenatal health
behaviors, 38% of women smoke cigarettes, 62%
drank coffee, 68% drank soda, 28% drank tea, and
81% took prenatal vitamins during their pregnancy.
OLS analyses showed the following prenatal health
behaviors were significantly associated with worse
postpartum depressive symptoms: smoking
cigarettes anytime during pregnancy and prenatal
drinking of soda. Moreover, having an infant that
refuses feedings, being stressed out by the
responsibility of being a parent, and having difficulty
balancing responsibilities because of one’s children,
were stressors associated with worse depressive
symptoms. Finally, having a lower level of
education, being born in a country other than the
US, and using both bottle-feeding and
breastfeeding as compared to only bottle-feeding
increased depressive symptoms.
Conclusions: This study found a high percentage
of women to be smoking during pregnancy and that
there exists an association between cigarette
smoking and worse postpartum depressive
symptoms. Due to the cross-sectional study design
that precludes causal inferences, it is possible that
depressed women use smoking as a selfmedication mechanism or that smoking cigarettes
results in changes in mood. Similar conclusions can
be applied to the relationship between drinking
soda and postpartum depression.
Implications for Policy, Delivery or Practice:
Issues for primary healthcare providers and
clinicians to consider include evaluating women for
risk of postpartum depression, identifying prenatal
health behaviors such as smoking and drinking
caffeine. Moreover, healthcare providers can advise
women on the importance of seeking social support
during the childbirth transition to cope with their new
role as mothers and decrease their feelings of
stress due to additional responsibilities. Future
research should investigate the mechanisms that
mediate the associations of smoking cigarettes and
drinking caffeinated products with postpartum
depression and explore other prenatal health
behaviors such as alcohol intake, drug use, and
exercise in relation with postpartum depression.
Funding Source(s): Maternal and Child Health
Bureau, Department of Health and Human Services
Poster Number: 232
“The Best of Both Worlds: Women's Preferences for
Medical Intervention in Childbirth
Presented by: Alexandra Dalton, Ph.D.,
Postdoctoral Fellow, Sheps Center for Health
Services Research, The University of North
Carolina at Chapel Hill, 725 Martin Luther King Jr.
Boulevard, CB# 7590, Chapel Hill, NC 27599;
Phone: (919) 843-3392; Email:
adalton@schsr.unc.edu
Research Objective: Vital statistics data reveal
trends towards both high- and low-intervention
childbirths over the past 30 years. Though
interesting, this data reveals little about women’s
preferences for medical intervention during
childbirth. This study sought to understand these
trends of medicalization and demedicalization within
the context of women’s decision-making and to
uncover any dissonance between women’s
preferences and birth realities.
Study Design: This study was conducted a part of
a larger, mixed-methods study on trends in the
medicalization of childbirth. This study utilized semistructured, face-to-face interviews with 35 pregnant
women to elicit information on their preferences for
medical intervention during childbirth. Participants
were recruited through emails sent to two listservs
for parents and through word of mouth. Interview
topics included childbirth preferences, factors
affecting those preferences, and resources utilized
in developing birth plans; some demographic
information was also collected. Interviews were
audio recorded and transcribed, and were
examined using content analysis.
Population Studied: The population of interest
was pregnant women in the Triangle region of North
Carolina. Participants ranged in age from 23-40 and
were predominately white, married, and welleducated. Slightly less than half of participants were
expecting their first child. Participants represented a
variety of planned birth experiences, including a
range of providers, locations, and preferences for
medical intervention.
Principal Findings: The majority of women
interviewed reported wanting to attempt a natural
(i.e. drug-free) birth. However, most of those
women also wanted to be in a hospital with
immediate access to medical interventions should
they change their mind about pain management or
require emergency care during childbirth. In other
words, they are looking for the best of both worlds.
Women who planned to give birth in a birth center
or at home typically were concerned that, in a
hospital setting, they would be pressured to
undergo medical procedures (episiotomy, induction,
etc.) that were not clinically necessary or pressured
to undergo a cesarean section before other
alternatives were exhausted. Some women who
planned to give birth in a hospital planned to
incorporate more holistic elements, such as hiring a
doula or working with midwives, to ensure that the
birth reflected their personal preferences to the
greatest extent possible. Women reported that their
preferences and plans for childbirth were most often
affected by the experiences and recommendations
of their friends, their partners’ preferences, logistic
or practical considerations, and prior life
experiences.
Conclusions: The findings of this study suggest
that many well-educated, white women are seeking
the best of both worlds, with rescue technology
available in a demedicalized environment. They
prefer to give birth in a setting offering immediate
access to medical intervention, but also want their
personal preferences to be respected, when
clinically appropriate.
Implications for Policy, Delivery or Practice:
Increased use of preference-sensitive, shared
medical decision-making, when clinically
appropriate, may help relieve the tension between
women’s preferences and common medical
practices. In addition, greater attention should be
paid to promoting the options available to women
during childbirth that may help address women’s
needs, including alternatives such as birth centers,
hospital-based midwifery practices, or
nonpharmacologic pain relief techniques.
Funding Source(s): Linda K. George Research
Fund, Duke University Department of Sociology
Poster Number: 233
Gender Differences in Use of Off-Pump and OnPump Cardiac Bypass Graft Surgery and
Associated In-Hospital Mortality: U.S. Community
Hospitals in 2007
Presented by: Bola Ekezue, M.P.H., Doctoral
Student, Public Health Sciences (Health Services
Research), University of North Carolina-Charlotte,
9201 University City Boulevard, Charlotte, NC
28223; Phone: (704) 905-0249; Email:
bekezue@uncc.edu
Co-authors: Jim Laditka, Ph.D, D.A., M.P.A; Sarah
Laditka, Ph.D, M.A, M.B.A.
Research Objective: A growing number of studies
of coronary artery bypass graft (CABG) surgery
suggest better outcomes for patients with off-pump
surgery, compared to conventional on-pump
surgery. There is some evidence that women may
have lower risk with off-pump surgery. More
research in this area is needed using nationally
representative data. We examined gender
differences in hospital mortality following on-pump
and off-pump CABG surgery. A secondary objective
was to examine whether differences in other
characteristics explain the benefits of off-pump
CABG in women.
Study Design: Data were from the 2007
Nationwide Inpatient Sample (NIS), which includes
all discharges from a 20% sample of U.S.
community hospitals. Analyses, stratified by onpump and off-pump status, included chi-square and
multivariate logistic regression weighted for national
representation. The outcome was in-hospital
mortality. The exposure variable of principal interest
was gender. Controls included the number of
coronary arteries bypassed, 21 comorbid
conditions, demographic characteristics (age, race,
health insurance, median household income), and
hospital characteristics (CABG volume, teaching
status, location).
Population Studied: 50,237 adults ages 40+ with
CABG surgery.
Principal Findings: In-hospital mortality was
significantly higher for women than for men for both
on-pump (3.82%, 2.16%, p<.0001) and off-pump
(4.71%, 2.90%, p<.0001). Women had 11% higher
adjusted odds of having off-pump CABG than men
(Odds Ratio, OR 1.11, 95% Confidence Interval CI
1.08-1.14); the analogous result for on-pump was
OR 0.90, CI 0.88-0.92. Women had significantly
higher rates of non-cardiovascular conditions and
indicators of coronary disease severity. Among
those with the off-pump procedure, adjusted
mortality odds were 39% higher for women (OR
1.39, CI 1.25-1.55); among those with on-pump, the
mortality odds for women were 44% higher (OR
1.44, 95% CI 1.35-1.54). Mortality risk factors for
both CABG techniques were comorbid congestive
heart failure, myocardial infarction, valve disorder,
nutritional deficiency, electrolyte disorder and
concurrent CABG/percutaneous coronary
intervention procedure or valve replacement.
Comorbid atherosclerosis in non-coronary vessels
and chronic renal failure were risks for on-pump
CABG. Among women: with off-pump CABG,
Blacks had 81% greater odds of mortality than nonHispanic Whites (OR 1.81, CI 1.31-2.52); but 44%
lower odds of mortality with on-pump (OR 0.56, CI
0.44-0.72). Among men: with off-pump CABG,
Blacks had 106% higher odds of mortality (OR 2.06,
CI 1.56-2.72); there was no analogous difference
for on-pump CABG.
Conclusions: Women having either on-pump or
off-pump CABG have substantially higher risk of inhospital mortality than men. Black women and men
both have particularly high risks of mortality
following off-pump CABG.
Implications for Policy, Delivery or Practice:
Increasing evidence suggesting beneficial effects of
off-pump over on-pump CABG is likely to cause
more patients to request the off-pump procedure.
Caution is suggested, as our findings indicate that
Blacks with the off-pump procedure may have
particularly high risk of in-patient mortality. Although
we controlled for the number of coronary arteries
bypassed and many relevant comorbid conditions,
unmeasured differences in disease severity may
have contributed to the results. More research is
needed to explore the relationship between surgeon
experience and adverse outcomes.
Poster Number: 234
Patient Follow-Up Post Emergency Visit for Heart
Failure (HF): Is there Gender Bias in Referral to
Specialized Clinics?
Presented by: Debbie Feldman, Ph.D., Associate
Professor, École de Réadaptation, Université de
Montréal, C.P. 6128 succ. Centre-ville, Montréal,
Canada 3C 3J7; Phone: +5145282400 ext. 3935;
Email: debbie.feldman@umontreal.ca
Research Objective: Multidisciplinary HF Clinics
are effective for managing this condition. On
average, there are two to three times as many men
than women in these clinics despite equal
prevalence of HF. The objectives of our study were
to: investigate whether men and women who had
an emergency department (ED) visit for HF were
subsequently referred to specialized HF clinics;
describe when this referral occurred and explore
factors that were related to prompt referral.
Study Design: Within 6 weeks of discharge,
participants were contacted and interviewed by
telephone. The interview consisted of questions
related to services (those received in the past, and
any consultation or referral post discharge),
sociodemographics, the Stanford self-efficacy
scale, and the Minnesota Living with Heart Failure
Questionnaire. The interview was repeated
subsequently at 3 and at 6 months. In addition, a
research nurse completed a standardized chart
review at recruitment, 3 and 6 months. Analysis
consisted of descriptive statistics regarding delays
to referral, logistic regressions which modeled
referral to the clinic at 6 weeks, 3 months and 6
months post ED visit as a function of gender,
sociodemographics, severity, comorbidities,
preferences regarding service providers and selfefficacy.
Population Studied: Patients with HF were
recruited by nurses in 8 hospital EDs in Quebec.
Principal Findings: Of the 416 patients included in
this study, 339 finished the 6 months follow-up
period. Mean age of these 339 patients was 76.0
years (±10.5) and 50.4% were women. At 6 weeks
post ED visit, 99 (24%) were referred to specialized
HF clinics. This figure increases to 121 (31%) at 3
months and 128 (38%) at 6 months. More men
were referred at each of these times (p < 0.0001).
Multivariate logistic regression indicated that men
were more than twice as likely to be referred within
6 months of their ED visit (adjusted OR 2.3, 95%CI
1.3, 4.2). Other factors associated with referral were
younger age (adjusted OR 0.95, 95% CI 0.92, 0.98)
and lower comorbidity (adjusted OR 0.84, 95% CI
0.75, 0.95).
Conclusions: More than a third of patients (38%)
with HF were referred to HF clinics within 6 months
following their ED visit. Younger male patients with
lower comorbidity are more likely to be referred to
clinics.
Implications for Policy, Delivery or Practice:
Research has shown that these clinics benefit all
HF patients. Standardized referral guidelines should
be developed in order to ensure that these clinics
are accessible to all patients who may require these
services.
Funding Source(s): CIHR
Poster Number: 235
Using Collaborative Modeling to Improve Birth
Outcomes
Presented by: Rachel Ferencik, M.P.A., Research
Associate, Georgia Health Policy Center, Georgia
State University, P.O. Box 3992, Atlanta, GA US
30302; Phone: (404) 413-0307; Email:
rachel@gsu.edu
Co-authors: Heather Devlin, M.A.; Karen Minyard,
Ph.D.; Chris Soderquist, B.S.; Angela
Snyder, Ph.D., M.P.H.
Research Objective: Despite being a rich nation,
low birth weight trends in the United States
resemble those in many of the world’s poorest and
most impoverished countries. Many ways to
improve healthy birth outcomes are known,
including reducing unintended pregnancies,
increasing birth spacing, reducing maternal
smoking and improving access to comprehensive
health care. Despite this knowledge, the number of
babies born at low and very low birth weights is
increasing. This project’s objective was to examine
policy and program alternatives from a systemic
perspective and to focus state efforts on the highest
leverage opportunities to reduce low birth weight
births, their complications and resulting economic
and societal costs.
Study Design: This collaborative modeling project
was initiated by a state health agency. The
collaborative systems modeling team includs state
agency representatives, legislative staff and experts
in maternal/child health, obstetrics, and
epidemiology. The model groups women into four
categories: women of child bearing age with and
without a reproductive plan and women of child
bearing age at risk or not at risk of a low birth
weight birth. The model simulats movement
between groups based on the impact of
interventions to reduce low birth weight.
Interventions were selected using a combination of
data from clinical trials and program evaluations
and expert consensus, where published evidence
was lacking. Interventions could either improve the
pre-conception odds of a woman having a normalweight birth or improve the post-conception odds
that subsequent births would be of normal weight.
Four preconception interventions were modeled: 1)
increasing a woman’s general health knowledge; 2)
increasing family planning coverage; 3) providing
primary care during inter-conception; 4) increasing
pregnancy health knowledge. Four post-conception
interventions were also modeled: 1) improving
pregnancy health knowledge; 2) implementing
Centering Pregnancy Models; 3) home visits for
high risk pregnancies; and 4) increasing Medicaid
coverage for pregnant women. A simple user
interface enables users to explore the interventions,
alone or in combination and at varying intensities, in
terms of their impact on low birth weight rates and
the associated health care cost over the coming
decade. The model is designed for real-time,
hands-on exploration in a learning lab environment.
Participants are encouraged to predict outcomes,
articulate theories and inquire into any differences
between their expectations and the model’s
outcomes.
Population Studied: Women of child bearing age
Principal Findings: Preliminary results show that
increasing the percentage of women with a
reproductive plan begins to reverse the trend in low
birth weight over the next ten years. We
hypothesize that focused inter-conception care for
women at risk of low birth weight will have a strong
effect on reducing low birth weight. However, it is
likely that multiple, intensive and sustained
interventions will be required over time to achieve
desired reductions in low birth weight. Complete
results will be available in April 2010.
Implications for Policy, Delivery or Practice:
Using a collaborative system modeling process and
perspective, this project brought together legislative
staff, researchers, and content experts to develop a
set of actionable policy options that can be used to
inform state policy to reduce low birth weight. The
resulting model provides a framework, a common
language and a credible and accessible tool that
enables interested stakeholders to engage in more
rigorous discussions about effective and feasible
interventions and policy options for reversing the
trend of low birth weight.
Funding Source(s): Georgia Department of
Community Health
Poster Number: 236
Emergency Department Utilization by Males May
Drive Overall ED Utilization Rates
Presented by: Kathleen Fuda, Ph.D., Sr.
Associate, Health Policy Practice, Abt Associates
Inc., 55 Wheeler Street, Cambridge, MA 02138;
Phone: (617) 520-2493; Email:
kathy_fuda@abtassoc.com
Co-authors: Rachel Fisher, M.S., M.P.H., R.D.
Research Objective: To determine whether gender
plays a role in determining a state’s overall per
capita utilization of emergency department (ED)
visits. In 2005, annual statewide per capita ED visit
rates varied widely, from a low of 262/1000 in
California to a high of 642/1000 in West Virginia.
Although per capita ED visit rates are higher among
women than among men when measured
nationally, it is not known whether gender-based
differences in utilization patterns are consistent
across the states, and whether total statewide
utilization rates may be more closely linked to
utilization by males or females.
Study Design: We analyzed Massachusetts
statewide data to determine age and gender
utilization rates for all ED visits, including those
resulting in admission, and compared them to
national ED utilization rates as estimated by the
NHAMCS. In addition, we used AHRQ’s HCUPNet
tool to determine the number of ED visits by gender
in the seven states for which data were available in
2005; used census estimates to derive gender
specific utilization rates for each state; and
compared them to the total ED utilization rates for
each state.
Population Studied: ED patients in
Massachusetts, Hawaii, Iowa, Minnesota,
Nebraska, New Hampshire, Utah, and Vermont.
Principal Findings: Analysis of Massachusetts ED
data showed that the state as a whole had a higher
than average per capita utilization rate (425/1000
vs. 372/1000 nationally). However, Massachusetts
females’ ED utilization was actually lower than the
national average, while Massachusetts males had
substantially higher per capita rates. This was not
driven primarily by age distribution; females were at
or below national averages in most age categories,
and males were higher than average in every
category except for children <15 years, so that the
overall male per capita rate for all ages was higher
than that for females, reversing the usual pattern
(430/1000 for males vs. 383 for females/1000).
Aggregate statistics for gender-specific ED visits in
Massachusetts in 2005, and seven additional states
available through HCUPNet, indicated that other
states’ overall ED utilization are similarly correlated
with disproportionately high percentages of ED
visits attributable to males (r=.30). Hawaii is an
outlier as the state with the second lowest ED
utilization rate in the country in 2005; when it was
removed from the data, the correlation was much
stronger (r=.92).
Conclusions: Gender may play an underrecognized role in driving ED utilization rates. The
reasons why males in some states are
disproportionately more or less likely to visit
emergency departments compared to female
residents of the same states deserve further
exploration.
Implications for Policy, Delivery or Practice:
Most hypothesized drivers of ED utilization, such as
poor access to primary care, geographical factors,
payer type, injury and disease patterns, etc. may
affect males and females differently, which should
be considered in research and policymaking. In
addition, there may be gender-specific beliefs or
attitudes that may vary geographically and help to
explain the patterns discovered. Finally, many
attempts to discourage “inappropriate” ED visits,
such as increased copayments, apply to all
population groups regardless of utilization patterns;
more targeted approaches might be more
successful while minimizing negative effects.
Poster Number: 237
Defining Caseload for Patient Navigation
Presented by: Mary Howard, B.S., Research
Assistant, Women's Health Unit, Boston University
School of Medicine, 801 Massachusetts Avenue,
Boston, MA 02118; Phone: (617) 638-8036; Email:
mbhoward@bu.edu
Co-authors: Karen Freund, M.D., M.P.H.; Christina
Santana, M.P.H.; Kristine Beaver, M.P.H.; Sharon
Bak, M.P.H.; Tracy Battaglia, M.D., M.P.H.
Research Objective: Patient navigation is an
innovative type of care management designed to
reduce cancer disparities. As many sites adopt this
model of care, a frequent query is how to assess
appropriate caseload for these new positions. While
data on the effectiveness of patient navigation
interventions appears promising, there is a lack of
literature that describes methods to assess
appropriate caseload for optimal effect of navigator
intervention. Boston University and its affiliated
Community Health Centers (CHCs) is one of nine
sites participating in the National Cancer Institute
Patient Navigation Research Program (PNRP),
which aims to develop and test navigation to reduce
delays in delivery of standard cancer services after
an abnormal breast cancer screening. The purpose
of this study is to develop and apply measure of
caseload comparable across settings, and describe
the caseload across three CHCs with breast health
navigators participating in the Boston PNRP.
Study Design: A subset of Boston PNRP sites with
patient navigation for breast cancer screening was
the focus of this study. Navigators target female
patients with abnormal breast cancer screening,
with fewer than 10 cases annually navigated during
treatment for breast cancer. Review of the literature
identified several concepts to consider in navigator
caseload: (1) a simple count of number of patients
(2) total patient-days in navigation, (3) documented
time spent navigating, and (4) complexity of
navigated patients, as defined by problem
(diagnostic evaluation versus cancer care) and
number of barriers to care. We examined the
following factors for each navigation site: absolute
number of cases navigated, patient-days in
navigation, and total documented patient-navigator
encounters. A formula was developed to calculate
caseload from the patient-days assessment: Total
patient-days in navigation divided by Total days of
observation divided by FTE of navigator. After a
caseload score was calculated for each health
center, we compared caseload by site to number of
navigated patients, encounters, and complexity of
navigated patients.
Population Studied: Eligible participants were
women over 18 with a Breast Imaging Reporting
and Data System (BIRADS) 0, 3, 4, or 5
mammogram at the participating CHCs from
7/1/2007 through 6/31/2008. Patients were
navigated until their screening abnormality was
resolved with a definitive diagnosis (cancer or noncancer) or until the end of cancer treatment.
Principal Findings: The calculated caseload
scores were 26.3, 71.8, and 75.8 for CHCs A, B
and C respectively. This large variation in caseload
can be somewhat accounted for by numbers of
navigated patients during this time frame (98, 175
and 248, respectively). The number of documented
encounters by navigators did not correlate with the
caseload score.
Conclusions: The caseload scores for breast
health navigation in Boston PNRP varied across
CHCs. The caseload score correlates somewhat
with number of patients, but may allow a better
reflection of the complexity of the care needed for
individual patients. Linking this measure with patient
outcomes will better illuminate the range in
caseload.
Implications for Policy, Delivery or Practice:
Developing an understanding of caseload is
essential to implementing viable patient navigator
programs. A calculation of caseload can also be
linked to outcomes and the size of a navigator’s
caseload can be controlled for in future analyses.
Funding Source(s): NCI
Poster Number: 238
Gender Differences in Management and Outcomes
of Acute Myocardial Infarction Patients- The Role of
Health Information
Presented by: Nicole Huang, Ph.D., Associate
Professor, Institute of Hospital and Health Care
Administration, National Yang Ming University, 155
Li-Nong Street, Section 2, Taipei, Taiwan 112;
Phone: +886228267372; Email:
syhuang@ym.edu.tw
Co-authors: Yiing-Jenq Chou, M.D., Ph.D.; ChengHua Lee, M.D., Dr.PH.
Research Objective: Providing information or
health knowledge to less advantaged patients has
been suggested as a major strategy to reduce
health disparities. Sex-based differences in access
to care and outcomes reported by some studies
have raised the concern of sexual discrimination
against women. We wonder whether a higher level
of medical information would help to narrow such
“disparities,” while many argue that gender
differences in medical care mainly result from
clinical manifestation. In such cases, we would
expect this difference to be observed among wellinformed women and men. Furthermore, whereas
the existing evidences are limited mostly to western
populations, due to heterogeneity in culture and
traditional gender role patterns in Asian societies,
gender issues in medical care may be more serious
in Asian populations. The objectives of this study
are to compare cardiac procedure use and mortality
between men and women in a typical Asian
population, and to determine whether such sex
differences exist among health professionals and
their relatives.
Study Design: Retrospective analyses were
conducted using the National Health Insurance
data. for patients, who had been admitted to a
hospital for acute myocardial infarction in Taiwan
between 1997-2002. Main outcome variables
included use of catheterization and
revascularization within 1 year of the index
admission, and 1-year mortality. Patient
characteristics (age, gender, relationship to health
professionals, comorbidity, residential region),
physician characteristics (age, gender, specialty),
and hospital characteristics (accreditation level,
case volume) were adjusted in the models.
Generalized Estimating Equations and Cox
proportional hazard models were applied.
Sensitivity analyses were also conducted for
different age strata, and general patients with a
higher socioeconomic status.
Population Studied: Patients who had been
admitted to a hospital for acute myocardial
infarction in Taiwan between 1997-2007.
Principal Findings: General Taiwanese women
not only underwent fewer cardiac catheterization
(+35.4% vs 59.7%; P<0.001) and revascularization
(25.4% vs 47.4%; P<0.001) than men, but also had
higher 1-year mortality rate (52.1% vs 19.1%;
P<0.001). In contrast, sex differences in procedure
uses observed among health professionals were
much smaller in magnitude (catheterization: 38.1%
vs 41.6%; P=0.59; revascularization: 23.8% vs
31.1%; P=0.24). Nonetheless, after adjusting for
potential confounders, either general women or
female professionals were considerably less likely
to undergo cardiac catheterization and
revascularization than their male counterparts.
General women and female professionals had
significantly higher mortality risks than their male
counterparts. Sex differences observed in
procedure use and mortality (P for interactions >
0.05) were not smaller when a patient was a health
professional.
Conclusions: Women who have had an AMI
undergo an invasive procedure less often, and
suffer a worse survival outcome than men in a
typical Asian population. More importantly, sex
differences persist among health professionals and
relatives who are considered better informed and
more privileged when they become patients.
Implications for Policy, Delivery or Practice: The
preliminary results suggest that the presence of
similar gender difference among informed
consumers may indicate factors such as clinical
manifestation may be more plausible explanations
for the observed gender differences in access to
invasive procedures and outcomes.
Funding Source(s): Taiwan National Science
Council
Poster Number: 240
Birth after Cesarean Section: Changes Over a TenYear Period in New Jersey.
Presented by: Ya-lin (Aileen) Huang, M.S., Ph.D.
candidate, Health policy and management, Emory
University, 1518 Clifton Road Room 623, Atlanta,
GA 30322; Phone: (404) 693-7546; Email:
yahuang@emory.edu
Research Objective: To determine whether the
variation of choices of delivery methods among
parturients with previous Cesarean section over
time has impacts on related maternal and neonatal
outcomes.
Study Design: A retrospective analysis of all
hospital births from 1997-2006 in New Jersey using
discharge data from New Jersey State Inpatient
Database (SID), part of Healthcare Cost and
Utilization Project (HCUP). We compared maternal
and neonatal outcomes between a trial of labor and
elective repeat Cesarean delivery among women
with previous Cesarean delivery. Outcomes of
interest were maternal and neonatal mortality,
major maternal complications such as hemorrhage,
infection and thrombotic condition, major neonatal
complications such as 5-minute Apgar score <7,
respiratory distress syndrome and neonatal injury.
All analyses are undertaken by year to assess
trends over time.
Population Studied: Women experiencing the next
birth after at least one Cesarean section delivered
in New Jersey hospitals between 1997 and 2006,
and their neonates, as linked through hospital
discharge data. We identify 142,121 deliveries with
previous Cesarean section and 108,012 motherneonate pairs.
Principal Findings: Over the ten-year period, the
rate of successful vaginal birth after cesarean
section (VBAC) declined significantly in New Jersey
(31%-7%). Both the rates of maternal complication
and neonatal complications declined over time. The
neonatal mortality rate in babies whose mothers
had a trial of labor was higher than those babies
born after an elective repeated Cesarean section
(OR=1.63; 95% CI: 1.11-2.64) although the
absolute numbers are very small. There was no
difference in maternal mortality risk between the
two groups. Neonatal complications were more
frequent in the trial of labor group, take respiratory
distress syndrome for example (OR=1.44; 95% CI:
1.29-2.87).
Conclusions: Rates of VBAC have declined over
this ten-year period. Rates of maternal and
neonatal mortality and morbidity are generally in the
low range and improved over time.
Implications for Policy, Delivery or Practice:
Women may benefit from shared decision-making
tools and enhanced physician-patient
communication on the subject of comparative
benefits and risks when determining their birth
preferences after Cesareans. More research to
predict which women are likely to achieve a
successful VBAC is essential.
Poster Number: 239
Understanding Women Veterans’ Use, and Unmet
Needs, for Reproductive Health Services in the
Veterans Administration
Presented by: Kristin Mattocks, Ph.D., Associate
Research Scientist, Internal Medicine, Yale
University, 950 Campbell Avenue, 11-ACSLG, West
Haven, CT 06510; Phone: (413) 548-6777; Email:
kristinmattocks@yahoo.com
Co-authors: Cara Nikolajski, B.A.; Sally Haskell,
M.D.; Cindy Brandt, M.D., M.P.H.; Jennifer McCallHosenfeld, M.D.; Sonya Borrero, M.D., M.S.
Research Objective: Women veterans constitute a
growing population of VA users. Among other
medical and mental health services, many women
utilize the VA for reproductive health care services,
including routine gynecological care,
mammography, and infertility services. However,
little is known regarding women’s knowledge of the
availability of these services, nor the extent to which
available services meet women’s needs. Therefore,
the objectives of this study were to: 1) explore
women veterans’ experiences with reproductive
health services in the VA; 2) understand what
reproductive health needs were not being met, and
3) assess how the VA could improve the scope and
quality of reproductive health services offered to
women veterans.
Study Design: A series of focus groups consisting
of 5-8 women veterans enrolled in the VA
Healthcare System were conducted at VA
Pittsburgh and VA Connecticut. The focus groups
were audio-taped, transcribed, and analyzed to
identify emergent themes. A standard protocol was
used to ask participants to discuss: Reproductive
health care services used at the VA, knowledge of
the full range of reproductive health services
available, barriers to receiving reproductive care,
and recommendations for improvement of
reproductive health care at the VA.
Population Studied: Research participants
included 21 women veterans, ranging in age from
29-46, receiving care in women’s health clinics at
VA Pittsburgh and VA Connecticut.
Principal Findings: Analyses revealed that many
women veterans were unaware of the range of
reproductive health care services available to them.
While nearly all the women had received routine
gynecological care from the VA, including Pap
smears, contraception, and mammography, women
had varying levels of knowledge and experience
regarding pregnancy and infertility care. Several
women were unaware that pregnancy and infertility
care were available from fee-basis providers
outside the VA. Women were generally satisfied
with the care they received from VA providers for
their reproductive health needs, though some
women expressed dissatisfaction with the degree of
gender-specific training among medical residents
working in women’s health clinics in the VA.
Recommendations for improved reproductive health
care within the VA included readily available
information detailing what reproductive health
services are available, more flexible hours to meet
women’s needs, and more extensive genderspecific training of medical residents providing care
in women’s health clinics in the VA.
Conclusions: Women utilize a broad array of
reproductive health care services in the VA.
However, we identified a number of important
reasons why women may not be accessing the full
range of reproductive health services available to
them. Chief among these reasons was a perceived
lack of available information regarding the full range
of VA benefits available for reproductive health
care.
Implications for Policy, Delivery or Practice: The
VA has made significant progress over the last
decade to improve the quality of care it provides to
its women veterans. The VA must make efforts to
ensure that women are aware of the reproductive
health services available to them, and that services
provided through fee basis providers outside the VA
are easily accessible and coordinated with ongoing
VA care.
Funding Source(s): VA
Poster Number: 242
Comparing Primary Care Appointment Wait Time
for Women Veterans When Both General Primary
Care Clinics and Women’s Health Clinics Are
Available
Presented by: David Mohr, Ph.D., Center for
Organization, Leadership and Management
Research, VA Boston Healthcare System, 150
South Huntington Avenue (152M), Boston, MA
02130; Phone: (857 364-5679; Email:
david.mohr2@va.gov
Co-authors: Justin Benzer, Ph.D.; Marjorie Nealon
Seibert, M.B.A.; Bert White, M.B.A., D.Min.
Research Objective: Women Veterans comprise
about 14% of the Veteran population. Many VA
health care centers have changed structures and
processes to provide care in women’s health clinics
to meet the needs of this population. This study
examines waiting times for VA medical hospitals
and community clinics that provide both a general
internal medicine clinic and a women’s health clinic
in their primary care service.
Study Design: We used administrative data for the
study that was collected between October 2006 to
September 2007. The outcome variables for this
study consisted of average appointment wait time
and the percentage of appointments completed
within 31 days of request. We ran a Poisson
regression to identify significant factors associated
with longer waiting times, including type of clinic,
four U.S. Census regional areas, urban or rural
indicator, total number of appointments, new to
established patient ratio, teaching hospital
affiliation, and hospital or community-based
designation.
Population Studied: To be included, a site needed
both a general internal medicine and women’s
health clinic and at least 100 appointments by
women Veterans during the 12 month period. A
total of 155 sites were included. Only women
patients seen in either of these two clinics were
included. A total of 185,227 appointments were
made to women’s health clinics and 313,150
appointments were made to general internal
medicine for the study sites.
Principal Findings: The average wait time was
7.21 days for women’s health clinics compared to
4.82 for general internal medicine clinics. The mean
wait was longer in women’s health clinics
(SD=4.14). Approximately 95.7% of appointments
were completed within 31 days of request in
women’s health clinics compared to 97.3% in
general internal medicine clinics. The regression
model for average appointment wait time indicated
lower appointment volume, a higher rate of new
patients, and western geographic region were
associated with longer appointment wait times.
Conclusions: For medical centers with both
general internal medicine and women’s health
clinics, the average wait time for women was
slightly longer in women’s health clinics. Additional
factors that influenced this relationship were
appointment demand variables. The supply of
providers was not modeled for each clinic in this
study, but could be a contributing factor to the
longer wait times. Another reason for the wait may
have involved greater time to conduct genderspecific tests or procedures. Differences in
procedures used between the clinics could be an
area for further investigation to explain this
difference as well as patient preference with choice
of clinic and wait time.
Implications for Policy, Delivery or Practice:
Although the women’s health clinics had slightly
longer waiting times, the difference was an average
of 2.4 days. Possible factors that could reduce this
wait may include staffing and resources in women’s
health clinics.
Funding Source(s): VA
Poster Number: 243
DACOWITS: An In-Depth Analysis of Women in the
Military
Plan B as Insurance: The Effect of OTC Emergency
Contraception on STIs and Abortions
Presented by: Laura Nelson, M.P.H., Research
Associate, Altarum Institute, 4401 Ford Avenue,
Suite 800, Alexandria, VA 22302; Phone: (703)
575-1892; Email: laura.nelson@altarum.org
Co-authors: Christina Borrelli, M.H.A.; Erin Haslag,
M.P.H.; Karen Metscher, Ph.D.; Lorraine Babeu,
Ph.D.
Presented by: Anjali Oza, The University of
Chicago, 211 E Ohio Street Apartment #2417,
Chicago, IL 60611; Phone: (281) 961-0757; Email:
aoza@uchicago.edu
Research Objective: The objective of this study is
to use quantitative data to identify significant
differences in reported mental well-being, physical
well-being and stress-levels between genders in the
Military.
Study Design: The Department of Defense's
Health Related Behaviors (HRB) responses were
used in an analysis (correlation, regression testing
of the survey questionnaire responses). Random
selection of 60 DoD Services installations with >
500 personnel assigned were sampled and grouped
by Service, Major Command, and Region
(CONUS/OCONUS) and 10 Coast Guard
Installations Active-duty personnel randomly
selected at installations based on pay grade and
gender. The survey was administered anonymously
on-site. The variables of interest included survey
questionnaire measures for mental well-being,
physical well-being and stress-levels. The sample
frame includes 28,546 responses, with a response
rate of 70.6% (vs. 51.8% in 2005 Survey).
Significant differences in gender responses and
patterns are detailed through the quantitative
analysis.
Population Studied: Active Duty Military Personnel
(females)
Principal Findings: The 2005 DACOWITS
analysis reported significant health, stress-level and
stressor differences existed between genders
amongst Military personnel. The 2008 HRB data
was recently released and the analysis of the
significant differences will be analyzed, reported
and compared to the 2005 data.
Conclusions: There are significant (p-value >=
0.05) differences in stress levels and stressors
encountered among the genders serving in the
Military. Identifying key stressors and enacting
policies to promote health and reduce stress is
critical to sustaining the mission of our Military.
Implications for Policy, Delivery or Practice: The
primary objective of the DACOWITS analysis is to
provide meaningful information that will enable
policy makers and Military leadership to enhance,
improve and continue to support women in the
Military.
Poster Number: 244
Research Objective: The United States has the
highest number of unintended pregnancies and
abortions among Western industrialized countries.
Emergency contraception is 89 percent effective in
preventing unintended pregnancy when taken
within 72 hours of unprotected sex. In August 2006,
the FDA announced that Plan B, commonly referred
to as the “morning after pill”, would be made overthe-counter for women 18 and older. This paper
argues that the introduction of over-the-counter
access to Plan B lowered the risk of unintended
pregnancy and thereby lowered the costs of
unprotected sex. Consequently, the policy played a
significant role in the decline in abortion between
2005 and 2007, but may have led to an increase in
risky sexual behavior on the margin as women
engaged in more sex and substituted away from
contraceptive methods such as condoms, birth
control pills, withdrawal, and abstaining from sex.
This may have resulted in an increase in sexually
transmitted infections (STIs).
Study Design: I employ a difference-in-difference
approach to explore the impact of pharmacy access
to Plan B on STIs and abortions. My identification
assumption relies on the trends for early and late
adopter states being the same prior to the policy
change. I include state level controls for public
health infrastructure, funding for family planning,
concentration of planned parenthoods, and number
of abortion providers. I include county level fixed
effects to control for time invariant county level
attributes that may be correlated with risk behavior,
policy uptake, or the effectiveness of the policy.
Population Studied: I have a balanced sample of
1.5 million women followed continuously over time. I
use claims data from MarketScan, the largest
collection of employer-based patient data.
Principal Findings: I find that improved access to
Plan B accounts for as much as 37.2 percent of the
decline in abortion and up to 17.8 percent of the
increase in STIs between 2005 and 2007 among
women ages 15-29. I perform a range of sensitivity
analysis and robustness checks to address the
concern that early and late states may not be
comparable. I find that the results are consistent.
The potential direct medical cost savings from Plan
B use is $1.47 billion and the increase in medical
cost from the subsequent rise in STIs would cost
$283 million.
Conclusions: I find that there are advantages and
disadvantages to making a low-toxicity drug like
Plan B available over-the-counter. Given the ethical
debate surrounding abortion and given that some
people believe Plan B is equivalent to termination,
how we weigh the consequences of the policy really
depends on how society places value on each
outcome.
Implications for Policy, Delivery or Practice: The
findings are relevant for understanding how public
health policy affects individuals’ incentives and for
discussing the policy’s impact on direct medical
costs to the healthcare system.
Funding Source(s): AHRQ
Poster Number: 245
Creating a Quality Measurement Strategy for
Women of Reproductive Age
Presented by: Sarah Scholle, M.P.H., Dr.PH.,
Assistant Vice President, Research, National
Center for Quality Assurance, 1100 13th Street NW
Suite 1000, Washington, DC 20005; Phone: (202)
955-1726; Email: Scholle@ncqa.org
Co-authors: Natalie Davis; Dana Rey, M.P.H.; Beth
Tapper, M.A.; Carol Weisman, Ph.D.
Research Objective: Performance measurement is
a powerful tool for driving improvements in the
quality of care. The most commonly used quality
measures for women of reproductive age focus on
gender- specific preventive screenings and
pregnancy-related visits; they do not consider the
full range of women’s preventive needs. The
purpose of this project was to identify a
comprehensive framework for evaluating the quality
of preventive care for women of reproductive age,
including pregnancy-related care.
Study Design: The National Committee for Quality
Assurance (with funding from Centers for Disease
Control and Prevention and Health Resources and
Services Administration), working with the American
Medical Association convened Physician
Consortium for Performance Improvement,
reviewed existing guidelines and measures for
preventive and pregnancy-related care for women
of reproductive age. A group of diverse
stakeholders was convened to create a framework
for prioritizing relevant measurement efforts. The
workgroup discussed goals, principles, key
priorities, as well as barriers, for quality
measurement and improvement.
Population Studied: Women of reproductive age
Principal Findings: Based on the recognition that
women’s overall health affects the woman and
pregnancy-related experiences and outcomes, the
panel recommended quality measurement should
address the full range of women’s health care
needs, including gender-specific needs as well as
conditions where gender differences exist in
prevalence, quality of care or outcomes. Second,
issues of care coordination across multiple health
care providers and across women’s life stages
should be considered. Third, women’s needs from
underserved/minority communities should be
addressed. The panel suggested that priorities for
measurement should focus on outcomes for healthy
women and normal childbirth. Key areas for
measurement include: mental and behavioral
health, substance use, obesity and nutrition, cancer
screening, chronic disease identification and
management, and reproductive life planning. Two
cross cutting issues were noted: the need for
engaging women in health care decision making
and health promotion, as well coordinated care
among different care providers (especially care
divided between generalists and reproductive
health specialists). For pregnancy-related care,
priorities for measurement include: availability of
prenatal record at the time of delivery, sharing risks
identified in pregnancy with the ongoing care
provider, and overuse of procedures (such as
elective induction of labor prior to 39 weeks and
cesarean section). A new conceptualization of
postpartum care is needed, as current measures do
not address post-pregnancy needs. Related
concerns include: methods for educating and
engaging consumers in their health care;
improvement of data sets and reducing barriers to
data sharing across providers and settings; and
reimbursement and liability concerns with regard to
preventive and pregnancy-related services.
Conclusions: Quality measurement for women of
reproductive age should focus on women’s
comprehensive health needs, should support care
coordination and data sharing across care providers
and life stages, and should engage women in their
health and health care.
Implications for Policy, Delivery or Practice:
Federal efforts to monitor quality in Medicaid and to
encourage adoption of health information
technology offer new opportunities for implementing
a broader framework for the quality of women’s
health care. NCQA, the AMA-PCPI and the
American College of Gynecologists are currently
developing a pregnancy-related care measure set.
Funding Source(s): CDC
Poster Number: 246
Do New Guidelines and Technology Make the
Routine Pelvic Exam Obsolete?
Presented by: Carolyn Westhoff, M.D., M.S.,
Family Planning Division Director, Obstetrics &
Gynecology, Columbia Presbyterian Medical
Center, 622 W. 168th Street PH 16-69, New York,
NY 10025; Phone: (212) 305-7070; Email:
clw3@columbia.edu
Co-authors: Heidi Johnson, M.P.H.; Maryam
Guiahi, M.D.
Research Objective: In the United States, the
routine pelvic exam is the core of the periodic
gynecological examination, and widely tolerated by
asymptomatic patients who accept this as a
necessary part of health maintenance. Is this
routine examination beneficial for the asymptomatic
woman? Specific justifications for the pelvic exam
include: 1. screening for sexually transmitted
infections (STIs), especially Chlamydia and N.
gonorrhea, 2. evaluation prior to the initiation of
hormonal contraception, 3. screening for cervical
cancer, and 4. detecting early cases of ovarian
cancer.
Study Design: We reviewed relevant literature and
professional guidelines to explore the validity of
these justifications for the standard use of the pelvic
exam in asymptomatic women.
Population Studied: English language literature.
Principal Findings: 1. Highly sensitive and specific
DNA amplification tests for Chlamydia and N.
gonorrhea now permit use of easy-to-obtain
specimens, including urine and self-administered
vaginal swabs. Most women prefer these
approaches over a pelvic exam. 2. Findings from a
pelvic exam do not affect the decision to prescribe
or withhold hormonal contraception; therefore,
relevant authorities now explicitly state that a pelvic
examination is not needed for initiation of these
contraceptives. 3. Recent American College of
Obstetrics and Gynecology guidelines recommend
no cervical cancer screening before age 21.
Cervical screening from age 21-30 can occur every
2 years, and then every 2-3 years until age 65 or
hysterectomy. These recommendations for less
frequent cervical cancer screening, along with use
of urine or swab-based STI testing, markedly
decrease the number of times that a clinician needs
to carry out a speculum exam to sample the cervix.
While the speculum exam is still the main approach
to obtain specimens for cervical cancer screening,
the new guidelines indicate that a woman might
need this only 15 times in her lifetime. 4. In the
United States, the speculum exam is routinely
accompanied by a bimanual pelvic exam for the
palpation of the uterus and ovaries, especially
intended to detect ovarian cancer. Clinical trial data,
however, show that bimanual exams do not lead to
earlier detection of ovarian cancer. In most
European countries asymptomatic women do not
undergo routine bimanual exams.
Conclusions: The pelvic exam is not needed to
screen for STIs, to initiate hormonal contraception,
nor does it assist in early detection of ovarian
cancer. The speculum exam is now needed far less
often for cervical cancer screening. No evidence
identifies benefits of a pelvic exam in the early
diagnosis of other conditions in the asymptomatic
woman.
Implications for Policy, Delivery or Practice:
Speculum and bimanual examinations are
uncomfortable, disliked by many women, and use
scarce time during a well woman visit. Eliminating
the pelvic exam from most of these visits will free
resources to provide other services of proven
benefit. Asymptomatic women may be able to seek
preventive services from a broader spectrum of
clinicians, including those who do not routinely
perform pelvic exams. Finally, such changes will
allow gynecologists more time for the diagnostic
evaluation of symptomatic women and for providing
specialist services.
Poster Number: 247
Gender Differences in Medication Therapy
Management Programs
Presented by: Donghui Wu, Ph.D., MEDai Inc.,
4901 Vineland Road, Suite 450, Orlando, FL 32811;
Phone: (321) 281-4561; Email: dwu@medai.com
Co-authors: Ognian Asproughov, Ph.D.; Anton
Berisha, M.D.
Research Objective: This study focuses the
gender differences of Medicare beneficiaries in
Medication Therapy Management Programs
(MTMP). We study the gender differences in
utilization patterns in drugs, occurences of
adversary drug interactions, and conditions/costs
progress over time.
Study Design: This is retrospective data study.
The study includes one year prescription history of
about 800,000 eligible Medicare members from
multiple states. The prescriptions are categorized
into 88 treatment classes, and adversary drug
interactions are defined based on treatment
classes. There severities of adversary drug
interactions based on potential harms/side effects
to members's health conditions.
Population Studied: About 800,000 eligible
Medicare members are selected from near two
million members from a nation Medicare sponsor.
39% of the 800,000 eligible members are male,
61% are female.
Principal Findings: We find major gender
differences in usage patterns in drug usage in terms
of treatment classes, adversary drug interactions,
costs, and further future prospects, i.e.
effectiveness of MTM programs.
Conclusions: It is very important to pay attention to
the gender differences in terms of drug usage
patterns, adversay drug interactions if MTM
program want truely improve quality of life Medicare
beneficiaries and reduce future costs.
Implications for Policy, Delivery or Practice: It is
very important for policy maker to put gender
specific and disease specific requirement into MTM
program requirement and guidelines for eligibility for
enrollmet in MTM program.
Funding Source(s): Company R&D
Poster Number: 248
Trends in the Incidence of IUD Insertion in the
United States: Data from a Population with
Employer-Sponsored Insurance
Presented by: Xin Xu, Ph.D., Prevention
Effectiveness Fellow, Economist, WHFB, DRH,
NCCDPHP, Centers for Disease Control and
Prevention, Atlanta, GA 30341; Phone: (770) 4886470; Email: iip5@cdc.gov
Co-authors: Maurizio Macaluso, M.D., Ph.D.; Lijing
Ouyang, Ph.D.; Scott Grosse, Ph.D.
Research Objective: The prevalence of
intrauterine devices (IUD) is low among women of
reproductive age in the United States. The objective
of this study was to examine the trend in IUD
insertions and related medical expenditures
between 2002 and 2007 in a population of women
covered by employer-sponsored health insurance
(ESI).
Study Design: We conducted a population-based
study using the MarketScan Commercial Claims
and Encounter Enrolled Population database. We
identified women, 15 to 49 years old, who filed a
claim for the insertion of an IUD or IUS (intrauterine
system, i.e., the Levonorgestrel-releasing IUD)
between Jan. 1, 2002 and Dec. 31, 2007. The
numerator for incidence calculations was the
number of patients having at least one insertion
reimbursement in that year, while the denominator
was the number of patients under ESI plans with
IUD/IUS coverage. We adopted the MarketScan
national weights to generate nationally
representative estimates.
Population Studied: Women of reproductive age
with employer-sponsored health insurance
coverage in the United States.
Principal Findings: The IUD/IUS insertion
incidence in the ESI-covered population increased
from 2/1,000 eligible women of reproductive age in
2002 to more than 8/1,000 in 2007, largely due to
the increase in IUS insertion incidence from
0.5/1,000 to 6.0/1,000 during the same time period.
The insertion incidence increased the most among
women aged 25 to 34 years, from 4/1,000 in 2002
to almost 16/1,000 in 2007. The IUD/IUS insertion
incidence was relatively high among states on the
West and East coasts. The average annual
increase in insertion incidence rates was more
pronounced in Southern states where the incidence
was low in 2002. Among the 10 states with the
largest absolute growth in IUD/IUS insertion, the
incidence increased on average from 3.1/1,000 in
2002 to 13.5/1,000 in 2007, while the incidence
among the 10 states with the smallest growth only
increased from 2.4/1,000 to 5.5/1,000. The annual
medical expenditures associated with IUD/IUS
insertions in this population increased from 26.7
million dollars in 2002 to 86.0 million dollars in
2007, after adjusting for inflation (both monetary
terms are in 2007 dollars).
Conclusions: The IUD insertion incidence in the
United States has increased substantially from
2002 to 2007, mainly driven by the remarkable
growth in the number of IUS insertions. Substantial
geographic disparity exists in insertion rates. The
increases in medical expenditures were also largely
driven by the growth in IUS insertions.
Implications for Policy, Delivery or Practice:
IUDs have lower contraceptive failure rates than
any other reversible contraceptive method, and
higher rates of IUD use should be associated with
fewer unwanted pregnancies. Additional research is
needed to understand whether the recent growth in
IUS insertions and the state disparity in IUD/IUS
insertion incidence are related to differences in
provider attitudes and local policies.
Poster Number: 249
A Comparison of Intrauterine Device Users and
Non-Intrauterine Device Contraceptive Users in a
Nationally Representative Survey
Presented by: Xin Xu, Ph.D., Prevention
Effectiveness Fellow, Economist, Women's Health
and Fertility Branch (WHFB), Centers for Disease
Control and Prevention; Email: xinxu@cdc.gov
Co-authors: Maurizio Macaluso, M.D., Dr.PH.;
Jennifer Frost, Ph.D.; John Anderson, Ph.D.;
Kathryn Curtis, Ph.D.; Scott Grosse, Ph.D.
Research Objective: Little is known about the
characteristics of intrauterine contraceptive device
(IUD) users in the United States. The objective of
this study was to describe demographic and
reproductive characteristics, family background,
and demand for reproductive health services
among IUD users, and their reasons for choosing
IUDs.
Study Design: We adopted women aged 15-44
years old who responded to the 2002 National
Survey of Family Growth (NSFG) and used any
reversible contraceptive methods during 01/199903/2003. We divided them into five groups
according to a hierarchy of methods: IUD, longacting hormonal, pills and patch, barrier, and
others. We compared users in these categories with
respect to self-reported demographic and
reproductive characteristics; family background;
and demand for reproductive services. Logistic
regression was used to evaluate multiple predictors
of IUD use. Analysis took into account the complex
sampling scheme of the NSFG. We employed data
from a telephone survey conducted by the
Guttmacher Institute in 2004 to investigate reasons
for choosing IUDs. The Guttmacher survey consists
of 1,978 randomly selected women aged 18-44 who
were currently at risk of unintended pregnancy. The
Guttmacher samples were also weighted to reflect
the U.S. female population.
Population Studied: Women of reproductive age
in the United States.
Principal Findings: Compared to women using
other reversible contraception, IUD users were
more likely to be 25-34 years old (odds ratio (OR
2.0)), to be foreign born (OR 2.1), to have lived in a
rented property (2.0), and to have never worked in
their life time (OR 2.0), but less likely to be never
married (OR 0.4), mot to have had sexual
intercourse every month during the previous 12
months prior to the interview (OR 0.3), and to have
had no live birth (OR 0.2). IUD users were more
likely to have had a mother with at least a high
school education (OR 2.6) and less likely to have
had teenage mothers (OR 0.6). Each of these
variables was statistically significant (95%) in the
multiple logistic regression analysis. IUD users
were also more likely to have had a medical checkup or test related to a birth control method (51.1%
vs 33.3%) and more likely to have received
counseling about birth control during the 12 months
prior to the interview (33.5% vs 26.5%). The three
most frequently reported reasons for choosing IUDs
were the method’s ability to prevent pregnancy
(86.4%), the method’s ability to not interrupt sex
(70.5%), and the provider’s recommendation to use
it (60%). IUD users also reported the highest level
of satisfaction among all users of reversible
contraceptive methods (82.3% vs 61.4%).
Conclusions: IUD users differ markedly from users
of other reversible contraception. The IUD was well
perceived among most users.
Implications for Policy, Delivery or Practice:
IUDs have lower contraceptive failure rates than
other reversible contraceptive method, and higher
rates of IUD use should lead to fewer unwanted
pregnancies. Women with the characteristics
described above may be more receptive to
promotional messages about IUD use. Health care
providers’ perceptions about the IUD could also be
crucial for the success of IUD promotion.
Poster Number: 250
A Mixed Methods Study Exploring the Value of an
Interprofessional Patient Safety Education Course
to Nursing Students
Presented by: Amy Abbott, Ph.D., R.N., Assistant
Professor, School of Nursing; Center for Health
Services Research and Patient Safety, Creighton
University, 2500 California Plaza Office 193A,
Omaha, NE 68178; Phone: (402) 280-2055; Email:
aabbott@creighton.edu
Co-authors: Kevin Fuji, Pharm.D.; Karen Paschal,
P.T., D.P.T., M.S.; Kimberly Galt, Pharm.D.; Ph.D.
Research Objective: Equipping nurses with
knowledge, skills, and attitudes to improve patient
safety is critical for reducing harm and injury due to
errors in patient care. Educators must prepare
students with foundational knowledge to keep
patients safe in challenging and ever-changing
healthcare environments. The purpose of this pilot
study was to understand nursing students’ attitudes
about the value of interprofessional patient safety
education to their professional development and its
value and role in health professions’ curricula.
Study Design: An exploratory mixed methods
design was used. Quantitative data was analyzed
using information obtained from all students’ course
evaluation questionnaire, final examination
performance, final case study performance, and
classroom assessment techniques. Critical case
sampling was used to identify interview participants.
Interviews were guided by three key ideas: 1) the
role of the nursing profession, the individual nurse,
and the nursing student in the context of patient
safety; 2) how skills and attitudes learned in the
course will change behavior as a student and future
practitioner; and 3) recommendations for
incorporation of patient safety content into nursing
curriculum. Interviews were transcribed, checked
for accuracy, and in vivo coded by each
investigator. Themes were developed from the
interviews using the quantitative data as a
contextual framework for the qualitative findings
and nursing student performance was compared
with the entire class to distinguish if learning
differed significantly from the other professions.
Population Studied: 27 students enrolled in an
interprofessional patient safety course participated
in the study (7 nursing, 19 pharmacy, and 1 law).
Quantitative data were collected from all students.
Six nursing students were interviewed.
Principal Findings: Students’ course average was
95.2% with no differences noted between nursing
students and the rest of the class. Students
averaged 88.7% on the final case study, designed
to assess students’ abilities to comprehensively
apply knowledge learned to a specific situation
involving multiple professions. The final
examination, designed to assess overall knowledge
and recall of core patient safety material, had an
average of 89.2%. Students estimated 80% of the
material was new knowledge and not taught in
other courses. All students felt the course material
was essential or enriching for all health professions.
The qualitative themes that emerged were
awareness, ownership, and action. The awareness
theme described students becoming aware of the
science of patient safety and what it meant in their
role as a professional. Ownership described role
identification of both self and others in patient
safety. Action described students spreading the
word, advocating for patients, and ensuring safe
patient care delivery. Those involved in patient care
were able to describe application of the course
content in their work environment.
Conclusions: Nursing students perceive high value
in using an interprofessional format to learn about
patient safety. Students perceive the science of
patient safety as foundational and core knowledge
needed by all health professionals.
Implications for Policy, Delivery or Practice:
Findings will inform educators about the meaning of
this interprofessional patient safety education
experience to professional nursing student
development. A values-oriented patient safety
education framework is responsive to nursing
students and may enrich opportunities for nursing
student education.
Funding Source(s): Creighton University Office of
Academic Excellence and Assessment
Poster Number: 252
Implementation of an Interdisciplinary Coordination
Intervention
Presented by: Justin Benzer, Ph.D., Post-doctoral
Fellow, Center for Organization, Leadership, and
Management Research, Department of Veterans
Affairs, 150 S. Huntington Avenue, Boston, MA
02130; Phone: (857) 364-5691; Email:
justin.benzer@va.gov
Co-authors: James Burgess, Ph.D.; Irene Cramer,
Ph.D.; David Mohr, Ph.D.; Jenny Sullivan, Ph.D.
Research Objective: The report of the President’s
New Freedom Commission on Mental Health
eM.P.H.asizes that mental health and physical
health problems are interrelated components of
overall health and are best treated in a coordinated
care system. The purpose of this study was to
determine the facilitators and barriers to
implementing a team-based, interdisciplinary model
of primary care-mental health integration (PC/MHI)
using a transformational change model
(VanDeusen Lukas et al., 2007).
Study Design: Semi-structured interviews were
conducted with a PC and a MH leader at each of
the 17 sites. Additional interviews were conducted
with 37 front line clinical staff, with at least one
mental health and one primary care person
interviewed at each site. Interviewees were asked
to describe how the process of coordinating primary
care and mental health care had changed since
they joined the clinic, and how processes could be
further improved. Interview transcripts were coded
with respect to the transformational change model
elements of leadership (commitment to change),
improvement initiatives (staff involvement),
alignment (goals and resources across
organizational levels), and integration (bridging
intra-organizational boundaries).
Population Studied: Eight medical facilities and
nine large outpatient clinics in one geographic
region within the Veterans Health Administration
were included in the study. In response to a
national mandate to provide integrative mental
health care, local leaders had implemented a model
whereby mental health workers were co-located
with primary care.
Principal Findings: Interviews revealed
differences between sites in both the degree of colocation, the degree of collaboration between
primary care and mental health providers, and the
structures, processes, and professions used to
bridge the boundary between primary care and
mental health. Leadership served as a facilitator to
the implementation of PC/MHI through planning,
procuring necessary resources, and encouraging
communication between clinics. Most sites did not
describe active involvement of frontline staff in the
development of improvement initiatives, but where
present, these efforts were seen as critically
important in developing effective procedures.
Integration problems included failure of PC
providers to utilize PC/MHI services and shifting
responsibility for patient care between PC and MH
providers. Some sites were found to have good
alignment, as evidenced by adequate training and
resources for PC/MHI implementation. Resource
constraints were commonly cited as barriers, but in
several sites resource availability was not observed
to affect quality of integration, as providers were
able to develop innovative solutions to
implementation challenges in resource constrained
facilities.
Conclusions: Local barriers and facilitators
influenced the implementation of the PC/MHI
model. This study supports the propositions of the
transformational change model, and in particular
reinforces the importance of leadership in the
implementation of integrative, interdisciplinary
interventions. Leadership was cited as a barrier to
change or as a facilitator that was reported to
overcome barriers of alignment or integration.
Implications for Policy, Delivery or Practice:
This study indicates that leadership is important for
bridging boundaries between professions, for
ensuring resource support, and for promoting staff
involvement in improvement initiatives. Planned
innovations, particularly those that bridge
organizational boundaries require leadership
support, and clinical staff should be actively
involved in the discussion and revision of innovation
procedures to fit clinic needs.
Funding Source(s): VA
Poster Number: 253
Teaching Service Census Caps and Unit-Based
Admission of Patients Decreases Resident
Workload, Improves Duty Hour Compliance, and
Maintains Patient Safety
Presented by: Rosa Cabanela, Ph.D., MS Health
Services Analyst, Health Care Policy & Research,
Mayo Clinic, Pavilion 3, Rochester, MN 55901;
Phone: (507) 538-0220; Email:
cabanela.rosa@mayo.edu
Lead Author: Darcy Reed, M.D.
Co-authors: James Naessens, Sc.D., M.P.H.;
Rosa Cabanela, Ph.D.; Andrew Halvorsen, M.S.;
Uma Thanarajasingam, M.D., Ph.D.; Furman
McDonald, M.D.
Research Objective: To reduce workload and
enhance education, we implemented a census cap
of 14 patients on the six Internal Medicine hospital
teaching services at Mayo Clinic. In addition, we
developed a unit-based admissions process in
which patients and their multidisciplinary care team
were geographically consolidated on a single
hospital unit. This study sought to examine the
impact of these two interventions on residents’
workload, educational conference attendance, duty
hour compliance, and patient safety.
Study Design: Difference in differences design
comparing internal medicine teaching services with
other general medicine hospital services for 12
month periods before and after the interventions.
Residents reported workload and duty hours via
monthly survey. Conference attendance was
recorded using a card-swipe system. Patient safety
variables including mortality, adverse events (e.g.
medication errors, falls), and AHRQ Patient Safety
Indicators (PSI) were determined for all patients
before and after the interventions. To control for
hospital-wide temporal trends, patient safety
variables were also collected for all patients
hospitalized on internal medicine non-teaching
services during the study period. Variables were
compared before and after the interventions using
Fisher’s exact test, t-tests, and multivariate logistic
regression. Significance was set at 0.01 to account
for multiple comparisons.
Population Studied: All 10,541 patients and 280
residents on all six Internal Medicine teaching
services one year before and one year after the
interventions were included.
Principal Findings: After implementation of the
interventions, residents’ perception of the
appropriateness of their workload improved from a
mean (SD) of 3.17 (1.09) on a 5-point scale to 3.89
(0.67), p< .0001. Residents’ ability to obtain 10
hours rest between shifts improved from 361.9 to
364.5 violation-free days/year (p< .0001), and their
ability to work no more than 24 consecutive hours
with 6 hours for education and patient hand-off
improved from 361.2 to 364.0 violation-free
days/year (p<.0001). Additionally, residents’
conference attendance increased by 12%, p<
.0001. In logistic regression analyses adjusted for
expected mortality, there were no significant
differences before compared to after the
interventions for resident versus control services in
the percentage change in mortality rate (0.90%
change vs. -0.18% change, p=.22), adverse events
(-0.48% change vs. 1.06% change, p=.03), or
Patient Safety Indicators (-1.64% change vs. 1.48% change, p=.24).
Conclusions: The implementation of teaching
service census caps and unit-based admissions
decreases resident’s workload, improves duty hour
compliance, increases conference attendance, and
does not adversely affect patient safety.
Implications for Policy, Delivery or Practice: In
the wake of broad-based policy changes reducing
residents’ work hours, systems-based solutions to
reduce work hours and workload without
compromising patient safety are needed.
Implementation of service census caps and unitbased admissions processes have beneficial effects
for resident learning and no deleterious impact on
patient quality or safety, but careful planning of
such changes is necessary. Further studies are
needed to examine the feasibility of these
interventions among different specialties and
patient populations.
Poster Number: 254
Psychological Ownership and Organizational
Change: Engaging Health Care Providers in New
Program Implementation
Presented by: Vivian Chan, M.P.P.,
Interdisciplinary Studies Graduate Program,
University of British Columbia, 953 West 16th
Avenue, Vancouver, Canada ; Phone: (604) 7614919; Email: vwychan@interchange.ubc.ca
Co-authors: Smillie, Kristen, M.A.; Kazanjian,
Arminee, Dr.Soc.
Research Objective: This paper explores the
theoretical concept of psychological ownership and
its potential effects during an organizational change
environment. Case studies from Canada will be
presented and discussed to elaborate recent
literature on the theoretical concepts of
psychological ownership and territoriality.
Study Design: The authors conducted secondary
analysis on data from a larger research program on
knowledge exchange in cancer survivorship
programming in diverse settings across Canada.
The research team utilized “thick description” to
organize the data, an ethnographic approach used
to elucidate an understanding of a specific event or
behavior within the social context that it occurs. The
results were compiled into experiential accounts
and presented in a series of reports to practitioners,
policy makers and academics at local and national
levels.
Population Studied: Clinical program team
members from six different cancer survivorship
initiatives during the implementation and planning
phases of the respective initiatives
Principal Findings: Our literature review indicated
that the inherent need for ownership and sense of
territoriality is pervasive in an organizational setting.
It seems natural that people instinctively defend that
which they see as theirs (Brown & Robinson, 2007).
Psychological ownership is the feeling of
possessiveness and of being psychologically tied to
the object (Pierce et al., 2001: 299). The object of
possession can be physical (e.g. a thing, a person,
a place) or social (e.g. an idea, a group, a work
role) in nature. Territorial behaviour such as
behaviours for constructing, communicating,
maintaining, and restoring territories around those
objects in the organization toward which one feels
proprietary attachment were evident when
examining the experiential accounts from the
cancer survivorship clinical intervention programs.
Paré, Sicotte, and Jacques’s (2006) posit that
feelings of psychological ownership towards the
clinical information system may be developed
through active physician involvement during the
implementation of the clinical system. They were
able to show that physician’s attitudes, as
conceptualized as psychological ownership toward
the clinical information system, are related to the
subsequent usage of the information system.
However, Brown and Robinson (2007) cautioned
that the feeling of ownership and territoriality might
not always lead to positive organizational
behaviour. Resistance to change as a result of
strong proprietary attachment to existing objects is
one of the dysfunctions of psychological ownership
and territoriality.
Conclusions: We were able to highlight how
clinical program team members exhibited positive
and negative aspects of psychological ownership
and territorial behaviour during the planning and
implementation phases of selected cancer
survivorship initiatives. We subsequently reflect on
how positive and negative aspects of territoriality
and psychological ownership influenced the
implementation of these programs.
Implications for Policy, Delivery or Practice: If
the need to possess and a sense of territoriality are
prevalent, how are we to identify these behaviours
when developing change management strategies?
What are the routes to positive aspects of
psychological ownership to support health care
providers through organizational and practice
change? How do we avoid or reduce negative
aspects from creating a resistant organizational
culture? Answers to these questions will lead to a
better understanding of successful organizational
change strategies and consequently, more effective
initiatives. Sense of territoriality and ownership are
universal human attributes which may be expressed
and enacted in a variety of ways within systemspecific conditions. Examining aspects of the
Canadian health system that yields positive results
in the process of change management would be
instructive to other health care systems and in
different care settings.
Funding Source(s): Canadian Institutes for Health
Research
Poster Number: 255
Perceived Facilitators and Barriers to Improving
Internal Medicine Residency Curricula
Presented by: Kristina Cordasco, M.D., M.P.H.,
M.S.H.S., Associate Scientist, Adjunct, Health, The
RAND Corporation, 1776 Main Street, P.O. Box
2138, Santa Monica, CA 90407-2138; Phone: (310)
393-0411 ext. 6364; Email: kcordasc@rand.org
Co-authors: Cristina Boccuti, M.P.P.; Craig Lisk,
M.S.; Nicole Lurie, M.D., M.S.P.H.; Marianna HortaCapelli, B.A.; Barbara Wynn, M.A.
Research Objective: Over the past 50 years, the
delivery of medical care has changed dramatically
with advances in technology, population aging and
diversification, and rising prevalences of chronic
diseases. Training in internal medicine, however,
has changed little. There have been multiple calls
for reforms in residency training to better prepare
our physician workforce in the knowledge and skills
needed. In response to these calls, accrediting
organizations of allopathic and osteopathic
residency programs have begun incorporating new
topics into required curricula. We conducted an
exploratory study for the Medicare Payment
Advisory Commission (MedPAC) of how internal
medicine (IM) residency programs are training their
residents in the competencies of systems-based
practice, practice-based learning and improvement,
and communication. As part of this study, we asked
IM residency program directors to comment on
facilitators and barriers to improving their curricula
in these competencies. Responses regarding
facilitators and barriers is the the focus of this
paper.
Study Design: Using semi-structured interviewing
techniques, we asked each program director to
describe their program’s approach to training in
topics in each of the three competencies, and then
asked them to describe the facilitators and barriers
they have experienced in improving the curricula in
that competency. At the conclusion of the interview,
we asked about overall facilitators and barriers for
curricular improvement. Each interview was audiorecorded. Using detailed notes taken from
reviewing each audio-recording, and content
analysis with techniques from grounded theory, we
identified and then described the common themes
that emerged from comments on facilitators and
barriers.
Population Studied: We invited a randomlyselected 59 program directors of allopathic and
osteopathic programs to participate in one-hour
interviews, 26 (44%) of whom agreed to and
completed the interview between November 25 and
December 29, 2008. No incentives were offered.
Principal Findings: Eight themes emerged from
directors’ comments on facilitators and barriers to
improving curricula in these competencies: (1)
presence or lack of information technology
resources; (2) presence or lack of faculty expertise
in these topics; (3) presence or lack of faculty time
protected for curricula development and
implementation; (4) characteristics and resources of
the program’s setting; (5) presence or lack of
institutional support for training in these topics; (6)
competing priorities for resident time; (7) residents’
baseline knowledge and interest in these topics;
and (8) scarcity of research in educational and
evaluation strategies for these topics.
Conclusions: Graduate medical education funding
policies, accreditation standards, certification exam
topics, undergraduate medical education and
investment in research of educational and
evaluation strategies act as facilitators and barriers
to curricula innovation in residency programs.
Implications for Policy, Delivery or Practice:
Examining and potentially intervening in these
factors, so that incentives are aligned with desired
changes, could have a significant positive impact on
how well IM programs are preparing our nation’s
physician workforce to care for our 21st century
population.
Funding Source(s): The Medicare Payment
Advisory Commission (MedPAC)
Poster Number: 256
Effect of Increased Insurance Coverage under
Health Care Reform on the Demand for Primary
Care and Specialist Physicians
Presented by: Tim Dall, M.S., Managing Director,
The Lewin Group, 3130 Fairview Park Drive, Suite
800, Falls Church, VA 22042; Phone: (703) 2695743; Email: tim.dall@lewin.com
Co-authors: Paul Hogan, M.S.; Yaozhu Chen,
M.P.A.
Research Objective: Estimate the implications of
increased health insurance coverage through health
care reform on the demand for physicians and the
projected future excess demand.
Study Design: Controlling for demographics (age,
sex, race/ethnicity), healthcare utilization patterns of
the insured are compared to utilization patterns of
the uninsured using 2007 data from the National
Ambulatory Medical Care Survey, National Hospital
Ambulatory Medical Care Survey, Nationwide
Inpatient Sample, and National Health Interview
Survey. Differences in health status and prevalence
of chronic diseases for the insured and uninsured
are analyzed using the National Health and
Nutrition Examination Survey and National Health
Interview Survey. Based on differences in utilization
patterns between the insured and other similar
uninsured, an estimate of the increase in utilization,
and the change in resources utilized are estimate
from the data. Using data from CDC's Behavioral
Risk Factor Surveillance System, we estimated
rates of uninsured by demographic group for each
state. These inputs are entered into the demand
component of a Physician Supply and Demand
Model. Estimates of the effects on current and
future demand, and the implications for the excess
demand (shortage) of physicians are examined
under multiple scenarios related to increase
insurance coverage. Scenarios include 100%
coverage, as well as expanding coverage to include
currently uninsured U.S. citizens (reaching
approximately 96% coverage).
Population Studied: The U.S. resident population.
Principal Findings: Likely effects of proposed
expanded medical coverage under health care
reform on utilization will result in about a 4-7%
increase in demand for physicians. The increase,
however, will vary by physician specialty and by
state. Because most of the over 65 population is
already covered under Medicare, and most children
are covered through private insurance, Medicaid or
CHIP, there is not likely to be a large increase in
utilization from children and the elderly by
expanding medical coverage. Among adults under
age 65, there is a mix of healthy uninsured that
likely will see little increase in healthcare utilization
and uninsured with chronic conditions such as
diabetes and cardiovascular disease that will likely
see substantial increases in utilization. Preliminary
findings will be updated to incorporate the latest
health care reform legislation on expanded
insurance coverage.
Conclusions: Health care reform legislation to
expand medical coverage will create a modest
increase in demand for physicain services at a time
when there are already concerns about an
inadequate supply at the national and local level.
Implications for Policy, Delivery or Practice: The
findings can serve to improve health workforce
policy and planning. The overall increase in
demand, and the variation in this increase across
areas of specialization, may have policy
implications for undergraduate and graduate
medical education resources and policy planning.
Poster Number: 257
Results from the First and Second Cohorts of the
VA Nursing Academy: Using VA-Nursing School
Partnerships to Reduce the Clinical Nursing Faculty
Shortage
Presented by: Aram Dobalian, Ph.D., J.D.,
Research Health Scientist, COE Healthcare
Provider Behavior, Department of Veterans Affairs,
16111 Plummer Street (152), Sepulveda, CA
91343; Phone: (818) 891-7711 ext. 7182; Email:
aram.dobalian@va.gov
Co-authors: Candice Bowman, Ph.D., R.N.; Tamar
Wyte, D.P.T., M.P.H.; Darya Friedman, M.P.H.;
Barbara Simon, M.A.; Jack Needleman, Ph.D.
Research Objective: According to HRSA, the U.S.
is experiencing a moderate shortage of registered
nurses (RNs). About 22,000 RNs will be eligible for
retirement within the Department of Veterans Affairs
(VA) this year. It has been suggested that much of
this shortfall is due to a lack of qualified nursing
faculty. Consequently, VA established the VA
Nursing Academy, a 5-year pilot program that funds
faculty positions at 15 competitively selected
partnerships of VA facilities and nursing schools.
The Academy is designed to expand nursing
faculty, enhance professional development,
increase student enrollment, promote innovations in
education and practice, and increase recruitment
and retention of nurses within VA. We are in year
three of five of our national evaluation of the
program.
Study Design: We conducted over 120 semistructured interviews (45-120 minutes) that
assessed the partnerships’ structures and the
consequences of their activities during the first ten
partnerships’ initial year.
Population Studied: Nursing leadership, local
partnership project directors, nursing faculty, nurse
managers, nursing students, and nursing staff.
Principal Findings: Most partnerships completed
faculty hiring by the end of the first year. Teaching
responsibilities were usually limited to clinical
instruction. A few Academy faculty were offered the
opportunity to teach an entire didactic class, most
often in skills or simulation labs. Faculty workload
was generally believed to be greater than what was
expected, although most stated that it was within
reasonable limits. Most partnerships met their
increased student enrollment quota for the year.
Initially, students were sometimes resistant to the
idea of having clinical rotations at the VA. However,
after exposure to the VA, most students were much
more positive about the experience and
enthusiastic about going back to the VA because, in
comparison with non-VA facilities, students were
allowed to participate more in hands-on patient
care, nursing staff were friendlier and more helpful,
and veteran patients liked receiving care from them.
Creatively selecting areas used for clinical
placements, such as perioperative areas,
rehabilitation units, nursing homes, and outpatient
clinics, was a source of innovation. We identified
critical themes that influenced “Launch-Year
achievement,” i.e. what was feasibly attainable after
one year of a partnership’s implementation. Key
domains identified that potentially influence
performance include local market conditions,
leadership involvement and commitment, scope of
faculty roles and development, intra-organizational
facilitation, communication, power sharing, problem
solving mechanisms, pre-existing relationship
between partners, presence of champions, stability
of partnership personnel, and perceived benefits of
the Academy.
Conclusions: Of these key domains, our
observations across partnerships have led us to
eM.P.H.asize inter-organizational collaboration
(e.g., teamwork between direct and indirect
participants across partnering institutions regarding
specific issues, strategies, or decisions) as a critical
factor in enabling these partnerships to be
successful. In designing and implementing future
partnerships, organizations should consider each of
these domains that influenced Academy partnership
performance during the Launch Year.
Implications for Policy, Delivery or Practice: The
Academy may serve as a model for how best to
design and structure partnership relationships
between two distinct institutions, with intrinsically
divergent needs.
Funding Source(s): VA
Poster Number: 258
Persistent Primary Care Health Professional
Shortage Areas (HPSAs) and Health Care Access
in Rural America
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, WWAMI Rural Health Research Center,
Family Medicine, University of Washington, 4311
Eleventh Avenue NE, Suite 210, Seattle, WA
98105; Phone: (206) 616-9207; Email:
mdoesche@u.washington.edu
Co-authors: Meredith Fordyce, Ph.D.; Susan
Skillman, M.S.; J. Elizabeth Jackson, M.A.; Roger
Rosenblatt, M.D., M.P.H.
Research Objective: Primary care Health
Professional Service Areas (HPSAs) help states
and communities increase their primary care supply
through eligibility for loan repayment, Federally
Qualified Health Center and Rural Health Clinic
designation, and other mechanisms. This study
examined whether persistence of primary care
HPSA designation in rural counties was associated
with lower population socioeconomic status (SES)
and deficiencies in access to health care services,
including provider supply.
Study Design: Cross-sectional analysis of U.S.
rural county characteristics by their primary care
HPSA designation, classifying counties as: never
HPSAs; intermittent partial-county HPSAs;
persistent partial-county HPSAs; intermittent wholecounty HPSAs; and persistent whole-county
HPSAs.
Population Studied: The primary care HPSA
categories were used to assess: per capita supply
of clinically active family physicians, general
internists, general pediatricians and obstetriciangynecologists using the 2005 American Medical
Association and American Osteopathic Association
Masterfiles; socioeconomic-demographic
characteristics using U.S. Census Bureau data; and
access to care using data from the 2004/5
Behavioral Risk Factor Surveillance System.
Principal Findings: Distribution: Of 2,051 U.S.
rural counties in 2005, there were 308 (15% of all
rural counties) “never” HPSAs, 370 (18.0%)
“intermittent partial-county” HPSAs, 490 (24%)
“persistent partial-county” HPSAs; 195 (105)
“intermittent whole-county” HPSAs; and 688 (33%)
“persistent whole-county” HPSAs.
Sociodemographics: A gradient occurred in which
SES was lowest in persistent whole-county HPSAs,
while counties never receiving a HPSA designation
had the highest SES status (e,g, persistent poverty
characterized 28.1% of persistent whole county
HPSAs vs. 5.8% of never HPSA counties). A
racial/ethnic gradient was observed in which
minorities made up the highest percentage of the
population in persistent whole-county HPSAs, while
the fewest minority group members lived in the
never HPSAs (e.g., minorities characterized 21.6%
of persistent whole county HPSAs vs. 11.4% of
never HPSA counties). Access: A gradient of
decreased access occurred in which access was
poorest in persistent whole-county HPSAs for:
primary care physician supply (39.3 per 100,000
population of persistent whole county HPSAs
vs.72.3 per 100,000 population in never HPSAs);
the percentage of adults lacking a regular primary
care provider (22% in persistent whole county
HPSAs vs. 17% in never HPSAs); the percentage
lacking health insurance coverage (21% in
persistent whole county HPSAs vs. 16% in never
HPSAs); and the percentage foregoing needed
health care because of its cost (18% in persistent
whole county HPSAs vs. 14% in never HPSAs).
Conclusions: This five-level classification of rural
counties measuring partial- vs. whole-county
persistence of primary care HPSA designation
stratifies rural populations by SES, race/ethnicity,
primary care supply, health insurance uptake and
access to needed health care services. Those U.S.
rural counties that were persistently designated as
whole-county HPSAs had much lower SES and
adults residing in these counties reported
substantial financial obstacles to obtaining needed
health care services. Rural counties persistently
designated as whole-county HPSAs faced severe
provider shortages and adults residing in these
locations were less likely to have a regular primary
care provider.
Implications for Policy, Delivery or Practice: The
ability to identify persistence and extent of HPSA
designation may be a valuable tool in selecting
counties with higher levels of need.
Funding Source(s): HRSA
Poster Number: 259
Dentist Supply, Access to Dental Care, and Oral
Health among Rural and Urban Residents: A
National Study
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, WWAMI Rural Health Research Center,
Family Medicine, University of Washington, 4311
Eleventh Avenue NE, Suite 210, Seattle, WA US
98105; Phone: (206) 616-9207; Email:
mdoesche@u.washington.edu
Co-authors: Gina Keppel, M.P.H.; Susan Skillman,
M.S.
Research Objective: Dentists in rural locations are
in short supply compared to urban locations.
However, whether low rural supply adversely
influences dental health care access and oral health
status is not well understood. To address this gap in
knowledge, we examined relationships between
dentist supply and dental access, preventive care,
and oral health among U.S. rural and urban adults.
Study Design: We conducted a cross-sectional
assessment of data from the 2006 Behavioral Risk
Factor Surveillance System of the Centers for
Disease Control and Prevention, a nationally
representative survey. These data were used to
identify dental access (visits to a dentist or dental
clinic in the past year), preventive care (teeth
cleaning in the past year), and oral health
(permanent tooth loss). Data from the 2008
American Dental Association Dentist Masterfile
were used to examine the influence of per capita
dentist supply on these measures.
Population Studied: A representative sample of
U.S. civilian, non-institutionalized adults aged 18 or
older was examined (n=355,720). Rural residence
was grouped as metropolitan (urban) or nonmetropolitan (rural) county of residence based on
the county-based Office of Management and
Budget taxonomy, and smaller/more isolated rural
locations were classified using the 2003 Urban
Influence Code groupings of the Economic
Research Service of the U.S. Department of
Agriculture.
Principal Findings: After adjustment for sociodemographic and other factors, rural residents were
significantly less likely than urban residents to
report having a dental visit (Adjusted Odds Ratio
[AOR] 0.87, 95% Confidence Interval [CI] 0.84 to
0.91) or teeth cleaning (AOR 0.83, CI 0.81 to 0.87)
in the past year, and they were significantly more
likely to report having had any permanent teeth
extracted (AOR 1.19, CI 1.14 to 1.23). As dentist
supply increased, the likelihood of having one or
more dental visits and teeth cleaning in the past
year increased, and the likelihood of undergoing
permanent tooth extraction due to dental decay or
disease decreased in both urban and rural locations
(p<0.001 for each model). Rural adults residing in
counties having the lowest quintile of dentist supply
were less likely than their counterparts living in the
highest quintile counties to report having a dental
visit (61% vs. 69%, respectively) and teeth cleaning
(59% vs. 68%), and they were more likely to report
having had any permanent teeth extracted (57% vs.
46%).
Conclusions: U.S. rural adults utilize fewer dental
services and have more permanent tooth loss, even
after adjusting for differences in socio-demographic
characteristics. However, in rural and urban
locations increases in dentist supply were
associated with improvements on these access and
outcome measures.
Implications for Policy, Delivery or Practice:
Dentists are an essential component of rural health
care and higher supply is related to improved dental
care and better oral health. Yet rural communities
struggle to recruit and retain sufficient numbers of
dentists to meet their oral health care needs.
Policies to increase rural dentist supply may lead to
improvements in dental access, preventive care
and oral health outcomes of US rural adults.
Funding Source(s): HRSA
Poster Number: 260
Enhancing the Delivery of Preventive Care at
Primary Care Practices in Underserved
Communities Using Shared Outreach Workers
Presented by: Carlene Duncan, M.P.H., Program
Manager, PCIP, NYC Department of Health and
Mental Hygiene, 161 William Street, 5th Floor, New
York, NY 10038; Phone: (347) 882-0186; Email:
cduncan1@health.nyc.gov
Lead Author: Carlene Duncan, M.P.H.
Co-authors: Daniel Halevy, M.D.; Amanda Parson,
M.D., M.B.A.; Kaleena Colon
Research Objective: 1) Examining the impact of
outreach workers as a shared resource among
electronic health record-enabled small practices on
the quality of care of patients and encouraging
patients to see their primary care physician (PCP)
on a regular basis to get recommended care. 2)
Examining the cost effectiveness of this outreach
strategy. This will be measured by analyzing
Revenue Value Units (RVU) through the number of
patient visits at the practices.
Study Design: The role of a prevention outreach
specialist (POS) was defined as a layperson without
specific healthcare certification, who received
training in the use of the electronic health record in
order to identify patients with the targeted chronic
conditions. One POS was recruited and was
embedded at a single primary care practice. The
first part of the pilot involved the identification of
patients with asthma during the peak asthma
season, and the generation of appointment
reminder letters. The second part of the pilot
involves the recruitment of four additional
specialists, each working with four practices. The
specialists will identify patients with all of the
targeted at-risk conditions and will coordinate with
the practices to prioritize and conduct outreach
activities, through phone calls and letters to patients
recommending that they make appointments, fill
prescriptions, and receive education about their lab
test results.
Population Studied: Primary care practices in New
York City’s underserved communities and their
patients who are identified as being at-risk for
chronic disease. The at-risk population consists of
patients with cardiovascular disease, hypertension,
hyperlipidemia, asthma, smoking, diabetes and
obesity. All practices are enrolled in the Primary
Care Information Project, in which they use an EHR
with an integrated registry tool, making it possible to
use the practices' own clinical data to systematically
identify patients with gaps in care.
Principal Findings: Of the 567 patients with the
targeted at-risk chronic conditions, over 60%
returned for recommended care during the
intervention period. The return rate varied
depending on the disease area, from 40% for
patients requiring improved blood pressure and
cholesterol control, up to 60% for patients requiring
an assessment for antithrombic therapy. Fifty five
percent of patients responded to the intervention by
returning for a diabetes management visit. Of the
199 asthma patients identified during the peak
asthma allergy season and were sent a reminder
letter, over 50% patients returned for recommended
care within eight weeks.Over a course of 6 months,
567 unique patients were enrolled and received the
intervention, leading to 475 phone calls and 199
letters.
Conclusions: This pilot demonstrated the
feasibility of embedding a POS in a primary care
practices located in underserved communities in
order to ensure the delivery of timely and
appropriate care. We were able to measure patient
follow up to the intervention and plan to compare
these rates with historical and contemporary
controls.
Implications for Policy, Delivery or Practice: The
goal of this project is to demonstrate the value, cost
savings and quality improvement of a shared
outreach worker. If such an approach is successful,
then the concept may be broadly embraced
throughout New York State and throughout the
country. We believe that the elements of this project
are highly replicable, particularly as it has been
proven among independent, highly under-resourced
practices.
Funding Source(s): Pfizer
Poster Number: 261
Where will North Carolinians with Cancer Find
Treatment? An Assessment of the Oncology
Workforce
Presented by: Danielle Durham, M.P.H.,
Research Associate, Research, The Carolinas
Center for Medical Excellence, 100 Regency Forest
Drive, Suite 200, Cary, NC 27518-8598; Phone:
(919) 380-9860; Email:
ddurham@thecarolinascenter.org
Lead Author: William Carpenter, M.H.A., Ph.D.
Co-authors: Franziska Rokoske, P.T., M.S.; Karen
Knight; Anna Schenck, M.S.P.H., Ph.D.
Research Objective: The purpose of this research
was to examine cancer treatments at the county
level to identify which counties face a potential
shortage of oncology physicians given their cancer
burden. The paper examines the burden of cancer
based on North Carolina cancer cases diagnosed
from 2003-2007, focusing on care provided for the
top ten cancers: Female breast, Colon and Rectum,
Kidney and Renal pelvis, Lung and Bronchus,
Melanomas of the skin, Non-Hodgkin LyM.P.H.oma,
Ovarian, Prostate, Urinary bladder and Uterine.
Study Design: A cohort study design was used to
identify cancer patients and their treatments. The
North Carolina Central Cancer Registry (NCCCR)
provided information for cancer cases diagnosed
from 2003 to 2007. Cases were linked with
Medicare data to identify Medicare enrollees.
Medicare claims data were obtained for Medicare
enrollees to identify their cancer treatments. Cancer
incidence was calculated for each North Carolina
county using the Medicare county enrollment as the
denominator and Medicare cancer cases as the
numerator. Cancers with the highest incidence
rates for male and females combined were included
in the study. Cancer treatments were identified by
the presence of cancer diagnoses and treatment
codes on the claims and were classified into three
categories: radiation treatments, surgical
treatments, medical treatments (which includes
chemotherapy). Physicians were identified by their
UPIN number and their geographic location was
determined based on their office address contained
in the Medicare files, which was geo-coded to the
centroid of the office zip code.
Population Studied: A total of 15,810 North
Carolina Medicare enrollees were diagnosed with
cancer in 2005. The study population was limited to
cancer cases diagnosed among North Carolina
residents and non-HMO Medicare enrollees age 65
and over as of January 1 with at least 11 months of
Part B coverage in 2005.
Principal Findings: Of the leading four cancers,
prostate cancer had the highest number of cases
(n=2664), followed by breast cancer (n=2304), lung
cancer (n=2367) and colon cancer (n=2004).
Across all cancers, 13% of NC counties had no
physicians providing cancer treatments for
Medicare enrollees. This occurred often in counties
with higher incidence rates. The mismatch between
cancer incidence and physicians providing
treatment varies by type of cancer and type of
treatment. Physicians providing cancer treatments
to Medicare enrollees with prostate cancer had the
greatest coverage across the state, although 18%
of counties had no treatments attributed to county
physicians. The largest gap in cancer physicians
was seen for lung cancer, with 42% of NC counties
having no treatments attributed to physicians with
addresses in that county. The lack of any
physicians providing radiation treatment was
observed more frequently than the absence of
physicians providing surgical or medical treatments.
Conclusions: Cancer burden in some North
Carolina counties is not matched by the availability
of treatment. An aging workforce may have
implications for availability of treatment in the future.
Implications for Policy, Delivery or Practice:
Funding Source(s): University Cancer Research
Fund, The University of North Carolina at Chapel
Hill
Poster Number: 262
Expanding Pharmacists’ Roles in Preventive Health
related to Infant Feeding
Presented by: Roger Edwards, Sc.D., Assistant
Professor, Bouvé College of Health Sciences,
Northeastern University, 360 Huntington Avenue,
206 Mugar Building, Boston, MA 02115; Phone:
(617) 373-8344; Email: ro.edwards@neu.edu
Research Objective: With their knowledge of
medication safety, highly visible roles in the
community, and frequent interactions with soon-tobe and new parents, pharmacists are in key
positions to expand their roles in the promotion and
support of breastfeeding as optimal infant feeding.
Improving rates of breastfeeding will have a
significant public health impact today and in the
years ahead. Only 11.3% of infants are breastfed
exclusively for six months (American Academy of
Pediatrics recommendation). The objective of this
research is to analyze existing contact that
pharmacists have related to infant feeding and to
identify opportunities to expand pharmacists’ roles
in breastfeeding and improved infant formula
feeding.
Study Design: We initially performed a needs
analysis to determine the nature of existing
shortfalls related to supporting breastfeeding/infant
feeding based on literature review. We then
conducted twenty structured interviews of pharmacy
and breastfeeding experts to obtain their insights
and feedback using flowcharts of all possible
contact opportunities during four periods: prepregnancy, prenatal, delivery, and postnatal. We
matched unmet needs with pharmacists’
opportunities and capabilities. We also analyzed
infant feeding/breastfeeding policy statements
worldwide to learn more about activities of
pharmacists in other countries.
Population Studied: Pharmacists
Principal Findings: Women want to breastfeed
longer but discontinue because of inadequate
support, notably from health professionals. Over
90% of women receive medication in the week
following giving birth. Pharmacists have multiple
opportunities in which to engage parents and the
general public on optimal infant feeding. Good
matches between unmet needs and capabilities
were identified in: provision of health promotion
resources and public awareness campaigns,
assistance with purchase of breastfeeding products
and pumps, and provision of information, support,
and referral related to commonly encountered
difficulties as well as medication use during
lactation. Absence of adequate pharmacy
education/knowledge was identified as a barrier.
Absence of a professional organization position
statement, as exists for professional pharmacist
organizations in other countries, also creates
hurdles.
Conclusions: Our qualitative research approach
identified that pharmacists are underutilized
resources that could play important roles in
supporting improved infant feeding behaviors. This
research established the foundation for a broader
research program including conducting interventionbased studies of roles pharmacists can play in
improving infant feeding.
Implications for Policy, Delivery or Practice:
Evidence for the health risks of not breastfeeding
continue to accumulate. Breastfeeding results in
substantive health improvements in the short term
and long run for both infants and mothers; yet,
community-level support for this important health
behavior is inadequate in much of the US.
Leveraging an underutilized resource such as
pharmacists to address unmet preventive health
needs is especially important in our current health
care delivery environment. Optimal intervention
efforts still need to be studied. Infant feeding
education received by pharmacy students and
continuing professional pharmacist education
should be expanded. A certification program for
pharmacists can enhance lactation expertise in
pharmacy. Position statements for consideration by
the American Pharmacist Association are currently
being developed. Medication therapy management
is a recent example of how pharmacist clinical
services have been effectively utilized to improve
patient outcomes and reduce health care costs;
improved infant feeding support by pharmacists has
that same potential.
Poster Number: 263
What about Everyone Else? Trends among the
Non-Physician and Non-Nurse Healthcare
Workforce
Presented by: Bianca Frogner, Ph.D., Assistant
Professor, Health Services Management and
Leadership, The George Washington University,
2175 K St. NW, Suite 320, Washington, DC 20037;
Phone: (202) 994-3816; Email: bfrogner@gwu.edu
Research Objective: This study examines the
trends among a largely ignored, but sizable and
growing portion of the healthcare workforce - nonphysician and non-nurse workers in healthcare – in
Australia, Canada, and the US. This study then
disaggregates the healthcare workforce trends in
the US into specific occupations and discusses the
impact of technology adoption on healthcare
workforce trends.
Study Design: The study is an exploratory analysis
of longitudinal data as an initial step in
understanding dynamics in this understudied
population. For the trend in “other” healthcare
workforce, the author calculates the remaining
healthcare workers after subtracting out nurses and
physicians from the total healthcare workforce
estimates. The author compiles annual data for
specific occupations among the “other” workers with
a focus on occupations that have contributed the
most to the growth of the healthcare workforce.
Population Studied: Healthcare workforce data
collected by the Organization for Economic
Cooperation and Development Health Dataset of
2009 provides national survey estimates in
Australia, Canada, and the US from 1970 to 2008.
For the US, the occupation specific data is from the
Occupational Employment Statistics, a semi-annual
panel survey of 1.2 million establishments collected
by the Bureau of Labor Statistics from 1999 to
2008.
Principal Findings: Non-physician and non-nurses
are over three-quarters of the US workforce, 70% in
Canada, and 60% in Australia in the most recent
year of data. The US always had more “other”
healthcare workers per capita and has been
growing faster than the other two countries over the
38 year period. However, all three countries have
seen a deceleration in the growth of “other” workers
over the time period. A closer examination of
occupations suggests that technicians/
technologists, administrators and support staff are
the largest categories of workers that have also
been rapidly increasing in the US. By the
conference date, technicians/technologists will be
further disaggregated to identify specific
technologies that may have been complementing
the growth of the healthcare workforce. Data to
perform similar occupation analysis in Australia and
Canada is not available
Conclusions: The rapidly growing healthcare
workforce in the US compared to Canada and
Australia are associated with rapidly growth among
“other” healthcare workers. Disaggregation of
occupations in the US shows that
technicians/technologists, administrators and
support staff are large contributors to the growth.
The growth of technicians/technologists is
consistent with previous studies that have identified
technology as a driver of the growth of healthcare
spending. The growth of administrators and support
staff is consistent with studies about administrative
waste due to the complex third party payment
system in the US as well as the insufficient use of
technologies that increase productivity, such as
health IT.
Implications for Policy, Delivery or Practice: By
focusing on physician and nurses, the literature has
ignored a significant portion of the healthcare
workforce. The “other” healthcare workers play an
important role in the delivery of healthcare and also
contribute to the bulk of labor costs. Disaggregating
occupational trends in the healthcare labor force is
a conduit for linking discussions of technology to
total healthcare spending.
Funding Source(s): NIMH
Poster Number: 264
The Influence of Local Area Physician Supply on
the Dispersion of Care among Medicare Patients
with a Consistent Diagnosis
Presented by: An-Chen Fu, M.S., College of
Pharmacy, The University of Iowa, 115 S. Grand
Avenue, Iowa City, IA 52242; Phone: (319) 5129734; Email: anchen.fu@gmail.com
Co-authors: John M. Brooks, Ph.D.
Research Objective: Commentators have
suggested that greater local area physician supply
leads to the use of more physicians by patients,
resulting in diffuse, uncoordinated, and wasteful
care. However, we have found no studies that
demonstrated direct relationships between local
area physician supply and the use of physicians by
individual patients. To fill this gap, we isolated a set
of patients with a consistent diagnosis and course
of treatment and investigated whether local area
physician supply leads to the use of more
physicians by these patients and more care
dispersed across physicians during their first
treatment course.
Study Design: We estimated multiple regression
models that relate local area physician supply in
four specialty categories to the number of
physicians used by patients within a consistently-
defined first course of treatment and the dispersion
of visits across physicians during the first course of
treatment. Visit dispersion is measured via a
Physician Visit Index that ranges from 0 to 1 and is
derived from the Herfindahl concentration index.
The closer the index is to 1, the more concentration
patient visits were to fewer physicians. The
explanatory variables are local area physician
supply by specialty measured as the number of
unique physicians in four specialty categories per
1,000 patients within 50-miles of patient zip codes.
Population Studied: Stage III colorectal cancer
patients diagnosed during 1992 to 2001 receiving
surgery and chemotherapy but not radiation therapy
during their first course treatment (n=6539) in
SEER-Medicare-linked database.
Principal Findings: Local area physician supply
had significant affect on the number of physicians
used by each patient and the dispersion of visits
across physicians. However, the directions of these
relationships varied by specialty. Local area primary
care and oncologist supplies had positive effects on
the number of physicians used and care dispersion.
Whereas, local area surgeon supply negatively
affected the number of physicians used and care
dispersion.
Conclusions: Our results show that local area
physician supply affects the use of physicians by
patients but that these relationships are specialtyspecific. Greater local area supply of primary care
physicians and oncologists lead to increased
physician utilization and care dispersion. However,
greater local area surgeon supply decreased the
number of physicians seen by patients and the
dispersion of care across physician.
Implications for Policy, Delivery or Practice:
Commentators have suggested that greater local
area specialist supply promotes greater care
dispersion leading to wasted healthcare resources.
Our study indicates that for certain physician
specialties (surgeons), greater local area supply
leads to the use of fewer physicians and less care
dispersion. Perhaps specialists in areas with
greater supply have more time to allocate to
individual patients, thereby, lower the need to pass
patients along to other providers. Future
investigation can use the variation in physician use
that is attributable to local area physician supply to
assess the effects of care dispersion on patient
outcomes and costs.
Poster Number: 265
High-Performance Work Practices in Healthcare
Management: Five Case Studies of Best Practices
in Healthcare Organizations
Presented by: Andrew Garman, Psy.D., Associate
Professor and Associate Chair, Health Systems
Management, Rush University Medical Center,
1700 W Van Buren, Suite 126 TOB, Chicago, IL
60612; Phone: (312) 942-5402; Email:
Andy_N_Garman@rush.edu
Lead Author: Ann Scheck McAlearney, Sc.D., M.S.
Co-authors: Paula H. Song, Ph.D.; Megan
McHugh, Ph.D.; Julie Robbins, M.H.A.; Michael
Harrison, Ph.D.
Research Objective: A capable workforce is
central to delivery of high-quality care. Evidence
from a breadth of sources outside and within
healthcare suggests that innovative human
resource practices (known as high-performance
work practices (HPWPs) - including systematic
personnel selection and incentive compensation -help attract and retain highly qualified health care
staff; HPWPs also enhance individual and team
performance. Hence, HPWPs may represent an
underutilized healthcare improvement strategy. We
investigated exemplary uses of HPWPs and their
possible link to care quality and patient safety.
Study Design: Following an extensive literature
review, we performed five case studies of U.S.
healthcare organizations purposively selected
based on known success with HPWP use. Site
visits in 2009 included in-person interviews with key
organizational and clinical informants and collection
of documents on HPWPs. Interviews were recorded
and transcribed verbatim to permit rigorous
qualitative analysis.
Population Studied: Case study sites were
purposively selected based on their reputation as
exemplars for innovative human resource practices
and to provide variability in organizational context,
approach to the selected HPWPs, and their
potential to serve as best practice examples worthy
of attention and possible emulation by other
healthcare organizations. Seventy-one key
informants included human resources
professionals, organizational leaders, clinical
leaders, quality improvement professionals,
information systems managers/directors, finance
and accounting professionals, and select
administrative and clinical personnel involved in HR
practices.
Principal Findings: Across sites and key
informants, senior leadership support, capabilities
of the implementers of the HPWPs, and
organizational financial conditions were reportedly
important in influencing the adoption and use of
HPWPs. The four HPWP subsystems we had
previously characterized as directly relevant to
healthcare—organizational engagement, staff
acquisition/development, frontline empowerment,
and leadership alignment/development were
reportedly eM.P.H.asized in each of the five
organizations. While direct evidence of links
between HPWPs and both employee outcomes
(e.g., turnover, higher satisfaction/ engagement)
and care system and organization-level outcomes
(e.g., fewer “never events”, innovation adoption,
lower agency costs, and lower turnover costs) was
largely unavailable, informants consistently noted
that they believed such links to exist, and that future
data collection could quantify the links.
Conclusions: These case studies provide
examples and preliminary evidence of how HPWPs
can improve operations and impact quality of care
and safety in healthcare organizations. While
further research is needed to assess the
applicability of HPWPs to diverse settings and to
assess the generalizability of our findings on HPWP
impacts, we would encourage development of highperformance work systems within healthcare
settings and assessment of their contributions to
quality, safety, and other dimensions of
organizational performance.
Implications for Policy, Delivery or Practice:
From a practitioner perspective, our findings
highlight the importance and potential impact of
HPWPs in healthcare organizations, and provide
direction about best practices for these
organizations to consider. From an academic
perspective, these findings lay the groundwork for
future research into a more definitive link between
HPWP and quality outcomes; additional research
will provide further insight as to which of the
practices will have the highest leverage for
improving quality and safety in healthcare.
Funding Source(s): AHRQ
Poster Number: 266
Psychosocial Stress at Work and Quality of Care
among Surgeons in Germany Hospitals - Analysis
of Survey Data using Structural Equation Modelling
Presented by: Kirstin Grosse Frie, M.A., research
associate, Center of Psychosocial Medicine,
University Medical Center Hamburg-Eppendorf,
Martinistr. 52, Hamburg, Germany 20246; Phone:
+004940741054509; Email: k.grosse-frie@uke.unihamburg.de
Co-authors: Jens Klein; Karl Blum, Dr.; Johannes
Siegrist, Prof.; Olaf von dem Knesebeck, Prof.
Research Objective: Psychosocial stress at work
among physicians is a research topic of increasing
importance. The effort-reward-imbalance model
(ERI) is an established model to analyze
psychosocial stress at work. The ERI-Model implies
that high efforts and low rewards lead to chronic
stress. Several international studies among health
care professionals tested the model’s assumptions
and proved that psychosocial work stress is a
predictor of impaired health, burnout and
depression. However, only a few studies examined
associations between work stress and quality of
care.
Study Design: A cross-sectional standardized mail
survey was conducted in German hospitals with a
general surgical and gynecological ward. The
questionnaire included the ERI-Questionnaire, a
validated and reliable instrument consisting of
scales for Overcommitment (specific cognitive and
motivational pattern), efforts and rewards. Rewards
were measured by the three sub-dimensions
esteem, job security/ job promotion and salary.
Respondent’s health care performance was
measured by self-evaluation. Questions were
phrased according to the SERVQUAL, a multipleitem scale for measuring service quality and the
Physician Achievement Review, a multisource
feedback questionnaire to assess the competencies
of practicing surgeons. Additionally, two questions
about the frequency of diagnostic and therapeutic
errors were introduced. To analyze the association
between work stress and health care quality
structural equation modelling (SEM) was
performed. Furthermore, gender and occupational
position were additionally introduced into the model.
Population Studied: 1,311 surgeons working in
489 hospitals (response rate 65%).
Principal Findings: Results of the SEM showed
significant chi-squares but fit indices for the model
were good (CFI>0,96, RMSEA<0,4). Five latent
health care dimensions were differentiated by
confirmatory factor analysis: service quality,
psychosocial health care, quality management,
diagnosis and errors. No significant associations of
high efforts and low rewards with these health care
dimensions were found. Significant associations
between high rewards and the three health care
dimensions service quality, quality management
and errors, implied better care in these dimensions.
Significant associations between efforts and
psychosocial health care as well as between
overcommitment and diagnosis respectively errors,
implied less quality of care.
Conclusions: This study shows that psychosocial
work stress according to the assumptions of the
ERI-Model is not associated with a poorer
evaluation of health care quality. It seems that there
are other mechanisms between job stress and
health care quality. Efforts and especially rewards
seem to be important on their own and efforts can
not be moderated by high rewards.
Implications for Policy, Delivery or Practice: For
redesigning physician's workplace environment it
should be accounted for that higher rewards are
associated with a higher evaluation of health care,
but they should not be established to compensate
high efforts.
Funding Source(s): Hans Böckler Stiftung
Poster Number: 267
The Changing Face of Primary Care Medicine:
Implications for Workforce Development
Presented by: Timothy Hoff, Ph.D., Associate
Professor of Health Policy and Management, Health
Policy, Management, and Behavior, University at
Albany, SUNY, Room 181, GEC Building, 1
University Place, Rensselaer, NY 12144; Phone:
(518) 402-6512; Email: thoff@albany.edu
Research Objective: Primary care medicine is
increasingly unattractive to medical students.
Primary care specialties are also undergoing
internal demographic shifts that have implications
for workforce development. These shifts include
increasing numbers of women physicians and
international medical graduates (IMGs) going into
primary care, an increasingly older primary care
physician (PCP) cohort, and a younger PCP cohort
that has different expectations regarding career and
work. There is a need to examine the work
experiences, values, and career expectations of
PCPs to better understand the workforce
recruitment and retention strategies that can attract
more of the demographic groups already picking
primary care in greater numbers.
Study Design: A qualitative approach was
employed involving semi-structured interviews with
88 PCPs practicing within a variety of distinct work
settings. The qualitative approach was chosen to
obtain rich description of how these newer PCP
demographic groups think about their work and
careers, and to explore what their perceptions and
experiences imply for primary care workforce
recruitment and retention strategies. Data were
coded and analyzed using Atlas.ti software and a
content analytic approach.
Population Studied: Eighty-eight primary care
physicians stratified on the basis of primary care
specialty, age, career stage, gender, race and
ethnicity, type of employment status, employment
setting, and geographic location. These sampling
strata were selected purposively, consistent with
the qualitative approach, based primarily on the
study focus which is exploring PCP cohorts
according to age, gender, and ethnic background.
By having an appropriate number of PCPs
interviewed across the different cohort groups,
constant comparison of interview data could occur
both within and across the different groups.
Principal Findings: The results suggest
differences in career and work expectations
between women, younger, and non-U.S. citizen
PCPs compared to older, male, and white PCP
cohorts. These differences relate to desired
everyday job structures, motivations for choosing
and remaining in primary care, preferred types of
primary care work, and preferences regarding
patient relationships. Several of these expectations
were similar across the newer PCP demographic
groups. Many of the expectations differed when
compared to those held by older PCPs, especially
concerning preferences among newer demographic
groups for flexible, compartmentalized work
schedules, adequate time off from work, and work
based purely in the office ambulatory care setting.
Conclusions: The descriptive findings inform the
development of specific workforce recruitment and
retention strategies in primary care medicine, which
is losing practitioners faster than it can replace
them.
Implications for Policy, Delivery or Practice:
From a practice level perspective, shifting
preferences for particular types of job structures
and time away from work encourages the
consideration of new employment arrangements for
PCPs that are non-traditional and may impact the
the physician-patient relationship. In addition,
younger cohort expectations regarding traditional
PCP work such as hospital medicine and
preferences for more office-based ambulatory care
work requires policy eM.P.H.asis on making the
ambulatory care setting the focal point for the right
mix of rewards and incentives to keep PCPs
satisfied and interested in continuing as primary
care practitioners.
Poster Number: 268
Analysis of the Effects of Primary Care Education in
Japan
Presented by: Genta Kato, M.D., Emergency
Physician, Department of Primary Care and
Emergency Medicine, Kyoto University Hospital, 54
Shogoin Kawahara-cho, Sakyo-ku, Kyoto, Japan
606-8507; Phone: +81757513126; Email:
qq9f8hn9@gmail.com
Research Objective: To clarify the existing
problem concerning new residency training system
in Japan, which shifts the focus from specialist
training to primary care education from 2004 under
the control of the Health, Labor and Welfare
Ministry.
Study Design: Analyzing qualitative data from
unstructured interviews involving doctors who train
new residents.
Population Studied: 18 advising doctors being
chosen by snowball sampling.
Principal Findings: Under the new training
system, residents belong to a general hospital in
the first two years, and rotate between several
specific departments. They learn each specialty for
around two months. Some advising doctors are
skeptical about whether the content of their
education and the shortness of the period to which
they are exposed to each specialty is effective.
Others criticize residents' motivation to learn, since
advising doctors must educate all residents, many
of whom will never practice the teaching doctors’
specialty in the future. However, emergency
physicians and general internees tend to be positive
about the new residency training system. On the
other hand, some young advising doctors positively
evaluate this system because residents can
broaden the scope of their learning.
Conclusions: Although protocol for instruction is
published by the Health, Labor and Welfare
Ministry, most of the teaching doctors themselves
are not primary care physicians, but specialists.
Many teaching doctors do not feel satisfied with
their instruction, because the new residency training
system is not clear about whether it aims to yield
many primary care physicians or to give residents
common and practical medical literacy.
Implications for Policy, Delivery or Practice: In
Japan, patients’ accessibility to medical institutions
is considered to be relatively convenient in that
patients can visit various doctors regardless of their
specialist field. However at the same time, some
specialists cover the same territory as other
physician with different specialties, including
primary care, since the borders between one
specialty and another are to some extent obscure. It
is often said that the concept of ‘primary care’ is not
well-established among patients and doctors in
Japan. For example, the certifying system for
primary care physicians is only 17 years old. In the
Japanese new residency training system, ‘primary
care’ practices are considered to be common
knowledge for all doctors. Educated senior
residents have no chance to make use of their skills
as primary care physicians because from the 3rd
year of their career they choose a specialty and
from that point forward are trained only in that
specialty. Even if we want to compare the new
system with an old system in which residents were
taught specific knowledge and skills from the
beginning, it is difficult to estimate how much
residents’ knowledge of primary care have been
improved and upgraded. Since residents must
spend their first two impressionable years acquiring
skills of a primary care physician, policy makers
should define the concept of ‘primary care’ and
settle a specific image of goal more clearly to
enhance the effectiveness of this education system.
Poster Number: 269
Relationship of Regional Variations in the Use of
ENT procedures for Children in NY State to
Regional Variations in Physician Distribution
Presented by: Lawrence Kleinman, M.D., M.P.H.,
Vice Chair, Health Evidence & Policy, Mount Sinai
School of Medicine, One Gustav Levy Place, New
York, NY 10029; Phone: (212) 659-9556; Email:
lawrence.kleinman@mssm.edu
Co-authors: Natalia Egorova, Ph.D.
Research Objective: Describe regional variations
in use of childhood otolaryngic (ENT) surgeries in
NY State and their relationship to the number of
primary care physicians (PCP) and surgical
subspecialists (SSS) in each region, controlling for
the number of children in each region.
Study Design: Analysis of administrative data
(SPARCS) for children <= 16 y.o. who had
ambulatory surgery in NY State in 2006; 11 regions
were defined by NY State Dept of Health. Physician
data for 2006 are from Annual NY Physician
Workforce Profile (SUNY Albany 2007). Linear
regression used 3 predictor variables: number of
children, PCP, and SSS per region. As outcomes
for three models we looked at tonsillectomy and/or
adenoidectomy(TnA), tympanostomy tube (TT)
insertion, and either TnA or TT. We also regressed
the square root of each outcome to confirm
distributional assumptions.
Population Studied: All children under 16 years of
age who had an ambulatory surgery at a facility
reporting to SPARCS, which is intended to be a
comprehensive statewide data reporting system
and serves as NY State's data source for the
federal Health Care Utilization Project data
collection.
Principal Findings: Region populations vary from
~250,000 to 650,000 children. 10.3% of SSS were
otolaryngologists and 24.6% of PCP pediatricians.
Regional variation is considerable: TT range 1.3-8.5
per 1000 children (statewide mean 4.3), TnA 1.79.9 (5.1), either 2.6-15.8 (8.3) Models for all three
outcomes had consistent results, as did those using
SQRT(outcome). Controlling for population, the
marginal increase in the number of surgeries
associated with each SSS was slightly more than
the decrease associated with each PCP. Betas for
the model regressing either TT or TnA were:
population of children 0.02; surgical specialists 8.8;
primary care provider -6.8, r-squared=0.88, all
betas <.05.
Conclusions: Childhood ENT surgery is known to
be highly discretionary. In NY in 2006 ENT surgery
use varied by region directly proportional to the
number of surgical specialists, and inversely
proportional to the number of PCP. If causal,
extrapolation would estimate an average marginal
impact of 84.5 more surgeries per year for each
ENT surgeon and of 27.7 fewer surgeries per year
for each pediatrician per region. Limitations include
ecological data from one state and lack of
granularity of specialty within each region.
Strengths include high quality full population data.
Implications for Policy, Delivery or Practice:
This early study suggests that the phsyician
disribution and specialty distribution may influence
the use of ENT surgery in children. Our findings
suggest that surgery use increases as the mumber
of surgical specialists increase and decrease as the
number of primary care providers increase. If
confirmed this could have important implications for
quality and cost of care and provides data to guide
physician workforce policy.
Poster Number: 270
The Frontline Healthcare Worker: An
Underrecognized Resource for a 21st Century
Healthcare System
Presented by: Thomas Konrad, Ph.D., Senior
Fellow, Cecil G. Sheps Center for Health Svcs
Research, University of North Carolina at Chapel
Hill, 725 MLK, Jr. Boulevard CB 7590, Chapel Hill,
NC 27599-7590; Phone: (919) 966-2501; Email:
bob_konrad@unc.edu
Co-authors: Emmeline Chuang, A.B.; Jennifer
Craft Morgan, Ph.D.; Janette Dill, M.A., M.P.H.
Research Objective: Frontline healthcare workers
(FLWs), consisting of 7 million workers in 32 health
care occupations, currently constitute about half the
US healthcare workforce. These workers have
median annual wages <$40K and often less than a
bachelor’s degree. Although FLWs play a critical
role in the delivery of “hands on” healthcare, little
systematic information exists about their life
circumstances, work motivations, job satisfaction,
and how they view their contributions to health
outcomes across different types of healthcare
organizations. This study fills that gap.
Study Design: Data used for this study were drawn
from an eight-page survey administered to 819
FLWs working in 42 health care employers. These
data were collected as part of a baseline
assessment of FLWs within Jobs to Careers, a
national program supporting partnerships of health
care employers, educational institutions, and other
organizations across the U.S. Survey items
included questions about FLW demographics, job
history, wages, and job perceptions. Job perception
items were adapted from existing instruments,
aggregated into multi-item scales, assessed for
validity reliability, and examined. Analysis of
variance of scale means and percentages and chisquare tests comparing categorical variables
(p<.05) were used to assess cross-segment
variability.
Population Studied: Frontline workers across four
major health care sectors: hospitals, long-term care
(LTC), community health centers, and behavioral
health centers.
Principal Findings: Frontline workers within our
sample are overwhelmingly female (90%),
minorities (50%), and dependent on their paycheck
as the sole source of family income (60%). A
significant percentage are also single mothers
(20%). These FLWs have positive perceptions of
quality of care that their employers provide and high
global job satisfaction (7.9 and 7.1 on 10-pt scales)
especially with intrinsic aspects of work (i.e.,
carework). Although there is widespread
dissatisfaction with work overload (2.6 on an 8-pt
scale), FLWs express high levels of commitment to
current employers--only 10% are considering
leaving their jobs. Some cross-sector differences
are observed in specific facets of job satisfaction. In
LTC settings, where entry level educational
requirements are lower, only 20% of respondents
have some college (as opposed to 40% in other
sectors). LTC workers have somewhat lower wages
and are less satisfied with career opportunities and
supports than FLWs in other settings.
Conclusions: FLWs--a sizable, ubiquitous, and
critical yet relatively invisible component of the
healthcare system--can be considered a single
workforce, with a common set of demographic and
socioeconomic challenges, positive perceptions of
their core work, and high levels of commitment to
their employers. However, these FLWs face many
common barriers to skill and career development,
suggesting that similar approaches may be used to
improve their situation and enhance their
contributions to health care quality and population
health. Particular challenges are found in the LTC
segment of the workforce.
Implications for Policy, Delivery or Practice:
Given high demand for FLWs and evidence of that
they can be committed to employers and patients,
increased private and public investment in the
career development of these workers by employers,
educational institutions, and workforce development
organizations seems warranted.
Funding Source(s): RWJF, The Hitachi
Foundation, US Dept of Labor
Poster Number: 271
Using Health Policy Education Targeted toward
Underserved Patients to Promote Interdisciplinary
Collaboration
Presented by: Helene Lipton, Ph.D., Professor,
Institute for Health Policy Studies, University of
California, San Francisco, 3333 California Street,
Suite #265, San Francisco, CA 94118; Phone:
(415) 476-2964; Email:
smitha@pharmacy.ucsf.edu
Co-authors: Amanda Smith, M.P.H.; Cindy Lai,
M.D.; Timothy Cutler, Pharm..; Marilyn Stebbins,
Pharm.D.
Research Objective: Given recent and anticipated
health reforms, it is critical to improve health policy
literacy among medical students, residents, and
other health professionals. One such reform is
Medicare Part D (“Part D”), a federal program
providing drug coverage for Medicare patients. In
2006, we developed an interdisciplinary Peer
Educator program, in which pharmacy students
educate health professionals about Part D. After
demonstrated success of this initiative at our own
institution, our objectives were: (1) to assess
whether this Peer Educator program can be
replicated at other institutions and (2) to evaluate
the impact of the program on learners’ Part D
knowledge, intent to collaborate with pharmacists,
and awareness of the special challenges
underserved Part D patients confront.
Study Design: The program was expanded
statewide by training faculty champions at six
additional California pharmacy schools to develop
their own Part D Peer Educator programs. Peer
Educators were pharmacy students selected
through a competitive application process. Part D
policy lectures ranged from one-to-two hours and
used a case-based approach to demonstrate
specific strategies that physicians and other
prescribers can use to ensure access to appropriate
drug plans and decrease out-of-pocket drug costs
for underserved patients. A pre-post survey was
administered to all attendees.
Population Studied: Targeted audiences included
students, residents, nurse practitioners and faculty
from health professional schools and academic
medical centers. From October 2008 to November
2009, Peer Educators delivered 36 Part D lectures
statewide to 1,011 health professionals. A total of
950 pre-post surveys were completed (response
rate = 94%).
Principal Findings: Lack of Part D training was
evidenced by the fact that before the lecture, 60%
of learners reported no prior exposure to Part D.
Using the Wilcoxon signed-ranks test, ten of eleven
outcome measures were improved significantly,
compared to baseline (p<0.0001). The biggest
gains were noted in learners’ knowledge of the Part
D benefit and their ability to identify clinically
relevant resources to help patients. Learners also
significantly increased their awareness of
challenges confronting underserved Part D patients
and their intention to collaborate with pharmacists
on patient activities including drug product
selection, formularies, Part D insurance issues, and
drug costs. Improvements were consistent across
all schools. Of the 950 learners, 86% rated the
overall quality of instruction on a 5-point Likert scale
(1=poor, 5=excellent) as either “very good” or
“excellent”. Similarly, 83% of participants rated the
information presented as “very useful” or “extremely
useful” (1=not at all useful, 5=extremely useful).
Conclusions: With a combination of dedicated
faculty champions and clinically relevant material,
the Peer Educator program was rapidly integrated
into six additional schools of pharmacy,
demonstrating that trained students can significantly
improve health policy knowledge, attitudes toward
other health care professionals, and increase the
likelihood of interdisciplinary collaboration.
Implications for Policy, Delivery or Practice:
Lack of Part D training for health professionals
limits their ability to help patients’ realize the
intended benefits of Medicare Part D. Our research
suggests that academic health centers, working
with policymakers, should initiate health policy
education programs that also foster interdisciplinary
collaboration, especially regarding care delivery for
underserved patients.
Funding Source(s): The Amgen Foundation
Poster Number: 272
An Exploration of the Relationship Between
Employee’s Personality Characteristics and the
Implementing Effect of Hospital Learning
Organization Models
Presented by: Chenchung Ma, Ph.D., Assistant
Professor, Healthcare Administration, I-Shou
University, 8, Yida Road, Yanchao Township,
Kaohsiung County, Taiwan 824; Phone:
+073836187; Email: up000238@isu.edu.tw
Lead Author: Chien-Chang Yang, M.H.A.
Co-authors: Patricia Moulton, Ph.D.; Paul Su,
Ph.D.
Research Objective: In the knowledge-based
economical age, entrepreneurs who apply the
learning organization model will enhance their
competitive advantage. A learning organization is
an organization that facilitates the learning of all its
members and continuously transforms itself.
Learning organization can lead better employee
satisfaction. The learning organization model has
become the successful model that business
desires. Whether the learning organization can be
implemented well or not depends on the staff in the
organization. The personality characteristics of
employees can deeply influence the effect of
learning within an organization. This research was
conducted to find out how employee’s personality
characteristics related to implementation
effectiveness in five Taiwan hospitals who have
implemented the learning organization model. The
results will provide valuable information to other
healthcare organizations as they start to implement
learning organization models in their facility,
especially when recruiting staff.
Study Design: This study used a cross-sectional
research design to identify the relationship between
employee’s personality characteristics and the
implementing effect of hospital learning
organization models. The structured questionnaire
for this study contained 57 items in three parts: (1)
Employee’s personality characteristics contained 30
items and measured by a five-point Likert-type
scale, (2) Learning organization models contained
20 items and measured by a five-point Likert-type
scale, and (3) demographics contained 7 items.
This cross-sectional design involves the collection
of data at one point in time.
Population Studied: The target population for this
study was employees at five hospitals in northern,
central, southern, and eastern areas of Taiwan.
Each hospital distributed 100 questionnaires via
human resource department staff to physicians,
nurses, medical technicians, and administrative
staff. A total of 500 questionnaires were distributed
and 419 completed questionnaires were returned
resulting in a response rate of 83.8%.
Principal Findings: Analysis of variance (ANOVA)
demonstrated that the implementing effect of
learning organization varied significantly based on
gender, age, marital status, level of education, job
position, job category, and year of service.
Openness to experience, conscientiousness,
extraversion, and agreeableness all had positive
correlations with the implementing effect of learning
organization. Only neuroticism showed a negative
correlation with implementation effectiveness ((R)=
-0.284, P<0.01).
Conclusions: This study indicates that the higher
the educational level, the greater likelihood of better
implementation of the learning organization model.
Medical technicians and administrative staff had
higher scores of the implementation of the learning
organization models. In the future, when healthcare
organizations recruit and select the employee, they
should take level of education into consideration.
Furthermore, they should also encourage their
employees to attend on-the-job training courses.
Another result indicated that the employee’s
personality characteristics will influence the
effectiveness of the learning organization model.
Therefore, when healthcare organizations recruit
and select employees, they should consider
employee’s personality characteristics in order to
find the appropriate employee to serve in
healthcare organizations.
Implications for Policy, Delivery or Practice:
This study provide valuable information to other
healthcare organizations as they start to implement
learning organization models in their facility,
especially when recruiting staff.
Poster Number: 273
Increasing the Supply of Nurse Educators
Presented by: John Marcotte, Ph.D. Center for
Health Outcomes & Policy Research, University of
Pennsylvania, PA 19104-4217; Phone: (215) 8987716; Email: marcotte@nursing.upenn.edu
Lead Author: Robyn Cheung, R.N., Ph.D.; Email:
cheungr@nursing.upenn.edu
Research Objective: The object of this study is to
assess what factors increase or decrease the
likelihood that nurses will work in nursing education.
In order to be an educator, a nurse must at least
obtain a master's degree; therefore, this study also
assesses what factors influence educational
advancement among nurses.
Study Design: The data are from multiple cohorts
of nurses dating from the 1960s through the mid2000s. The design has two stages. The first stage
examines what affects whether nurses obtain
advanced degrees (master's and above) while the
second stage looks at what influences whether
nurses with advanced degrees work in nursing
education. Since the data contain degree dates, we
are able to analyze how long until nurses earn
second or third degrees and how long before
nurses engage in nursing education. Because the
data cover multiple decades, we also examine age,
period and cohort effects.
Population Studied: Nurses from the 1960s
through today are the study population.
Principal Findings: Initial degree is a strong
determinant of whether nurses ultimately obtain a
degree that qualifies them for teaching. Nurses
whose first degree is an Associate's as opposed to
a Bachelor's rarely advance. Results also show
differentials about ethnic groups in the rate of
obtaining advanced degrees and engaging in
teaching.
Conclusions: While both Associate's and
Bachelor's degrees qualify nurses to work. Only the
Bachelor's is a potential gateway to obtaining an
advanced degree. As a result, a study of what
determines the supply of nurse educators must
examine factors affecting both the earning of
advanced degrees and the engaging in teaching.
Implications for Policy, Delivery or Practice: The
key policy question is how to increase the number
of nurses who work in nursing education. From a
policy standpoint, nurses should be encouraged to
obtain a BSN as the first degree. Without an initial
BSN, nurses are effectively locked out from
teaching. Increasing the number of qualified nurse
educators will ultimately produce more nurses.
Poster Number: 274
Economic Burden of Illness for the United States
Workforce With Chronic Musculoskeletal Pain
Conditions: Arthritis, Back and Fibromyalgia Pain
Presented by: Margaret McDonald, Ph.D.,
Director Health Economics and Outcomes
Research, Global Market Access, Pfizer Inc, 235
East 42nd Street, New York, NY 10017; Phone:
(212) 733-3093; Email:
margaret.m.mcdonald@pfizer.com
Co-authors: Marco daCosta DiBonaventura, Ph.D.;
Stacey Ullman, M.H.S.
Research Objective: Few studies have
simultaneously assessed the impact of pain and
productivity loss for US workers with arthritis, back
problems or fibromyalgia. This study updates the
current literature by comparing absenteeism,
presenteeism, and average annual direct and
indirect health care costs for US workers reporting
pain from these chronic musculoskeletal conditions
to workers without these painful conditions.
Study Design: Cross-sectional study design. Work
productivity was assessed using the Work
Productivity and Activity Impairment (WPAI)
questionnaire. Four subscales (absenteeism,
presenteeism, overall work impairment, and activity
impairment) are generated in the form of
percentages of time missed due to health. Direct
costs were total employer medical costs and
indirect costs were costs associated with
absenteeism and presenteeism.
Population Studied: Analysis of national data
collected from employed (full-time, part-time, or
self-employed) US adults aged 20 to 64 years
participating in the US National Health and
Wellness Survey (NHWS) 2008 (N=30,828).
Principal Findings: Of the 122 million US workers
aged 20 to 64 years, 34% reported pain in the past
12 months. Among these workers, 24%, 46%, and
4.1% reported arthritis, back or fibromyalgia pain,
respectively, in the past month. Absenteeism
among workers with arthritis, back or fibromyalgia
pain is 8.2%, 7% and 15.3%, respectively. For
workers without arthritis, back or fibromyalgia pain,
absenteeism is 4.3%, 4.2% and 4.5%, respectively
(p<0.0001 for all absenteeism differences).
Presenteeism (reduced productivity while working)
is highest for workers with fibromyalgia pain
(45.3%), followed by workers with arthritis or back
pain, 31.9% and 27.3%, respectively. Among
workers without fibromyalgia, arthritis or back pain,
presenteeism is 16.2%, 15.2% and 14.6%,
respectively (p<0.0001 for all presenteeism
differences). Average annual direct health care
costs per worker with fibromyalgia pain vs. without
fibromyalgia pain are $6,082 vs. $2,437; indirect
costs are $15,966 vs. $6,269; total costs are
$22,048 vs. $8,706 (p<0.001 for all cost
differences). Arthritis pain sufferers had 2 times the
annual total health care costs per worker than nonsufferers ($16,787 vs. $8,148; p<0.0001); back pain
sufferers and non-sufferers had average total health
care costs of $14,151 and $7,899, respectively
(p<0.0001).
Conclusions: Chronic, painful conditions are costly
to employers. Pain sufferers generally reported
twice the level of absenteeism and presenteeism,
and higher associated annual health care costs
relative to non-sufferers. Fibromyalgia pain
sufferers had the highest impairment metrics and
associated costs.
Implications for Policy, Delivery or Practice:
Increasing awareness and effective programs and
polices can encourage proper risk management,
treatment and prevention.
Poster Number: 275
Frontier America's Health Workforce Supply and
Population Health Outcomes
Presented by: Preethy Nayar, M.D., Ph.D., Asst.
Professor, Health Services Research and
Administration, University of Nebraska Medical
Center, 984350 Nebraska Medical Center, Omaha,
NE 68198; Phone: (402) 559-1981; Email:
pnayar@unmc.edu
Co-authors: Bettye Apenteng; Ahn Nguyen,
M.P.H.; Keith Mueller, Ph.D.
Research Objective: The objective of this study
was to examine and compare the county level
characteristics including demographics, workforce
supply and population health outcomes of frontier
and non-frontier counties in the United States.
Study Design: The study design is a crosssectional descriptive study.
Population Studied: All counties in the United
States were studied using the Community Health
Status Indicators (CHSI) 2008 data. The definition
of frontier county used is the conventional definition
of less than 7 persons per square mile. In 2005, of
a total of 3141 counties in the US, 419 were frontier
counties.
Principal Findings: Frontier counties have a
significantly higher proportion of elderly and white
population. Although not statistically significant,
frontier counties also have a higher poverty
rate.Frontier counties have a lower average life
expectancy as compared to non-frontier counties
(66 vs. 76). Frontier counties also have a
significantly lower ratio of primary care physicians
per 100,000 population and dentists per 100,000
population(not statistically significant)and a higher
rate of Health Professional Shortage Area or HPSA
single-county, primary medical care, health
professional shortage area designation.
Conclusions: “Very little exists in the published
literature on the health, mental health and related
problems in these [frontier] areas” (Wagenfeld,
2000).This study decribes the unique challenges
that frontier counties face in terms of demographics,
population health outcomes and health workforce
supply.
Implications for Policy, Delivery or Practice:
Frontier counties in America face significant
challenges in terms of access to health care
delivery due to geographical isolation. Policy
makers should take in to consideration these
unique challenges and design innovative solutions,
including a greater use of telehealth initiatives, to
address these issues.
Funding Source(s): Other Govt
Poster Number: 276
Perceptions of Work Environment, Burnout, and
Job Dissatisfaction: Are There Differences Between
International and U.S. Nurse Graduates?
Presented by: Donna Felber Neff, Ph.D., R.N.,
Assistant Professor, College of Nursing, University
of Florida, PO Box 10087, Gainesville, FL 326100187; Phone: (352) 273-6334; Email:
dneff@ufl.edu
Co-authors: Jeannie Cimiotti, D.N.S., R.N.; June
Nogle, Ph.D.; Linda Aiken, Ph.D., R.N., F.A.A.N.
Research Objective: In the US, the state of Florida
employs the second highest number of
internationally educated nurses. The purpose of this
study was to determine if differences exist in
international nurse graduates and US educated
nurses’ perceptions of the professional practice
environment, job-related burnout, and job
satisfaction.
Study Design: This was a secondary analysis of
2007 survey data collected from a random sample
of registered nurses licensed and residing in the
state of Florida. Nurses’ were queried on a number
of items that included demographics, work setting,
aspects of the work environment, satisfaction with
their work, and perceived job-related burnout. The
quality of the nurse work environment was
measured by the Practice Environment Scale of the
Nursing Work Index. Nurse job-related burnout was
measured using the Maslach Burnout Inventory –
Human Services Survey.
Population Studied: Nurses met inclusion criteria
if they provided direct patient care (n=6628) in an
adult acute care hospital. Nurses were categorized
based on their basic nursing education as obtained
in the US or elsewhere.
Principal Findings: Of the hospital nurses included
in our sample 5,734 (87%) were educated in the
US. When compared to US educated nurses,
significantly larger proportions of internationally
educated nurses had a BSN degree and were
significantly more satisfied with their jobs. While
there were no significant differences, more than a
third of the nurses in both groups reported high
levels of job-related burnout as defined for health
care providers. Furthermore, internationally
educated nurses reported a significantly more
favorable mean score on work environment on four
of the five subscales of the Practice Environment
Scale - staffing and resource adequacy; nurse
participation in hospital affairs; nursing foundations
for quality of care; nurse manager ability,
leadership, and support. There were no significant
differences between the two groups in mean scores
of the subscale collegial nurse physician
relationships.
Conclusions: When compared to US educated
nurses, internationally educated nurses are more
likely to have a BSN, have a better perception of
the work environment and are more satisfied with
their jobs. However, both groups of nurses report
high level of job-related burnout.
Implications for Policy, Delivery or Practice:
Currently in the US, there are 116,000 nurse
vacancies in hospitals, and it is projected that the
demand for registered nurses will continue to grow
by 2 to 3% each year. Active recruitment of
international nurse graduates has increased
substantially to address this shortage. Based on
previous work describing the work environment of
nurses and patient outcomes, the findings from this
study suggest that the use of internationally
educated nurses should complement the quality of
care provided by US trained counterparts. However,
job-related burnout continues to plague registered
nurses regardless of their country of origin.
Poster Number: 277
Does Use of Supplemental Nurses Increase
Hospital Costs?
Presented by: Katia Noyes, Ph.D., Associate
Professor, Department of Community and
Preventive Medicine, University of Rochester
Medical Center, 601 Elmwood Avenue Box 644,
Rochester, NY 14642; Phone: (585) 275-8467;
Email: katia_noyes@urmc.rochester.edu
Lead Author: Ying Xue, D.N.Sc.
Co-authors: Andre Chappel, Ph.D. candidate;
Linda Aiken, Ph.D.; Deborah Freund, Ph.D.; Katia
Noyes, Ph.D.
Research Objective: The United States is in the
midst of a nursing shortage, which is expected to
intensify as baby boomers age and the need for
health care grows. At the same time, there is
tremendous pressure on hospital administrators to
design organizational systems that are both efficient
in cost and offer high quality of care. The purpose
of this study is to examine the impact of use of
supplemental registered nurses (SRNs) on quality
of care and cost.
Study Design: Data on patient satisfaction, inhospital mortality, and nurse turnover are regularly
collected for performance assessment. Medication
errors, falls, and pressure ulcers were reported by
clinicians. Costs associated with admissions, total
reimbursement (net revenue) and direct costs per
patient stay were obtained from the hospital’s billing
and administrative records. These financial
measures were then aggregated by quarter to the
unit level. The length of stay in each unit was also
derived from the billing records and aggregated by
quarter to the unit level. The quarterly financial
measures for each unit were then divided by the
total patient days per unit. The quarterly per day
financial measures were adjusted for patient casemix (mean age, percent male, and patient severity
and comorbidities) and clustered at the unit level.
We used the risk-adjusted per day financial
measures and risk-adjusted quality indicators for
the cost-effectiveness analysis (CEA).
To calculate the incremental cost-effectiveness ratio
(ICER) for each unit (n=17), we used the unit with
the lowest percentage of SRNs among all RNs
across all quarters (0.43%) as a control.
Population Studied: The study examined patient
discharges in adult medical, surgical, and ICU units
in a large Upstate New York teaching hospital
between 2003 and 2006.
Principal Findings: Higher charges were
associated with higher revenue and higher costs
(correlation 0.9). Very low and very high SRN use
(up to 13%) was associated with higher costs
compared to units with near average SRN use.
Patients in most of the units had lower overall
satisfaction with care than patients in the control
unit with the lowest SRN use and, hence, these
units were dominated by the control unit. Units with
higher SRN use had lower mortality and higher per
diem costs than the lowest SRN use unit, with
ICERs between $2,953 per life saved to $75,295
per life saved. In respect to medication errors, the
units with higher SRN use spent between $166 and
$2,252 per error reduced compared to the unit with
the lowest SRN use. Finally, the majority of units
had about the same or worse RN turnover than the
control unit (with lowest use of SRNs), but had
higher per diem costs.
Conclusions: Our results indicate that the
underlying reason for why certain hospital units use
SRN may be more important than overall percent of
SRN and total SRN time. We found no evidence
that use of SRN leads to higher hospital costs.
Implications for Policy, Delivery or Practice: Use
of SRNs could be one strategy for hospitals to
supplement nurse staffing without experiencing
negative economic consequences.
Funding Source(s): RWJF
Poster Number: 278
Recruiting and Retaining Primary Care Physicians
in Urban Underserved Communities: The
Importance of Having a Mission to Serve
Presented by: Kara Odom Walker, M.D., M.P.H.,
M.S.H.S., Robert Wood Johnson Clinical Scholar,
University of California, Los Angeles, 911 Broxton
Avenue, 3rd floor, Los Angeles, CA 94110; Phone:
(215) 292-4963; Email: kodom@mednet.ucla.edu
Co-authors: Gery Ryan, Ph.D.; Robin Ramey,
M.S.; Felix Nunez, M.D., M.P.H.; Robert Splawn,
M.D., M.P.H.; Arleen Brown, M.D., Ph.D.
Research Objective: Improving the supply of the
physician workforce in the safety net must remain a
national priority. In Los Angeles County, however,
recent funding cuts and hospital closures have
reduced the number of providers serving minority,
underserved and uninsured populations. Working
with a community advisory board (CAB), comprised
of researchers, medical educators, physician
leaders, and community clinic administrators, our
objective was to identify strategies to enhance
recruitment and retention of primary care physicians
in urban underserved settings.
Study Design: We conducted a community based
participatory research qualitative study through indepth, semi-structured interviews. We recruited
participants using snowball sampling, starting with a
first wave of referrals from the CAB followed by 2
waves of referrals from interviewed physicians.
Research team members trained in qualitative
interviewing techniques used a standard interview
guide. Transcriptions were reviewed independently
by two research team members and coded into
themes by the investigators using pile-sorting
methods (intercoder reliability, Kappa=82.6%).
Population Studied: Forty-two non-Latino White,
African American and Latino physicians who
reported their specialty as primary care (internal
medicine, family medicine or pediatrics) and
practiced over 20 hours/week in a clinical setting,
were equally stratified into underserved and nonunderserved areas, defined by a census tract or
comprehensive clinic designated as a Health
Profession Shortage Area within Los Angeles
County.
Principal Findings: The 42 participating primary
care physicians ranged in age from 31 to 73 years
(mean age 48 years, SD=12); 45% were female;
56% had no educational debt; and 21% had loan
repayment obligation. By specialty, 53% were in
internal medicine, 30% were in family medicine,
17% were in pediatrics. Three major themes
emerged in relation to selection of current
geographic and population based practice
decisions: 1) personal motivators, 2) career
motivators, and 3) clinic support. We found that
themes describing personal motivators (e.g.
personal mission and self-identify) for choosing a
practice were more common among physicians who
worked in underserved areas than those who did
not. In contrast, physicians in non-underserved
areas were more likely to cite work hours and
lifestyle as reasons for selecting their current
practice location or leaving an underserved area.
Regardless of their race or ethnicity, the majority of
physicians who practiced in underserved areas
reported feeling a unique connection to the
particular community in which they practiced.
Notably, none of the physicians who trained in a
non-underserved setting went to work in an
underserved setting underscoring the importance of
training in underserved locations.
Conclusions: Ultimately, a concerted health care
workforce policy can address distribution disparities
in underserved areas and comprehensively address
challenging issues such as benefits, salary,
lifestyle, work schedule, and physician specialty
distribution. Enlightened and informed recruitment
strategies should seek out, develop and train a
corps of motivated, mission-driven, and committed
primary care physicians and retaining them by
employing strategies to improve work/life balance.
Implications for Policy, Delivery or Practice: The
current health care reform debate provides unique
opportunities to develop and implement policies that
support both a larger primary care workforce and
incentives that encourage physicians to practice
into shortage areas.
Funding Source(s): RWJF, University of California
Los Angeles Resource Center for Minority Aging
Research/Center for Health Improvement of
Minority Elderly (RCMAR/CHIME) under NIH/NIA
Grant P30AG021684
Poster Number: 279
Workforce Challenges in Delivering Healthcare to
Elderly and Low-Income Populations: Wyoming
Medical Providers’ Acceptance of Medicaid and
Medicare Patients
Presented by: Davis Patterson, Ph.D., Research
Scientist, Department of Family Medicine, WWAMI
Center for Health Workforce Studies/Rural Health
Research Center, 4311 11th Avenue NE, Suite
210B, Seattle, WA 98105; Phone: (206) 543-1892;
Email: davisp@uw.edu
Co-authors: Susan Skillman, M.S.; C. Holly
Andrilla, M.S.; Mark Doescher, M.D., M.S.P.H.
Research Objective: National studies show
differences in healthcare providers’ acceptance of
patients according to type of insurance. Wyoming, a
highly rural state, offers an intriguing case study of
state and federal healthcare policies targeting low
income and elderly patients. EqualityCare,
Wyoming’s Medicaid program, offers “Pay for
Participation” provider incentives and reimburses
providers at high levels compared with Medicare
and other states’ Medicaid programs. EqualityCare
also covers a narrower segment of its low-income
population than do many other states. This study
sought to (1) determine the extent to which
Wyoming medical professionals currently serve
Medicaid and Medicare enrollees, (2) identify
factors associated with provider acceptance of new
patients, and (3) consider policy options for
strengthening the health workforce and patient
access to high-quality care.
Study Design: Cross-sectional surveys of all
Wyoming healthcare facilities and multiple types of
licensed professionals, using contact lists
maintained by state licensing boards, were
conducted in late 2008 and early 2009 via postal
mail, internet, fax, and e-mail.
Population Studied: This study analyzed survey
data (response rates in parentheses) from 261
medical clinics (83%), and three types of providers:
561 physicians (56%, including osteopathic
physicians), 108 physician assistants (62%), and
132 advanced practice nurses (56%).
Principal Findings: Wyoming ambulatory care
providers accepted both Medicaid and Medicare
patients more readily than did U.S. physicians as a
whole. More Wyoming ambulatory care providers
reported accepting at least some new Medicaid
patients (96%) than Medicare patients (91%). This
pattern is reversed among all U.S. physicians (72%
v. 86% in 2008). Clinics that accepted new
Medicaid patients (mean of 12.4 staff FTEs) were
larger than those that did not (5.2 FTEs).
Specialists were more likely than primary care
providers, and rural were more likely than urban
physicians, to accept new Medicaid and Medicare
patients. In clinics that served more rather than
fewer Medicare patients, new patients waited longer
for an appointment.
Conclusions: Providers’ higher rates of accepting
new Medicaid patients, relative to both national
rates and Medicare acceptance rates in Wyoming,
provide preliminary evidence that Wyoming’s
EqualityCare policies have had a positive effect.
Providers vary significantly, however, and
systematically, in their acceptance of both types of
insurance.
Implications for Policy, Delivery or Practice:
Wyoming offers insights into the effectiveness of
state and federal policies to insure low-income and
elderly populations, particularly relevant to states
with significant rural populations. As Medicaideligible populations grow in an economic downturn,
maintaining coverage levels is a formidable
challenge for declining state budgets. Smaller
clinics, particularly those that are the sole
healthcare providers in a community, may require
special incentives to ensure Medicaid patient
access. Meanwhile, as elderly populations grow in
Wyoming and elsewhere, healthcare access may
become increasingly limited without changes to
Medicare reimbursement policies, such as the
Dorgan-Conrad amendment to health reform
legislation. Providers in isolated rural communities
may feel socially or financially obligated to accept
all patients, contributing to the higher rural Medicaid
and Medicare acceptance rates found in this study.
Further stress to reimbursement in rural areas due
to an increasing proportion of Medicare patients or
other factors ultimately could exacerbate shortages
of rural primary care providers.
Funding Source(s): Wyoming Healthcare
Commission
Poster Number: 280
Policy Implication of the Narrowing Scope of
Primary Care Practice
Presented by: Stephen Petterson, Ph.D., Sr
Health Policy Researcher, Robert Graham Center,
1350 Connecticut Avenue, Washington, DC 20036;
Phone: (202) 331-3360; Email:
spetterson@aafp.org
Co-authors: Nicole Johnson, M.D.; Jason Rinaldo,
Ph.D.
Research Objective: As health care reform
focuses on primary care as the solution to the
current national health care climate, policy will
declare which services are eligible for primary care
incentive reimbursements. However, the proposed
services per the Medicare Payment Advisory
Commission (MedPAC) may be too restrictive.
Squeezed on one side by nurse practitioners and
physician assistants and on the other side by
subspecialists, the de facto scope of practice of
U.S. primary care physicians has narrowed
substantially in past decades. MedPAC’s proposals
to limit a definition of primary care to Evaluation and
Management CPT codes may worsen this trend. To
inform current policy discussions, this study
presents evidence about the range and variation
the practices engaged in by family physicians
(FPs).
Study Design: We use questions about the percent
of time providing emergency care, urgent care,
major surgery, maternity care, office surgery, pain
management, palliative care, pre- and post-op care,
and hospital care. Diplomates were also asked
whether they provided prenatal care, care of
newborns and the average annual number of
deliveries. Our independent variables include
practice structure (solo, group, partnership, other),
years in practice, gender and a four-category
measure of rurality (urban, large rural, small rural,
isolated rural and rural frontier) based on the RuralUrban Commuting Area (RUCA) classification.
Methods: STATA 9.2 was used to obtain simple
descriptive statistics and multivariate logistic
regression models.
Population Studied: 2009 ABFM Administrative
data. ABFM obtains detailed practice characteristics
from FPs. Starting in 2005, this information is
available from nearly every diplomate taking the
certification exam (n=26,186).
Principal Findings: FPs are engaged in a wide
range of activities, but any particular activity
accounts for a relatively small percentage of their
work week. One exception is urgent and emergency
care, due mainly to a small fraction of diplomates
(5%) who devote nearly 100% to either type of care.
For instance, while about 40% of FPs devote some
time to office surgery, on average this amounts to
less than 1.5% of their time. The multivariate results
show considerable variation across practices. FPs
in rural areas are substantially more likely to
perform procedures—major and minor surgery as
well as delivering babies—and are more likely to
practice emergency care, pain management and
palliative than their urban counterparts. Female FPs
are significantly more likely practice maternity and
newborn care than male FPs, but less likely to
perform procedures. Physicians with fewer years of
practice are significantly less likely to deliver babies
or perform procedures in their office. They are
significantly more likely to practice urgent care.
Conclusions: Reform measures that attempt to
define primary care via claims-based definitions of
scope risk harming some providers absent
knowledge of the current breadth and scope of
practice. This is particularly true for rural providers
whose broad scope includes procedural care not
currently included in reform bill definitions.
Implications for Policy, Delivery or Practice: The
disparity between MedPAC’s defined scope of
primary care and those proposed by the
professional societies that comprise and represent
primary care physicians must be reconciled. Policy
that allows exemptions, lower thresholds, or
considers a broader scope of practice of primary
care services must be considered as health care
reform unfolds.
Funding Source(s): HRSA, ABFM
Poster Number: 282
General Internal Medicine and Medicare Spending
Variation
Presented by: Stephen Petterson, Ph.D., Senior
Health Policy Researcher, Robert Graham Center,
1350 Connecticut Avenue Suite 201, Washington,
DC 20036; Phone: (202) 331-3360; Email:
spetterson@aafp.org
Co-authors: Robert Phillips, M.D., M.S.P.H.;
Bridget Teevan, M.S.
Research Objective: Healthcare costs for people
naming a general internist as their usual source of
care are significantly higher compared to those with
a family physician (FP), and a much higher
proportion of general internists would not be eligible
for proposed primary care bonuses. This study
sought to understand their relationship to variations
in healthcare spending.
Study Design: Dartmouth data for hospital referral
region (HRR, n=307) and primary care service
areas (PCSA, n=6542) allow ecologic observation
of relationships between clinicians, utilization and
costs. At the PCSA level, we examined the
relationship of Part A Medicare costs to quintiles of
per capita general internists and FPs and to
quintiles of per capita non-primary care physicians.
For HRRs, we examined the association of nonprimary care specialists with general internists and
FPs; and the association of Medicare expenditure
with general internists and FPs. We also examined
the association between primary care as a
percentage of the physician workforce and growth
in Medicare spending between 2000 and 2006
using least-squares linear regression, weighting
observations by HRR 2000 population, and
adjustments for baseline expenditures and potential
ecologic confounders, and finally this association in
HRRs with more non-primary care specialists.
Population Studied: General internal medicine
physicians and family physicians.
Principal Findings: Within PCSAs, Medicare Part
A costs per beneficiary increased as general
internists increased, regardless of non-primary care
physician count; whereas, costs per beneficiary
declined as FP counts increased, particularly in
areas with more specialists. In the HRR analysis,
there was a strong positive association between
non-primary care specialists and general internists
(r=.848, p<.001), but a negative association with
FPs (r=-.234, p<.001). There was a significant
positive association between Medicare spending
(Part A and B) and availability of general internal
medicine (GIM) within HRRs (r=.404, p=<.001), but
a significant negative association for FPs (r=-.593,
p<.001). There is a strong association between the
percentage of the total physician workforce that is
primary care in an HRR and supply of FPs (.767,
p<.001) Adjustments for baseline expenditures and
ecologic variables made the cost-reducing effects of
primary care growth insignificant, except in regions
with more specialists where a 10% increase in the
percent primary is associated with a 1.06%
decrease in the annual growth rate (p=.056).
Conclusions: Ecologic analyses are consistent
with patient- and physician-level analyses,
suggesting that the general internists’ role may
contribute to cost variation. General internists are
more likely to locate like non-primary care
specialists, but this does not explain the ecologic
association with increased costs. The reduction in
Medicare spending associated with growth of
primary care, particularly in areas with more
specialists, is largely explained by family medicine.
Implications for Policy, Delivery or Practice:
Differences in cost outcomes for general internal
medicine and family medicine is beginning to blur
the cost savings previously attributed to primary
care. The remarkable difference compared to family
physicians begs further studies of practice
differences. One suggestion is that internal
medicine training may lead to practicing to specialty
epidemiology and practice patterns rather than
those more typical to primary care.
Poster Number: 281
Medical Directors’ Evaluation of Physician
Workload, Organizational Problems and
Organizational Climate in Hospitals: The Results of
a Survey of 551 German Hospitals
Presented by: Holger Pfaff, Ph.D., Professor,
Medical School & Human Sciences Faculty,
University of Cologne, Eupener Straße 129, Köln,
Germany 50933; Phone: +004922147897100;
Email: holger.pfaff@uk-koeln.de
Co-authors: Antje Hammer, M.A.; Nicole
Ernstmann, Ph.D.; Oliver Ommen, Ph.D.; M.D.,
M.P.H.
Research Objective: To assess, how a hospital’s
top management, especially medical directors, see
the current state of physician workload,
organizational problems and organizational climate
in German hospitals.
Study Design: In 2008, a cross-sectional
representative study was conducted. Among the
scales included in the standardized written
questionnaire were “physician workload” (6-item
scale), “organizational problems” (6-item scale),
and “organizational climate” (4-item scale). For
each scale, respondents were given four response
options, with scores ranging from 1 (do not agree)
to 4 (completely agree).
Population Studied: 1224 medical directors from
all hospitals in Germany with at least one internal
medicine and one surgery unit. A total of 551
medical directors filled out the questionnaire, giving
a response rate of 45.2%.
Principal Findings: The average score on the
"physician workload” scale was 19 points (range: 6–
24 points). For example, 90% of the medical
directors surveyed agreed with the statement that
physicians are required to work at very high speed
and are often put under a lot of time pressure. In
addition, over 65% felt that physicians' work
involves a great deal of physical strain. The
average score on the "organizational problems”
scale was 13 points (range: 6–23 points). Over 35%
of the medical directors agreed with the statement
that the patient admission process usually results in
organizational problems and the rescheduling of
patient check-ups at the hospital. 50% of the
medical directors also indicated that hospital checkups often involve long waiting times. Over 20% of
those surveyed felt that there are communication
problems between the hospital wards and between
physicians and nursing staff. The average score on
the "organizational climate" scale was 11.3 points
(range: 4–16). Over 30% of the medical directors
surveyed disagreed with the statement that there is
a sense of unity and understanding in the hospital
and over 20% disagreed that there is a sense of
mutual trust, or a "we" feeling, among employees in
the hospital. Furthermore, approximately 19%
considered the overall work climate in the hospital
to not be good.
Conclusions: The results of the survey show that
the medical directors see hospital physicians as
being subjected to a heavy workload. Additionally,
the medical directors surveyed also find hospitals'
internal operations to have significant shortcomings,
especially with regards to appointment scheduling
and waiting times. Nearly one-fifth of those
surveyed considered the essential collaboration
between individual wards and between physicians
and nursing staff to be at an insufficient level.
Approximately 20% also found the hospitals’
organizational climate to be deficient.
Implications for Policy, Delivery or Practice: The
results of this study provide us with an initial
assessment of physician workload, organizational
problems and organizational climate in hospitals
from the eyes of medical directors throughout
Germany. Since medical directors are key persons
in the health care system, their assessment is
highly relevant to policy and practice in hospitals.
The results of this survey should therefore be taken
seriously.
Funding Source(s): German Medical Association
Poster Number: 283
Employment Options for Low-Skilled Workers in
Baltimore: Workforce Training Issues and
Challenges
Presented by: Luis Pinet-Peralta, M.Sc.,
Research Analyst, Health Policy Research Center,
Optimal Solutions Group LLC, 5825 University
Research Court, Suite 2800, College Park, MD
20737; Phone: (310) 918-5038; Email:
lpinet1@umbc.edu
Co-authors: Mark Turner, Ph.D.; Christabel
Dadzie, M.I.A.
Research Objective: The objective was to identify
viable occupations for low-skilled workers and offer
alternatives in the Baltimore Metropolitan Statistical
Area.
Study Design: This was a retrospective,
descriptive study using Standard Occupational
Classification System data from the Bureau of
Labor Statistics and Baltimore metropolitan
statistical area employment data from the U.S.
Census. Using a set of inclusion and exclusion
criteria, we performed a 3-stage analysis that
included an evaluation of occupational clusters, risk
analysis and semi-structured interviews.
Occupations were evaluated based on net job
openings (availability), education requirements for
entry into the occupation (accessibility), income and
wage growth (advancement potential) for each
occupational cluster. Risk analysis followed the
methodology previously used on a national study
and focused on outsourcing, technological
innovation and industry decline for each
occupational cluster. Lastly, a qualitative
assessment was performed using semi-structured
interviews, which were arranged with 25 of 46
potential key stakeholders, including local leaders,
educators and employers, for a response rate of
70%. The interviews were conducted between the
months of June and August 2008.
Population Studied: Baltimore Metropolitan
Statistical Area according to the U.S. Census
Bureau
Principal Findings: Ten professional occupations,
concentrated in six employment clusters were
identified, with average earnings of $46,874.00,
wage growth of $45,486.00 and 3,029 available
positions. Most entry-level occupations require
high-school education and offer opportunities for
professional development. Employer-based and
school-based training is available for all
occupations. Soft skills, science, reading and
writing can enhance growth opportunities and
upward mobility.
Conclusions: The occupations identified provide
reasonable entry wages and wage growth potential,
are accessible and have career advancement
options. The emerging green-job industry may also
offer viable opportunities, particularly in the
construction and utility sectors, with appropriate
training programs.
Implications for Policy, Delivery or Practice:
Policies meant to assist low-skilled workers gain the
needed hard and soft skills necessary to secure
better work need to address both the technical and
the social deficiencies that represent barriers to
better employment. Social deficiencies are more
difficult to address because of the need for more
inter-network collaboration, social participation and
political action. The challenge remains to integrate
specific local policy interventions with
comprehensive workforce reform and, more
importantly, evaluate the effects these policies have
on labor markets for low-skilled workers in the
United States.
Funding Source(s): The Ford Foundation
Poster Number: 284
Institutional Volume Shifting of Pediatric Surgical
Care in North Carolina, 1997-2006
Presented by: Stephanie Poley, B.A., Research
Associate, Health Policy Research Institute,
American College of Surgeons, 725 MLK Jr.
Boulevard, CB # 7590, Chapel Hill, NC 275997590; Phone: (919) 843-7661; Email: stephanie_
poley@unc.edu
Co-authors: Daniel von Allmen, M.D.; Thomas
Ricketts, Ph.D., M.P.H.; Elizabeth Walker, M.S.P.H.
Research Objective: In recent years, concern has
intensified regarding the adequacy of the U.S.
pediatric surgical workforce. Pediatric surgeons
have noted increasing case volume, which may be
a result of general surgeons in community hospitals
avoiding pediatric cases due to concerns about
liability, lack of confidence due to by minimal
exposure to pediatric cases during the general
surgical residency training, or pressure to take on
more “bread and butter” adult cases due to an
emerging general surgery workforce shortage. This
study examined North Carolina inpatient discharge
data between 1997 and 2006 to identify trends in
pediatric surgical utilization, to test the hypothesis
that pediatric surgical cases are becoming
concentrated in that specialty.
Study Design: Discharges with one of 934 ICD-9
procedure codes considered general surgery cases,
for patients under 19 years old were examined.
Records were summarized and analyzed by
institution, provider, and county of patient origin.
Discharge data were linked to files containing
information about institutions and providers using
the facility ID and UPIN. Descriptive and
multivariate analyses were performed to examine
longitudinal trends in the characteristics of patients
and type of procedures associated with facilities
and providers.
Population Studied: Patients under 19 years old
presenting at a North Carolina hospital between
1997-2006 requiring a surgical procedure.
Principal Findings: Throughout the study period,
tertiary care facilities accounted for just over half of
the state’s pediatric surgical discharges. Between
1997 and 2006, pediatric surgical case volume
increased by 11% overall, with a concentration in
the State’s six tertiary care facilities. Surgical
discharges for all ages increased by 31% in the
same period, with no statistically significant
difference in location in tertiary versus community
hospitals. Tertiary care facilities experienced an
18% increase in pediatric surgical cases over the
ten-year period, while the 109 community hospitals
discharged only 5% more pediatric surgical cases in
2006 than in 1997. Appendectomies accounted for
nearly one-third of pediatric surgical procedures in
community hospitals, while tertiary care facilities
had a broader mix of procedures. Eighty-eight
percent of procedures in community hospitals were
therapeutic, whereas more than 20% of all pediatric
surgical procedures in tertiary care facilities were
diagnostic. Patients in tertiary care hospitals were
younger and the average length of stay and total
charges were higher than in community hospitals.
Conclusions: Between 1997-2006, pediatric
surgical cases became more concentrated in large
tertiary care facilities.
Implications for Policy, Delivery or Practice:
Documenting the changes in pediatric surgical
operative volume for facilities and pediatric
surgeons is important for workforce and facility
planning. Volume shifts observed in this study are
unlikely to be isolated to North Carolina. Pediatric
surgical residency programs may need to expand
training slots to meet future demand for services.
Further, general surgery residency programs may
need to consider changes in the volume and/or
content requirements for pediatric surgical cases
during residency in order to prepare new trainees to
confidently and appropriately treat pediatric
patients. Additionally, significant changes in the
volume of pediatric surgical cases seen in tertiary
care (training) facilities may have implications for
surgical training programs if the trend affects
training capacity or quality.
Funding Source(s): American College of Surgeons
Poster Number: 285
The Impact of Resident Duty Hour Reform on
Hospital Readmission Rates
Presented by: Matthew Press, M.D., Clinical
Scholar, Robert Wood Johnson Foundation Clinical
Scholars Program, University of Pennsylvania, 13th
Floor, Blockley Hall, 423 Guardian Drive,
Philadelphia, PA 19104-6021; Phone: (401) 3741928; Email: matthew.press@uphs.upenn.edu
Co-authors: Jeffrey Silber, M.D., Ph.D.; Amy
Rosen, Ph.D.; Patrick Romano, M.D., M.P.H.; Yanli
Wang, M.S.; Kevin Volpp, M.D., Ph.D.
Research Objective: A key goal of resident duty
hour reform by the Accreditation Council for
Graduate Medical Education (ACGME) in 2003 was
to improve patient outcomes. Prior research has
demonstrated little to no impact of this reform on a
variety of outcome measures. The effect on hospital
readmission rates, which may have been
particularly influenced by disruptions in continuity of
care that resulted from the reform, is unknown. The
objective of this study was to assess whether
resident duty hour reform led to a relative
improvement or worsening of readmission rates
among Medicare patients in hospitals of different
teaching intensity.
Study Design: Observational study using
difference-in-differences analysis with hospital
discharge data from July 1, 2000 to June 30, 2005.
Fixed effects logistic regression was used to
examine the change in the odds of readmission in
more versus less teaching-intensive hospitals
before and after the duty hour reform, adjusting for
patient comorbidities, age, sex, admission source,
common time trends, and hospital. Primary
outcomes were 30-day all-cause readmission, allcause readmission or death within 30 days of
discharge, and all-cause readmission or death
within 30 days of admission.
Population Studied: All unique Medicare patients
(n=8,282,802) admitted to short-term acute-care
nonfederal hospitals with principal diagnoses of
acute myocardial infarction, congestive heart
failure, gastrointestinal bleeding, stroke or a DRG
classification of general, orthopedic, or vascular
surgery.
Principal Findings: In medical and surgical
patients, no significant relative increases or
decreases in the odds of readmission were
observed in more versus less teaching-intensive
hospitals in either the first or second year following
the reform. For the combined medical group,
postreform year 1 odds ratio (OR) = 0.99, 95%
confidence interval (0.94-1.03), and postreform year
2 OR = 0.99 (0.95-1.04). For the combined surgical
group, postreform year 1 OR = 1.03 (0.98-1.08) and
postreform year 2 OR = 1.02 (0.98-1.07). The
findings for the other outcomes—-readmission or
death within 30 days of discharge and readmission
or death within 30 days of admission—-were
qualitatively similar. Our results remained
unchanged after the following stability analyses:
excluding patients admitted to hospitals in New
York State, expanding the inclusion criteria for
index admissions, adjusting for patients discharged
against medical advice, and removing the
adjustments for comorbidity, hospital transfer, and
nursing home admission source.
Conclusions: Teaching hospital readmission rates
neither improved nor worsened in association with
ACGME duty hour reform. These findings were
robust to the use of composite measures of
readmission and mortality and were insensitive to
changes in patient selection and alterations in
severity adjustment.
Implications for Policy, Delivery or Practice: Any
potential adverse consequences of resident duty
hour reform on continuity of care did not lead to
observable changes in readmission rates. This
finding adds to the evidence that the reform in 2003
did not generally improve or worsen patient
outcomes.
Funding Source(s): NHLBI
Poster Number: 286
Attitudes Toward Drug-Eluting Stent Use and
Distribution of Regulatory Focus Type among
Interventional Cardiologists in New York State
Presented by: Feng Qian, M.D., M.Sc., Ph.D.
candidate in Health Services Research and Policy,
Community and Preventive Medicine, University of
Rochester, 601 Elmwood Avenue, Box 644,
Rochester, NY 14642; Phone: (585) 275-9496;
Email: feng_qian@urmc.rochester.edu
Co-authors: Charles Phelps, Ph.D., M.B.A.;
Frederick Ling, M.D.; Edward Hannan, Ph.D.; Peter
Veazie, Ph.D.
Research Objective: The safety of drug-eluting
stent (DES) use was called into question in 2006.
Literature in psychology suggests that regulatory
focus type (promotion vs. prevention) orients
physicians toward using different strategies to
approach a task: ensuring against errors of
omission (promotion) vs. ensuring against errors of
commission (prevention). However, the attitudes
toward DES use after safety concerns were
expressed and the distribution of regulatory focus
type among interventional cardiologists are
unknown. The objectives of this study are: (1) to
examine the attitudes toward DES use among
interventional cardiologists in New York State
(NYS) after DES safety concerns were expressed,
and (2) to investigate the distribution of regulatory
focus type among these interventional cardiologists.
Study Design: We conducted a questionnaire
survey of interventional cardiologists in New York
State from October 2008 to April 2009 by mail,
email and fax. The questionnaire included face valid
items to measure the attitudes toward DES use,
valid Regulatory Focus Questionnaire to measure
the regulatory focus type, and items collecting
demographic information.
Population Studied: Interventional cardiologists in
New York State.
Principal Findings: We received total 119 valid
returned questionnaires and the response rate was
47%. There are no statistically significant
differences regarding the demographic factors
between the respondents and the non-respondents.
The vast majority of interventional cardiologists
(92%) agreed or strongly agreed that DES is a
revolutionary technology and that DES use will
increase in the future (70%). The regulatory focus
type of interventional cardiologists was
predominantly of the promotion type (89%). We
found that age and gender were significant
predictors of the promotion score, but we did not
find any demographic factors were significant
predictors of the prevention score.
Conclusions: Despite the DES safety issues
reported in 2006-2007, interventional cardiologists
have a very positive attitude regarding DES use
and most of them predict the continual growth of
DES use in the future. Such information is valuable
in predicting the modality change in coronary
revascularization therapy. Furthermore, the vast
majority of interventional cardiologists were found to
be promotion focused, which means that they are
concerned about achieving positive outcomes and
they want to prevent errors of omission.
Implications for Policy, Delivery or Practice: To
the best of our knowledge, this is the first study to
report the distribution of regulatory focus type
among physicians. As suggested by the theory in
psychology, this finding can lead to more effective
interventions in conveying medical messages to
physicians and changing their behavior.
Poster Number: 287
A Research Agenda for the Physician Assistant
Profession
Presented by: Christal Ramos, M.P.H., Data
Research Analyst, Data Management and
Research, American Academy of Physician
Assistants, 950 N. Washington Street, Alexandria,
VA 22314; Phone: (703) 836-2272; Email:
cramos@aapa.org
Co-authors: Sabrina Smith, D.H.A.; Bill Leinweber,
M.B.A.
Research Objective: To outline a research agenda
for the physician assistant (PA) profession.
Study Design: The American Academy of
Physician Assistants (AAPA) will host a research
summit on March 4-5, 2010. Research summit
participants will work in groups by topic of expertise
to identify and reach consensus on a framework for
the research agenda. After the summit, AAPA, its
board of directors, and key participants from the
research summit will work within the framework
developed at the summit to finalize strategies and
specific items for the research agenda. AAPA will
gain buy-in on the agenda from the three other PA
organizations (The National Commission on the
Certification of Physician Assistants, the Physician
Assistant Education Association, and the
Accreditation Review Commission on Education for
Physician Assistants).
Population Studied: This research summit will
bring together representatives from practicing PAs,
the PA academic and regulatory communities, PAs
engaged in research, health services researchers,
health economists, thought-leaders from federal
agencies, industry, think tanks, and private
foundations that have a demonstrated commitment
to health care delivery and health policy research.
Principal Findings: AAPA anticipates the research
agenda will address:
1. Key questions regarding the PA profession such
as: (a) The contributions and impact of the PA
profession on health care broadly, on specific
health outcomes and on providing care for
vulnerable populations (b) The positioning of the
profession on the health care delivery team (c) The
current and future workforce needs of the
profession (d) How the changing health care
landscape will impact the profession (e) Utilization,
cost and financing issues pertaining to the PA
profession. 2. Core strategic, specific, measurable,
achievable, realistic and time-specific (SMART)
goals. 3. Individual and institutional stakeholders
within the PA community and beyond, that have a
strong interest in and motivation for advancing and
supporting a research agenda for the profession.
Conclusions: The research agenda will outline key
issues and research strategies to inform the future
of the PA profession in American health care.
Implications for Policy, Delivery or Practice: In a
time where our health care system is in the midst of
reform, PAs have the potential to be part of the
solution to improve quality, cut costs, to serve an
increasingly diverse and aging population, and to fill
shortages. As the country seeks to open access to
care through the expansion of insurance coverage,
the need for health care providers will increase as
well. To ensure our nation’s health workforce meet
our population’s needs, there must be an organized
effort to collect and analyze data to inform
workforce policies. The research agenda will serve
as the first step towards this type of organized effort
for the PA profession.
Poster Number: 288
Physician Location Decisions Across States
Between 2000 and 2007
Presented by: Charles Roehrig, Ph.D., Vice
President, Altarum Institute, 3520 Green Court,
Suite 300, Ann Arbor, MI 48105; Phone: (734) 3024646; Email: charles.roehrig@altarum.org
Research Objective: There is a growing concern
among states about looming physician shortages,
both in primary care and selected specialties. This
has led to the development of strategies to increase
physician supply, including expanding in-state
medical school enrollment capacity, increasing instate residency training positions, and taking steps
to increase retention of physicians trained within the
state. Our research objective is to examine the
broad pattern of physician state-level location
decisions between 2000 and 2007 and determine
the extent to which they are driven by changes in
state-level physician demand as opposed to other
factors that can be influenced by states.
Study Design: We use AMA Masterfile data to
determine each state’s share of active physicians
by specialty in 2000 and 2007. We estimate each
state’s share of physician demand by specialty for
these two years using state-level population by age
and sex combined with specialty-specific per capita
physician demand estimates adjusted for age and
sex. We then use regression models to estimate the
extent to which changes in a state’s share of the
national physician supply are determined by
changes in that state’s share of physician demand.
We also model the extent to which other factors
influence changes in state physician shares.
Population Studied: Physicians actively practicing
in the U.S. as of December 31, 2000 and December
31, 2007.
Principal Findings: Between 2000 and 2007,
Texas and Arizona saw the largest gains in their
share of physicians, while New York and New
Jersey had the largest losses. Our initial regression
models use changes in state shares of population,
a primary determinant of physician demand, as a
substitute for changes in state physician demand.
The coefficient on the change in demand shares is
.88; thus, the change in supply closely tracks the
change in demand. We also find evidence that
states with relatively high physician-to-population
ratios at the start of the analysis period, and states
with older physician populations, tend to have a
lower supply share in 2007 than what would be
predicted based upon changes in demand. We are
currently increasing the precision of our regression
results by replacing the change in state population
with the change in actual physician demand, and
examining the impact of changes in state shares of
medical school graduates and residency positions.
Conclusions: These findings suggest that the
primary determinant of changes in state shares of
the national supply of physicians is the change in
state shares of demand for physicians. Our results
also suggest that states that are richly supplied with
physicians are slowly regressing toward the mean,
and that states with relatively older physician
populations may be lagging in their replacement of
those that retire.
Implications for Policy, Delivery or Practice:
While states may be attempting to influence their
share of the national physician pool, physicians
move to states where demand for their services is
growing. This finding clearly informs state policy
decisions regarding the location or expansion of
medical schools and residency training positions.
Funding Source(s): HRSA
Poster Number: 289
Factors Related to the Recruitment and Retention
of Primary Care Physicians at Community Health
Centers: A Statewide Physician Survey
Presented by: Judith Savageau, M.P.H.,
Epidemiologist/Biostatistician, Center for Health
Policy and Research, Univniversity of
Massachusetts Medical School, 333 South Avenue,
Shrewsbury, MA 01545; Phone: (508) 856-4333;
Email: judith.savageau@umassmed.edu
Co-authors: Linda Cragin, M.S.; Donna Johnson,
L.I.C.S.W.; Joan Pernice, R.N., M.S.; Joan Bohlke,
M.S.; Warren Ferguson, M.D.
Research Objective: Massachusetts ranks sixth
nationally for the number of federally-funded
community health centers (CHCs), with 34 located
in medically underserved communities. 52 CHCs,
including hospital-licensed, provide care at 185+
sites, making CHCs a significant part of the
Massachusetts health care system. When
Massachusetts implemented universal health
insurance coverage in 2006, CHCs were on the
primary care frontline. National news reported
success in insurance coverage yet highlighted the
exacerbation of primary care shortages as many
community-based physicians closed their practices.
Our objective was to characterize the
Massachusetts CHC primary care physician (PCP)
workforce and identify factors related to
preparedness, recruitment and retention.
Study Design: The University of Massachusetts
Medical School, Massachusetts League of
Community Health Centers (MLCHC), and
Massachusetts Department of Public Health
developed and implemented an on-line crosssectional survey of PCPs at MLCHC sites.
Population Studied: 572 physicians (58%
response rate) practicing in 46 CHCs.
Principal Findings: CHC PCPs are predominantly
female (59%), white (75%), non-Hispanic (88%),
work full-time (72%), speak at least one additional
non-English language (61%), and have been in
practice an average of 13 years. 60% were hired
since 2000; 39% practice Family Medicine, 31%
Internal Medicine, and 27% Pediatrics. 16% are
foreign-trained medical graduates. 22% had postgraduate medical education debt over $100,000
and 9% currently have similar debt. 48% are current
or former visa/loan repayment program participants.
Our mixed-model regression analyses identifying
factors related to the likelihood of PCPs continuing
to work in a CHC environment and with
underserved populations in the next five years
noted that females, those practicing in more urban
areas and those in practice for 10+ years were
more likely to report plans to remain in a CHC.
PCPs who reported being very prepared to practice
in this setting at the end of residency training, that
the interview process was important, and that the
compensation package was not important in their
first consideration to join their CHC were also more
likely to report remaining in a CHC. Higher levels of
satisfaction with their current compensation,
however, and increased satisfaction with the CHC’s
mission, and the diversity of patients seen were
also significantly related to a higher likelihood of
remaining in the CHC setting. Finally, a higher
likelihood of retention was negatively correlated
with opportunities for teaching and research and
spousal job assistance.
Conclusions: With the nation’s focus on health
care reform and results from the reform efforts in
Massachusetts and its impact on primary care, our
findings provide insights into opportunities to better
prepare medical students and residents for careers
in CHCs and to recruit and retain this critical
workforce.
Implications for Policy, Delivery or Practice: For
any health care reform effort to succeed, access to
a strong and sufficient primary care workforce for
underserved populations is needed. PCP
compensation is identified as one key strategy;
equally important is addressing priorities related to
work and personal life that will contribute to the
retention of PCPs who plan to remain in a CHC and
those currently undecided about their future.
Funding Source(s): MassAHEC Network
Poster Number: 290
Do Internists and Family Practitioners Have
Differing World Views? Variations in the Diagnosis
and Management of the Same Patient with Heart
Disease
Presented by: Rebecca Shackelton, Sc.M.,
Associate Research Scientist, Health Services and
Disparities Research, New England Research
Institutes, 9 Galen Street, Watertown, MA 02472;
Phone: (617) 972-3119; Email:
rshackelton@neriscience.com
Co-authors: John McKinlay, Ph.D.; Lisa Marceau,
M.P.H.; Carol Link, Ph.D.
Research Objective: Variations in the processes
and quality of medical care in the U.S have been
extensively documented and are a major concern
for health services researchers and policy experts.
There are documented variations between medical
specialties. However, variations within a medical
specialty, such as differences between internists
and family practitioners within primary care, have
received considerably less attention. Internists and
family practitioners have been described as having
distinct “disease” perspectives with family
practitioners having a bio-psychosocial perspective
and internists having a patho-physiological
perspective of disease. This study aims to answer
the following questions: First: are there any
differences between internists and family
practitioners in their suggested diagnoses and level
of diagnostic certainty when encountering exactly
the same “patient” with coronary heart disease
(CHD)? Second, are there any noteworthy
differences in the management of disease between
internists and family practitioners?
Study Design: In two factorial experiments,
primary care providers were shown clinically
authentic filmed vignettes with actors portraying
“patients” with identical signs and symptoms of
CHD. The physician participants were asked how
they would diagnose and manage the case.
Population Studied: Internists (n=175) and family
practitioners (n=192) were randomly sampled from
throughout Massachusetts, North Carolina, and
South Carolina.
Principal Findings: Internists were more certain
about the correct diagnosis of CHD than family
practitioners and were more likely to act on this
diagnosis (order a stress test and prescribe beta
blockers). Family practitioners were more likely to
diagnose a mental health issue and were more
certain about this diagnosis.
Conclusions: Clear variations are evident in the
way internists and family practitioners would
manage the “patient” they encountered. Such
variations may reflect fundamental differences in
the “disease” model used by the two primary care
professional groups (bio-psychosocial vs. pathophysiological).
Implications for Policy, Delivery or Practice: The
question of training and socialization to a disease
model in diagnostic accuracy and certainty is an
essential, and often overlooked, element of medical
care.
Funding Source(s): National Heart, Lung, and
Blood Institute
Poster Number: 291
Washington State's Oral Health Workforce: Who
Provides Services Now and What is Needed for the
Future?
Presented by: Susan Skillman, M.S., Deputy
Director, Center for Health Workforce Studies,
University of Washington, 4311 11th Avenue NE,
Suite 210, Seattle, WA 98105; Phone: (206)5433557; Email: skillman@u.washington.edu
Co-authors: C. Holly Andrilla, M.S.; Joseli AlvesDunkerson, D.D.S., M.P.H., M.B.A.; Wendy
Mouradian, M.D., M.S.; Mark Doescher, M.D.,
M.S.P.H.
Research Objective: To describe all components
of the oral health workforce in Washington State,
the programs by which state residents receive oral
health care and preventive services, current and
future workforce gaps, and policy options to help
ensure an adequate future oral health workforce.
Study Design: Using data from surveys,
professional licensure records, educational
institutions, and a variety of oral health-related
organizations and programs in the state, the study
provides estimates of the size and distribution of
Washington’s oral health workforce; demographic,
education and practice characteristics of the
workforce; projections of state dental school
graduates required to maintain statewide supply to
2027; and discussion of the state’s oral health
workforce needs.
Population Studied: Providers of oral health care
services and oral health improvement programs in
Washington State.
Principal Findings: There are growing numbers of
dentists, dental hygienists, expanded function
dental hygienists, dental assistants, denturists,
medical providers (physicians, advanced practice
nurses and physician assistants), and supporting
caregivers who provide oral health services in
Washington. There remain, however, underserved
populations in much of the state. The size of the
workforce appears unlikely to keep pace with state
population growth, and some provider types will
decline in size as baby boomers retire. Expanding
dental school capacity by 25% will contribute to
growth of the state's supply of dentists, but not
enough to keep pace with the number of dentists
needed to maintain Washington’s current
dentist:population ratio. Because of projected
increases in state population, meeting this
benchmark would require having 24% more
licensed dentists in Washington in 2027 than were
licensed in 2009.
Conclusions: More providers will be needed (from
in-state educational programs as well as from inmigration) to increase Washington's oral health
workforce capacity in the future. There is great
need for providers who will care for underserved
populations and provide preventive services that
will reduce future oral health care demand
associated with the growth and aging of
Washington’s population.
Implications for Policy, Delivery or Practice:
Strategies for state health planners, policymakers
and educators to improve the distribution and
capacity of the oral health workforce include
promoting early interest in health careers, improving
the capacity of the workforce to care for
underserved and special populations, supporting
early prevention, supporting collaborative and interprofessional education and practice, and exploring
ways to expand the capacity and scope of services
provided by oral health care professionals.
Funding Source(s): State of Washington
Department of Health
Poster Number: 292
Projections of Washington State Licensed Practical
Nurse Supply and Demand: 2007-2026
Presented by: Susan Skillman, M.S., Deputy
Director, Center for Health Workforce Studies,
University of Washington, 4311 11th Avenue NE,
Suite 210, Seattle, WA 98105; Phone: (206) 5433557; Email: skillman@u.washington.edu
Co-authors: C. Holly Andrilla, M.S.; Davis
Patterson, Ph.D.; Linda Tieman, R.N., M.N.,
F.A.C.H.E.
Research Objective: This study estimates trends
in licensed practical nurse (LPN) supply and
demand for Washington State from 2007 through
2026.
Study Design: Using available data from state
licensing records and a survey of LPNs, the study
estimated future in-state LPN graduations, inmigration from other states, re-activation of
licensure after license expiration, deaths, agerelated retirements, career changes, and outmigration to other states. Future LPN demand was
estimated from state labor department projections.
Projections scenarios included estimates of both
increases and decreases to the current LPN
graduation rates.
Population Studied: Licensed and practicing LPNs
in Washington State.
Principal Findings: The largest single contributor
to LPN supply was graduations from in-state
education programs. A major factor causing exits
from supply was retirement of “baby boomers” from
the current workforce. If Washington’s LPN
education completion rates increase by 200 LPNs
in 2011 (nearly 20% more than the 2007 levels) and
this increase is sustained through 2026, the
projected supply of practicing LPNs will increase
over the next two decades but still be 2,300 LPNs
(14%) short of demand in 2026. None of the supply
scenarios keep pace with the number of LPNs
needed to maintain Washington’s 2008
LPN:population ratio: a benchmark for which
increasing numbers of LPNs are needed because
of projected increases in state population.
Conclusions: Demand for LPNs in Washington will
outpace the supply in the state over the next 20
years, even if very large increases to LPN
education capacity are implemented.
Implications for Policy, Delivery or Practice:
Rural locations, including those in Washington, are
likely to see greater shortages of LPNs because
their populations are older, on average, than urban
areas. This growing elderly population has greater
need for services, including long-term care, and
LPNs are the primary nursing providers in elder
care settings. Major changes in health care delivery
systems or the economic environment, however,
could alter the rate of increase in LPN demand. The
registered nurse and LPN workforces are closely
related, and growing RN demand may exacerbate
the LPN shortage if hiring practices shift to replace
RNs with LPNs. LPN education is a critical entry
point into the nursing profession, and clear
articulation paths between LPN education and other
nursing education programs could improve interest
in nursing overall. While encouraging LPNs to
pursue RN careers may shorten their tenure as
LPNs, the potential for professional growth may
encourage more people to enter nursing career
pathways as LPNs.
Funding Source(s): Washington Center for
Nursing
Poster Number: 293
Hospital RN Job Satisfaction and Nurse Unions
Presented by: Joanne Spetz, Ph.D., Professor,
Nursing, University of California San Francisco,
3333 California Street, Suite 410, San Francisco,
CA 94118; Phone: (415) 502-4443; Email:
jojo@thecenter.ucsf.edu
Lead Author: Jean Ann Seago, R.N., Ph.D.
Co-authors: Michael Ash, Ph.D.; Carolina Herrera,
M.A.; Dennis Keane, M.P.H.; Joanne Spetz, Ph.D.
Research Objective: Maintaining an adequate
nursing workforce has posed an ongoing challenge
to policy-makers and hospital managers. While
many studies have pursued the determinants of job
satisfaction and turnover among hospital-based
RNs, none have systematically examined the role of
labor unions. Labor unions that represent nurses
often argue that their advocacy is central to
addressing problems with nurse satisfaction and
retention. The broader industrial relations literature
suggests unionized workers are more willing to
express dissatisfaction and less likely to change
employers. This paper addresses this contradiction
and considers the role of unions in nurse
satisfaction and job seeking in the United States.
Study Design: We use data from the 2004 National
Sample Survey of Registered Nurses (NSSRN). We
limited our sample to RNs who worked in acute
care hospitals as their principal nursing
environment. We specified nurse satisfaction, jobseeking behavior, and union representation through
three NSSRN survey questions. The nursing
satisfaction question (“How would you best describe
your feelings about your principal nursing position?)
offer response options on a five-point Likert scale,
while job-seeking and union representation
responses were recoded to binary variables. We
take three approaches to analyzing the impact of
unionization on satisfaction and job seeking. First,
we estimate differences in socio-demographic
variables by union status and by RN satisfaction,
and conduct t-tests and non-parametric tests to
determine whether differences are statistically
significant. Second, we assess the possibility that
unions affect dissatisfaction and job seeking using
probit regressions with dichotomized satisfaction
(satisfied=1) and job-seeking as the outcome
variables. In our third analysis, we examine nursing
satisfaction and job seeking in the context of a
differentiated attrition model. Differential attrition
occurs when job turnover rates differ significantly
between two similar cohorts – in this case, between
unionized and non-unionized nurses. We stratify
those cohorts by age in a probit regression. In all
regressions, socio-demographic variables act as
controls.
Population Studied: United States nurses in 2004.
Principal Findings: The population of unionized
nurses in the United States in 2004 did not look like
the population of non-unionized RNs. Nurses in a
union had higher average incomes from their
principal nursing positions, worked fewer hours per
year and per week, were older, and were more
likely to be a minority. They were less likely than
non-unionized RNs to be married. They were
slightly more likely to report that their highest
nursing or nursing-related education was a
bachelor’s degree, and less likely to have a
graduate degree. RNs reporting higher levels of job
satisfaction were more likely to be to be married,
female, have higher incomes, and have graduate
degrees. T-Testing and non-parametric analysis
confirmed a negatively relationship between union
status and the distribution of satisfaction. After
controlling for other factors in our regression
analysis, union representation of all age groups was
not a significant predictor of satisfaction. When
testing the differentiated attrition model, we find the
cohort of older unionized nurses were significantly
more likely to be dissatisfied than younger or nonunionized nurses. Preliminary analysis of job
seeking suggested it is negatively correlated with
unionization; further analysis of job seeking
behavior is ongoing.
Conclusions: We find evidence that unions are
associated with lower nurse satisfaction
Implications for Policy, Delivery or Practice:
Nurse managers and administrators may find this
study useful in that there is no evidence that unions
either reconcile nurses to the workplace or foment
grumbling. This is in line with expectations about
the relationship between unions and satisfaction
outside of nursing. Nursing managers may want to
note along with intractable demographics such as
age and marital status, that lower income and
overwork were significantly associated with greater
dissatisfaction. This suggests that retention of
nurses may devolve to the hospital unit level, where
work load, management characteristics, and other
environmental factors can be altered by nurse
managers. Hospital administrators should evaluate
nursing-unit work environments on an individual
basis and alter characteristics that dissatisfy their
RNs.
Funding Source(s): AHRQ
Poster Number: 294
Teamwork, Experience and Temporary Registered
Nurses Impact on Patient Outcomes and Length of
Stay
Presented by: Patricia Stone, Ph.D., M.P.H.,
Professor of Nursing, School of Nursing, Columbia
Univeristy, 630 W. 168th Street, New York, NY
10040; Phone: (212) 305-1738; Email:
ps2024@columbia.edu
Co-authors: Ann Bartel, Ph.D.; Ciaran Phibbs,
Ph.D.
Research Objective: There is evidence on the
importance of registered nurse (RN) staffing in the
provision of quality care. However, little evidence
exists on the characteristics of the RN needed to
provide quality care. To have sufficient staffing
some hospitals hire a higher proportion of less
experienced and/or temporary RNs (i.e., those from
outside agencies). Guided by the theory of human
capital, we examined the impact various RN
characteristics had on patient outcomes and length
of stay (LOS). We chose to measure LOS as it
increases with many different adverse events and is
a measure of efficiency
Study Design: A longitudinal analysis (2003-2006)
of administrative data. RN characteristics included
the percentage of hours provided by agency
nurses, team overlap, unit specific tenure, prior
work experience and education. Patient outcomes
included the staffing sensitive patient safety
indicators (i.e., decubitus ulcer, failure-to-rescue,
infections, and post-op deep vein thrombosis) and
LOS. All variables were aggregated at the monthly,
unit level. Fixed effect, multivariate regressions
were conducted for both general and Intensive Care
Unit (ICU) settings; we controlled for level of
staffing, nursing skillmix (i.e., percentage of RN
hours over total nursing hours) and patient severity
of illness.
Population Studied: Nurses and patients from all
Veteran’s Administration (VA) hospitals.
Principal Findings: Using 11,637 monthly
observations from 266 general wards, the average
LOS was 4.9 days (s.d. = 3.7) in general wards and
3.0 days (s.d. = 1.5) in ICUs (7,148 monthly
observations from 161 ICUs). On average, in
general wards RN unit tenure was 4.0 years (s.d. =
1.3), years of prior experience was 3.4 (s.d. = 1.7)
and the percentage of hours of care provided by
agency RNs was 2.5 (s.d.= 6.0). In ICUs, RN unit
tenure was 5.0 (s.d. = 1.3), years of prior
experience was 5.3 (s.d. = 2.3) and the percentage
of hours of care provided by agency RNs was 1.6
(s.d. = 5). While some of the RN characteristics
were significantly associated with the patient safety
outcomes, the results were somewhat inconsistent.
We consistently found that lower use of agency
nurses, increased length of time RNs have worked
on the unit and the larger the overlap of these RNs
were all independently associated with decreased
LOS in both general wards and ICUs (all p values <
0.05).
Conclusions: The characteristics of RNs matter.
Implications for Policy, Delivery or Practice: Not
only do we need sufficient RNs in the workforce,
our findings point to the need for institution specific
policies. Hospitals should limit the use of contract
nurses and implement strategies to recruit and
retain nurses as well as implement exercises
designed to encourage RNs to stay with well
functioning teams (e.g., team building). A limitation
to this study is that it was conducted in VA
hospitals; however, because VA RNs have the
same credentials and licensing requirements as
non-VA nurses and the patient outcomes we are
studying are important for all hospitalized patients, it
is unlikely that the results would change in non-VA
hospitals. Nevertheless, more research is
warranted.
Funding Source(s): RWJF
Poster Number: 295
The Effect of Medical Malpractice Liability on
Access to Care
Presented by: Daniel Weinberg, Ph.D., Research
Associated, IMPAQ International, LLC, 10420 Little
Patuxent Parkway, Suite 300, Columbia, MD
21044; Phone: (443) 539-9765; Email:
dweinberg@impaqint.com
Research Objective: The purpose of this paper is
to estimate the effect of medical malpractice liability
on access to care measures such as ED utilization,
putting off care, having a usual source of care,
having unmet health care needs, and others. A
primary way in which liability may diminish access
is through reductions in provider workforce in highliability markets.
Study Design: I employ individual level access to
care data from four rounds of the Community
Tracking Study (CTS) along with metropolitan
statistical area liability measures (a panel at the
metro level) to estimate the effect of malpractice
liability on access to care. I use a fixed-effects
model with a rich set of individual-level covariates to
control for metropolitan area unobservables and
micro-level factors affecting access.
Population Studied: Individuals aged 18-64 living
in 47 metropolitan statistical areas in 34 states.
National estimates were obtained using the
appropriate weighting variables provided by the
CTS.
Principal Findings: Uninsured individuals
experience diminished access to care for some
access to care measures (ED utilization, preventive
care, mammography, putting off needed care, and
travel time) but not for others (flu shot, unmet
medical care needs, having a usual source of care,
and number of visits with a health care provider).
Access measures tend to be more sensitive to the
size of medical malpractice claims payments than
to the frequency per physician of such payments.
Possible explanations for diminished access
resulting from liability include reductions in
physician workforce and changes in physician
behavior such as rejecting high-risk patients,
refusing to perform high-risk procedures, and
focusing on high-profit patients whose insurers are
relatively generous payers. The evidence from this
paper suggests that changes in physician supply
are not the primary cause of the effect of liability on
access. Medicaid patients’ access measures are
much less sensitive to changes in malpractice
liability.
Implications for Policy, Delivery or Practice:
Policymakers whose goals include expanding
access to care among the uninsured (by, for
example, reducing unnecessary ED utilization and
increasing preventive care) might consider
expanding Medicaid since Medicaid patients’
access measures appear to be less sensitive to
changes in malpractice liability. Also, reducing the
size of malpractice settlement payments will likely
improve access among the uninsured. Caps on
noneconomic damage awards may help to
accomplish this goal.
Poster Number: 296
Certified Nurse Midwives, Nurse Practitioners and
Physician Assistants as Abortion Providers:
Preliminary Findings from the California Primary
Care Initiative
Presented by: Tracy Weitz, Ph.D., M.P.A.,
Associate Professor, ANSIRH, University of
California, San Francisco, 1330 Broadway, Suite
1100, Oakland, CA 94612; Phone: (510) 986-8939;
Email: weitzt@obgyn.ucsf.edu
Lead Author: Diana Taylor, R.N., Ph.D., N.P.;
Co-authors: Kristin Nobel, M.P.H.; Molly Battistelli,
B.A.; Erin Schultz, J.D.; Tracy Weitz, Ph.D., M.P.A.
Research Objective: Early abortion is one of the
safest medical procedures, with an average
complication rate between 1-5%. The 1.2 million
abortion procedures performed in the United States
annually are provided by only 178 facilities.
Consequently, many women have to travel long
distances to access abortion care, resulting in
unintended births and abortions at later gestational
age. Solutions to this access problem include
appropriate training and development of the
workforce. Advancing New Standards in
Reproductive Health (ANSIRH) initiated a Health
Workforce Pilot Project in California in 2007 to train
and evaluate whether certified nurse midwives
(CNM), nurse practitioners (NP) and physician
assistants (PA) as early aspiration abortion
providers can help fill the workforce gap. This
presentation reports selected findings from data
gathered from August 2007 to December 2009.
Study Design: This is a multi-site prospective
evaluation of CNMs, NPs and PAs trained to
competency in early aspiration abortion procedures
and followed over time. CNM/PA/NP data are
compared with published standards. Data are
collected on patient outcomes (safety, satisfaction,
access to care) and clinician outcomes
(competency, satisfaction) across the peri-abortion
care continuum.
Population Studied: Women seeking first trimester
aspiration abortion in 14 clinical location in
California.
Principal Findings: Preliminary results from 23
CNMs, NPs and PAs and 2,685 patient procedures
indicate that the average rate to date of abortionrelated complications (primarily incomplete
abortion, hematometra, mild infection) for CNMs,
NPs and PAs is 1.5%. Using an extensive postabortion follow-up protocol, no major complications
among CNM/NP/PA patients have been reported,
and all minor complications have been resolved
without further problems. Patients report a high
satisfaction rate of 9.4 (scale 0-10).
Conclusions: California CNMs, NPs and PAs
provide all aspects of early pregnancy care with the
exception of aspiration abortion. Preliminary
evidence demonstrates that post-abortion problems
can be treated promptly and hospital visits avoided
when more abortion-providing clinicians are
available. Patients report high satisfaction with the
care they receive from CNMs, NPs and PAs.
Additional benefits to patients of incorporating
aspiration abortion care into CNM/NP/PA practice
include: increased continuity of care; ability to
receive follow-up care or treatment of
com¬plications quickly at local clinics; and reduced
stigma by accessing abortion services in primary
care and family planning settings.
Implications for Policy, Delivery or Practice:
Substantial data on abortion provision by CNMs,
NPs and PAs are essential to improving patient
access to abortion care and to translating evidence
into policy, professional education, and clinical
practice.
Funding Source(s): Private Foundations
Poster Number: 297
Tracking the Growth of Hospitalist Medicine in
Ontario: Trends in Inpatient Physician Specialty and
Continuity-of-Care from 1996 to 2009
Presented by: Heather White, M.Sc., Ph.D.
Candidate, Health Policy, Management and
Evaluation, University of Toronto, 169 Street
George Street, Apartment 506, Toronto, Canada
M5R 2M4; Phone: (647) 439-0271; Email:
heatherlynn.white@utoronto.ca
Co-authors: Dr. Richard Glazier, M.D., M.P.H.
Research Objective: Providing high quality, costeffective services within the acute-care setting is
becoming increasingly complex and hospitals are
seeking to adapt by exploring new delivery models,
including the use of hospitalists, which may allow
them to improve quality of care while remaining
financially sustainable. Defined as physicians
specializing in the general medical care of
hospitalized patients, hospitalists have emerged as
the fastest growing medical specialty in North
America with more than 28,000 practitioners and
2,300 programs currently operating across Canada
and the United States. Under this discipline,
unattached patients and patients whose GP/FPs do
not provide inpatient services are transferred to the
care of a hospitalist upon admission. Acting as
case-manager, the hospitalist’s primary role is to
co-ordinate the care of all assigned patients and
facilitate their access to post-acute services. Upon
discharge, patients are returned to the community
and into the care of their GP/FP - if they have one.
Although hospitalist medicine has become the
dominant model for inpatient care in many
Canadian hospitals, data characterizing the
prevalence of hospitalists or their influence on
inpatient specialty volume and continuity-of-care
has not been explored in the Canadian literature to
date. This study aims to describe the growth of
hospitalist medicine within Ontario hospitals, viewed
alongside contemporary changes in inpatient
physician specialty and continuity-of-care from 1996
to 2009.
Study Design: Using retrospective times-series,
this analysis draws on hospitalization abstracts and
physician billing claims to examine temporal
patterns in inpatient specialty and continuity-of-care
for non-elective adult admissions to acute-care
hospitals in Ontario from April 1, 1996 to March 31,
2009. Annual service volume and location were
extracted from the Ontario Health Insurance Plan
database to derive prevalence estimates of
hospitalists, GP/FPs, internists and subspecialists
providing inpatient services within Ontario hospitals.
Clinical information from the Discharge Abstract
Database were then linked to billing claims to
assess continuity-of-care and hospitalist penetration
according to hospital type, size and geographic
location. Outcomes were entered into separate
univariate regression models, accounting for
autocorrelation with a lag set to one.
Population Studied: All non-elective, nonobstetrical admissions to Ontario hospitals for
adults aged 18 and older from April 1, 1996 to
March 31, 2009.
Principal Findings: Analyses are currently
underway; however results will be available for
presentation and discussion in June.
Implications for Policy, Delivery or Practice:
This study will provide the first descriptive
assessment of hospitalist prevalence and
penetration within Ontario Hospitals, presented
alongside contextual analyses describing overall
trends in inpatient service delivery and continuity-ofcare since its inception. This analysis will aid in
defining inpatient volume cut-points for identifying
hospitalist providers within administrative
databases; a necessary first step for future work
aimed at evaluating and monitoring hospitalist
performance. As hospital CEOs, policy analysts and
government agencies continue to explore the
viability of these models for improving inpatient care
for Canadians, quantitative evidence supporting
their penetration and value needs to enter into the
policy agenda.
Funding Source(s): Canadian Institutes of Health
Research
Poster Number: 298
The Effects of Data Aggregation on the Link
Between Primary Care Workforce and Health Care
Utilization Rates
Presented by: Brad Wright, M.S., Doctoral
Student, Health Policy and Management, University
of North Carolina Chapel Hill, 1300 Laurel Springs
Drive, Apartment 1301, Durham, NC 27713;
Phone: (202) 465-4815; Email:
bradwright@unc.edu
Co-authors: Thomas Ricketts, III, Ph.D., M.P.H.
Research Objective: Most health care workforce
studies do not account for data aggregation effects
that can significantly affect the direction and
magnitude of findings. Building on a 2008 study by
Kravet and colleagues (Am J of Med (2008)121:
142-148), which found a higher proportion of
primary care physicians to be associated with a
decrease in per capita rates of inpatient
admissions, emergency room visits, and total
surgeries, we sought to re-examine these
associations within both metropolitan statistical
areas (MSAs) and the counties that compose them
to investigate the potential impact of geographic
data aggregation on the findings.
Study Design: We estimated four distinct crosssectional multivariate regression models to predict
health care utilization rates at the county level and
the metropolitan statistical area (MSA) level using
Area Resource File data from 2007. Our study
focused on health care utilization in the United
States using inpatient admissions, hospital-based
outpatient visits, emergency room visits, and total
(inpatient and outpatient) surgeries as dependent
variables in separate regressions. These outcomes
were selected because we felt that appropriate
primary care use would reduce their incidence. The
key independent variable was the proportion of
primary care physicians out of all active, officebased, non-federal physicians in the area engaged
in direct patient care. Controls for regional
differences in physician practice patterns were
included along with several community-level
controls (e.g., physician and population density,
percent over age 65, percent female, percent
African-American, per capita income, per capita
hospital bed supply, and per capita death rate.)
Population Studied: We studied county-level and
MSA-level relationships between the proportion of
primary care physicians in an area and the
population’s utilization of health care in that area.
For the purposes of this analysis, non-MSA
counties were excluded.
Principal Findings: We found that a higher
proportion of primary care physicians in the area’s
physician supply was associated with a decreased
number of inpatient admissions at the MSA level,
but not the county level, and a decreased number
of emergency room visits at the county level, but
not the MSA level. Outpatient visits and total
surgeries were not associated with the proportion of
primary care physicians.
Conclusions: There is some evidence that a
higher proportion of primary care physicians is
associated with a decrease in health care
utilization, but these findings depend on the level of
aggregation.
Implications for Policy, Delivery or Practice:
Primary care physicians may help to control health
care costs by encouraging more appropriate service
use and preventing persons from developing more
serious health conditions that require use of more
resource-intensive health care services. Further
studies are warranted and investigators should be
aware of the implications that aggregating data may
have on their results, and acknowledge any
resultant limitations.
Funding Source(s): AHRQ, This research was
partially supported by a National Service Research
Award Pre-Doctoral Traineeship from the Agency
for Health Care Research and Quality sponsored by
the Cecil G. Sheps Center for Health Services
Research, University of North Carolina at C
Poster Number: 299
Maintenance of Certification: The Case of Family
Medicine
Presented by: Imam Xierali, Ph.D., Research
Scientist, Robert Graham Center, American
Academy of Family Physicians, 1350 Connecticut
Avenue Suite 201, Washington, DC 20036; Phone:
(202) 331-3360; Email: ixierali@aafp.org
Co-authors: Jason Rinaldo, Ph.D.; Andrew
Bazemore, M.D.; Bob Phillips, M.D.; Stephen
Petterson, Ph.D.
Research Objective: Although medical licensure is
a state requirement, specialty certification in the
U.S. remains a voluntary process. This study tests
hypotheses regarding family physicians in direct
patient care whose board certification had expired.
Study Design: We merged multi-year American
Board of Family Medicine Maintenance of
Certification (MOC) data with 2009 American
Medical Association Master File data, and U.S.
census data. We then completed cross-sectional
spatial, descriptive, and logistic regression analyses
comparing family physicians who are versus who
are not maintaining their board certification.
Population Studied: Direct patient care family
physicians whose board certification expired by the
end of 2008
Principal Findings: At the end of 2008, there were
10,759 direct patient care family physicians whose
ABFM board certification expired. Physicians
younger than 65 years of age whose certification
expired were more likely to be male (76.52%
vs.62.63%, p<.0001), more likely to be older (p <
.0001), more likely working in areas of dense
poverty (53.83% vs.47.75%, p < .0001), in Health
Professional Shortage Areas (26.17% vs.21.16%, p
< .0001), in rural areas (19.95% vs. 19.56%, p
<.0001), in Medically Underserved Areas (25.67%
vs.21.55%, p < .0001), in Primary Care Service
Areas with fewer physicians (5.12% vs.4.12%, p <
0.05). Board certification maintenance is also
significantly associated with physician practice type,
IMG status, and employment type.
Conclusions: Although about 87% of direct patient
care family physicians in the United States had
maintained their board certification, there were still
a substantial number of direct patient care family
physicians whose certificate had lapsed by the end
of 2008. The study found that family physicians
located in poorer and underserved areas would be
less likely to have maintained board certification.
Implications for Policy, Delivery or Practice:
Although the causal relationship between practicing
in underserved areas and tendency to lapse board
certification is not well established in this analysis,
the significant association between them begs for
more research into this issue. Also the effect of
employment type of physicians show significant
association with the likelihood to lapse board
certification, indicating policy actions may be
needed to reach out to physicians in certain
employment setting, especially those in solo
practices.
Poster Number: 300
Quality Outcomes of Hospital Supplemental Nurse
Staffing
Presented by: Ying Xue, D.N.Sc., Assistant
Professor, School of Nursing, University of
Rochester, 601 Elmwood Avenue Box SON,
Rochester, NY 14642; Phone: (585) 276-3136;
Email: ying_xue@urmc.rochester.edu
Co-authors: Linda Aiken, Ph.D.; Deborah Freund,
Ph.D.; Katia Noyes, Ph.D.
Research Objective: Many hospitals use
supplemental nurses (SRNs) hired from agencies to
compensate for their shortage of nurses. As the
nursing shortage continues, the demand for SRNs
may persist or even increase, but little is known
about the use and impact of such staff. This study
was designed to better determine how one hospital
used SRNs and to examine the impact of SRNs on
patient and nurse outcomes.
Study Design: A retrospective study was
conducted using hospital administrative data
between 2003 and 2006. Data were collected on
discharge patient demographics, primary insurance,
principal diagnosis, six secondary diagnosis, nurse
staffing characteristics, nursing work environment,
and patient and nurse outcomes. Primary patient
outcomes included patient satisfaction, patient
satisfaction with nurses, inpatient mortality, falls,
pressure ulcers, and medication errors. The primary
nurse outcome was nurse voluntary turnover. Use
of SRNs was defined as the proportion of total
nursing hours provided by SRNs. The unit of
analysis is unit quarter. Multilevel modeling was
employed to model longitudinal effects. False
discovery rate was used to adjust multiple
procedure comparison and control for type I error.
Population Studied: Discharge patient records
from 19 adult medical, surgical, and intensive care
units in a large urban teaching hospital in the
upstate New York during year 2003 and 2006.
Principal Findings: The hospital had used SRNs
over time with a decreasing trend. Use of SRNs
varied widely by units. Of 19 units, 18 units had
used SRNs some time during year 2003 to 2006.
Use of SRNs ranged from 0.16% to 20.02% of total
nursing hours per unit quarter. Units that relied
more on SRNs had worse nursing work
environments. Higher use of SRNs was associated
with lower patient satisfaction and higher RN
turnover when only adjusted for patient and nurse
staffing characteristics. After controlling nursing
work environment, however, use of SRNs didn’t
have any negative impact on patient & nurse
outcomes.
Conclusions: Theses results suggested that heavy
use of SRNs might be an indicator of a poor nursing
work environment that other literature has linked to
poor patient and nurse outcomes.
Implications for Policy, Delivery or Practice:
Improving working environments could improve
patient and nurse outcomes. Future study is
needed to further explore the relationship between
patterns of using SRNs and nursing work
environment.
Funding Source(s): RWJF
Poster Number: 301
National Trends of Supplemental Nurses’
Characteristics from 1984 to 2004
Presented by: Ying Xue, D.N.Sc., Assistant
Professor, School of Nursing, University of
Rochester, 601 Elmwood Avenue Box SON,
Rochester, NY 14642; Phone: (585) 276-3136;
Email: ying_xue@urmc.rochester.edu
Co-authors: Joyce Smith, M.S.; Deborah Freund,
Ph.D.; Linda Aiken, Ph.D.
Research Objective: Use of supplemental nurses
(SRNs) as a strategy to address nurse shortages is
a long-standing practice. The purposes of this study
were to describe patterns of SRNs’ demographics,
nursing educational preparation, and nursing work
experience; and examine whether these
characteristics are different from those of nurses
with a permanent position (PRNs).
Study Design: A time series cross-sectional
analysis of the National Sample Survey of
Registered Nurses 1984-2004 was employed.
Population Studied: SRNs were defined as nurses
who were working in their principal or secondary
nursing position in a supplemental role at the time
of survey. PRNs were defined as nurses who were
working as a facility employee in their principal
position with exclusion of those who had secondary
positions as SRNs.
Principal Findings: SRNs accounted for 3%-5% of
RN population during 1988 to 2004. SRNs and
PRNs were different in demographics. The
proportion of nurses who were older than 40
increased from 41% in 1984 to 58% in 2004 in the
SRN group, and from 51% in 1992 to 72% in 2004
in the PRN group. The proportion of nurses who
were male increased from 6% in 1984 to 11% in
2004 in the SRN group, and from 4% in 1992 to 6%
in 2004 in the PRN group. The proportion of nurses
with non-white racial background increased from
15% in 1984 to 21% in 2004 in the SRN group, and
from 8% in 1992 to 11% in 2004 in the PRN group.
Non-currently married status was about 42-47% in
the SRN group, and about 27% in the PRN group
over time. SRNs were more likely to receive initial
RN education in a foreign country (4% in 1988 to
6% in 2004) than PRNs (about 3% over time).
Trend of nurses who had Baccalaureate degree or
higher were similar in both groups, it increased from
35% in 1984 to 47% in 2004 in the SRN group, and
from 39% in 1992 to 49% in 2004 in the PRN group.
Nursing work experience was defined as years
since received highest nursing or nursing related
degree. SRNs had fewer years of experience
(about 13 years in both 2000 and 2004) than PRNs
(15 years in 2000 and 16 years in 2004).
Conclusions: SRNs are different from PRNs in
demographic characteristics. SRNs are
characterized by a higher percentage of nurses who
are younger, male, non-White, not currently
married, and received a basic nursing education in
a foreign country. SRNs are equally qualified for
work as PRNs in terms of level of nursing
educational preparation.
Implications for Policy, Delivery or Practice:
SRNs are a more very diverse group than PRNs.
As the population needing health care in the U.S.
becomes more diverse there is urgency to bring
more diversity to the PRN population. Future
research is needed to explore why some nurses are
more likely to work as a SRN and based on further
understanding of the SRN population, strategies
may be developed to attract these nurses to a
permanent position.
Funding Source(s): RWJF
Poster Number: 302
Physician Perspectives on Electronic Prescribing
Using a Locally Developed Versus Commercial
Electronic Health Record
Presented by: Erika Abramson, M.S., M.S.,
Assistant Professor, Pediatrics, Weill Cornell
Medical College, 402 E. 67th Street, New York, NY
10065; Phone: (646) 962-8058; Email:
err9009@med.cornell.edu
Co-authors: Vaishali Patel, Ph.D., M.P.H.; Sameer
Malhotra, M.D., M.A.; Elizabeth Reynolds, B.A.; S.
Nena Osorio, M.D.; Rainu Kaushal, M.D., M.P.H.
Research Objective: Federal incentives to adopt
interoperable, certified electronic health records
(EHRs) with electronic prescribing (e-prescribing)
are motivating organizations previously using locally
developed systems to transition to commercially
developed EHRs. Understanding physicians’
perspectives on the transition and experiences with
e-prescribing on these different systems can
improve the design and implementation of
commercial EHRs. We use the term e-prescribing
to describe electronic ordering of medications,
regardless of whether prescriptions were printed or
electronically transmitted to a pharmacy.
Study Design: We conducted a mixed methods
study consisting of a closed-ended survey, field
observations of prescribing using a new commercial
system, and semi-structured interviews with
physicians at an academic ambulatory care office
practice from January through November 2009.
Using these complementary approaches allowed us
to develop a rich description of a small group of
experienced users. Data collection began nine
months after the transition. We analyzed survey
data using descriptive statistics and analyzed field
notes of observations and transcripts of semistructured interviews using qualitative methods
guided by a grounded theory approach.
Population Studied: We studied internal medicine
faculty (n=19) at an academic-affiliated ambulatory
care office practice. Use of the commercial system
was mandatory. The information systems (I.S.)
team conducted a large-scale, intensive effort to
transition physicians from a longstanding locally
developed system that had been in place for 15
years to a new commercial system, as well as to
transfer and integrate medication data from multiple
additional sources.
Principal Findings: We identified key themes
describing physician experiences transitioning from
a locally developed to a commercial EHR and
perceptions about the advantages and
disadvantages of prescribing with each type of
system. Overall, we found that even experienced eprescribers considered the transition difficult. Most
physicians did not perceive the commercial system
as improving safety, despite having more advanced
clinical decision support (CDS). Many physicians
were concerned about the potential of CDS to
contribute to errors, particularly amongst trainees.
Additionally, physicians felt the commercial system
was too complex, reducing their efficiency.
Conclusions: In spite of the intensive training and
effort by I.S. to provide a smooth transition between
systems, physicians perceived a negative impact on
their productivity. Physicians’ attitudes towards eprescribing using commercial systems appear
strongly related to perceived effects on workflow.
Designing systems that make prescribing efficient
and limiting alerts to those that are clinically
relevant may reduce physician resistance to
switching to commercial EHRs. Ironically,
sophisticated commercial interfaces may need to be
simplified to best fit physician workflow.
Longitudinal studies should be conducted to
examine how physician perspectives evolve over
time as physicians become more familiar with new
systems and iterative refinements are made.
Implications for Policy, Delivery or Practice:
Given national focus on increasing the usage of
ambulatory interoperable EHRs with e-prescribing,
our findings have important implications for the
design and widescale implementation of
commercial systems. This is amongst the first
mixed-method studies to examine physician
experiences transitioning between a locally
developed EHR with e-prescribing and a
commercial one and provides insight on physician
preferences for prescribing applications. Developing
technical standards to ensure interoperability will
also be critically important given the challenges in
transferring medication data between systems.
Funding Source(s): AHRQ
Poster Number: 307
Impact of Information Sharing Through Electric
Medical Network Between Pharmacist and
Physician on Patient Care in Japan
Presented by: Miki Akiyama, Ph.D., Associate
professor, Faculty of policy management, Keio
University, 5322 Endo Fujisawa, Kanagawa, Japan
2520882; Phone: +81466493448; Email:
miki@sfc.keio.ac.jp
Co-authors: Takebayashi, Toru, M.D., M.P.H.,
Ph.D.; Hirai, Aizan, M.D., Ph.D.
Research Objective: This study aims to explore
influences and effects of use of regional medical
information communication network upon doctors
and pharmacists communication and their patients’
medication adherence, by conducting clinical
epidemiology study of osteoporosis patients and an
interview survey with their pharmacists in Sanbu
district in Chiba prefecture of Japan where the
electronic medical information networking system
was implemented in. We conducted cohort study to
examined medical adherence of osteoporosis
patients in Sanbu district in Chiba prefecture in
Japan. The system can be characterized as
asynchronous media and store-and-forward
techniques liberating users from temporal
constraints.
Study Design: Triangulation of retrospective cohort
study, questionnaire survey and semi-structured
interview was carried out. In the cohort study,
efficacy of information network use on osteoporosis
treatment was evaluated by comparing changes in
urinary cross-linked N-telopeptides of type I
collagen (NTX), turnover marker of bone, over the
study period between patients on the information
network and those without the network. A follow-up
questionnaire survey was also conducted. In
addition, semi-structured interviews with all
pharmacists using the network were carried out and
transcribed to explore how they perceive the effects
of the network. NTX data was compared by MannWhitney U-test, and questionnaire survey data was
analyzed by chi-square or Fisher’s exact method.
Qualitative data was collected and analyzed
according to a modified Grounded Theory
approach.
Population Studied: The cohort was consisted of
297 female osteoporosis patients in a regional core
hospital. Of those, 179 were patients whose records
were on the network and 118 were those without
the network. There was no difference between the
age distributions between two groups (with-network
and without-network); 3% /3% for under 40,
40%/41% for 40-64) 58%/56% for over 65. We sent
questionnaires to 230 out of 297, who were eligible
for questionnaire survey and 136 (82/54) responded
(response rate=59.1%). In addition, 21 pharmacists
(male 13, female 8) who were using the information
network were interviewed.
Principal Findings: The born turnover marker NTX
was significantly improved among the patients on
the network compared to those without network as
the medication period got longer. Among patients
whose medication period was over 270 days,
improvement rate of NTX from baseline was
significantly higher (P=0.02) in the network group
(69.7%) compared to that of non-network group
(48.2%). Patients’ survey indicated that the use of
the network enabled better communication, mutual
understandings and trusts between pharmacists
and patients. Interview data obtained from 21
pharmacists indicated that the network enabled
pharmacists to understand physicians’ intentions
more clearly and consequently, brings pharmacists
well-grounded decisions, effective learning, and
motivation as a medical professional.
Conclusions: Triangulation of the results obtained
from the studies revealed that the medical
information network improved patients’ medication
adherence. The interviews with pharmacists and
the patients’ questionnaire survey explained how
the information network ensured better
communication between doctors, pharmacists and
patients, which lead to better compliance based on
mutual understandings and trusts between
pharmacists and patients.
Implications for Policy, Delivery or Practice:
Regional medical information communication
network can contribute to good communications
between doctors, pharmacists and patients, which
would result in good medical practice especially in
chronic diseases.
Funding Source(s): partly funded by Pfizer Health
Research Foundation Japan
Poster Number: 308
Impact of Health Information Technology Portfolio
on Hospital Quality: A National Study of U.S. Acute
Care Facilities
Presented by: Ajit Appari, Ph.D., Research
Fellow, Center For Digital Strateties, Dartmouth
College, 100 Tuck Hall, Hanover, NH 03755;
Phone: (603) 646-9017; Email:
Ajit.Appari@Tuck.Dartmouth.Edu
Co-authors: Denise Anthony, Ph.D.; M. Eric
Johnson, Ph.D.
Research Objective: This study empirically
examines the impact of a portfolio of health
information technologies, compared to individual
applications, on quality in acute care hospitals.
Study Design: We analyze data from the 2006
annual survey of health information technologies
(HIT) conducted by Dorenfest Institute-HIMSS
Analytics and hospital quality performance data
from the 2006 CMS Hospital Compare database.
We consider the impact on quality of care of
adopting a portfolio of six clinical health IT
applications in U.S. acute care hospitals: clinical
data repository (CDR), clinical decision support
(CDS), computerized physician order entry (CPOE),
order entry with communications (OEC), enterprise
electronic medical records (EMR), and electronic
medication administration record (EMAR) systems.
Hospital quality is measured using the eight CMS
benchmark care indicators for acute myocardial
infarction (AMI). Each AMI quality measure is
regressed on the individual HIT applications, and on
overall HIT portfolio (controlling for bed size, profit
status, academic status, and affiliation with
integrated health delivery system). Since the quality
measures are bounded within the interval [0,1], we
use a generalized linear model (GLM) regression
with the logistic function as linking function. To
account for potential spatial correlation of errors, we
use state-level error clustering. Our approach
examines the effect of overall HIT portfolio
compared to prior studies examining the
relationship between individual HIT applications and
quality.
Population Studied: 2,624 non-federal acute care
hospitals with at least 100 beds.
Principal Findings: In our sample, the average
HIT portfolio is 3.4 applications (range 0-6). We find
limited evidence of an association between
individual HIT applications and AMI quality
measures. For example, CPOE and EMAR have a
positive impact on two of the six medication-based
AMI indicators. In analysis of overall HIT portfolio,
we find a positive association between size of HIT
portfolio and quality. Particularly, hospitals with a
larger HIT portfolio are more likely to give patients
smoking cessation advice (p-value: 1%), administer
aspirin at discharge (p-value: 5%), give PCI
intervention within two hours of arrival, (p-value:
5%) and administer beta blocker at arrivals (pvalue: 10%). However, there was no significant
relationship between HIT portfolio and the
remaining four AMI quality measures. Our findings
suggest that the effects of HIT use should not be
viewed by individual applications; instead a portfolio
view may enable greater quality improvement.
Conclusions: We assessed the impact of overall
HIT portfolio on the quality of AMI care provided in
U.S. hospitals and found statistically significant
effects that hospitals with larger HIT portfolios have
better process quality measures.
Implications for Policy, Delivery or Practice: HIT
adoption has become a priority for U.S.
policymakers, especially in light of the increasing
focus on healthcare cost and quality. Our findings
suggest that adoption of a portfolio of key clinical
technologies does have a positive impact on quality
of care. Developing appropriate policy measures
and incentives to enhance the wide-spread diffusion
of healthcare technologies, particularly CDR, CDS,
CPOE, EMR, and EMAR systems will be critical to
realize ulterior goals of improving care quality for all
Americans.
Poster Number: 309
Physicians’ Eligibility for Medicare and Medicaid
EHR Incentives and Current Use
Presented by: Brian Bruen, M.S., Lead Research
Scientist, Dept. of Health Policy, George
Washington University, 2021 K Street NW, Suite
800, Washington, DC 20006; Phone: (202) 9944125; Email: brian.bruen@gwumc.edu
Co-authors: Leighton Ku, Ph.D., M.P.H.; Peter
Shin, Ph.D., M.P.H.; Patricia MacTaggart, M.B.A.,
MMA; Sara Rosenbaum, JD; ,
Research Objective: Under the American
Recovery and Reinvestment Act (ARRA),
physicians will be eligible for incentive payments to
adopt, improve and maintain electronic health
records (EHRs). Medicare and Medicaid incentives
are available to those who attain “meaningful use.”
Medicare physicians who do not use an electronic
health record (EHR) system would eventually be
penalized through payment reductions. This study
estimates the number and characteristics of
physicians eligible for Medicare or Medicaid
incentives and compares this to their current use of
EHRs.
Study Design: We analyzed the National
Ambulatory Medical Care Survey (NAMCS), which
provides data on a nationally representative sample
of office-based physicians, including a sample of
ambulatory visits and description of physicians’
EHR use. Any physician billing Medicare is eligible
for Medicare incentives, while Medicaid incentives
require a minimum level of Medicaid volume. A
physician must choose whether to receive Medicaid
or Medicare incentives; both cannot be received.
Since Medicaid incentives tend to be larger, we
assume those eligible for both would tend to take
the Medicaid payments.
Population Studied: Practicing non-federal, officebased physicians
Principal Findings: Based on preliminary analyses
of the 2007 NAMCS, approximately 257,000 of the
nation’s 320,000 physicians (or 80%) could qualify
for HIT incentives through either Medicare or
Medicaid. Of those physicians who could qualify,
about 47,000 physicians (18%) reported having
“comprehensive” EHRs already. Another 15% of
qualifying physicians have “partial” EHRs and the
remaining two-thirds (67%) have no EHRs.
Pediatricians, OB/Gyns and psychiatrists were less
likely to be eligible for incentives. (Note: "partial"
and "comprehensive" do not necessarily
correspond to "meaningful use" of EHRs.)
Conclusions: Nearly 80% of practicing officebased physicians could be eligible for ARRA
incentives. Most did not have comprehensive EHRs
as of 2007, so the incentives could spur substantial
adoption or improvement of EHR systems. Eligibility
for incentive payments does not necessarily mean
those physicians will adopt EHRs; that will also
depend on factors such as the total cost of EHR
systems, perceived difficulty of attaining meaningful
use and other factors.
Implications for Policy, Delivery or Practice: The
ARRA incentives could encourage most officebased physicians to adopt or upgrade EHR
systems, which may ultimately improve the quality
and cost-effectiveness of care. However, the 20%
of physicians who would not be eligible for
incentives may represent a special policy challenge
to attaining universal deployment of EHRs.
Funding Source(s): To Be Determined
Poster Number: 310
Assessing the Value of the Veterans' Health
Administration Investments in Health Information
Technology
Presented by: Colene Byrne, Ph.D., Senior
Analyst, Center for IT Leadership, Partners
Healthcare, One Constitution Center, Charlestown,
MA 02119; Phone: (617) 643-4157; Email:
cbyrne1@partners.org
Co-authors: Blackford Middleton, M.D., M.P.H.,
M.Sc.; Douglas Johnston, M.T.S; Eric Pan, M.D.,
M.Sc.; Adam Vincent, M.P.P; Lauren
Mercincavage, M.H.S.
Research Objective: To assess the value of the
Veterans' Health Administration (VHA) investment
in the Veterans Health Information Systems and
Technology Architecture (VistA) using two
complementary methods: benchmarking the VHA
against industry norms and modeling the costs and
benefits of selected VistA applications/components.
Study Design: Two analytic methods were used: 1)
comparing the VHA against industry norms on IT
spending, adoption, and IT-related quality
benchmarks; and 2) stochastic simulation modeling
of the monetized costs and benefits of four selected
VistA applications/components. Sources of VistA
value were categorized: workload reductions; freed
space; eliminated redundancy; avoided utilization;
and decreased expenses (e.g., supplies, contracted
services).
Population Studied: The national VHA healthcare
system and selected VistA applications were
studied - Bar Coding Medication Administration;
Computerized Patient Record System (CPRS)
including order entry and results reporting; Picture
Archiving and Communication System (PACS); and
Laboratory Electronic Data Exchange.
Principal Findings: Benchmarking: While the VHA
historically has spent a higher proportion of its
budget on IT, it has achieved a very high level of
HIT adoption when compared to industry norms,
using several adoption frameworks. The VHA
greatly exceeds industry norms in outpatient
EHR/EMR adoption and in selected outpatient
quality measures that are reflective of the use of
automated clinical reminders. Modeling Due to the
lack of systematic tracking of costs and benefits of
the VistA applications, retrospective, high-level
estimates of annual and cumulative costs and
benefits were calculated. The total monetized
benefit of the 4 selected VistA applications - all first
implemented between 1997 and 2001 - was
estimated to exceed total costs around 2003. At the
end of the evaluation period, when VA had reached
full adoption of all the applications studied, the
estimated annual net benefit was conservatively
estimated at $690 million per year, with a total
cumulative estimated benefit during the study
period of $3.1 billion.
Conclusions: The findings suggest that the VHA
investment in IT is associated with increased
healthcare value through healthcare cost savings,
efficiencies, and improved quality of care from the
VistA applications. The VHA spent approximately
1% more than the private sector on IT as a percent
of its total budget. This investment has resulted in
high inpatient and outpatient adoption of the EHR,
and high clinical functionality as compared to the
private sector, and likely contributed to better
inpatient and outpatient quality measures. The
cumulative net benefit from these VistA applications
during the study period is estimated at $3.1 billion.
The CPRS, which was 100% adopted by 2002, is a
major driver of the costs as well as the value. The
extent to which the VHA has been able to capture
and realize this created value is unknown.
Implications for Policy, Delivery or Practice:
Cost savings from this investment likely occurred
from economies of scale due to in-house software
development and support, followed by rapid and
high inpatient and outpatient adoption. The VHA is
now implementing initiatives to better monitor and
track the impact of IT investments, which will ideally
add to the body of knowledge about IT cost benefit
and value analyses, and IT adoption and evaluation
measures. Establishing a framework and methods
for evaluating health IT is essential to
understanding actual gains and/or consequences
from IT use across a system. Public and privatesector policies and programs alike should work
towards developing a common set of measures.
Funding Source(s): VA
Poster Number: 311
The Impact of Health Information Technology
Expenditures on Length of Stay and In-hospital
Mortality for Acute Myocardial Infarction- Evidence
from the Texas Hospital Inpatient Database
Presented by: Hsueh-Fen Chen, Ph.D., Assistant
Professor, Health Management and Policy,
University of North Texas Health Science Center,
3500 Camp Bowie Boulevard., Grapevine, TX
76107; Phone: (817) 735-0311; Email:
hschen@hsc.unt.edu
Co-authors: Fernando Wilson, Ph.D.; Nuha
Lackan, Ph.D.; Abby Kazley, Ph.D.
Research Objective: To examine the effect of
health information technology (HIT) expenditures by
hospitals on the length of stay and in-hospital
mortality for acute myocardial infarction (AMI)
patients.
Study Design: This study uses patient-level, crosssectional data from the 2007 Texas Hospital
Inpatient Discharge Public Use Data File (THIDPUDF). The THID-PUDF contains clinical data on
all inpatient hospital stays occurring in Texas
hospitals. The outcome measures are length of
stays and in-hospital mortality for AMI patients. The
primary independent variable is the HIT expenditure
from the 2006 Texas Hospital Annual Survey. This
variable is dichotomized into high expenditure
versus low expenditure. A propensity score
stratification approach is used with linear regression
for length of stay outcome and logit regression for
in-hospital mortality. This approach adjusts for
possible endogeneity resulting from the possibility
that hospitals with high quality of and efficient care
are more likely to invest in HIT. The control
variables are hospital, market, and patient
characteristics from the Texas Hospital Annual
Survey, Bureau of Economic Analysis, and THIDPUDF.
Population Studied: The study population is
individuals aged 18 or older who resided in Texas
and were admitted into acute general community
hospitals in Texas with a primary diagnosis AMI.
Subjects were excluded from the study if they
transferred into or out of other acute care hospitals,
were admitted into a hospital with less than five
admissions, had a length of stay less than two days
or more than 30 days, or had missing disposition of
the patients at discharge, race, or age information.
Additionally, patients with any secondary diagnoses
related to HIV or substance abuse were excluded
because age information was not available for
these patients. The study sample contains data on
18,357 AMI patients admitted to 162 hospitals in
Texas in 2007.
Principal Findings: After controlling for the lagged
effect of HIT expenditure and the endogeneity
between HIT expenditures and the two outcome
measures, AMI patients admitted to hospitals with
above 50 percentile of HIT expenditure significantly
experienced shorter length of stay by 3 percent
(p<0.01), compared to the ones admitted to
hospitals with below 50 percentile of HIT
expenditure; however, differences in in-hospital
mortality rates between hospitals with high and low
HIT expenditures is not statistically significant.
Conclusions: Our results suggest that hospitals
with high HIT expenditure provide more efficient
AMI care, as measured by inpatient length of stay
than hospitals with low HIT expenditure. Shortening
lengths of stay for AMI inpatients may serve as a
plausible mechanism for hospitals to reduce costs
without compromising quality of care.
Implications for Policy, Delivery or Practice:
Investment in HIT by hospitals not only results in
benefits to patients, but to the health system as a
whole. Given the health sector’s rising proportion of
the GDP, any strategies to improve efficiency and
reduce costs within the healthcare system should
be carefully considered. Although HIT is expected
to increase quality of care, our result for the inhospital mortality is not significant. Identifying other
quality indicators to measure the HIT effect is
needed.
Poster Number: 312
A Qualitative Study of Technology Acceptance in
Primary Care Practices
Presented by: Karen Danna-Lynch, Ph.D.,
Research Fellow, Family Medicine, UMDNJ-Robert
Wood Johnson Medical School, 1 World's Fair
Drive, Somerset, NJ 08873; Phone: (609) 2403122; Email: lynchda@umdnj.edu
Co-authors: Deborah Cohen, Ph.D.; Rebecca Etz,
Ph.D.; Jesse Crosson, Ph.D.
Research Objective: Appropriate health
information technology (HIT) is needed to deal with
the wide level of care provided in the primary care
(PC) setting. Yet, despite the expected benefits,
there is evidence of numerous implementation
failures and low adoption rates. Thus, to optimize
HIT adoption in PC settings, it is crucial to better
understand the implementation process. This study
explores the extent to which a widely used model in
the information systems literature (Davis’ 1986
Technology Acceptance Model or TAM) predicts
successful implementation.
Study Design: This study represents a secondary
analysis of 18 qualitative case studies of PC
practices in the process of adopting a new HIT.
Original data were collected using semi-structured
interviews, open-ended questions, and
observations. We examined the data with specific
reference to TAM, looking at the implementation of
3 different forms of HIT used for the purpose of
delivering (direct) patient care: electronic medical
records, e-prescribing software, and populationspecific electronic registries. We conducted a
content analysis of all data using ATLAS.ti.
Population Studied: Eighteen PC practices in 3
states (CA, MI, and NJ) involved in the process of
implementing a new HIT. The end-users of the HIT
were health-care professionals providing medical
care.
Principal Findings: Results show that while TAM
predicts a portion of factors related to the use and
acceptance of HIT (such as ease of use and
perceived usefulness), other context-specific factors
play an important role in facilitating or hindering HIT
adoption in the PC setting, including: 1) personal
experience (including prior knowledge of and
comfort levels with computers); 2) social support
(such as practice-specific champions or team
support); 3) social norms (e.g., is HIT use required
in the practice? Are computers routinely used for
other functions within the practice?); and 4)
contextual factors (including government
announcements or other outside
recommendations). A method for incorporating
these additional variables into subsequent theory
and research is presented.
Conclusions: While TAM has a strong track record
in other industries, questions remain concerning its
applications to health care. Results show that TAM,
used in its generic form, does not capture some of
the unique contextual features related to successful
HIT implementation in the PC setting. By focusing
on end user reactions to the implementation and
use of various HIT, this study presents empirical
evidence of the subjective norms, personal
experiences, and contextual factors that act as
actual barriers and/or facilitators to HIT use. We
use this evidence to refine and contextualize TAM
with respect to the PC context. We tentatively
suggest that contextualized TAM measures replace
the practice of using generic measures to increase
the theory’s compatibility to the health care field.
Implications for Policy, Delivery or Practice:
This study suggests ways of tailoring TAM so as to
help end-users in PC practices prepare for the
implementation of various HIT. We present
evidence for a contextualized version of TAM that
will allow both researchers and practitioners to
become aware of PC relevant variables. As such,
the results of this study are in line with the growing
interest in the concept of “fit” as a critical need for
successful HIT design and successful
implementation.
Funding Source(s): AHRQ, National Institute for
Diabetes and Digestive Kidney Diseases
Poster Number: 313
IT-Enabled, Non-Invasive Automated Detection of
Diabetes Mellitus from a Pulse Morphology Pattern
in the Impedance Plethysmographic Pulse
Waveform
Presented by: Medha Dhurandhar, Ph.D., Head &
Program Coordinator, High Performance
Computing, Centre for Development of Advanced
Computing, Pune University Campus, Ganeshkhind
Road, Pune, India 411007; Phone: +2025704222;
Email: mdhurandhar@gmail.com
Co-authors: Girish Tillu, M.D.
Research Objective: To develop IT-enabled tools
for detection of metabolic syndromes based on
impedance plethysmographic (IPG) pulse signal
harnessing low cost, non-invasive technology and
mathematical modeling
Study Design: Multidisciplinary approach
deploying IT-enabled data mining, medical
instrumentation and mathematical modeling.
Observational, Double-blinded Study. Recording of
Clinical Investigations like BP, Lipid profile, Glucose
level of the volunteers. Simultaneously capturing
IPG-based Pulse signal for a period of 300 seconds
from the wrist of the volunteer. Pulse signal analysis
based on data mining and mathematical modeling.
Multi-centric Study at various hospitals, clinics and
institutes. Methodology: Scrubbing of pulse signal
data by filtering high frequency noise; Data
conversion to ASCII format; Data analysis using
numerical methods for identifying a) Proper systolic
peaks b) On-set points of the pulse waveform c)
Identification of various complexes d) Classification
of the complexes into "dominant" and "nondominant" (ectopic) types e) Identification of major
deflection points in each of the complexes f)
Feature extraction measuring amplitudes and
intervals g) Automated pattern recognition based on
the correlation amongst the major deflection points
h) Determining pattern distribution over the 300
secs. interval i) Finding the most dominant pattern j)
Diagnostic classification
Population Studied: 557 volunteers: Male &
Female; Age Group- 35 to 65. 241 Healthy
volunteers. Patients: 91 Diabetic and 225 Nondiabetic but suffering from other diseases
Principal Findings: Sufficient condition obtained
for a peculiar pulse morphology pattern
“MedhaMorph” indicating the volunteer to be
diabetic with 98.74% accuracy. The volunteers
having MedhaMorph as the dominating pulse
pattern were found diabetic with the following
exceptions: 1. Only 2 healthy volunteers out of 241
had this pulse pattern (99.17% accuracy), and 2.
Only 5 non-diabetic patients out of 225 had pulse
pattern MedhaMorph. (97.78% accuracy).
Conclusions: Hyperglycemia induced vascular
changes in Diabetic patients lead to several
complications like retinopathy, nephropathy and
neuropathies. It is important to detect these
changes early to avoid complications. Proposed
method simulates pulse examination and classifies
an IPG-based pulse morphology pattern for DM
from the pulse signal captured from the peripheral
vessel. Even when a patient is under treatment for
DM, this pattern can be detected. Follow-up study is
being done to see if the exceptions prove to be DMprone. Though currently DM is targeted similar
method may be adopted to detect and predict other
diseases/disorders especially related with Metabolic
Syndrome from a particular pulse morphology
pattern.
Implications for Policy, Delivery or Practice:
Current diagnostic methods for Diabetes detection
are invasive. They require blood examination with
antiseptic precautions and a laboratory set-up
causing venipuncture. Glucometer can be used at
home but is expensive and gives approximate blood
concentration of glucose. Proposed tool is noninvasive and cost-effective. Also the procedure is
not time-consuming and can be carried out at home
with a little practice. This minimizes the need for
time-consuming visits to pathological laboratories
and can help detect DM related medical problems
before they become severe or life-threatening.
Moreover, such out of medical center testing could
reduce unnecessary burden on healthcare
professionals and facilities, as well as costs for
payers.
Funding Source(s): Ministry of ICT, India
Poster Number: 314
Reliability and Validity of HIMSS and Leapfrog Data
on Computerized Provider Order Entry
Presented by: Mark Diana, Ph.D., M.B.A.,
M.S.I.S., Assistant Professor, Health Systems
Management, Tulane University, 1440 Canal Street,
Suite 1900, New Orleans, LA 70112-2699; Phone:
(504) 988-5359; Email: mdiana@tulane.edu
Co-authors: Abby Kazley, Ph.D.; Nir Menachemi,
Ph.D.
Research Objective: The objective of this research
is to assess the reliability, validity, and
responsiveness of the HIMSS and Leapfrog dataset
measures of computerized physician order entry.
Study Design: We created dichotomous measures
of the presence of CPOE in hospitals for each data
set in each year using the most complete status of
CPOE adoption in each survey. We assessed the
reliability of the CPOE measure individually for each
dataset alone. We tested the validity of the CPOE
measure by calculating sensitivity, specificity,
positive predictive value, and negative predictive
value for each dataset, using the other dataset as
the gold standard. We also tested the
responsiveness of each dataset by examining the
percent increase in complete CPOE adoption from
one year to the next in HIMSS and Leapfrog data
respectively.
Population Studied: We used secondary data on
US hospitals collected by HIMSS Analytics, the
Leapfrog Group, and the American Hospital
Association from 2005 through 2007. We created a
dataset from each survey for each year for hospitals
that reported CPOE status, a dataset that combined
hospitals across all three years for each source,
and a dataset that combined hospitals across all
three years from both sources.
Principal Findings: Hospitals with known CPOE
status in both the HIMSS and Leapfrog datasets
ranged from 1,053 to 2,355. Reliability was similar,
ranging from 76 – 78% in the HIMSS data and 81 –
86% in the Leapfrog data. HIMSS data sensitivity
ranged from 48.4 – 71.8%, data specificity ranged
from 84 – 89.0%, positive predictive value ranged
from 10 – 20%, and negative predictive value was
in the range of 98%. Leapfrog data sensitivity
ranged from 10 – 20%, specificity was consistently
98%, positive predictive value ranged from 48 –
72%, and negative predictive value ranged from 84
– 89%.
Conclusions: Our findings indicate minimal
agreement between the two datasets regarding
which hospitals have adopted CPOE, but adequate
agreement within a given dataset from year to year.
The HIMSS data tends to overestimate increases in
adoption over time, while the Leapfrog data has
more downward trending estimates for year over
year increases in CPOE adoption. The Leapfrog
data are better suited to for studying outcomes
associated with CPOE adoption, while studies
whose purpose requires the certain classification of
non-adopting hospitals can utilize HIMSS data
effectively.
Implications for Policy, Delivery or Practice: The
disagreement between the HIMSS and Leapfrog
datasets regarding hospital CPOE use creates a
challenge for researchers, practitioners, and
policymakers who wish to understand CPOE use
from a national perspective. A possible way to
overcome this limitation is for future studies to
utilize both datasets and create multiple groups
based on agreement or disagreement in the HIMSS
and Leapfrog datasets. Specifically, researchers
could examine how the results of their study differ
when comparing hospitals that reported CPOE use
in both datasets to their counterparts that positively
reported in only one of the datasets. This sensitivity
analyses would be desirable to the presentation of
results for only one of the data sources.
Poster Number: 315
The Effect of the Implementation of an Electronic
Health Record on Nursing-Sensitive Patient
Outcomes
Presented by: Dawn Dowding, Ph.D., Harkness
Fellow, Institute of Health Policy, Kaiser
Permanente, One Kaisa Plaza, 22nd Floor,
Oakland, CA 94612; Phone: (510) 271-6853;
Email: dawn.w.dowding@kp.org
Research Objective: To examine the impact of the
introduction of an Electronic Health Record (EHR)
on nursing sensitive patient outcomes
Study Design: An Interrupted Time Series Design.
The intervention of interest is the implementation of
an EHR in acute clinical areas, specifically acute
medical, surgical and critical care units. Data will be
collected on the following nursing sensitive patient
outcomes and aggregated at facility level; failure to
rescue, hospital acquired pressure ulcers, falls and
falls with injury, restraint use, peripherally inserted
central catheter related blood infections, surgical
site infection rates, ventilator associated infections,
central line catheter-associated blood stream
infection rate for ICU and high risk nursery (HRN)
patients. Data will be obtained from three sources;
the Collaborative Alliance for Nursing Outcomes
(CalNOC), hospital administrative data bases and
the national quality, resource and risk management
department for Kaiser Permanente. As the
implementation of the EHR has been staged, the
time point at which the intervention has occurred is
different in different facilities. Facilities where the
EHR has not been implemented will act as a control
group, to assess the potential effect of any external
influences on changes in the outcome measures.
Population Studied: 33 health care facilities in one
Health Care Organisation.
Principal Findings: Data will be plotted to examine
trends in changes in outcomes over time. Data will
also be analyzed using established techniques for
the analysis of ITS (analysis of variance,
segmented regression analysis, autoregressive
integrated moving average models) depending on
the number of time points before and after
intervention. Data will be corrected for
autocorrelation and any seasonal variations in
outcome measures.
Conclusions: Results of the ITS can identify what
impact (if any) the introduction of a comprehensive
Electronic Health Record has on patient outcomes
associated with the quality of nursing care.
Implications for Policy, Delivery or Practice:
Many countries, including the U.S. and U.K., are
investing in health information technology as a way
to improve the quality and safety of care that
patients receive. Whilst nurses comprise a
significant part of the health care workforce,
comparatively little is known about how they use
information technology to inform the decisions they
make, and the impact this has on the care patients
receive. The results of this study will provide
valuable evidence to inform how electronic health
records can be used to improve nursing care.
Funding Source(s): CWF
Poster Number: 316
Implementation of Personal Health Records and the
Oregon Medicaid Experience: An Evaluation Study
Presented by: Oliver Droppers, M.S., M.P.H.,
Research Assistant, Hatfield School of
Government, Portland State University, PO Box
751, Portland, OR 97207-0751; Phone: (503) 24164692; Email: odropper@pdx.edu
Co-authors: Sherril Gelmon, Dr.PH.; Jill Rissi,
Ph.D.
Research Objective: The State of Oregon is
implementing a personal health record system
(PHR) using a health record bank model for its
Medicaid population, supported by a federal
Medicaid Transformation Grant. This unique
consumer-centric approach could potentially
enhance Medicaid clients’ ability to monitor and
manage their own health, helping individuals
become more active participants in their own
wellness and self-care. An evaluation team from
Portland State University was contracted by the
State to assess the utility of the PHR based on the
experiences of Medicaid clients during the initial
implementation in 2009-2010.
Study Design: The study is quasi-experimental,
employing mixed methods through a communityengagement research strategy including use of
online surveys, interviews, focus groups, direct
observations, and documentation review. Data will
be collected during three stages of PHR adoption:
design, implementation and optimization. An
evaluation conceptual framework of key concepts
and indicators will guide the research, helping to
illustrate barriers to and facilitators of PHR
adoption.
Population Studied: Potential participants include
all publicly insured individuals within the Oregon
Medicaid system who create a personal health
record during the study period. As of October 2009,
Oregon Medicaid covered approximately 384,000
low-income children, pregnant women, disabled
and senior adults. Adults are enrolling in the first
phase; followed by children and their associated
caregivers, case managers and/or foster parents.
All individuals creating a PHR account are eligible
to receive the baseline survey. In the first six
months (i.e. up to the 2010 ARM), all enrollees are
being surveyed.
Principal Findings: Preliminary findings from
baseline enrollment surveys indicate the following
factors are critical in terms of engaging and
facilitating consumers in the use of PHRs:
environment of trust, privacy and security;
consumer choice regarding enrollment; presence of
a trusted not-for-profit community organization or
state agency as the facilitator of the PHR system;
and sharing and accessibility of patient information.
Factors likely to influence accessibility and use of
PHRs among the study population are limited
computer and health literacy, challenges with
informed decision-making using their own medical
information, and usability of health information
accessed through Internet Web sites.
Conclusions: PHR systems, particularly those that
are consumer-oriented and controlled, are in their
infancy. The PHR is intended to facilitate
opportunities for Medicaid clients to effectively
control and share their personal health information.
Findings from this research may inform predictions
about the widespread use of PHRs, as we come to
appreciate and understand how Medicaid clients
interact with PHRs, given fundamental issues
related to Internet access and health literacy.
Implications for Policy, Delivery or Practice:
PHRs must be designed to be accessible to all
population groups in order to address existing
digital disparities across vulnerable populations.
Enhanced understanding of this new technology is
a first step. Findings from this study may inform
consumers, providers, policymakers, and others, as
well as complement the efforts of numerous
national, regional, and state health information
exchange projects seeking to control health care
costs, improve safety, quality, and efficiency, and
ultimately achieve better health outcomes through
the use of PHRs.
Funding Source(s): CMS
Poster Number: 317
Measurement of Utilization of an Installed EHR
Presented by: Prashila Dullabh, M.D., Health IT
Program Manager, Health, National Opinion
Research Center, 4350 East-West Highway, Suite
800, Bethesda, MD 20814; Phone: (202) 280-9294;
Email: dullabh-prashila@norc.org
Co-authors: Elizabeth Babalola, B.A.; Adil
Moiduddin, M.P.P.; James Sorace, M.D., M.S.
Research Objective: In 2009 the National Opinion
Research Center (NORC) conducted a study to
assess the feasibility of systematic automated
approaches of tracking EHR usage.
Study Design: NORC conducted a review of the
literature using PubMed to search for peer reviewed
publications on national studies of EHR adoption.
Internet searches were also conducted to identify
government reports and articles relevant to
assessing EHR adoption in ambulatory settings. A
series of phone discussions were held with
stakeholders representing major ambulatory EHR
vendors, health center representatives, ambulatory
practices and informants involved with quality
improvement and EHR adoption.
Population Studied: This study included a review
of EHR adoption broadly but specific focus was
given to federally supported health centers and
other safety-net providers.
Principal Findings: Limitations in the use of
surveys to assess EHR adoption. In a report
comparing existing surveys assessing EHR
adoption up till the year 2008 it was found that the
surveys were of varying quality. EHR Use Limited to
Basic Functions. Review of current EHR use
suggests that in all practices (small, mid and large)
there are certain basic clinical and administrative
functions that are commonly used notably
encounter notes, medication lists, allergy lists,
problems lists, and order entry functions largely
focused around lab order entry and results delivery.
E-Prescribing (eRX) use is limited but growing
rapidly. In addition practices of all sizes report that
they support clinical decision support functions
related to eRx largely in the form of drug interaction
and drug-allergy checking. Tremendous Variability
in How EHR Use is Being Monitored. Tracking of
EHR utilization data vary in the granularity of
tracking, the features being tracked, metrics that
are used to assess utilization, the reporting
database and the kinds of reports that are being
generated. Smaller practices do not appear to have
the IT resources and infrastructure to readily
support utilization tracking. Feasibility of an
Automated Reporting Mechanism. Practices are
using different mechanisms of varying degrees of
automation to report on quality measures. Some
EHR vendors are already providing options for their
install base to upload de-identified clinical data on a
monthly basis to a centralized research server. The
capability of generating automated reports is
technically feasible and currently exists.
Conclusions: Study findings indicate that there is a
common set of EHR functions that all practices are
likely to have purchased and that sites would be
capable of implementing. Most EHR vendors have
capability to track usage of these common functions
even though they may not all support the same
robust reporting capabilities. By providing tools to
health centers and networks to monitor EHR
utilization these organizations would be better
equipped to promote meaningful use and adoption
proactively. These reports can also be used to
augment information derived from traditional survey
instruments
Implications for Policy, Delivery or Practice:
“The views expressed in this report/presentation are
solely those of the authors, and do not necessarily
reflect the views of the Department of Health and
Human Services.”
Funding Source(s): The Assistant Secretary for
Planning and Evaluation (ASPE)
Poster Number: 318
Health Information Technology and Physician
Career Satisfaction
Presented by: Keith Elder, Ph.D., M.P.H., M.P.A.,
Assistant Professor, Health Services
Administration, University of Alabama at
Birmingham, LRC 325A 1530 3rd Ave South,
Birmingham, AL 35294; Phone: (205) 996-9808;
Email: ktelder@uab.edu
Co-authors: Nicole Quon, Ph.D.; Ronica Rooks,
Ph.D.; Jacqueline Wiltshire, Ph.D.; Rhonda Belue,
Ph.D.; Lisa Gary, Ph.D.
Research Objective: Health information
technology (HIT) and physician career satisfaction
are associated with higher quality medical care.
However, the link between HIT and physician
career satisfaction, which could potentially reduce
provider burn-out and attrition, has not been fully
examined. This study uses a nationally
representative survey to assess the association
between key forms of HIT and career satisfaction
among primary care physicians (PCP) and specialty
physicians.
Study Design: We performed a retrospective,
cross sectional analysis of physician career
satisfaction using the Community Tracking Study
Physician Survey, 2004-2005. Nine specific types of
HIT as well as the overall adoption of HIT in the
practice were examined using multivariate logistic
regression.
Population Studied: The study sample included
5198 physicians, of which 3016 were primary care
physicians and 2182 were specialists.
Principal Findings: Physicians who used 5 to 6
(OR=1.46) or 7 to 9 (OR=1.47) types of HIT were
more likely than physicians who used 0 to 2 types
of HIT to be “very satisfied” with their careers.
Information technology usages for communicating
with physicians (OR =1.31) and emailing patients
(OR=1.35) were positively associated with career
satisfaction. PCPs who used technology to write
prescriptions were less likely to report career
satisfaction (OR=0.67) while specialists who wrote
notes using technology were less likely to report
career satisfaction (OR=0.75). Physicians who
treated chronically ill patients (OR range = 0.570.87) or patients with language barriers (OR
range=0.47-0.58) were also less likely to report
being “very satisfied” with their careers.
Conclusions: Using more information technology
was the strongest positive predictor of physicians
being very satisfied with their careers.
Implications for Policy, Delivery or Practice:
Health care organizations working in conjunction
with providers should consider exploring ways to
integrate various forms of HIT into practice.
Funding