Aging, Long Term Care and End-of-Life
Public Support for Financial Incentives for
Physicians and Patients for Several Different
Health Behaviors
Catherine Auriemma, University of
Pennsylvania; Christina Nguyen, Harvard
University; Aaron Delman, Fostering
Improvement for End-of-Life Decision Science;
Nicole B Gabler, Fostering Improvement for
End-of-Life Decision Science; Elizabeth Cooney,
Fostering Improvement for End-of-Life Decision
Science; Scott D Halpern, Fostering
Improvement for End-of-Life Decision Science
Presenter: Catherine Auriemma, Research
Fellow, University of Pennsylvania
caurie@mail.med.upenn.edu
Research Objective: Financial incentives have
potential to motivate behavior. The public’s
attitude towards financial incentives in
healthcare is not well understood. While there is
some evidence to support the use of financial
incentives to motivate outcomes such as
smoking cessation and colonoscopy screening,
this avenue has not been explored for
encouraging completion of advance directives
(ADs). ADs offer the opportunity for patients to
express preferences about their care near the
end of life. However, rates of completion remain
low. We sought to determine whether the public
would support the use of financial incentives to
encourage physicians and patients to
accomplish the outcomes of AD completion,
smoking cessation, and colonoscopy screening.
Study Design: Cross-sectional study utilizing
survey data to assess the amount of money the
public deems appropriate to be used as a
financial incentive for patients and physicians to
encourage completion of three health outcomes.
Participants selected a value between zero and
$1000 as an acceptable financial incentive for
either a physician or a patient for the completion
of an AD, colonoscopy screening, and smoking
cessation. The order of health outcome queried
was randomized to mitigate ranking effects.
Participants also completed a series of
demographic questions.
Population Studied: We recruited Englishspeaking adults from six public parks in
Philadelphia, Pennsylvania. Staggering the time
of day and day of week at each park,
investigators approached consecutive parkgoers and solicited participation in a brief
research survey. Participants gave verbal
consent and received a $5 gift card for
participation.
Principal Findings: We approached 553
eligible individuals and successfully recruited
380 participants on 10 days July-August, 2013
(response rate 68.7%). Just under half of
participants indicated that physicians ought to
receive some amount of monetary incentive
when either they discuss ADs with patients
(43.7%) or when one of their patients completes
an AD (46.0%). Approximately half of individuals
also responded that physicians ought to receive
financial incentives when a patient successfully
quits smoking (57.3%) or successfully completes
colonoscopy screening (50.7%). A majority
responded that patients ought to receive
financial incentives when he or she completes
an AD (62.6%), quits smoking (64.6%), or has a
colonoscopy (76.6%). Among individuals who
responded with a non-zero value as an
appropriate financial incentive, the median value
selected across behaviors ranged from $100 to
$300.
Conclusions: Approximately one half to two
thirds of participants responded with a non-zero
value as an appropriate financial incentive for
motivating patient and physician behavior in a
variety of health behaviors. Within all health
behaviors (AD completion, smoking cessation,
colonoscopy screening), more respondents
supported paying patients compared to
physicians. Median values among non-zero
responders tended to be higher for patients than
for physicians.
Implications for Policy, Delivery, or Practice:
A majority of individuals support financial
incentives targeted at physicians and patients to
motivate smoking cessation and colonoscopy
screening. While a majority also supports
motivating patients with financial incentives to
complete ADs, there is considerably less
support for financially incentivizing physicians to
encourage AD completion. In general, the
reported financial incentives suggest the
Philadelphia public is more in favor of financially
incentivizing patients than physicians to motivate
health outcomes.
Funding Source(s): Other Center for Health
Incentives and Behavioral Economics Pilot
Grant
Poster Session and Number: B, #631
Stochastic health and Family Structure on
Long-term Care Insurance Decision
Rashmita Basu, Scott&White Healthcare;
Presenter: Rashmita Basu, Assistant Professor,
Scott&White Healthcare
rbasu@sw.org
Research Objective: The uncertainty about the
need for private long-term care Insurance (LTCI)
is a major policy concern as it may lead to
higher expenditure on public programs such as
Medicaid. Due to population aging, the need for
long-term care (LTC) will inevitably increase in
the next few decades while the supply of
informal caregivers will likely to decrease due to
smaller family structure. An important but
overlooked issue in existing empirical studies of
LTCI is the relative importance of perceived
health uncertainty and family structure including
availability of informal care on the purchase of
private LTCI. The current study examines the
effects of stochastic health, family structure and
availability of informal care on the decision to
purchase LTCI among individuals aged 50 or
above.
Study Design: Retrospective longitudinal panel
study.The current stochastic health is assumed
to depend on prior health status (2010, 2008
and 2006) and other personal characteristics
including age, gender marital status, education
and race/ethnicity. Health status is a selfreported health with 5 categories (1=excellent,
2=very good, 3=good, 4=fair and 5=poor). For
the purpose of the analysis, self-reported health
variable was reverse coded, higher scores
reflect better health. Evidence based on the
HRS data suggests that self-reported health can
consistently predict future health outcomes
among HRS respondents which validates the
credibility of using this variable in the current
analysis. The current health is considered as a
continuous dependent variable and estimated
via ordinary least square method. The decision
to purchase LTC care is a binary variable
indicating whether respondents had purchased
private LTC insurance during the past two years
prior to the 2012 survey. The residual from the
forecast stochastic health model is utilized as a
measure of health uncertainty in modeling the
LTCI purchase decision using the logistic
regression method. Other variables of interest in
this model are availability of informal care from
relative or friends except spouse, number of
one’s own children and stepchildren, a dynamic
factor such as having a daughter moved close
by. A set of known individual characteristics
including age, gender, financial wealth, marital
status and race/ethnicity were adjusted in the
regression.
Population Studied: Nationally representative
individuals aged 50 years or above.The current
study focuses on the data from the Health and
retirement Study (HRS) (2006-2012) to model
impact of health uncertainly, family attributes
and availability of informal care on the LTC
purchase decision.
Principal Findings: The prior three time-lag
health variables (self-rated health, 2010, 2008,
2006) all have significant positive effect on
current health reflecting that higher prior health
predicted better perceived future health (0.13,
0.18 and 0.31). However, the perceived health
uncertainty (residual from the health-forecast
equation) did not have significant impact on the
decision to purchase LTCI. Individuals are less
likely to purchase private LTCI if they believe
their family is likely to take care of them in the
future (OR=0.83; p=0.004) and if they have
children living close by (OR: 0.86; p=0.02).
Conclusions: Family structure and availability
of informal care play more important role than
perceived health uncertainty to purchase private
long-term care insurance
Implications for Policy, Delivery, or Practice:
Greater demand and need for publicly funded
LTCI since the supply of informal care will
decrease in next few decades due to smaller
family structure and increased participation of
women in the labor market.
Funding Source(s): No Funding
Poster Session and Number: B, #632
Comparison of Long-Term Care in Nursing
Homes Versus Home Health: Costs and
Outcomes in Alabama
Justin Blackburn, University of Alabama at
Birmingham; Julie Locher, Department of
Medicine, Division of Gerontology, Geriatrics, &
Palliative Care University of Alabama at
Birmingham, School of Medicine; Meredith
Kilgore, Department of Health Care Organization
& Policy University of Alabama at Birmingham
School of Public Health
Presenter: Justin Blackburn, Assistant
Professor, University of Alabama at Birmingham
jblackburn@uab.edu
Research Objective: With an aging US
population, finding the optimal setting for longterm care is an issue many older Americans and
their families must confront. Policymakers must
not only increase the overall capacity of longterm care systems, but address the issue of
resource allocation to provide long-term care in
other settings than nursing homes. The purpose
of this study was to compare outcomes and
associated costs for acute care among patients
receiving care in nursing homes with those
receiving care in the community by home health
agencies.
Study Design: A retrospective cohort study was
conducted using high-dimensional propensity
score matching. Using assessment data from
the Minimum Data Set 2.0 (MDS) and the
Outcome and Assessment Information Set
(OASIS), beneficiaries were identified as new
admissions to nursing homes or new recipients
of home health care provided under Medicare if
they had at least 90 days without a record of an
assessment. Medicare claims were compared
one year after admission into either a nursing
home or home health. Death, emergency
department and inpatient visits, inpatient length
of stay, and acute care costs were compared
using t-tests. Medicaid long-term care costs
were compared for a subset of matched
beneficiaries with follow-up terminating in 2008.
Population Studied: Alabama residents aged
>=65 years admitted to a nursing home or home
health care between March 31, 2007 and
December 31, 2008 (N=1,291 matched pairs).
Principal Findings: After up to one year, 77.7%
of home health beneficiaries were alive
compared to 76.2% of nursing home
beneficiaries (p=0.3746). Home health
beneficiaries averaged 0.2 hospital visits and 0.1
emergency department visits more than nursing
home beneficiaries, differences that were
statistically significant. Overall acute care costs
were not statistically different; home health
beneficiaries’ costs were $31,423 on average,
while nursing home beneficiaries’ costs were
$32,239 (p=0.5032). Among 426 pairs of
beneficiaries, the average Medicaid long-term
care costs were $4,582 greater for nursing home
residents than home health recipients (p<0.001).
Conclusions: Using the data from Medicare
claims, beneficiaries with similar functional
status, medical diagnosis history, and
demographics had similar acute care costs
regardless of whether they were admitted to a
nursing home or utilizing home health care.
Long-term care costs to Medicaid appear to be
greater among nursing home residents analyzed
in our subset of beneficiaries. Additional
research is needed to control for exogenous
factors relating to long-term care decisions.
Implications for Policy, Delivery, or Practice:
There may be opportunities to provide skilled
nursing facility long-term care in the home.
While some older adults will have care needs
requiring intuitional care assistance, providing
in-home care can be cost-effective if
circumstances are optimized. As states consider
expansion of home and community-based
services (HCBS), comparisons of outcomes and
cost-effectiveness will be crucial for
policymakers.
Funding Source(s): Other Alabama Nursing
Home Association
Poster Session and Number: B, #633
Seeking High Reliability Practices to Reduce
Transmission of Infections in Long Term
Care
Barbara I. Braun, The Joint Commission;
Salome Chitavi, The Joint Commission; Linda
Kusek, The Joint Commission; Kristine Donofrio,
The Joint Commission; Beth Ann Brehm, The
Joint Commission
Presenter: Barbara I. Braun, Associate Director,
The Joint Commission
bbraun@jointcommission.org
Research Objective: 1) Identify high reliability
practices that can be used to reduce infection
transmission in long term care, 2) convene a
roundtable meeting comprising a panel of topic
experts and leaders from long term care
organizations to develop practical
recommendations for implementation, and 3)
prepare professionally developed educational
materials to disseminate information resulting
from the project.
Study Design: The methodology comprised a
scoping literature review and cross-sectional
survey of nursing homes and assisted living
facilities. The literature review process searched
for three sets of materials 1) peer-reviewed
publications, 2) clinical practice guidelines and
3) implementation guidance documents and
toolkits. Initial screening criteria for inclusion
were that the primary setting of interest was a
nursing home (NH) or assisted living facility
(ALF) AND that the content pertained to
infection prevention OR high reliability/ safety
culture/ learning culture.
The call for effective practices was sent
electronically to over 30,000 subscribers of an
online list serve and further disseminated in
partnership with 14 organizations involved in
long term care or infection prevention.
Roundtable meeting participants reviewed
submitted practices to determine the extent to
which they considered the practice to be a)
innovative, b) effective, and c) systematic.
Population Studied: Infection prevention and
control practices in nursing homes and assisted
living facilities
Principal Findings: The literature review
identified 88 articles, 28 guidelines and 50
guidance documents. Though many materials
existed, few specifically addressed high
reliability per se. Instead many referred to safety
culture, person-centered care and
transformational change. This triggered the
research team to develop a crosswalk of the key
concepts in high reliability with those long term
care concepts which had been the focus of
existing initiatives such as the Advancing
Excellence campaign and the Quality Assurance
Performance Improvement program.
The roundtable meeting was held July 26, 2013
during which 18 invited experts discussed high
reliability and infection prevention expectations
from the perspective of different stakeholders
(residents, physicians, administrators and
infection preventionists). The call for practices
yielded 61 submissions eligible for review, 34
from free-standing NHs, 9 from hospital-based
NHs, 7 from ALFs and 17 from combined or
other facilities. Five submissions scored highest
overall against the criteria. Most submissions
reflected standard infection control practices
such as hand hygiene and contact precautions.
Several of the submitted practices were
incorporated into a free on-line educational
module scheduled for release mid 2014. Also,
an on-line searchable database derived from the
scoping literature review will be available on a
website accessible to the public.
Conclusions: While the term high reliability is
relatively unfamiliar to the long term care field,
many of the concepts and practices are already
incorporated into ongoing quality improvement
initiatives.
Implications for Policy, Delivery, or Practice:
It is important to highlight the commonalities
across long term care quality improvement
models and strategies to avoid confusion and to
promote the adoption of a learning environment
and quality improvement mindset among
persons expected to implement the practices
Funding Source(s): AHRQ
Poster Session and Number: B, #634
Antipsychotic Use and Effectiveness of
Physician Accountability and
Communication in Nursing Homes
Shubing Cai, University of Rochester; Orna
Intrator, University of Rochester
Presenter: Shubing Cai, Assistant Professor,
University of Rochester
shubing_cai@brown.edu
Research Objective: Antipsychotics are
frequently used in nursing homes (NHs) despite
increasing concerns about the safety of such
medication for older adults. Although there is an
increasing recognition of the importance of
physician involvement in NH quality, it is unclear
whether effective communication and
collaboration between a physician and NH staff
is associated with reduced antipsychotic use.
Study Design: A Director of Nursing (DON)
survey was conducted among a stratified
random sample of NHs in the U.S. between
2009 and 2010. The survey data were linked
with 2009 Minimum Data Set (MDS) resident
assessments and NH federal certification
information from the OSCAR.
We identified long-stay NH residents (in NH >90
days) aged 65 years or older who resided in the
surveyed NHs in 2009. Residents with bipolar,
schizophrenia, hallucination and delusion were
excluded from the final analysis.
Outcome variable (dichotomous) was defined as
whether a resident had any use of antipsychotic
medications reported on MDS resident
assessments in 2009.
The main independent variable of interests was
a measure of the DON’s perception of whether
and how NH staff's needs of physician expertise,
experience and regulatory required services
were met. This was measured by the
Effectiveness of Physician Accountability and
Communication (EPAC) score, which was
derived from DON responses to 10 survey
questions relating to physicians’ attentiveness,
physician communication, and staff concerns
about physician practice. For example,
questions ask how often physicians
communicate with other staff about the care of a
resident, and how often the DON has to recontact physicians because of unclear
medication orders. Based on the distribution of
the score, we categorized NHs into three
groups, indicating high (highest quartile),
medium (2nd and 3rd quartiles), or low EPAC
(lowest quartile).
Covariates included individual characteristics
(e.g. age, gender, CPS, ADL, bedfast, use of
physical restraints, depression, anxiety,
dementia and other comorbidities), and NH
characteristics(e.g. ownership status, facility
size, staffing level).
A logistic regression model at the resident level
with NH random effects was estimated to
examine the relationship between EPAC and the
presence of antipsychotics, controlling for
individual and facility level characteristics and
accounting for survey weights.
Population Studied: 179,062 eligible longstayers were identified in 1,961 surveyed NHs.
Principal Findings: The prevalence of
antipsychotic use for this cohort was 22.3% in
2009. Individual characteristics were associated
with antipsychotic use. For example, longstayers with antipsychotic prescriptions were
more likely to use physical restrains (5.7%
versus 2.4%), more likely to have anxiety (8.8%
versus 5.5%), depression (63.2% versus 51.7%)
and dementia (17.1% versus 13.4%). Controlling
for personal and NH characteristics, the randomeffects logistic regression model indicated that
long-stay residents in NHs in the upper quartile
of EPAC scores were 8% less likely to use
antipsychotics than other residents who were in
NHs in the lowest quartile of EPAC scores.
Conclusions: Effective communication between
physician and NH staff was associated with a
lower prevalence of antipsychotic use in NHs.
Implications for Policy, Delivery, or Practice:
Policy interventions targeted at improving
physician’s involvement and collaboration with
NH staff may ameliorate some of the potentially
harmful antipsychotic prescriptions to NH
residents.
Funding Source(s): NIH
Poster Session and Number: B, #635
Healthcare Worker Influenza Vaccination in
Oregon Nursing Homes: Correlates of
Facility Characteristics
Lauren Campbell, University of Rochester
Medical Center; Qinghua Li, University of
Rochester Medical Center; Yue Li, University of
Rochester Medical Center
Presenter: Lauren Campbell, Doctoral
Candidate, University of Rochester Medical
Center
lauren_campbell@urmc.rochester.edu
Research Objective: Nursing home [NH]
employee influenza vaccination is associated
with reductions in morbidity and mortality among
NH residents. Little is known regarding
associations between facility characteristics and
employee vaccination rates [EVRs]. Resource
dependence theory was used to identify facility
and market characteristics representing
dependency, organizational, and environmental
components that may be associated with EVRs.
It was hypothesized that, generally, facilities with
greater resource availability, higher efficiency,
and higher quality would have higher EVRs and
would be more likely to achieve Healthy People
vaccination goals, controlling for market
characteristics. These facilities should be better
poised to provide key components of vaccination
programs from the literature, such as free and
convenient access to vaccine, and formal
education.
Study Design: Data on employee vaccination
rates and education programs were gathered
from reports from the Office for Oregon Health
Policy and Research for three influenza seasons
from 2009-2012. Oregon NH data were merged
with OSCAR and AHRF files, from which facilitylevel characteristics and market controls were
obtained. Outcome variables of multivariate
linear and logistic regression models were EVR
per facility/year, whether formal education was
conducted, and whether 2010, 2015, and 2020
Healthy People employee vaccination targets
were met [greater than or equal to 60-percent,
70-percent, and 90-percent, respectively].
Population Studied: NHs reporting sufficient
data to calculate an EVR were included. Based
on information obtained from 2009-10, 2010-11,
and 2011-12 surveys, EVRs were calculated for
81-percent, 91-percent, and 98-percent of NHs,
respectively.
Principal Findings: On average, chain-affiliated
NHs had 9-percent higher EVRs [P=0.01] and
were 73-percent more likely to achieve 60percent EVR [P=0.05], compared to individual
facilities. For-profit NHs had, on average, 8percent lower EVRs [P=0.04], compared to notfor-profit NHs. As bed size increased, facilities
were more likely to achieve 60-percent EVR
[OR=1.01, P=0.05] and had increased odds of
conducting formal education [OR=1.01, P=0.01].
Hospital-based NHs were more likely to achieve
90-percent EVR [OR=12.13, P=0.04], compared
to free-standing NHs. Although literature
suggests that NHs with higher proportions of
Medicaid residents have lower quality, results
showed that a 10-percent increase in proportion
of Medicaid residents was associated with a 2percent increase in EVR [P=0.01] and increased
odds of achieving 60-percent [OR=1.20,
P=0.004] and 70-percent [OR= 1.14, 0.05] EVR.
Conclusions: Results show an association
between facility characteristics and EVR. NH
EVRs are inadequate, which may jeopardize
residents’ health. Although some NHs may
consistently meet recommended EVRs, the
majority does not, suggesting that this is an area
for improvement.
Implications for Policy, Delivery, or Practice:
OHPR administered surveys to publicly report
EVRs to improve patient safety by giving NHs an
incentive to improve vaccination. However,
EVRs did not improve, decreasing slightly over
time, and proportion of facilities conducting
formal education decreased substantially. It
may be necessary to target low-performing
facilities to achieve significant improvements.
Identifying facility characteristics associated with
lower than recommended EVRs would help
target facilities for improvement. However,
significant correlates of this study cannot be
easily addressed by NH management or
policymakers. Without policy change
encouraging key components of vaccination
programs, such as free and convenient access
to vaccine and formal education, public reporting
may be insufficient to improve EVRs.
Funding Source(s): NIH AHRQ
Poster Session and Number: B, #636
Frailty Trajectories in an Elderly PopulationBased Cohort
Alanna Chamberlain, Mayo Clinic; Lila Rutten,
Mayo Clinic; Sheila Manemann, Mayo Clinic;
Barbara Yawn, Olmsted Medical Center; Debra
Jacobson, Mayo Clinic; Chun Fan, Mayo Clinic;
Veronique Roger, Mayo Clinic; Jennifer St.
Sauver, Mayo Clinic; Brandon R. Grossardt,
Mayo Clinic
Presenter: Alanna Chamberlain, Assistant
Professor of Epidemiology, Mayo Clinic
chamberlain.alanna@mayo.edu
Research Objective: Frailty, a concern for
aging populations, has been shown to predict
hospitalizations, admission to long-term care
institutions, and death. Little is known about
longitudinal changes in frailty. Therefore, the
goal of this study was to identify distinct frailty
trajectories (clusters of individuals following a
similar progression of frailty over time) in an
aging population and to estimate associations
between frailty trajectories and all-cause
mortality.
Study Design: Data from the 2005 population of
Olmsted County, MN aged 60-89 (n=16,443)
obtained from the Rochester Epidemiology
Project records-linkage system was analyzed. A
frailty index, a measure of variation in health
status based on an accumulation of deficits, was
calculated as the proportion of deficits present
using 32 deficits: body mass index, 17 chronic
conditions, and 14 activities of daily living (range
of frailty index: 0-1). Repeated measures of
frailty were calculated yearly from 2005 through
2012 and individuals with at least 3 frailty
measures were retained (n=12,205). K-means
cluster modeling for longitudinal data (Genolini C
and Falissard B, Computational Statistics, 2010)
was used to define the number and shape of
frailty trajectories by sex and decade of age.
Trajectories did not differ by sex; therefore, sex
groups were combined for further analyses.
Associations between frailty trajectories and
death within each age stratum were assessed
with logistic regression.
Population Studied: The 2005 population of
Olmsted County, MN, aged 60-89 (n=16,443).
Principal Findings: Median baseline frailty
increased with increasing age (60-69 years:
0.11, 70-79 years: 0.14, and 80-89 years: 0.19).
In the 60-69 year olds, 3 trajectories were
identified, while 2 trajectories were identified in
the 70-79 and 80-89 age groups. Among the
60-69 year olds, 11.8% were in the highest
frailty trajectory and the mean frailty index
increased from 0.27-0.41 over the 8 years of
follow-up. In the 70-79 year olds, 33.9% were in
the highest frailty trajectory and the mean frailty
index increased from 0.24-0.41. In the 80-89
year olds, 44.3% were in the highest frailty
trajectory and the mean frailty index increased
from 0.28-0.49. An increased risk of all-cause
mortality was found with increasing frailty
trajectory in all ages. Among the 60-69 year
olds, those in the middle and highest frailty
trajectories were respectively 3 and 15 times
more likely to die during follow-up compared to
those in the lowest frailty trajectory (middle OR:
3.06, 95% CI: 2.31-4.05 and highest OR: 15.39,
95% CI: 11.51-20.57). For the 70-79 year olds,
those in the highest frailty trajectory experienced
a 6-fold increased odds of death (OR: 5.91, 95%
CI: 5.02-6.98). Finally, among the 80-89 year
olds, those in the highest frailty trajectory were
5-fold more likely to die compared to the lowest
frailty trajectory (OR: 4.85, 95% CI: 3.99-5.89).
Associations remained significant after
adjustment for age, sex, and baseline frailty
index.
Conclusions: The optimum number of frailty
trajectories differed by age group. Those in the
highest frailty trajectories experienced worse
survival, even when baseline frailty was taken
into account.
Implications for Policy, Delivery, or Practice:
Frailty trajectories may offer a way to target
aging individuals at high risk of death for
therapeutic or preventive interventions.
Funding Source(s): Other Rochester
Epidemiology Project and Center for the Science
of Health Care Delivery
Poster Session and Number: B, #637
Differences in Communication Performance
Perceptions Between Caregivers’ Selfevaluation and Elders’ Satisfaction
Shu-Yuan Chao, Hungkuang University; YingChih Chen, Hungkuang University; Mary Jo
Clark, Hahn School of Nursing and Health
Science
Presenter: Shu-Yuan Chao, Professor,
Hungkuang University
sychao@sunrise.hk.edu.tw
Research Objective: Communication skill is the
foundation for an effective professional
relationship between caregivers and care
receivers. Effective communication is particularly
important in providing quality care to a growing
population of older clients. The purpose of this
study was to compare caregivers’ self evaluation
and elders’ satisfaction with caregivers’
communication performance.
Study Design: A cross sectional survey was
conducted using a stratified sampling method.
We randomly recruited two nursing homes, two
senior care centers, and two home care
agencies in the Taichung area of central Taiwan.
All residents and their caregivers in each
institution were invited to participate. The “
caring-oriented communication inventory” was
used to measure caregivers’ communication
performance when they interacted with the
elderly. Data were collected by means of selfadministration of the tool by caregivers and faceto-face interview with the elders.
Population Studied: 120 elders and 115
caregivers participated in this study. Mean ages
were 77.60 years and 52.08 years, respectively.
Most of the elders were educated at the primary
school level, and most caregivers graduated
from high school.
Principal Findings: Elders’ degree of
satisfaction was higher than caregivers’ ratings
for some items, such as “having a friendly
attitude”, and “greeting elders”. On items related
to “sensitive to the elders’ feelings, empathy for
elders emotional reaction”,"considering elders’
culture background when communicating with
them” , "considering elders’ visual or hearing
ability when interacting” and“considering elders’
cognition function”, however, elders rated their
level of satisfaction lower than the caregivers’
self evaluation scores. These differences were
statisticallysignificant.
Conclusions: Differences existed in
communication performance perceptions
between the caregivers’ self-evaluation and
elders’ satisfaction. Caregivers were adept at
expressing a friendly attitude, but inadequate
with respect to empathy and functional
considerations when they interacted with the
elderly.
Implications for Policy, Delivery, or Practice:
Additional training may be needed for caregivers
to effectively incorporate functional
considerations and empathy in their
communications with the elderly.
Funding Source(s): Other National Science
Council, Taiwan
Poster Session and Number: B, #638
Assessing Training Standards for Personal
Care Aides: Which States are Leading the
Way?
Susan Chapman, University of California, San
Francisco; Abby Marquand, PHI; Dorie Seavey,
PHI
Presenter: Susan Chapman, Associate
Professor, University of California, San
Francisco
susan.chapman@ucsf.edu
Research Objective: The study is being
conducted as core research for the Health
Workforce Research Center focused on long
term care in collaboration with PHI
(Paraprofessional Healthcare Institute). In the
absence of federal minimum training standards
or guidelines for personal care aides (PCAs), as
exist for other direct-care occupations such as
home health aides and certified nurse aides,
states have been left to develop their own
standards which often differ from state to state
and even between programs in a given state.
The objectives are to assess the rigor and
uniformity of state training standards for
personal care aides (PCA) using a 50-state
database developed by PHI in prior research,
and to identify leading states with the most
developed requirements. This study will also
analyze federal and state policy issues related to
advancing training standards for PCAs
Study Design: The PHI database relies on
documents search and analysis of state
administrative codes, licensing laws, department
regulations, and Medicaid provider manuals and
waiver documents. Search was limited to
standards within state Medicaid programs,
particularly state plan Personal Care Option and
home and community based waiver programs
for the elderly and persons with disabilities.
Documents were coded for rigor and uniformity
of standards across the state’s Medicaid
programs. Study investigators will independently
rate each state and develop a consensus on
indices for rigor and uniformity.
Population Studied: 50 states in the U.S. and
the District of Columbia
Principal Findings: Previously collected PHI
data indicate that few states have welldeveloped, or rigorous training standards for
PCAs. In fact, while 25 states (49%) require
training in at least one program, half of those
programs require 40 hours or less; 23 states
(45%) have at least one program with no training
requirements at all; 21 states (41%) have
uniform requirements across states however
only a few specify skills or curricula required for
PCA training. Ten states are rated as “leaders”
based on factors such as standardized curricula,
list of competencies required, number of hours,
exam and/or certification requirements.
Conclusions: State training requirements for
PCAs are underdeveloped compared to other
entry level occupations in long term care.
Training requirements is leader states may be
useful to other states as they build training
programs in response to state or federal
mandates.
Implications for Policy, Delivery, or Practice:
There is growing interest at the state and federal
level to improve and standardize training
requirements for PCAs. The growth of home and
community based services, Medicaid waiver
programs, dual-eligible demonstration projects,
and capitation of long term care services and
supports has fueled interest in ensuring a high
quality workforce of caregivers providing home
and community based services. PCAs are
projected to be the nation’s second-fastest
growing occupations over the coming decade
due to an increasingly older population and
consumer preferences for HCBS rather than
institutions. Training may increase the status
and pay of this workforce who are often low
paid, without benefits, and often living in poverty.
State certification of personal and home care
aides could also help to attract workers into this
high demand field.
Funding Source(s): HRSA CMS, Ford
Foundation, SCAN Foundation, NIDRR
Poster Session and Number: B, #639
Prognostic Communication Needs as
Reported by Widowed Fathers
Devon Check, University of North Carolina at
Chapel Hill; Eliza Park, University of North
Carolina at Chapel Hill; Justin Yopp, University
of North Carolina at Chapel Hill; Donald
Rosenstein, University of North Carolina at
Chapel Hill
Presenter: Devon Check, Doctoral Student,
University of North Carolina at Chapel Hill
devon.check@gmail.com
Research Objective: Effective communication
about prognosis is a critical aspect of providing
high quality care for patients with terminal
illnesses. It is crucial for patients to understand
their prognoses in order to facilitate informed
decision-making about current and future care.
This may be especially important for terminally ill
parents and their partners, who may have
unique needs for communication about
anticipated death because of concerns related to
their parental roles. The purpose of this
investigation was to explore experiences of end
of life communication with providers, as reported
by bereaved fathers whose wives died from
cancer.
Study Design: As part of an open-access
educational website for widowed fathers due to
cancer, we conducted an online survey about
men’s experiences during their wives’ illnesses.
The survey included an open-ended question
regarding prognostic communication between
the respondents and their wives’ providers:
“What is the most important thing you would like
us to know about whether/how your wife’s
doctors communicated with you about her
anticipated death? What do you wish had been
different, if anything?” We performed traditional
content analysis of men’s responses. Two
researchers coded and categorized the data.
Topics identified in the text were assigned
descriptive codes and categories were created
from groups of codes using NVivo10.
Population Studied: 242 men who selfidentified as fathers who had dependent-age
children at the time of their wives’ deaths
answered this question. Average respondent
age was 47 (range 28-69). Approximately 90%
of respondents were Caucasian, and
approximately 70% were college graduates.
61% of respondents reported that their wives
had received hospice services. The majority of
respondents completed the survey within six
months of their wives’ deaths. Prevalence of
depressive symptoms in this sample was high,
with a mean Center for Epidemiologic Studies
Depression Scale score of 22.6 ± 12.4 (range 060; scores >16 reflect clinically significant
depressive symptoms).
Principal Findings: Men’s responses
highlighted several areas of need. For example,
28% of men discussed the importance of honest
communication from their providers. Although
most men felt that their providers were
transparent, 11% felt misled by information that
was vague or overly optimistic. 26% of men
discussed positive or negative examples of
physician bedside manner, or the way in which
their providers delivered prognostic information.
Specifically, 12% cited the importance of
provider compassion. 33% felt they needed
more information about what to expect, including
disease course, life expectancy, and palliative
and hospice care.
Conclusions: Our data suggest several
potential areas for improvement in prognostic
communication between providers, patients, and
their partners. Future research should seek to
identify interventions (e.g., programs specifically
for parents) that may improve satisfaction with
prognostic communication for this population.
Implications for Policy, Delivery, or Practice:
Bereaved caregivers are at risk for major
depression, and parents with terminally ill
partners may have unique psychosocial needs.
Increased provider focus on the communication
preferences of both patients and their partners
may improve outcomes for bereaved family
members.
Funding Source(s): Other University Cancer
Research Fund
Poster Session and Number: B, #640
Producing and Evaluating the Effectiveness
of a Video as an Hospice Educational Tool
for Hispanics
Kyusuk (stephan) Chung, California State
University, Northridge; Samira Moughrabi,
California State University, Northridge; Frankline
Augustine, California State University,
Northridge
Presenter: Kyusuk (stephan) Chung, Associate
Professor, California State University, Northridge
stephan.chung@csun.edu
Research Objective: Hospice care is becoming
increasingly important in the care of the elderly,
accounting for nearly half of Medicare decedents
in 2011. However, half of the decedents die
within a week—a figure that has remained
unchanged over the past three decades.
Research suggests that doctors find it hard to
admit when no curative treatment is available
and by the time the hospice option is discussed,
it is typically too late. This is particularly true for
the Hispanic elderly, who are more likely than
non-Hispanic whites to be at the
hospital/emergency room at the time of hospice
enrollment decisions, emphasizing the
importance of empowering Hispanic
patients/family caregivers to initiate end-of-life
care discussions. Nonetheless, it is challenging
to educate the Hispanic elderly, who tend to
have language barriers and lower education
levels.
Study Design: This work consists of two
phases: 1) video production and 2) video
evaluation. As an educational tool, we have
produced a video featuring a Mexican American
patient at the end stage of Alzheimer’s. The
video also films the patient’s husband as the
primary caregiver and her Spanish speaking
clinical caregivers, including a doctor, a RN and
LVN. An Alzheimer’s patient and a family
caregiver were featured in the video for two
reasons: 1) Hospice care is intended both for a
patient and his/her caregivers; and 2)
Alzheimer’s disease and other dementias are
more prevalent in the Hispanic population. The
Principal guiding the production of the video was
to show, not tell the answers to the three
questions that the literature identified as the
most wanted by patients/families prior to hospice
enrollment decisions: How often will hospice
send someone out to visit?; Who pays for
hospice?; and What kinds of practical support
and assistance does hospice provide? For the
evaluation of the video, we recruited students of
Mexican origin from a public university and staff
members from a local hospice agency in the Los
Angeles-Long Beach area where the Mexican
American population exceeds 4.5 million. We
compared the questionnaires administered
before and after viewing the video.
Population Studied: Mexican Americans
Principal Findings: Forty-six participants
completed pre-viewing questionnaires. Eightythree percent of the viewers reported an
‘average’ or ‘more than average’ level of
knowledge of hospice. However, after watching
the video, all but two of them reported that their
‘previous’ knowledge had been shallow. These
two participants, whose relatives had received
hospice care, confirmed that they harbored the
same experience/feeling toward hospice care as
the patient’s husband in the video. Qualitative
analyses of open-ended questions confirmed
that the video met the Principal guiding the
production of the video—show, not tell. With
regards to the best aspect of the video, the most
frequently used descriptions were personal,
authentic, trustful, real, real testimony, culturally
credible, can see, and showing instead of
talking. At pre-viewing, fifteen individuals
responded, ‘no’ or ‘I am not sure’, to the
question of whether or not they would
recommend hospice care to their relatives.
Fourteen of them changed their minds upon
viewing the video and reported that they would
strongly recommend hospice care, suggesting
that the required knowledge helped change their
minds. When asked how favorably the video
compares with typical hospice-awareness
material in Spanish, six of the twelve local
hospice workers indicated that they had never
seen the Spanish material before.
Conclusions: Our video has been favorably
accepted as useful as an educational tool in
sharing the benefit levels of hospice care with
Mexican Americans.
Implications for Policy, Delivery, or Practice:
While hospice awareness in the US has
increased, there is a need for a more effective
educational tool like our video to reach out to the
Hispanic population and facilitate timely actions.
Funding Source(s): NIH
Poster Session and Number: B, #641
Hospice Benefit Levels in Covered California
Managed Care Plans
Kyusuk (stephan) Chung, California State
University, Northridge; Joelle Jahng, California
State University at Northridge; Syuzanna
Petrosyan, California State University at
Northridge; Victoria Yim, Marlborough School
Presenter: Kyusuk (stephan) Chung, Associate
Professor, California State University, Northridge
stephan.chung@csun.edu
Research Objective: This study contains
findings from a survey of managed care plans
currently participating in the California Health
Benefit Exchange in order to determine hospice
benefit levels. Our study is timely in that it
follows up on how California, Massachusetts,
and many other states have mandated hospice
benefits as the minimum requirement for
participating health plans. There is currently a
lack of information about hospice coverage for
privately insured patients. Our study’s aim is to
help fill this gap.
Study Design: Focusing on California, we
conducted two surveys – one for large hospice
agencies in operation in California and the other
for health care insurance providers that
participate in Covered California. Using the
2011 California Annual Home Health and
Hospice Utilization Database spreadsheet
obtained from the Office of Statewide Health
Planning and Development, 75 hospice
agencies were chosen that reported ‘more than
500’ as the number of hospice patients served in
2011. In addition, California has selected and
certified 13 individual health plans, called
“Covered California” health plans, which serve
19 pricing regions. Some plans are locally
based, while others serve the full spectrum of
the pricing regions. We surveyed these 13
Covered California Individual Health Plans.
Population Studied: 75 hospice agencies were
chosen that reported more than 500 as the
number of hospice patients served in 2011. The
13 Covered California Health Plans
Principal Findings: Compared with the benefit
levels for Medicare and Medicaid patients, we
found that privately insured patients are subject
to more stringent requirements both for hospice
enrollment and for prior approval of specific
hospice services after hospice enrollment.
Mandating hospice coverage alone may not be
sufficient to facilitate hospice access among the
younger population increasingly covered by
managed care health plans.
Conclusions: Our survey of five health plans
certified by Covered California and twenty-five
large sized hospice agencies in California
revealed that hospice benefit levels by managed
care private insurances are more stringent than
those by Medicare/Medicaid hospice benefits.
Particular concerns arise from the gatekeeping
practice of delaying hospice enrollment, further
aggravating the on-going issue of shortened
hospice lengths of stay.
Implications for Policy, Delivery, or Practice:
Requirements of prior approval from specific
hospice services and limited availability of
contract for hospice agencies may act as
barriers to access hospice care among the
privately insured younger population. If
gatekeeping practices continue, mandating
hospice coverage for newly insured individuals
alone may not be sufficient to increase the
current low hospice utilization rates among the
younger population.
Funding Source(s): NIH
Poster Session and Number: B, #642
Integration of Palliative Care into the PatientCentered Medical Home
Elizabeth Ciemins, Billings Clinic; Deric Weiss,
Confluence Health; Diane Arkava, Billings Clinic;
Wendy Riehl, Billings Clinic; Dustin Dickerson,
Billings Clinic Center for Clinical Translational
Research
Presenter: Elizabeth Ciemins, Director, Billings
Clinic
eciemins@billingsclinic.org
Research Objective: The primary objective of
this project was to integrate palliative care
services into the patient-centered medical home
at Billings Clinic through early identification and
management of medically complex patients in
need, completion of advance directives in
primary care, and increased quality of life, and
symptom management. A second objective was
to improve relational coordination among the
care team as a way to facilitate improved
communication between disciplines and
successful integration of services.
Study Design: A longitudinal cohort design
followed enrolled patients for six months and
measured symptoms and quality of life at
baseline, one, three, and six months. The
intervention consisted of the identification of
patients in need of palliative care services,
integration of patient identification into the
current “nurse navigator” role, and provision of
needed services, including Dignity Therapy, a
brief psychotherapy approach based on an
empirically validated model of dignity in the
terminally ill.Patients were screened, identified,
and referred by physicians, nurse practitioners,
nurse navigators, and social workers.
Population Studied: Using a modified
screening tool and direct referral, adult primary
care patients were identified who were
determined to have unmet needs and who were
suffering from a life-limiting or serious illness.
Principal Findings: Fifty-six patients were
identified with palliative care needs. Of these,
forty-six were referred for a palliative care
consultation. A total of 22 patients (41%)
patients consented to participate in the study.
Ten patients received a palliative care
consultation and twelve received Dignity
Therapy. At three months, symptoms of pain,
tiredness, drowsiness, appetite, dyspnea, and
wellbeing improved, although statistical
significance was not obtained due to small
sample size. Self-reported quality of life scores
improved at one-month (p=.05) but declined to
baseline scores at 3-months post-intervention.
However, 25% of respondents reported ‘poor’ or
‘worst possible’ overall quality of life at baseline
and none reported ‘poor’ or ‘worst possible’ at 1or 3-months post intervention. The relational
coordination measure of accurate
communication improved among team members
(p=.04); other domains, e.g., frequent
communication, and problem-solving
communication trended toward improvement
(p=0.17 – 0.18).
Conclusions: Palliative care, including Dignity
Therapy, can be successfully integrated into the
patient-centered home if focus is placed on
practice team relationships. Patient quality of life
may be improved but may not be sustained both
due to the nature of serious illness, and without
continued patient contact.
Implications for Policy, Delivery, or Practice:
This work has broad implications for the patientcentered medical home and outpatient palliative
care. Earlier identification of patients in need of
palliative care services has great potential to
improve care of patients with serious illness and
avoid costly and undesired medical interventions
at the end-of-life. Integration of these services
into the medical home is a logical approach to
ensure the seamless delivery of health care
services and the provision of high quality, costeffective care.
Funding Source(s): Other Pacific Source
Foundation
Poster Session and Number: B, #643
A Qualitative Study of Attributes
Contributing to a Successful Palliative Care
Interdisciplinary Team
Elizabeth Ciemins, Billings Clinic; Jeannine
Brant, Billings Clinic; Elizabeth Mullette, Billings
Clinic Center for Clinical Translational Research;
Dickerson Dustin, Billings Clinic Center for
Clinical Translational Research; Kersten Diane,
Billings Clinic
Presenter: Elizabeth Ciemins, Director, Billings
Clinic
eciemins@billingsclinic.org
Research Objective: The primary objective of
this project was to discover the experiences,
roles, and characteristics of health care
professionals who work with patients in need of
palliative or end-of-life care services that are
necessary for a positive team and patient
experience. A secondary objective was to
determine the factors that promote a positive
team experience.
Study Design: In-depth, semi-structured
palliative care health care professional
interviews were conducted and transcribed. Five
qualitative investigators independently reviewed
data using grounded theory methodology and
preliminary interpretations. A combined
deductive and inductive iterative qualitative
approach was used to identify recurring themes
following five steps: (1) develop a priori template
of codes based on prior research and interview
guide; (2) test codes for reliability; (3) identify
meaningful units of text to represent themes; (4)
define emerging themes; and (5) compare and
contrast themes across interviews. Themes and
patterns were further refined and new themes
co-generated. The process was facilitated by the
Atlas.ti data analysis software.
Population Studied: A purposive sample of 10
health care professionals who regularly provide
palliative care services were interviewed.
Interviewees included physicians, nurses, nurse
practitioners, chaplains, and social workers
working on a palliative care team at an
integrated health care delivery system.
Principal Findings: Four individual and five
team attributes emerged as necessary for a
positive team/patient experience. Individual
attributes included self-awareness, spirit of
inquiry, humility, and comfort with dying.
Interdisciplinary team attributes included shared
purpose, holistic thinking, relational coordination,
trust in the process, and propensity to inquire
humbly. Professional and personal motivations,
such as helping the suffering as a spiritual
calling, personal fulfillment, encouragement by
peers, and a reinforcing team experience,
contributed to, and were reinforced by, a positive
team/patient experience.
Conclusions: Interdisciplinary palliative care
teams have the potential to significantly impact
patient and team experiences during the care
process for the seriously or terminally ill. Specific
individual and team attributes are necessary for
successful interaction with patients that are
satisfying to patients, families, and care team
members.
Implications for Policy, Delivery, or Practice:
This work has profound implications for
organization-level understanding of the
importance of interdisciplinary teams in the
provision of palliative care services. This work
should help promote the necessity of these
teams and their subsequent support. Further,
individuals who provide palliative care services
in the absence of a team may be informed by
this work of the potential benefits of providing
services as part of an interdisciplinary team, and
the important attributes of those teams that
result in success. Potential impacts of increased
use of interdisciplinary teams for the provision of
palliative care services include greater patient,
family, and team satisfaction, and a reduction in
provider and staff burn-out. Findings from this
study support interventions that focus on
relational coordination and application of a
complex systems theory approach to team
development.
Funding Source(s): No Funding
Poster Session and Number: B, #644
Profiling the Relationship of the Burden of
Informal Caregivers and Caregiving Mental
and Physical Health
Chris Claeys, KJT Group; Lynn Kistner, KJT
Group; Dan Wasserman, KJT Group; Kenneth
Tomaszewski, KJT Group
Presenter: Chris Claeys, Associate
Methodologist, KJT Group
chrisc@kjtgroup.com
Research Objective: More than sixty-five
million adults in the United States provide
informal care (unpaid) to a relative or friend. This
study examines the physical and mental health
of informal caregivers as it relates to their
burden levels. Developing deeper understanding
of how burden relates to caregiver health is
critical to the design of programs or services to
assist caregivers.
Study Design: Longitudinal online surveys
among nationally representative independent
cross-sectional cohorts were conducted. 3,000
adults in the United States were surveyed in
2013 with 705 indicating that that they provided
informal care sometime within the past year; 154
of these were removed due to missing data
values. The remaining caregivers (n=551) were
segmented based on level of burden (e.g., time
spent providing care, number of adults assisted,
money spent). Model-based cluster analysis was
used to segment caregivers. Bayesian Ordered
Probit Regression fitted probabilities were
included in the cluster analysis to help
differentiate segments based on mental and
physical health (measured on an ordered 1-5
scale). Descriptive statistics were used to
compare segments.
Population Studied: American adults were
surveyed in KJT Group’s LightSource Poll. A
stratified sampling plan was designed based on
U.S. Census figures and the Statistical Abstract
of the United States. Stratification variables
included gender, household income, age and
region. Respondents were classified as informal
caregivers if they personally provided unpaid
care to a relative or friend 18 years or older.
Principal Findings: Several cluster solutions
were evaluated and a four segment solution was
chosen to maximize differences across
segments and minimize differences within
segments while providing pragmatic
interpretations: 1. Low frequency minimally
involved caregivers (n=78, 14%), 2. High
frequency moderately involved caregivers
(n=100, 18%), 3. Part-time moderately involved
caregivers (n=199, 36%), and 4. Full-time
heavily involved caregivers (n=174, 32%).
Caregivers in Segment 4 report the worst
physical and mental health, while caregivers
from Segment 3 report the best physical and
mental health. Segments are also differentiated
based on use of outside support services, and
demographics (e.g., employment, healthcare
coverage).
Conclusions: This study suggests informal
caregivers are varied in how much assistance
they provide to adults; however, they can be
grouped together based on their efforts. Based
on employment rates and health ratings, the
data suggest that caregiving may be more
stressful than other full-time jobs. Outside
support service assistance is related to
caregiver mental and physical health.
Implications for Policy, Delivery, or Practice:
Based on this research, we identified a link
between informal caregiving burden and the
caregiver’s physical and mental health. Policy
focus should be directed to those caregivers
who are providing the greatest assistance; this
will help them in their caregiving efforts and with
their physical and mental health. In addition,
policymakers should consider the needs of new
caregivers. Many show strains on their physical
and mental health as they have not developed
appropriate coping mechanisms. Policymakers
should direct services (e.g., coaching, training)
to those new to caregiving on what to expect in
their role as a caregiver and to ensure they
focus on their own personal health in addition to
the health of the adult they assist.
Funding Source(s): No Funding
Poster Session and Number: B, #645
Assisted Living Facilities and Nursing Home
Case-Mix in Stable Markets
Jan Clement, Virginia Commonwealth
University; Jaya Khushalani, Virginia
Commonwealth University
Presenter: Jan Clement, Professor, Virginia
Commonwealth University
jclement@vcu.edu
Research Objective: Assisted living facilities
(ALFs) can offer an alternative to nursing home
(NH) care for the elderly with less intensive care
needs. As a result, in markets with ALFs, NHs
may attract residents with more complex needs,
increasing case-mix. Research regarding
competition between these two sectors has
been limited. To see how nursing homes have
adapted to potential competition from ALFs, this
study examines the relationship between ALF
market capacity and nursing home case-mix in
markets in a state with a stable ALF and NH
presence.
Study Design: The multivariate cross-sectional
study examines average nursing home RUGS
Nursing Case Mix Index and average Activities
of Daily Living (ADL) scores from the Brown
University Long-Term Care Focus web site. In
addition to ALF bed capacity (assisted living
beds per 1,000 population over 65 in the market
area), other market variables include nursing
home competition, number of home health care
agencies, and per capita income. Nursing home
variables include ownership, chain membership,
continuing care retirement community status,
and payer mix. Data for these variables come
from the Virginia Department of Social Services,
US Department of Commerce Bureau of
Economic Analysis (BEA), Online Survey
Certification and Reporting (OSCAR) system,
among others. Market areas are defined as the
county or combined city-county definition used
by the BEA.
Population Studied: All nursing homes and
assisted living facilities in Virginia in 2010, a
state where the majority of markets have
experienced no or very limited growth among
assisted living facilities. Measures of
competition among nursing homes have also
been stable.
Principal Findings: The relationship between
ALF capacity and NH average ADL was not
linear. Average nursing home ADL score was
significantly higher for NHs in markets in the
mid-range of ALF beds per 1,000 population
over 65 than in markets without ALFs (p < .05).
Average NH ADL scores in markets with the
highest ALF bed capacity did not differ from
those in markets without ALFs but were
marginally significantly lower than those for the
mid-ALF market NHs (p< .06). The average
RUGS NCMI was unrelated to competition from
assisted living facilities.
Conclusions: The relationship between
assisted living facility market capacity and
nursing home resident case-mix may be more
complex than the linear model previously
hypothesized. NHs where there is a higher ALF
capacity may have learned how to compete for
the least complex residents. Or ALFs may have
adapted to offer higher levels of care, thereby,
decreasing the case mix of NHs in the market.
Implications for Policy, Delivery, or Practice:
Our findings suggest that where there ALF
capacity is limited, some NH residents may be
candidates for care in assisted living facilities
where it may be less costly. More research is
needed to understand the nature of ALF and NH
competition in markets with high ALF capacity.
Funding Source(s): No Funding
Poster Session and Number: B, #646
Treatment Intensity during the Terminal
Hospitalization among Patients with a Do Not
Resuscitate Order: Do Economic Incentives
Play a Role?
Anna Davis, University of California at Los
Angeles Department of Health Policy and
Management
Presenter: Anna Davis, PhD Student, University
of California at Los Angeles Department of
Health Policy and Management
annadavis@ucla.edu
Research Objective: Costs at the end of life
constitute a substantial portion of total US
medical expenditures. Patients who die in the
hospital setting increasingly have DNR orders to
limit their care. However, even in the setting of a
DNR order there is broad variation in the
intensity of care delivered to dying patients,
driven largely by patient and family preferences
and the patient’s severity of illness. This study
investigated whether financial incentives are
associated with intensity of treatment at the end
of life, such that hospitals and/or providers
pursue more intensive treatment for patients
with high-value private third party insurance
coverage.
Study Design: This is an observational
retrospective cross-sectional cohort study. The
dependent variable for this analysis is the total
charges during the hospitalization, and my
primary regressor of interest is the patient’s
insurance status. This study uses data from the
California Office of Statewide Health Planning
and Development (OSHPD) public use Patient
Discharge Dataset (PDD) and Annual Financial
Dataset (AFD) for 2010. The analysis uses a
log-OLS model with correction for clustering at
the hospital level. Estimates were retransformed
to facilitate interpretability on the original scale,
and post-estimation with bootstrapping was
used to generate predictive margins and 95%
confidence intervals for each payer category.
Population Studied: The analysis includes
17,298 individuals who died in an acute care
setting in California during 2010, and who had a
DNR order in place during the first 24 hours of
their final hospitalization. The population was
also limited to adults aged 18 and older.
Principal Findings: After controlling for other
covariates in the model, indigent patients had on
average $8,323 higher total charges than
privately insured patients. Bootstrapped 95%
confidence intervals confirm that this difference
was statistically significant at the p<0.05 level.
There were no statistically significant differences
in expenditures between privately insured
patients and those with Medicare or Medicaid.
Conclusions: These findings refute the
hypothesis that economic incentives may drive
providers to offer more intensive treatment to
privately insured patients compared to those
without private insurance but with similar
severity of illness. This is an encouraging
finding, but it might be weakened by the fact that
this analysis was unable to completely control
for many of the key conceptual domains. In a
model that more completely controlled for these
factors, it is possible that future analyses might
identify economic incentives and resulting
differential treatment by payer category.
Implications for Policy, Delivery, or Practice:
Costs of care at the end of life continue to rise in
the US, and must be addressed. Additional
research is needed to elucidate the factors
associated with these costs, to assess whether
opportunities exist to improve patient and family
satisfaction while limiting unnecessary
expenditures. Future analyses should include
more detailed controls for patient health status
as well as explicit measures of patient/family
preferences, to better understand the interplay
between insurance status and treatment
intensity at the end of life.
Funding Source(s): NIH
Poster Session and Number: B, #647
Re-conceptualizing Medical Decisions: How
Home Hospice Care Fosters Patient and
Family Engagement and Decision-Making
Ellis Dillon, Palo Alto Medical Foundation
Presenter: Ellis Dillon, Post-doctoral Fellow,
Palo Alto Medical Foundation
ellis.c.dillon@gmail.com
Research Objective: Despite growing
acceptance that patient engagement and shared
decision-making should be goals of medicine,
organizational attempts to support these ideals
are limited and larger institutional structures
often constrain patient autonomy. Home hospice
care is a subset of medical care that has
consciously developed a philosophical and
practical approach to encouraging patient and
family engagement. This research examines
how home hospice care provides a different
strategy for improving patient engagement and
shared decision-making.
Study Design: This article is drawn from an
ethnographic study of home hospice care at a
large non-profit hospice in California. Data
include in-depth interviews with a range of
participants in hospice care and observation of
hospice workers home visits with patients and
family members. An original convenience
sample of patients was enrolled first.
Subsequently the hospice workers and family
members involved in that patient’s care were
recruited for the study.
Population Studied: The sample includes 55
home hospice participants: 18 patients, 11
family members/caregivers, and 26 hospice
workers. Data collection involved 48 interviews
and 23 episodes of observation of home hospice
work.
Principal Findings: I find that hospice
institutional structure and daily work practices
embody a holistic approach which assumes
patients and family members are the critical
experts in most instances of decision-making.
Differences in institutional structure, such as
providing care in patient homes and having an
interdisciplinary team approach focused on the
“whole person”, empower the patient and family
members. Likewise the interactions between
hospice workers and patients and family
members narrow the field of purely “medical”
decisions and broaden the field of decisions
open to patients. Hospice workers accomplish
this re-conceptualization by framing many
medical decisions as being more about what is
best for the patient and family and less about
medical expertise.
Conclusions: By broadening and reconceptualizing the idea of decision-making,
hospice workers enable patients and their
families to have more control over their medical
care and in many cases the process of dying
itself.
Implications for Policy, Delivery, or Practice:
While some attributes of hospice care are
unique, many techniques could be introduced or
accentuated in other models of medical care.
Funding Source(s): Other University of
California, San Diego
Poster Session and Number: B, #648
Examining States’ Efforts in Achieving a
Person-Centered, Balanced Long-term
Services and Supports (LTSS) System in a
Changing LTSS Environment
Susan Flanagan, IMPAQ International; Jennifer
Howard, IMPAQ International; Kerry Lida,
Centers for Medicare and Medicaid Services
(CMS)
Presenter: Susan Flanagan, Principal
Associate/LTC Practice Lead, IMPAQ
International
sflanagan@impaqint.com
Research Objective: Federal mandates and
initiatives (e.g., Americans with Disabilities Act,
Olmstead Decision, Affordable Care Act and the
DHHS Community Living Initiative) provide an
impetus for states to pursue community
integration for all individuals. However, there is a
gap in the availability of common indicators to
examine States’ efforts in achieving a personcentered, balanced long-term services and
supports (LTSS) system that offers older adults
and individuals with disabilities and chronic
conditions access to a full array of quality
services that assure independence, optimal
health and quality of life. As states continue to
reform their LTSS systems, there is growing
interest in examining their progress in attaining
and maintaining a person-centered approach to
service delivery and achieving a more equitable
balance between the provision of institutional
and home and community-based services. To
address this gap, the National Balancing
Indicator Project (NBIP) is assisting CMS in
developing and refining indicators that examine
states’ progress in offering person-centered and
balanced LTSS systems.
Study Design: A review of relevant literature
and research was conducted on existing LTSS
indicators. In addition, a state systems-level selfassessment survey tool that included the NBIP
Principals, indicators and related questions was
developed and field-tested in 2012 with seven
State Profile Tool (SPT) grantees (AR, FL, ME,
MA, MI, MN & KY). The data collected, along
with literature and research reviewed and
feedback from Technical Expert Panel, Federal
Partners and Stakeholder Group members were
used to refine and add to the NBIP Principals,
indicators and state self-assessment survey tool.
A Technical Assistance Guide to NBIs also was
created.
Population Studied: A state self-assessment
survey tool was field-tested in 2012 with seven
SPT grantees (AR, FL, ME, MA, MI, MN & KY).
Principal Findings: A set of systems-level NBI
Principals and indicators were developed for use
by CMS and states to examine the effectiveness
of federal LTSS policies that focus on LTSS
systems change and states’ efforts in achieving
person-centered and balanced LTSS systems.
Conclusions: While the NBIs, as currently
developed, have the potential to examine the
effectiveness of federal LTSS policies related to
LTSS systems change and states efforts in
implementing and maintaining person-centered
and balanced LTSS systems, there are
significant challenges that must be considered
before the NBIs are finalized. These challenges
include: 1) consensus building regarding
whether a certain system infrastructure being
examined is the best solution for all LTSS
programs across all states; 2) difficulties in
achieving and maintaining the cross-agency
collaboration necessary to gather data; 3)
differences in key taxonomy across states and
agencies; 4) limitations in both process and
outcome NBIs; 5) development of a core set of
NBIs that tells a compelling story, while including
key pieces of information that are essential to
understanding a LTSS system; and 6) methods
of collecting data in a sustainable way.
Implications for Policy, Delivery, or Practice:
The NBIs provide CMS with a tool to examine
the effectiveness of federal policies related to
LTSS systems change. They also provide CMS
and states with a tool to assess states’ efforts in
developing and maintaining person-centered
and balanced LTSS systems in a changing
environment.
Funding Source(s): CMS
Poster Session and Number: B, #649
Sources of New Workers and Job Mobility in
Long-Term Care
Bianca Frogner, George Washington University;
Joanne Spetz, University of California, San
Francisco
Presenter: Bianca Frogner, Assistant Professor,
George Washington University
biancafrogner@gmail.com
Research Objective: The past decade has
seen substantial job growth in the health care
industry overall, and the long-term care (LTC)
sector in particular. Limited research is available
as to where the LTC industry draws its supply of
workers. The characteristics and experiences of
LTC workers after they leave LTC jobs also are
unknown. In addition, there are high rates of
turnover in LTC jobs, and there is little
knowledge about the characteristics that make
people more likely to leave LTC positions or
what happens to them after they leave. The
study objective is to examine the pipeline of
workers into the LTC industry and LTC
occupations, and the factors associated with
workers leaving LTC employment.
Study Design: This project uses 12 years (2000
to 2011) of the publicly-available Current
Population Survey (CPS) Annual Social and
Economic Supplement (“March Supplement”),
which is an annual cross sectional survey of
approximately 200,000 individuals. This study
uses descriptive statistics to analyze trends on
where LTC draws their workforce, where leavers
of LTC go, the jobs individuals take in the LTC
sector, and the skills/education they bring into
LTC. We use multivariate logistic regression to
estimate the predictors of individuals entering or
leaving the LTC sector. All data are analyzed
using weights to ensure statistics represent the
full LTC workforce.
Population Studied: U.S. workers who enter
and exit the long-term care industry and longterm care occupations.
Principal Findings: Our results are preliminary
with final results expected in March 2014.
Eighteen percent of entrants into health care are
entering the LTC sector. However, a third of
those who leave health care are from the LTC
industry. Over a quarter come from another part
of the health care industry and another quarter
were not previously working in the labor force.
The LTC industry has also been drawing from
the leisure and hospitality industry (11.8%)
followed by from the retail trade industry (6.8%).
Sixty percent of the entrants are also making
occupation changes. Among care related
occupations, over a quarter of entrants are
taking on jobs as a nursing, psychiatric, and
home health aide followed by 8% as a personal
care aide. The nursing, psychiatric and home
health aides also had the highest exit rate,
though it was half the rate of entry. Entrants are
slightly younger and more diverse than current
workers in LTC.
Conclusions: There is substantial industry and
occupational turnover in LTC, and many people
who enter LTC jobs were not in the labor force
before entering LTC work.
Implications for Policy, Delivery, or Practice:
It is important to understand the characteristics
of people who enter LTC jobs – especially entrylevel occupations such as CNAs, HHAs, and
PCAs – to ensure that an adequate supply of
workers will be available and that training
programs can adequately prepare them for work
in this setting. An understanding of the
demographic and socioeconomic characteristics
that are associated with individuals being more
likely to enter or leave LTC, as well as the
industries that draw them away from LTC, will
provide guidance in designing retention
strategies.
Funding Source(s): HRSA
Poster Session and Number: B, #650
Impact of Advance Care Planning on Quality
and Cost of Care at the End of Life
Melissa Garrido, James J. Peters VA Medical
Center/Icahn School of Medicine at Mount Sinai;
Tracy Balboni, Dana-Farber Cancer Institute /
Harvard Medical School; Holly Prigerson, DanaFarber Cancer Institute / Harvard Medical
School /Weill Cornell Medical College
Presenter: Melissa Garrido, Research Health
Science Specialist /Assistant Professor, James
J. Peters VA Medical Center/Icahn School of
Medicine at Mount Sinai
melissa.garrido@mssm.edu
Research Objective: Advance care planning
(ACP) is intended to improve patients' end-of-life
outcomes, but retrospective analyses, surrogate
recall, and selection bias have hampered efforts
to determine the effects of ACP on patient
quality of life or costs of care near death. We
used prospective, propensity score-weighted
data that adjusts for patients’ values about endof-life care, illness severity, and
sociodemographics to examine the impact of
ACP (do not resuscitate [DNR] orders, living
wills and durable powers of attorney) on quality
of life and costs of care in the week before
death.
Study Design: Data are from the Coping with
Cancer study, which enrolled patients with
advanced cancer and their primary informal
caregivers from outpatient clinics in five states.
Data from patient baseline interviews and postmortem caregiver evaluations were analyzed.
Cost estimates were adjusted to 2013 dollars
and are from the Healthcare Cost and Utilization
Project Nationwide Inpatient Sample and from
published Medicare payment rates and cost
estimates. Our outcome measures were quality
of life in the week before death (an average of
three questions about overall quality and
physical and psychological distress) and
estimated costs of care received in the last week
of life. To account for observable characteristics
associated with both ACP engagement and
outcomes, we weighted our sample by a
propensity score that included illness severity,
sociodemographics, and preferences for lifeprolonging and heroic care at the end-of-life.
After weighting the sample, we included the
covariates from the propensity score in
generalized linear models of the treatment
effects (gamma family and log link for costs,
Gaussian family and identity link for quality of
life).
Population Studied: ACP data were available
for 342 patients and their caregivers.
Principal Findings: The average quality of life
in the week before death was 6.3 (standard
deviation [SD]=2.6, range 0-10 with higher
numbers indicating better quality), and the mean
estimated cost of care in the last week of life
was $7,741 (SD= $9,495; range $1,022$38,819). In the propensity score-weighted
sample, the adjusted incremental effect of a
DNR order on average quality of death in the
last week of life was ß=0.96 (standard error
[SE]=0.34, p=.004). DNR orders were
significantly associated with reduced costs of
care in our weighted sample (mean incremental
effect= -$2,554, bootstrapped SE= $1,213,
p=.004), but this relationship did not persist after
controlling for illness severity and other
confounders (-$2,695, bootstrapped SE=
$1,467, p=.07). In contrast, there were no
significant relationships between living
wills/durable powers of attorney and quality of
life or costs of care in the last week of life.
Conclusions: DNR orders, which reflect a
specific preference for less life-prolonging care,
result in significantly better quality of life in the
last week of life. Living wills and durable powers
of attorney were not associated with end-of-life
outcomes and did not have the impact that DNR
orders did.
Implications for Policy, Delivery, or Practice:
The positive impact of DNR orders on patient
quality of life supports the need for continued
patient education about benefits and risks of
cardiopulmonary resuscitation as well as health
policies encouraging ACP.
Funding Source(s): NIH
Poster Session and Number: B, #651
Patterns of Resident Discharge from Nursing
Homes and the Role of Facility
Characteristics in Community Transition
Zachary Gassoumis, University of Southern
California; Amanda Holup, University of South
Florida; Kathleen Wilber, University of Southern
California; Kathryn Hyer, University of South
Florida
Presenter: Zachary Gassoumis, Doctoral
Candidate/Research Assistant, University of
Southern California
gassoumi@usc.edu
Research Objective: Despite the emphasis on
rebalancing the nation’s long-term services and
supports (LTSS) from institutional toward homeand community-based (HCBS), research on
nursing home (NH) characteristics that support
community transition is sparse. Further, no
study has examined these transitions across
various state policy contexts. This study
responds to that gap by examining facility
characteristics that influence the transition of NH
residents to the community across two states
with markedly different LTSS systems and
policies: California and Florida. An examination
of these states allows for a discussion of how
differing LTSS systems as well as facility
characteristics influence community transition
among a sizeable proportion of the older adult
population in the United States. This study
responds to another gap by looking at facilities’
influence on discharge among both short-stay
and long-stay NH residents.
Study Design: Data from the Minimum Data Set
2.0 were used to construct episodes of care,
beginning with resident admission to the NH and
ending either with death or with discharge
without reentry to the same NH within 30 days.
Kaplan-Meier survival estimates were
constructed to track residents for up to one year
across four outcomes: community discharge
(home, group home, board-and-care, or assisted
living); discharge to death; discharge to other
care facilities (e.g., acute care); and remaining in
the NH more than 365 days post-admission. We
then predicted community discharge within 90
days (short-stay) and between 91-365 days
(long-stay) using individual, facility, and market
variables structured in three-level hierarchical
generalized linear models.
Population Studied: A cohort of residents aged
65 and older (N=189,437) admitted to all freestanding NHs in California (n=1,127) and Florida
(n=657) from July 2007 to June 2008.
Principal Findings: Survival curves indicated
that considerably more Florida residents were
discharged to the community within one year
(67.8% vs. 59.9% in California), but that the
speed at which residents returned to the
community was comparable across the two
states. Discharge to death and remaining in the
NH were more common in California. The
hierarchical models revealed robust significance
of individual-level variables. Significant facility
predictors for community transition during shortstay included occupancy and number of beds;
for-profit ownership and the proportion of
residents on Medicare only predicted transitions
from long stays. Notably, proportion of residents
on Medicaid reduced transitions across states
and discharge windows. State differences were
primarily seen at the market (NH concentration
and population density predicted long-stay
transitions, but only in Florida) and individual
level (recently falling and having mental health
diagnoses predicted transitions only in
California).
Conclusions: Findings suggest that community
transition of short- and long-stay NH residents is
affected by resident, facility, and sometimes
market characteristics, with Medicaid influencing
discharge across states and duration of stay.
Survival curves suggest that California and
Florida differ in their use of LTSS and HCBS or
the acuity of NH admissions, perhaps explained
in part by the higher numbers of diversion
programs that exist in California.
Implications for Policy, Delivery, or Practice:
It seems that current rebalancing policies may
fail to adequately address the availability of and
access to HCBS, thereby preventing Medicaiddependent NH residents from transitioning to the
community.
Funding Source(s): Other Florida’s Agency for
Health Care Administration
Poster Session and Number: B, #652
Medical Innovation and the Changing Health
and Health Care Costs of Obesity Among the
Elderly
Etienne Gaudette, University of Southern
California; Dana P. Goldman, Leonard D.
Schaeffer Center for Health Policy and
Economics; Andrew Messali, Leonard D.
Schaeffer Center for Health Policy and
Economics; Neeraj Sood, Leonard D. Schaeffer
Center for Health Policy and Economics
Presenter: Etienne Gaudette, Postdoctoral
Fellow, University of Southern California
etienne.gaudette@usc.edu
Research Objective: To measure the impact of
medical innovation on the health and health care
costs of obesity among the elderly, using statins
as a case study.
Study Design: Life trajectories and medical
costs of elderly Americans are forecasted using
the Future Elderly Model (FEM) – an established
dynamic microsimulation model of health of
Americans aged over 50. The change in the
health and health care costs of obesity due to
the introduction and widespread use of statins
are estimated by introducing a scenario in which
statins have not been discovered, using wellrecognized estimates of the health impact of
statins.
Population Studied: The American population
aged over 50.
Principal Findings: Simulations reveal that the
life expectancy of 50-year-olds with a healthy
BMI (18.5-<25) is 0.92 year longer than it would
be in a world without statins. Among the obese
population, the life expectancy gains due to
statins are of 1.00 year for type 1 obesity (BMI
30-<35), 1.05 year for type 2 obesity (BMI 35<40) and 1.07 year for type 3 obesity (BMI>=40).
These life expectancy gains augment the
present value of per capita lifetime health care
costs by $15,000 for individuals with a healthy
BMI, $18,000 for type 1 obesity, $19,100 for
type 2 obesity and $19,800 for type 3 obesity.
About 33% of these costs are shouldered by the
Medicare program.
Conclusions: While the widespread use of
statins is beneficial for individuals of all weight
types, their health impact is highest among the
obese population. Additional health care costs
from statin use are small relative to the value of
life expectancy gains, and mostly paid for by
individuals.
Implications for Policy, Delivery, or Practice:
Policies promoting development and access to
technologies to treat obesity-related sequela,
such as heart disease and stroke, can effectively
reduce the health costs of obesity.
Funding Source(s): No Funding
Poster Session and Number: B, #653
Patient-Reported Quality of Life Satisfaction
with Services among Money Follows the
Person Participants
Amie Goodin, University of Kentucky; Sandi
Kiteck, University of Kentucky; Ann Williamson,
University of Kentucky
Presenter: Amie Goodin, Graduate Student,
University of Kentucky
amie.goodin@g.uky.edu
Research Objective: The Money Follows the
Person (MFP) waiver has enabled some
institutionalized Medicaid recipients to transition
back to community-based living for long-term
care. This study evaluated whether MFP
participation was associated with changes in
reported quality of life (QoL) and satisfaction
with services.
Study Design: Participants were interviewed
pre-transition (baseline n=393) and at one year
post-transition (n=294) in person or via phone.
Two outcomes were selected as dependent
variables: satisfaction with services and QoL,
both of which were operationalized as response
of “Happy” or “Unhappy”. Bivariate analyses
examined the relationship between the
dependent variables and residence type (private
residence or group home). Multivariate analyses
were conducted with a pre/post design using
logistic regression to generate odds ratios and
associated 95% confidence intervals (CI)
controlling for gender, age, physical disability,
brain injury, behavioral or intellectual disability,
group home residence, and urban residence.
Population Studied: Kentucky Medicaid
beneficiaries enrolled in MFP from 2008-2013
aged 18 and older.
Principal Findings: At follow up, participants
were 2.80 times more likely to report satisfaction
with services (95% CI: 1.51-5.19) and 3.73 times
more likely to report improved QoL (95% CI:
1.65-8.44). Females (OR: 0.62, 95% CI: 0.410.94) and group home residents (OR: 0.31, 95%
CI: 0.17-0.58) had lower QoL and group home
residence was associated with lower satisfaction
with services (OR: 0.35, 95% CI: 0.17-0.75).
Participants with behavioral or intellectual
disabilities were more likely to report satisfaction
with both services (OR: 2.61, 95% CI: 1.28-5.34)
and QoL (OR: 4.00, 95% CI: 2.23-7.17). Several
participants were not interviewed post-transition
due to non-completion of the program or death.
Conclusions: MFP participants report greater
satisfaction with services and QoL after
transition to private residences from institutional
care when compared to those transitioned to
group homes.
Implications for Policy, Delivery, or Practice:
Long-term care waivers, such as MFP, are
intended to reduce Medicaid costs from
institutionalization and improve QoL. Transition
to group homes if private residence alternatives
are available may decrease satisfaction,
thereby, increasing participant’s risk for reinstitutionalization.
Funding Source(s): No Funding
Poster Session and Number: B, #654
Physician Extenders in Nursing Homes: Role
of Market Factors in the Employment
Decision
Shivani Gupta, The University of Alabama at
Birmingham; Nitish Patidar, The University of
Alabama at Birmingham; Josue Patien Epane,
University of Nevada Las Vegas; Robert WeechMaldonado, The University of Alabama at
Birmingham
Presenter: Shivani Gupta, Graduate Teaching
Assistant, The University of Alabama at
Birmingham
sgupta9@uab.edu
Research Objective: Physician staffing patterns
can significantly affect the quality of care
delivered in nursing homes. One of the potential
strategies to meet the increasing demand for
physician care in nursing homes could be
employment of physician extenders, such as
nurse practitioners or physician assistants. The
purpose of this study is to examine the various
market factors that influence the decision to
employ physician extenders in nursing homes.
Study Design: The data was derived from
Long-term Care Focus (Ltcfocus). The
dependent variable represents nursing homes’
decision to employ physician extenders (nurse
practitioners and/ or physician assistants) (1 =
did employ, and 0 = did not employ). The
primary independent variables include:
competition measured by Herfindahl-Hirschman
Index (HHI), states with certificate of need
(CON), hospital-based facilities in the county,
proportion of residents in the county whose
primary support is Medicaid, Medicare managed
care penetration and supply of RNs per 1000
elderly. Control variables include: case mix
index, occupancy rate, proportion of white and
minority residents, CNA hours per resident day,
LPN hours per resident day, RN hours per
resident day, RN to nurses ratio, ownership,
size, system affiliation, and proportion of
Medicaid and Medicare patients. A panel logistic
regression with facility random effects and state
and year fixed effects was used for analysis.
Population Studied: The studied sample
consisted of all U.S. nursing homes (N= 16,341)
between 2000 and 2010.
Principal Findings: Results show that nursing
homes operating in markets with higher
competition (lower HHI) (O.R = 0.41; p = 0.001),
higher levels of Medicare managed care
penetration (O.R. = 1.01; p = 0.001), and higher
supply of RNs per 1000 elderly (O.R. = 1.01; p =
0.001) are more likely to employ physician
extenders. In contrast, nursing homes in
counties with higher proportion of residents
whose primary support is Medicaid (O.R. = 0.98;
p = 0.001) are less likely to employ them.
Nursing homes with higher occupancy rate (O.R.
= 1.01; p = 0.001), higher LPN hours per
resident day (O.R. = 1.17; p = 0.001), and lower
RN hours per resident day (O.R. = 0.84; p =
0.001) are more likely to employ physician
extenders. Similarly, larger (OR= 1.01; p= 0.001)
and system-affiliated nursing homes (OR= 1.35;
p= 0.001), and for-profit nursing homes (O.R =
1.18; p = 0.001) are more likely to employ
physician extenders. Lastly, facilities with higher
proportion of Medicare patients (O.R. = 0.99; p =
0.04) are less likely to employ physician
extenders.
Conclusions: Results suggest that nursing
homes employ physician extenders as a
strategy to differentiate themselves in more
competitive markets and those with higher
Medicare managed care penetration. Slack
resources (size and chain affiliation) also play a
role in adoption of this strategy.
Implications for Policy, Delivery, or Practice:
Employing physician extenders in nursing
homes may be a strategy to address the limited
engagement of physicians in nursing homes,
and potentially impact quality of care.
Policymakers should explore strategies that may
incentivize nursing homes in the use of
physician extenders, particularly among smaller
nursing homes and for those located in less
competitive areas.
Funding Source(s): No Funding
Poster Session and Number: B, #655
Impact of Organizational Factors on Hospice
EMR and Telemedicine Use
Mengying He, University of Alabama at
Birmingham; William Opoku-Agyeman,
University of Alabama at Birmingham; Jennifer
Crimiel, University of Alabama at Birmingham
Presenter: Mengying He, Graduate Assistant,
University of Alabama at Birmingham
merry429@gmail.com
Research Objective: The 2009 federal Health
Information Technology for Economic and
Clinical Health (HITECH) Act created incentives
for physicians and hospitals to adopt electronic
health records (EHR). However, hospice
agencies were not on the incentive list. Hospice
care is an important component in the delivery of
care in the healthcare system of the US,
especially in care coordination and the continue
spectrum of long-term care. As of 2011 there
were 5,300 hospice programs actively in place
with over 1.6 million patients receiving care from
these programs. The purpose of this study is to
examine the specific organizational characters
that are related to hospice agency EMR and
telemedicine use.
Study Design: 2007 National Home and
Hospice Care Survey (NHHCS) employed a
stratified probability design method to select
home health and hospice agencies from more
than 15,000 providing home health care and
hospice services in the US. 1,036 home health
and hospice agencies participated in the 2007
NHHCS. “Hospice agencies only” (N=359) in the
data set were kept for further analyses, while
341 “home health agencies” only and 336 “both
home health and hospice agencies” were
excluded. A complex sample design survey
method was used first to adjust sampling
weights. Secondly, a complex sample design
based descriptive analyses and logistic
regressions were adopted to assess the basic
hospice organizational characteristics and how
these factors influence hospice EMR and
telemedicine use.
Population Studied: 359 hospice agencies
were studied, which represent 2,029 hospice
agencies in the US in 2007 by using complex
sample design method.
Principal Findings: In logistic regression
models controlling for sample weights, hospices
agencies that are part of chain are more likely to
use telemedicine (OR=16.223; 95% confidence
interval CI: 4.467-58.917). For-profit hospices
are less likely to use EMR (OR=0.241; 95% CI:
0.061-0.954) and telemedicine (OR=0.028; 95%
CI: 0.003-0.231). Joint Commission for
Accreditation of Healthcare Organizations
(JCAHO) accredited hospices have higher odds
in adopting EMR (OR=5.871; 95% CI: 1.84618.675). Hospice agencies that have formal
contracts with outside agencies to provide
services to patients are more likely to use EMR
(OR=6.828; 95% CI: 1.957-23.828). Hospices
spend more years in the industry have lower
odds in adopting EMR (OR=0.936; 95% CI:
0.879-0.998). Hospices that provide more
services to patients have higher odds in
telemedicine use (OR=1.196; 95% CI: 1.0181.405). Hospices with higher volume of patients
are more likely to use EMR (OR=1.015; 95% CI:
1.004-1.025).
Conclusions: Various organizational factors
may influence hospice EMR and telemedicine
use. For-profit hospices are less likely to adopt
EMR and telemedicine. Older hospices are also
less likely to participate in organizational
innovation (EMR adoption). Furthermore, being
part of a hospice chain and providing more
services enables hospices to use telemdicine.
Finally, hospices' size, JCAHO accreditation,
and formal contracts with other hospices are all
indicators for EMR adoption.
Implications for Policy, Delivery, or Practice:
Understand the influence of organizational
factors on EMR and telemedicine adoption helps
hospices understand their position on EMR and
telemedicine use in hospice industry and helps
hospitals and other health care organizations
make decision in about which hospice agency
will be more efficient in the continuity care
coordination process.
Funding Source(s): No Funding
Poster Session and Number: B, #656
Prevalence of Clostridium difficile and Six
Month Follow-up in Nursing Home Patients
Admitted from an Acute Care Hospital
Nina Joyce, Brown University; Vincent Mor,
Brown University
Presenter: Nina Joyce, Student, Brown
University
ninarjoyce@gmail.com
Research Objective: lostridium difficile (C.
difficile) is the most common cause of acute
diarrheal infections in the hospital setting, as
well as in Nursing Homes (NH). As a
population, NH residents are especially
vulnerable to C. difficile infection (CDI) due to
their close living quarters, shared facilities,
advanced age and prevalence of comorbid
conditions. The prevalence of C. difficile among
NH residents admitted from an acute hospital is
currently unknown and, given the frequent
movement between NH and acute care
hospitals, is important for identifying areas of
intervention.
The objective of our study is to compare six
month follow up outcomes among NH patients
admitted from a hospital by C. difficile diagnosis
and the presence of a NH stay 30 days prior to
hospital admission.
Study Design: We conducted a retrospective
analysis of patient status (discharged to the
community, a nursing home, died or
hospitalized) six months after admission to a NH
from an acute care hospital using Medicare
inpatient claims linked to the NH minimum data
set assessment.
Population Studied: Our study population
included all Medicare patients admitted to a NH
from an acute care hospital in 2011.
Principal Findings: Among patients discharged
to a NH from an acute care hospital, 22% of
patients with a C.diff diagnosis on hospital
admission had a NH visit in the prior 30 days as
compared to 15% of patients with no C.diff
diagnosis (p<0.001). At six months follow up,
patients with a C.diff diagnosis were more likely
to have been hospitalized (47% vs. 41%) or died
(9.5% vs. 8.3%), and were less likely to have
been discharged back into the community (22%
vs. 27%). Although the difference between
patients with and without CDI did not vary by the
presence of a NH visit in the past 30 days,
overall, patients with CDI were more likely to die
(13% vs 9%) or be in a NH (25% vs 20%) at the
six month follow up if they had a prior NH visit.
Conclusions: Our results demonstrate that
among patients admitted to a NH from an acute
care hospital, those with a prior NH stay are
more likely to have a CDI. Additionally, among
patients diagnosed with a CDI in the hospital
and admitted to a NH, we found a higher rate of
death and hospitalization six months post NH
admission in patients with a prior NH stay as
compared to no prior stay.
Implications for Policy, Delivery, or Practice:
The revolving door between acute care hospitals
and NHs represents a particular challenge to
intervention as infections may be picked up and
spread between the different institutions. Our
results suggest that efforts to reduce CDI should
focus on patients with a prior NH stay as they
are more likely to be infected and worse
outcomes at a six-month follow up.
Funding Source(s): Other Unrestricted grant
from the American Health Care Association to
examine the predictors of hospitalization and rehospitalization
Poster Session and Number: B, #657
Satisfaction with Nursing Home Care: The
Impact of Ownership Change
Qinghua Li, University of Rochester Medical
Center; Yi Tang, University of Rochester
Medical Center; Yue Li, University of Rochester
Medical Center
Presenter: Qinghua Li, Phd Candidate,
University of Rochester Medical Center
qinghua_li@urmc.rochester.edu
Research Objective: To determine the impact
of ownership change of nursing homes on
satisfaction with nursing home care in Maryland.
Study Design: We obtained data from a survey
that measured the experience and satisfaction of
family members and other designated
responsible parties of long-term care residents
conducted in Maryland’s nursing homes from
April to June in 2012. In the survey, 17 Likertscale items (4-point) were used to assess five
domains of satisfaction with residents’ life and
care: staff administration of the nursing home,
care provided to residents, food and meals,
autonomy and residents’ rights, and physical
aspects of the nursing home. Domain scores
range from 1 to 4 with higher score indicating
higher satisfaction. A 1 to 10 scale item was also
used to measure overall satisfaction of survey
respondents, as well as the percentage of
respondents who would recommend the nursing
home to his/her friends needing nursing home
care. These seven measures are the primary
outcomes in this study. The key independent
variable is whether the nursing home had a
change in ownership in 2011.
Survey data were linked to the Nursing Home
Compare data to identify facility characteristics,
staffing information, and quality deficiency, and
to the Area Health Resource File to identify
county-level information. Both bivariate and
multivariate OLS regression models were
estimated to examine the association between
ownership change and family satisfaction with
nursing home care.
Population Studied: In total, 216 NHs were
included in the study.
Principal Findings: Overall, 22 out of 216
(10.19%) nursing homes changed ownership
type. Among these facilities, 17 converted from
not-for-profit to for-profit, while 5 from for-profit
to not-for-profit. Family satisfaction with nursing
home care was high, with an average score of
8.38/10 (SD=0.70) in the overall rating, and
89.97% of the respondents would recommend
the nursing home. The mean satisfaction scores
in the five domains of resident care and life
ranged from 3.44 to 3.68 and scores varied
across nursing homes (SD=0.16-0.24).
Bivariate analyses showed that ownership
change was significantly related to a decrease in
family satisfaction in all seven measures. After
controlling for facility characteristics, staffing
level, overall quality level, and county-level
market factors, we found that ownership change
was significantly related to a 0.08 decrease in
the score for staff administration, a 0.11
decrease for care provided to residents, a 0.11
decrease for autonomy and residents’ rights,
and a 0.10 decrease for physical aspects of the
nursing home. The overall satisfaction was 0.38
lower in nursing homes that experienced an
ownership change, and the overall
recommendation rating was 5.2% lower
(adjusted p<0.05 in all cases).
Conclusions: Family satisfaction with nursing
home care was generally high in Maryland in
2012. Ownership change of nursing facilities
was related to a decrease in family experience
and satisfaction.
Implications for Policy, Delivery, or Practice:
In the process of nursing home ownership
conversion, quality of care, measured by family
satisfaction with resident care in this study, may
be negatively impacted. Efforts to ensure
appropriate quality of care and protections of
vulnerable long-term residents are warranted
during facility ownership change.
Funding Source(s): NIH
Poster Session and Number: B, #658
The Influence of Facility-Level Dementia
Prevalence on Nursing Home Quality of Care
for Residents With and Without Dementia
Qinghua Li, University of Rochester Medical
Center; Yue Li, University of Rochester Medical
Center; Yeates Conwell, University of Rochester
Medical Center; Thomas Caprio, University of
Rochester Medical Center; Helena TemkinGreener, University of Rochester Medical Center
Presenter: Qinghua Li, Phd Candidate,
University of Rochester Medical Center
qinghua_li@urmc.rochester.edu
Research Objective: We examined how facilitylevel prevalence of dementia affects quality of
care for residents with and without dementia in
nursing homes (NHs).
Study Design: Data sources included the
Medicare beneficiary enrollment file, Medicare
hospital claims, the Minimum Data Set, and data
from LTCFocus.org.
Deterioration in urinary incontinence (UI),
deterioration in depressive/anxious symptoms
(DA), and potentially avoidable hospitalizations
(PAH) were three quality measures (QM)
examined.
The key independent variable, prevalence of
dementia, was categorized into three groups,
low (<mean-1SD:37%), middle
(mean±1SD:37%-67%, reference), and high
(>mean+1SD:67%).
We fit logistic regression (QM=UI, DA) and
Poisson regression (QM=PAH) models with
facility random-effects. All analyses were
conducted at the resident level.
Population Studied: All long-term care
residents (n=1,143,027) in 14,195
Medicare/Medicaid NHs in 2007.
Principal Findings: In 2007, 52% of NH
residents had dementia. About 71% of NHs
(n=10,102) were at the middle level of dementia
prevalence, 14% (n=2,011) at low prevalence,
and 15% (n=2,082) at high prevalence.
Approximately 25% of the residents had
deteriorated in UI status from admission to the
first quarterly assessment, and 14% became
more depressed or anxious between two
consecutive non-admission assessments. Within
one year, 18% the residents had at least one
PAH.
In NHs with low prevalence of dementia,
residents with dementia had a reduced risk of
deterioration in UI and DA (UI: OR=0.89,
95%CI=0.83-0.96; DA: OR=0.90, 95%CI=0.850.94), as did residents without dementia (UI:
OR=0.93, 95%CI=0.89-0.98; DA: OR=0.89,
95%CI=0.85-0.93). In NHs with high prevalence
of dementia, residents with dementia had a
reduced risk of PAHs (IRR=0.97, 95%CI=0.941.00), while residents without dementia had an
increased risk of deterioration in DA (OR=1.08,
95%CI=1.03-1.13).
Conclusions: High facility-level prevalence of
dementia is associated with significantly lower
quality of care for residents without dementia
and higher quality of care for residents with
dementia. Low prevalence of dementia is
associated with higher quality of care for both
populations.
Implications for Policy, Delivery, or Practice:
Facilities specializing in dementia care (high
dementia prevalence) may be able to provide
better quality of care to residents with dementia.
At the same time, facilities with low prevalence
of dementia may be able to provide better care,
for selected QMs, for residents without
dementia. Future research should explore the
feasibility and the benefits of operating
specialized dementia care facilities.
Funding Source(s): NIH
Poster Session and Number: B, #659
Pediatric End of Life Care under Healthcare
Reform: Factors Influencing State
Implementation Concurrent Care for Children
Lisa Lindley, University of Tennessee Knoxville; Don Bruce, University of Tennessee;
Sheri Edwards, Western Carolina University
Presenter: Lisa Lindley, Assistant Professor,
University of Tennessee - Knoxville
llindley@utk.edu
Research Objective: In the wake of
Presidential elections and Supreme Court
decisions, states now face the reality of
implementing the Patient Protection and
Affordable Care Act of 2010. ACA Section 2302
or Concurrent Care for Children was enacted
upon the signing of healthcare reform on March
23, 2010 and yet, little is known about the
implementation of ACA 2302 at the state level.
This mandatory provision states that children
enrolled in Medicaid or Children's Health
Insurance Plan may receive care related to their
terminal illness concurrently with hospice care.
Section 2302, therefore, eliminated the hospice
eligibility requirement that children must forgo
curative care upon admission to hospice. The
purpose of our study was to identify the states
that implemented ACA 2302 and to examine the
influence of economic, political, and legal factors
on state implementation of ACA 2302.
Study Design: This was a retrospective, crosssectional study using data gathered from
multiple publicly available sources from 2010 to
2012. Our outcome of interest was whether a
state implemented the Concurrent Care for
Children provision in a given year. A group of
variables was composed of economic, political,
and legal factors at the state-level. The analytic
strategy included descriptive analyses and
multiple regression models to estimate the
association between economic, political, and
legal factors and ACA 2302 implementation for
2010, 2011, and 2012.
Population Studied: All 50 states were
included in the study. We conducted individual
analyses for each year of the study and
excluded states from the analysis after they
implemented ACA 2302.
Principal Findings: From 2010 to 2012, we
identified 31 out of 50 states that implemented
ACA 2302. In 2010, nine states implemented
ACA 2302: Alabama, Arizona, Hawaii, Maine,
Massachusetts, Missouri, Oklahoma, Texas, and
Wisconsin. Fourteen states implemented ACA
2302 in 2011: California, Delaware, Idaho, Iowa,
Kansas, Maryland, Michigan, Mississippi, New
Jersey, New York, North Carolina, Ohio,
Oregon, and Washington. In 2012, ACA 2302
was implemented in eight states: Arkansas,
Illinois, Indiana, Kentucky, South Carolina, Utah,
Vermont, and West Virginia. Economic factors
were significantly related to implementation of
ACA 2302. In 2011, states that were engaged in
Medicaid cost containment were more likely to
implement ACA 2302, compared to states that
were not cutting Medicaid costs (ß = 0.01,
p<0.01). In 2012, states experiencing a
budgetary crisis were less likely to implement
ACA 2302 compared to states not in budget
crisis (ß = -0.38, p<0.05). There was no
association between political and legal factors
and state implementation of ACA 2302 in any
year of the study.
Conclusions: Our analysis revealed that for
early implementers economic, political and legal
factors did not influence implementation of ACA
2302 in 2010. However, by 2011, states that
were engaged in Medicaid cost containment
efforts were more likely to implement ACA 2302
and in 2012, states experiencing a budgetary
crisis were less likely to implement ACA 2302.
Implications for Policy, Delivery, or Practice:
These findings suggest that state-level
implementation of Concurrent Care for Children
may be an important bellwether for future
healthcare reform implementations.
Funding Source(s): NIH
Poster Session and Number: B, #660
Use and Patterns of Osteoporosis
Pharmacotherapy in Community-Dwelling
and Long-Term Care Facility-Residing
Medicare Beneficiaries with Osteoporosis
F. Ellen Loh, University of Maryland School of
Pharmacy
Presenter: F. Ellen Loh, Graduate Research
Assistant, University of Maryland School of
Pharmacy
floh001@umaryland.edu
Research Objective: To assess use and
patterns of osteoporosis pharmacotherapy in
community-dwelling and long-term care facility
(LTCF)-residing elderly women and men.
Study Design: A pooled cross-sectional study
examining (1) any use of the 5 classes of drugs
(bisphosphonates, calcitonin, parathyroid
hormone, estrogen and selective estrogen
receptor modulator) approved by FDA for
osteoporosis treatment or prevention and (2)
use of a specific class of drug among those 5
drug classes in community-dwelling and LTCFresiding Medicare beneficiaries with
osteoporosis. We captured drug use in Part D
prescription claims. Residential status was
defined as community only, LTCF only, and
both. We used a modified Poisson regression to
assess the impact of residential status on any
use of the 5 classes of osteoporosis
pharmacotherapy and a multinomial logistic
regression to measure the effect of residential
status on use of a specific class of osteoporosis
pharmacotherapy.
Population Studied: A random 5% sample of
the Medicare population with osteoporosis aged
70 years and older enrolled in stand-alone
prescription drug plans from January 1, 2006
through December 31, 2008, or death. The final
sample included 90,956 women and 8,465 men.
Principal Findings: Prevalence of use of any
osteoporosis medication ranged from 46.6% in
2006 to 44% in 2008 in women and 25.2% in
2006 to 24.5% in 2008 in men. When stratified
by residential status, prevalence of osteoporosis
medication use was the lowest among women
and men in LTCF only (35.4% to 31.8% in
women and 24.3% to 21.6% in men from 2006
to 2008). Bisphosphonates were the top choice
of medication prescribed for all medication
users. Prevalence of bisphosphonate use
ranged from 76.4% in 2006 to 78.3% in 2008
among female medication users and 91.3% in
2006 to 93.3% in 2008 among male medication
users. Prevalence of use of other osteoporosis
medications was below 12% among medication
users. However, calcitonin was much more likely
to be prescribed to women and men in LTCF
only and in both settings than in the community
only. After adjusting for confounding, any length
of stay in a LTCF would decrease the probability
of receiving any osteoporosis medication for
women (Prevalence Ratio (PR) 0.83, 95% CI
[0.81, 0.85] for women in LTCF only; PR 0.95,
95% CI [0.94, 0.97] for women in both settings)
but had no significant effect for men. Any length
of stay in a LTCF also increased the probability
of receiving calcitonin compared to
bisphosphonates in both women and men
(Relative risk ratio (RRR) 3.19, 95% CI [2.97,
3.43] for women in LTCF only; RRR 1.92, 95%
CI [1.78, 2.08] for women in both settings; RRR
1.54, 95% CI [1.11, 2.13] for men in LTCF only;
RRR 2.05, 95% CI [1.53, 2.76] for men in both
settings).
Conclusions: Prevalence of use of any
osteoporosis medication is low, especially in
LTCF residents, and patterns of osteoporosis
treatment differ by residential status, age, race,
geographic region, socio-economic status,
comorbidities and other medications used.
Implications for Policy, Delivery, or Practice:
Policy makers and clinicians should pay
attention to the low use of osteoporosis
pharmacotherapy and high use of calcitonin in
LTCFs and encourage compliance with
guidelines.
Funding Source(s): No Funding
Poster Session and Number: B, #661
Development of a Physician Resource
Packet to Assist with Advance Care Planning
at an Ambulatory Primary Care Practice
Leslie Peterson, Thomas Jefferson University;
Nancy Chernett, Thomas Jefferson University;
Susan Parks, Thomas Jefferson University
Presenter: Leslie Peterson, Thomas Jefferson
University
lpeterson.mph@gmail.com
Research Objective: Advance Care Planning
(ACP) involves a process of understanding,
planning for, and interpreting complex, often
difficult, healthcare choices based on personal
preferences and values for future healthcare
needs. ACP includes documentation in an
advance directive (AD), designation of a
healthcare decision-making proxy, and multiple
conversations to understand and update the
patient’s goals of care. Empirical data shows
that patients want ACP conversations to be
initiated by their healthcare provider prior to or in
early stages of disease onset and in the primary
care setting. Despite documented benefits of
AD and ACP, AD completion rates range from 515% of the general population. These rates are
alarmingly low and have been associated with
provider barriers to having ACP conversations.
This study aimed to expand on existing data to
determine provider barriers to offering ACP in an
ambulatory setting and preferences for reducing
those barriers. Tailored ACP resources were
developed to assist physicians in navigating
ACP conversations with their patients.
Study Design: Qualitative methodology in the
form of a focus group discussion was conducted
with physicians from a small primary care
practice affiliated with an academic medical
center in Philadelphia, PA. A focus group
discussion was chosen for data collection under
the assumption that an iterative conversation
would provide more in-depth data than other
qualitative or quantitative methods. The focus
group was facilitated by one of three coinvestigators; the two remaining co-investigators
served as note-takers. Thematic analysis of
major themes and sub-themes was reviewed
and revised until consensus was reached by all
three investigators.
Population Studied: Attending physicians and
post-residency physicians completing a
fellowship in geriatric medicine were asked to
participate in the focus group discussion. There
were seven participants out of a possible eight.
The particular practice used in this study was
chosen for its relatively small size, primarily
geriatric patient population, and specialization of
practicing physicians in either geriatric or
internal medicine.
Principal Findings: Thematic analysis showed
three main barriers to physicians offering ACP:
time constraints, physician logistical concerns,
and discomfort in having ACP conversations.
The three main barriers were divided into subthemes followed by specific approaches for
reducing the barriers as suggested by
participants.
Conclusions: The findings of this study provide
specific recommendations for future
interventions at the patient, provider, and
organizational levels to improve the ACP
process and increase rates of documented
Advance Directives. Data showed the need for
comprehensive ACP programs that include
systematic methods for patient/provider
education, multiple ACP conversations,
documentation procedures, and provider
training. ACP resources were tailored to
recommendations from participants in the focus
group, and continued evaluation is needed to
determine their efficacy and utility.
Implications for Policy, Delivery, or Practice:
Findings of this study clearly highlight barriers to
physicians for offering ACP, but also provide
specific strategies for reducing those barriers.
While the data collected is specific to a single
practice and may not be generalizable, it is the
belief of the researchers that these findings
support the need for systematic, comprehensive
programs in providing quality ACP that
translates into quality end-of-life care.
Funding Source(s): No Funding
Poster Session and Number: B, #662
Improving Transitions from Acute Care to the
Extended Care Setting
Heather Powell, Christiana Care Health System;
Kimberly Williams, Christiana Care Health
System; Melinda Acevedo, Christiana Care
Health System; Jomy Mathew, Christiana Care
Health System; Jeanmarie Okoniewski,
Christiana Care Health System; Jen Toto,
Christiana Care Health System; Aimee Vincent,
Christiana Care Health System; John McMillen,
Christiana Care Health System
Presenter: Heather Powell, Patient Care
Facilitator, Christiana Care Health System
HPowell@Christianacare.org
Research Objective: Preventable readmissions
consistently are cited as a source of healthcare
system inefficiency and directly result in poor
patient outcomes. In an effort to reduce
readmissions, we examined the process for
inpatient transitions to the extended care setting
in our delivery system. Our hypothesis was that
addressing deficiencies in the discharge process
would reduce readmission rates. The objective
of this study was to understand gaps in postacute care patient transitions from the hospital
setting to an extended care facility, improve the
process using a streamlined discharge process,
and measure the impact on reducing
readmission rates.
Study Design: In 2011, a multi-disciplinary
team of physicians, registered nurses, social
workers, case managers, unit clerks and
colleagues in the community was formed to
redesign the transition process for discharge
inpatients to extended care facilities. The team
first examined our existing process and then
implemented changes based on deficiencies
identified by the community, in conjunction with
recommendations from the Institute for
Healthcare Improvement. Changes included
creating disease specific communication tools
and a new discharge packet in an effort to
streamline documentation received by extended
care facilities. We also developed a survey and
administered it before and after implementation
of the new process to elicit extended care facility
feedback for ongoing process improvement.
Population Studied: Post-acute care patients
being discharged from an inpatient unit within a
regional independent academic medical center
to a community-based extended care facility.
Principal Findings: The redesigned discharge
process resulted in a sustained reduction in
readmission rates compared with baseline for all
post-acute inpatients that were transferred to an
extended care facility. In FY 2010, FY 2011 and
FY 2012, seven day overall readmission rates
decreased from 10.4% to 7.4% to 6.1%
respectively. Readmission rates for heart failure/
chronic obstructive pulmonary disease (COPD)
patients also decreased during this evaluation
period. However, a consistent downtrend in
readmission rates was not sustained over time
with rates slightly rising after a baseline
reduction from 17.6% to 5.6% to 7.2%
respectively. Survey results revealed consistent
positive feedback from extended care facility
practitioners and hospital staff regarding the
process improvement initiative. Specifically,
post-implementation survey results
demonstrated that extended care facility
satisfaction with process changes and transition
enhancements increased with over 90% of
facilities expressing satisfaction with the new
processes.
Conclusions: The redesigned transition
process for post-acute care inpatients to an
extended care facility resulted in a sustained
reductions of annual readmission rates for
patients overall. A reduction in readmission rates
was also maintained for heart failure/ COPD
patients, but annual readmission rates reflect a
slight upward trend in readmission rates from
the first to second year post-implementation.
Surveys revealed provider satisfaction with the
redesigned process.
Implications for Policy, Delivery, or Practice:
Patient transfers from hospitals to extended care
facilities represent a pivotal period in the
continuum of care. Streamlining the process for
this transition and obtaining provider buy-in for
the revised process has the capacity to reduce
inpatient readmission rates and positively
impacting patient care.
Funding Source(s): No Funding
Poster Session and Number: B, #663
Impact of Nursing Home Prospective
Payment System under Certificate-of-Need
Momotazur Rahman, Brown University;
Jacqueline Zinn, Temple University; David
Grabowski, Harvard University; Omar Galarraga,
Brown University; Vincent Mor, Brown University
Presenter: Momotazur Rahman, Investigator,
Brown University
momotazur_rahman@brown.edu
Research Objective: Nursing home certificate
of need (CON) legislation that were adopted to
limit the supply of nursing home beds with the
goal of reducing expenditures have not changed
in most of the states during last two decades.
Medicare adopted the Prospective Payment
System (PPS) in 1998 to reduce spending on
skilled nursing facility (SNF) care. This study
examines the extent to which the potential
effectiveness of the PPS may have been
affected by the presence CON laws. The central
hypothesis of this paper is that the long-term
presence of CON regulation results in nursing
homes with more market power that are less
sensitive to price-based interventions.
Study Design: We compared nursing home bed
supply and spending per beneficiary between
states with and without CON legislation in the
pre- and post-PPS periods using a difference-indifferences model. We adjusted for several
variables at the county and state levels.
Population Studied: We used nursing-facility
and state-level data for 1992-2008 from 44
contiguous states that did not change their CON
laws during the study period. The sample is
comprised of 3,334 nursing homes in non-CON
states and 8,992 nursing homes in CON states.
Principal Findings: After the adoption of the
SNF PPS, nursing homes in CON states had a
1-percentage point lower likelihood of exiting the
market, and they gained over 3 beds relative to
nursing homes in non-CON states. After PPS,
Medicare nursing home spending per enrollee
increased by $54in CON states relative to non-
CON states, while Medicaid nursing home
spending increased by $119 per enrollee. On
the other hand, home health care spending per
enrollee decreased at a higher rate (by $78 from
Medicare and by $99 from Medicaid) in CON
states relative to non-CON states.
Conclusions: CON laws during the cost-based
payment system period enabled incumbant
nursing homes to gain market power that
protected them from the effects of adoption of
PPS. Compared to states with CON laws,
nursing homes closed at a higher rate and
became smaller in states without CON laws after
adoption of the PPS. CON laws caused nursing
home expenditure to increase at a higher rate
and impeded the growth of home based care
following the adoption of PPS.
Implications for Policy, Delivery, or Practice:
A supply based policy can offset and even
reverse some of the effects of a payment based
policy. Policy makers and other stakeholders
should be most concerned about the total effects
of the policies, and not only about the potentially
very limited impacts of isolated policy
components.
Funding Source(s): NIH
Poster Session and Number: B, #664
Disenrollment from Medicare Advantage
Plans following Nursing Home Admission
Momotazur Rahman, Brown University; Amal
Trivedi, Brown University; Vincent Mor, Brown
University
Presenter: Momotazur Rahman, Investigator,
Brown University
momotazur_rahman@brown.edu
Research Objective: While a large number of
studies documented that Medicare Advantage
(MA) plans selectively enrolled relatively
healthier Medicare beneficiaries, relatively
recent studies found that the 2005 Medicare
Modernization Act (MMA) greatly reduced
measurable differences between fee-for-service
(FFS) and MA enrollees. This paper focuses on
disenrollment of Medicare beneficiaries from MA
plans following nursing home admission. The
central hypothesis of this paper is that nursing
home admission among community residing
patients changes expected future health care
utilization and incites patients to switch to more
generous plans. Thus we hypothesize that MA
patients are more likely to switch plans than FFS
patients following nursing home admission. We
also hypothesize that such switching rates
varied with patients nursing home length of stay
and acuity and became smaller after the
enactment of MMA.
Study Design: Using longitudinal Minimum data
set (MDS) and Medicare enrollment data, we
followed patients for one year from the day of
admission and compared the prevalence of
switching from the plan during admission among
patients who survived at least one year following
admission. We examined the relationship
between one year switch rate and length of
nursing home stay and how such relationship
changed in year 2008 compared to year 2000.
We tested these relationships using logit model
controlling for patient characteristics and nursing
home fixed effects.
Population Studied: We examined about 2
million patients who were admitted to nursing
homes in year 2000 and 2008 and did not reside
in nursing home during one year prior to
qualifying admission.
Principal Findings: While there are no
discernable difference clinical conditions
between MA and FFS patients during nursing
home admission in any year, differences in
demographic and socio-economic
characteristics (for example, fraction married,
fraction with at least high school education,
share of white) between FFS and MA patients
are statistically significant and smaller in 2008
than in 2000. Among the FFS patients who
survived for at least one year following nursing
home admission, fraction switched to MA within
one year was 2% in 2000 and 3% in 2008. On
the other hand, among similar MA patients,
fraction switched to FFS within one year nursing
home admission was about 21% in 2000 and
10% in 2008. Switch rates were about two times
higher for the MA patients who became longstay nursing home resident following admission
than short-stay MA patients.
Conclusions: Disenrollment from MA plans
following nursing home admission is prevalent
and positively related to patients’ nursing home
care utilization. However, such disenrollment
declined over time.
Implications for Policy, Delivery, or Practice:
Medicare Modernization Act of 2005 reduced
differences between MA and FFS patients as
well as the selective disenrollment of the MA
patients. However, there is room for
improvement in delivering long-term care among
MA patients.
Funding Source(s): NIH
Poster Session and Number: B, #665
Impact of Hospital-Based Geriatric Services
on Readmissions, Time to Readmission, and
Readmission Costs
Deborah Redmond, UPMC Center for HighValue Health Care; Pamela Peele, UPMC
Center for High-Value Health Care; Don Yoder,
UPMC Center for High-Value Health Care;
Donna Keyser, UPMC Center for High-Value
Health Care
Presenter: Deborah Redmond, Vice President
Clinical Affairs, UPMC Center for High-Value
Health Care
redmonddk@upmc.edu
Research Objective: To evaluate the impact of
the UPMC Supportive Services program, which
delivers hospital-based geriatric services with
continued outpatient support to a 65 year old
and older hospitalized population, on 30-day,
any Diagnosis Related Group (DRG)
readmissions.
Study Design: Individuals receiving hospitalbased geriatric support services (intervention
group) were compared to a propensity-matched
group who did not receive services (comparison
group) using a difference-in-difference design.
The primary outcomes of interest were 30 day,
any DRG hospital readmission rates; secondary
outcomes included post-discharge emergency
room use, time to readmission, and cost of
readmissions. The comparison group was
matched on age, sex, type of insurance
coverage, Charlson Comorbidity Index, and
reason for hospitalization. The study included
initial medical admissions with discharges
occurring January 1, 2011 through May 31,
2012. The difference-in-differences was
estimated using a logistic regression for
readmission and emergency room utilization and
a Poisson regression for days to readmission.
Readmission cost data were analyzed using a
linear regression model. All models were
adjusted for age, sex, type of insurance,
Charlson comorbidity index, and admission
diagnosis. Insurance claims data were the
primary source of information for all variables.
Population Studied: The intervention group
comprised 818 individuals aged 65 years or
older at the time of their initial hospitalization
who were insured by UPMC Health Plan and
assessed as needing supportive services via
palliative care assessments. A matched
comparison group comprised 972 individuals
who were insured by UPMC Health Plan but not
part of the study population. Study consent for
the intervention group was obtained from both
the individual and the attending physician.
Principal Findings: There was no significant
change in the readmission rate or the use of
emergency room services for individuals in the
intervention group compared to the comparison
group. However, the time to readmission was
significantly longer, by almost 11 days, for
individuals in the intervention group compared to
those in the comparison group (p=0.056).
Further, readmissions that occurred among the
intervention group were significantly less
expensive by an average of $4,800 per
readmission than readmissions in the
comparison group (p=0.01).
Conclusions: Hospital-based geriatric support
services did not decrease 30-day readmission
rates but did result in significantly more
community-dwelling days before readmission
and significantly less expensive readmissions,
suggesting that geriatric support services can
positively impact the post-hospital course for
patients.
Implications for Policy, Delivery, or Practice:
The results of this study highlight the potential
for hospital-based geriatric support services to
impact the time to readmission and the cost of
readmission. Given Medicare’s priority on
decreasing readmission rates, these findings
can provide guidance to hospitals around
managing readmissions.
Funding Source(s): Other UPMC Health Plan
Poster Session and Number: B, #666
End of Life Care Policy Reform: Terminally Ill
or Due for Resuscitation?
Hannah Schreibeis-Baum, Veterans
Administration; Lea Xenakis, Samueli Institute;
Joanne Lynn, Altarum Institute; Gina Brown, VA
Greater Los Angeles Healthcare System; Gery
Ryan, RAND Corporation; Karl Lorenz, VA
Greater Los Angeles Healthcare System
Presenter: Hannah Schreibeis-Baum, Health
Science Specialist, Veterans Administration
hannah.schreibeis-baum@va.gov
Research Objective: 1. Understand the most
emphasized topics in current legislation
regarding palliative and end-of-life care and how
that legislation is distribution nationally.
2. Discover the gaps between current policy and
the major problems identified by experts and
stakeholders in palliative and end-of-life care.
Study Design: We systematically reviewed and
thematically coded state and federal legislation
(2010-2012) regarding palliative, EOL care, and
hospice. We interviewed 22 palliative and EOL
experts using “free association” to identify
problems and solutions for improving palliative
and EOL care and characterized themes using
content analysis. We compared legislation with
the solutions suggested by experts to identify
policymaking gaps.
Population Studied: A snowball sampling
method was used to contact and interview policy
experts in end of life and palliative care. These
participants ranged from leaders of hospice
organizations, end of life care lawyers, and
palliative care physicians among others.
Principal Findings: Of 193 bills (166 state, 27
federal), Connecticut (n = 20) and
Massachusetts (n = 16) enacted the most state
bills, and 50% of state bills addressed hospice
issues. More than one-half of federal bills
focused on payment for and access to hospice
and the quality of hospice care.Interviews
identified the most prevalent problems as: 1)
lack of provider training and education, 2)
cultural-societal beliefs surrounding palliative
and EOL care, 3) lack of payment for palliative
services, and 4) lack of public awareness. The
three most common solutions from experts
were: 1) providing training and education for
providers, 2) providing payment and redesigning the payment system, and 3)
conducting research and development. Except
for the conjunction of legislation and expert
opinion on improving hospice payment, the two
activities were not convergent.
Conclusions: Although legislation focused on
hospice, experts endorsed that improving care
requires broad action on structural payment
reform, provider training, and public awareness.
Implications for Policy, Delivery, or Practice:
The Affordable Care Act will initiate innovations
in payment, but provider training and public
awareness require governmental, professional,
and philanthropic action.
Funding Source(s): Other California
HealthCare Foundation
Poster Session and Number: B, #667
Is Asymptomatic Carotid Endarterectomy
Appropriate for Octogenarians? An
Evaluation of Five-Year Survival among
Medicare Beneficiaries
Marcus Semel, Brigham and Women's Hospital;
Thomas Tsai, Brigham and Women's Hospital,
Harvard School of Public Health; Edward
McGillicuddy, Brigham and Women's Hospital;
C. Keith Ozaki, Brigham and Women's Hospital;
Michael Belkin, Brigham and Women's Hospital;
Ashish Jha, Harvard School of Public Health
Presenter: Marcus Semel, Vascular Surgery
Fellow, Brigham and Women's Hospital
msemel1@partners.org
Research Objective: Prophylactic surgical
procedures such as asymptomatic carotid
endarterectomy are expensive and associated
with substantial morbidity and mortality.
However, they can be beneficial if we select
patients based on their likelihood of achieving
long-term survival. For asymptomatic carotid
endarterectomy, patients generally need to
survive to five years to recoup the clinical cost of
undergoing surgery. We sought to understand
how often Medicare octogenarians undergoing
asymptomatic carotid endarterectomy survive
five years.
Study Design: We calculated overall mortality
and built Kaplan-Meier curves to examine 5-year
survival. We next built Cox proportional hazards
models to examine the independent predictors
of survival after asymptomatic carotid
endarterectomy.
Population Studied: We used national
Medicare data to examine patients over age 80
who underwent elective asymptomatic carotid
endarterectomy in 2006 without evidence of
transient ischemic attack or stroke 180 days
prior to surgery.
Principal Findings: In 2006 16,616 patients
over age 80 underwent asymptomatic carotid
endarterectomy and 1.3 percent died within 30
days of surgery. Overall, the probability of 5year survival was 0.56, 95 percent confidence
interval of 0.55-0.57. A multivariate Cox
proportional hazards model found that patients
who had chronic lung disease, diabetes, valvular
heart disease and renal failure all had much
lower 5-year survival than patients who lacked
these comorbidities. Increasing age was
associated with lower 5-year survival. Both age
and comorbidity affected 5-year survival with the
probability of survival ranging from 0.71, 95
percent confidence interval of 0.69-0.74, for an
80-year-old with no comorbidities to 0.18, 95
percent confidence interval of 0.04-0.38, for an
85-year-old with 3 comorbidities, p less than
0.01. Conversely, an 85-year-old with no
comorbidities had survival superior to an 80year-old with 3 comorbidities, 0.53 with a 95
percent confidence interval of 0.51-0.55 versus
0.34 with a 95 percent confidence interval of
0.23-0.45, p less than 0.01.
Conclusions: While asymptomatic carotid
endarterectomy is appropriate for many older
Americans, we found that we do a poor job of
appropriate selection as nearly half of all
patients over age 80 died before five years.
Implications for Policy, Delivery, or Practice:
Given that a major cause of high costs and
inappropriate care is the use of expensive,
invasive procedures among patients who are not
likely to benefit, our work suggests that we need
to do a better job identifying the right patients for
surgery.
Funding Source(s): No Funding
Poster Session and Number: B, #668
Further analyses by provider type and qualitative
quotes will be provided.
Conclusions: Quality EOL care extends beyond
managing the physical pain, but includes a
holistic perspective of care, patient control, and
a dedicated healthcare team.
Implications for Policy, Delivery, or Practice:
Tailoring the provision of care to consider these
elements can improve the EOL experience.
Findings from this study help denote areas for
focusing future quality improvement initiatives.
Funding Source(s): Other Canadian Institutes
of Health Research
Poster Session and Number: B, #669
What Matters Most During End-of-Life Care:
Perspectives from Palliative Care Providers
Hsien Seow, McMaster University
Addressing the Complexity of Policy
Implementation: Evaluations of the National
End of Life Care Programme in England
Jane Seymour, University of Nottingham
Presenter: Hsien Seow, Associate Professor,
McMaster University
seowh@mcmaster.ca
Research Objective: Research has
documented what matters most in end-of-life
(EOL) care from the perspective of seriously ill
patients and their families. However few studies
have described this from the perspective of
palliative care providers, who have daily
encounters with death and dying. This study’s
objective is to address this knowledge gap.
Study Design: We used in-person, semistructured interviews with front-line, managerial,
and administrative staff involved in EOL care
across 15 regions in Ontario. Qualitative data
were interpreted using thematic coding analysis
and grounded theory.
Population Studied: Palliative care providers in
Ontario
Principal Findings: Data from 107 respondents
were analyzed, from which 40 unique themes
emerged, further grouped into 9 parent themes.
44% of our respondents were nurses, 19%
physicians, and 37% other. The three most
frequently cited themes were 1. Fulfilling Patient
Wishes (e.g. aligning care plan to respect and
honor patient preferences; enabling patient
control), 2. Pain and Symptom Management
(e.g. addressing pain), and 3. Supporting Family
Needs (e.g. providing education and respite to
the family). The two most frequent parent
themes were 1. Addressing More than the
Physical Needs (e.g. communication; facilitating
dignity, peace, and closure) and 2. the Nature
and Quality of Palliative Care Delivery (e.g.
knowledgeable, caring, responsive team).
Presenter: Jane Seymour, Sue Ryder Care
Professor Of Palliative And End Of Life Studies,
University of Nottingham
jane.seymour@nottingham.ac.uk
Research Objective: The National End of Life
Care Programme was developed in England as
a service and practice-development initiative to
improve equity of access and quality of end of
life care for all adults with palliative care needs.
It supports local innovation. Its early work
influenced the development of the first End of
Life Care Strategy for England (2008); it was
subsequently responsible for implementation of
the Strategy. The aims of this paper are to
present key findings from an evaluation of the
the process, sustainability and impact of the
Programme in terms of its intended and
achieved outcomes relating to quality end of life
care for adults in England.
Study Design: A pragmatic evaluation
approach was employed to study the first (20047) and second (2008-11) phases of the
Programme. There were three elements to the
evaluation:
1. A national stakeholder enquiry using
qualitative interviews
2. Documentary analyses to study variations in
uptake of recommended end of life tools and
use of Programme publications;
3. Case studies of practice at local levels
using mixed methods of data collection (medical
records review; focus groups and interviews;
survey), with an emphasis on end of life care
planning.
Population Studied: Stakeholder enquiry: 57
individuals involved in developing and
implementing the Programme at local, regional
and national levels.
Documentary analyses: a)regional variations in
implementation of three care planning tools: The
Liverpool Care Pathway for Care of the Dying;
The Gold Standards Framework in End of Life
Care; The Preferred Priorities of Care tool; b)
uptake of 77 publications produced by the
Programme since 2008.
Case studies of practice at local levels: medical
notes review of 65 decedents; 42 health and
social care staff.
Principal Findings: Stakeholders perceived
that the Programme had been critical in the
development and implementation of the End of
Life Care Strategy. Factors seen as integral to
the roll out of the Strategy included perceptions
that the Programme staff team were flexible,
accessible, responsive and dynamic leaders.
Documentary analyses showed great regional
variation in tool uptake but higher uptake was
associated with larger proportions of home
deaths. Extensive dissemination of educational
materials had occurred. Locality based case
studies showed that there was no systematic
approach to the recording of discussions with
patients or carers about end of life care issues
and that tracking of this information in care
systems was poor. Raising end of life care
issues in communication practice was difficult for
staff.
Conclusions: The impact and outcomes of the
Programme are indivisible from the processes
and style of the working practices of Programme
staff; impact and outcome must be linked to the
activities and changes occurring at local levels
for which the Programme is a catalyst.
Implications for Policy, Delivery, or Practice:
The National End of Life Care Programme in
England initiated a process of unprecedented
growth and development in change in the field of
end of life care, involving many partners and
taking place at multiple levels. Our evaluation
shows that it is essential that work continues to
ensure greater understanding of and
engagement with end of life care among the
national health and social care workforce, to
improve outcomes for patients.
Funding Source(s): Other Department of
Health, UK
Poster Session and Number: B, #670
Infections and Risk Factors in Home Health
Care
Jingjing Shang, Columbia University; Jianfang
Liu, Columbia University
Presenter: Jingjing Shang, Assistant Professor,
Columbia University
js4032@columbia.edu
Research Objective: Home health care (HHC)
has been the fastest growing healthcare sector
for the past three decades and this growth is
expected to continue as the population ages.
The uncontrolled home environment, increased
use of indwelling devices, and the complexity of
illnesses among HHC patients lead to increased
risk for infections. Previous studies on infections
among HHC patients are limited by small
sample sizes, conducted at the local scope, and
lack controls for covariates. This study examined
the infection rates in HHC setting using the 2010
national Outcome and Assessment Information
Set (OASIS) data and identified risk factors for
infections among HHC patients.
Study Design: This is a cross-sectional study
using a 1% random sample of 2010 national
OASIS data. Infections were identified if records
indicate that patients were hospitalized or
received emergency care for the following
reasons: respiratory infection, urinary tract
infection, IV catheter-related infection, or wound
infection. Bivariate analysis between infection
and patient’s demographics, health status, and
admission condition were conducted. Variables
that were significantly (P < 0.10) related to
infection were chosen. Stepwise logistic
regression model using the selected variables
were conducted to identify risk factors for
infection.
Population Studied: The final analysis included
34,432 adult patients from 5,674 home care
agencies nationwide. The average age of the
study sample was 74 years old. 62% were
female.
Principal Findings: Of 34,432 HHC patients,
1,026 (2.98%) had infections during their HHC
treatment period. Among 5,965 (17.3%) patients
who were transferred to inpatient facilities during
the HHC treatment, 15.7% were caused by
infections, making infection as one of the top
reported reasons for hospitalization in HHC
patients. The identified risk factors for infection
included history of smoking, recent history of
indwelling catheters and UTI treatment before
HHC admission, having unhealed ulcers, or
open wound, receiving oxygen treatment at
home, having impaired ADL/IADLs functioning,
and with caregivers who needed training in
providing medical procedure or treatment.
Conclusions: Our findings suggest that HHC
patients are at risk for infection. Patients with
underlying medical conditions and limited
ADL/IADLs are more likely to get infection. The
caregiver’s lack of training in providing the
needed care at home also poses HHC patients
at high risk for infection.
Implications for Policy, Delivery, or Practice:
As hospital stays are getting shorter and HHC
service expands, many acutely ill and vulnerable
patients will receive more invasive and
advanced care at home, and patients’ risk for
infections will continue to increase. Infection
prevention and control, which is critical in
assuring high-quality, safe HHC services need
to receive more attention. A surveillance system
based on the identified risk factors from our
study will help HHC agencies to identify patients
who are at high risk for infection. Infection
prevention and control practice at HHC setting
should include an initial evaluation of patients
and caregivers’ ability to provide the needed
care. Patient and caregivers education should
be tailored based on their literacy level to ensure
a complete understanding.
Funding Source(s): NIH
Poster Session and Number: B, #671
The Prevalence of Infections and Patient
Risk Factors in Home Health Care: A
Systematic Review
Jingjing Shang, Columbia University; Chenjuan
Ma, University of Kansas School of Nursing;
Lusine Poghosyan, Columbia University School
of Nursing; Dawn Dowding, Columbia University
School of Nursing; Patricia Stone, Columbia
University School of Nursing
Presenter: Jingjing Shang, Assistant Professor,
Columbia University
js4032@columbia.edu
Research Objective: Home health care (HHC)
has been the fastest growing healthcare sector
for the past three decades. The uncontrolled
home environment, increased use of indwelling
devices, and the complexity of illnesses among
HHC patients lead to increased risk for
infections. Studies have been conducted since
the 1990’s to examine infections rates and risk
factors among patients in HHC setting. This
systematic review critically reviewed and
synthesized published evidence on infection
prevalence and risk factors among adult patients
who received HHC services and to evaluate the
methodological quality of these studies.
Study Design: The Preferred Reporting Items
for Systematic Reviews and Meta-Analyses was
used to guide this systematic review. Three
electronic databases, Medline, PubMed, and
CINAHL were used and the search terms
included “home care”, “home health care”,
“hospice”, “home infusion” in various
combinations with “infection”, “sepsis”,
“pneumonia”, “infectious disease”, and
“communicable diseases”. Hand searching of
reference lists was also conducted to identify
relevant citations. Study quality was assessed
by two reviewers independently using two
validated observational research checklists
respectively, one for studies only describing
infection rates, the other for studies examining
risk factors.
Population Studied: Original research
published in English through May 2013 that
primarily examined the infection rates and/or
identified risk factors of infections in adult
patients receiving HHC services were screened
for eligibility. Studies with very small sample
sizes (< 20) were excluded.
Principal Findings: Twenty-five studies met the
inclusion criteria and were reviewed. The
infection rates varied dramatically between
studies. The highest rate was reported in one
HHC site in which 81% patients receiving home
parental nutrition (HPN) treatments had central
line associated bloodstream infection. In
general, patients receiving HPN had higher
infection rates than patients receiving home
infusion therapy. HHC patients with indwelling
devices were at high risk for infections. The
majority of studies were conducted in single-site
HHC settings with limited generalizability. The
ability of studies to identify risk factors for
infections was often limited by methodological
flaws including small sample sizes and a lack of
control for potential covariates.
Conclusions: This systematic review is the first
to examine infection rates and risk factors in
HHC setting. The results suggest infection is
prevalent in HHC settings. As HHC service
expands, many acutely ill and vulnerable
patients will receive more invasive and
advanced care at home, and patients’ risk for
infections will continue to increase.
Implications for Policy, Delivery, or Practice:
Future studies should use a national
representative sample and multivariate analysis
for the identification of risk factors for infections.
Establishing a surveillance system for HHC
infections will help HHC agencies to begin
benchmarking. Utilizing research evidences,
HHC agencies will be able to identify patients at
high risk for infections, and tailor home health
care and patient education based on patient
living conditions. It is also important to facilitate
communication between different health care
facilities in order to enhance infection control in
HHC settings.
Funding Source(s): No Funding
Poster Session and Number: B, #672
Hospitalization and risk factors among Home
Health Care Patients
Jingjing Shang, Columbia University; Chenjuan
Ma, University of Kansas School of Nursing;
Jianfang Liu, Columbia University School of
Nursing; Jinjiao Wang, Columbia University
School of Nursing; Lusine Poghosyan, Columbia
University School of Nursing
Presenter: Jingjing Shang, Assistant Professor,
Columbia University
js4032@columbia.edu
Research Objective: Home health care (HHC)
is growing in past decades. There were over 12
million Americans receiving HHC nationwide in
2010. While HHC is supposed to keep patients
away from hospitals, studies showed that HHC
patients were admitted back to hospitals for
varied reasons. As hospitalization of home care
patients has serious physical, mental, and
financial consequences, it is important to
understand what put HHC patients at risk for
hospitalization. However, the existing research
on hospitalization of HHC patients has been
conducted exclusively at the local level, lacks
current data, and is often limited by small
sample sizes. Using the 2010 national Outcome
and Assessment Information Set (OASIS) data,
this study examined the hospitalization rates and
identified the risk factors for hospitalization
among HHC patients.
Study Design: This is a cross-sectional study
using 1% random sample of 2010 national
OASIS data. HHC patients’ health status change
was captured by the indication of hospitalization,
death, or discharge from home care. Bivariate
analysis between outcome variable
(hospitalization) and patient’s demographics,
health status, and admission condition, which
were all collected at admission phase, were
conducted. Variables that were significantly (P <
0.10) related to outcome variable were chosen
and entered into the stepwise logistical
regression which revealed the significant risk
factors for hospitalization among HHC patients.
Population Studied: There were totally 34,432
adult patients from 5,674 home care agencies
nationwide. The average age of the study
sample was 74 years old. 83% of patients were
over 65 years and 62% were female.
Principal Findings: 5,982 out of 34,432
(17.28%) HHC patients were hospitalized during
the 60-day HHC stay. The top 5 reported
reasons for hospitalization were respiratory
problem other than respiratory infection,
respiratory infection, heart failure, dehydration or
malnutrition, and wound infection or
deterioration. Stepwise logistic regression
identified the risk factors for hospitalization as
being African American, smoking history, with
pre-conditions such as IV catheter, impaired
decision making, or urinary incontinence, and
needing assistance on ADL/IADL and
medication management (P<0.05).
Conclusions: Our findings suggest a high
hospitalization rate among HHC patients. We
revealed that African American patients, smoker,
and patients with pre-conditions have high risk
for hospitalization. Our study also suggests that
patient who were not capable in ADL/IADL
functioning, and managing medication were
more likely to get hospitalized.
Implications for Policy, Delivery, or Practice:
As HHC service expands, more patients will be
receiving care at home. As the hospitalization
may cause significant burden to patient and
society physically, psychologically and
financially, identifying high risk patients and
implementation of strategy to reduce the risk will
be beneficial. Our findings indicate that
information collected when patients entered
HHC can be utilized to identify the ones who are
at high risk for hospitalization. Our study also
suggests that HHC should be tailored according
to patients’ individual needs in order to prevent
hospitalization and reduce relevant health care
cost.
Funding Source(s): NIH
Poster Session and Number: B, #673
Long-Term Care Insurance and Health Care
Financing in South Korea
Jaeun Shin, KDI School of Public Policy and
Management
Presenter: Jaeun Shin, Professor, KDI School
of Public Policy and Management
jshin@kdischool.ac.kr
Research Objective: This paper is to evaluate
how the new element of public health insurance
for long-term care services (henceforth, LTCI)
financially affects the preexisting National Health
Insurance program (NHI). We focus on
examining the role of the LTCI benefit in service
use and costs of the elderly members in the NHI
program.
Study Design: Using longitudinal data of 245
municipalities of South Korea for the period of
2008-2011, we conduct fixed-effects panel
estimation of the NHI service use (total number
of patients, total number of admissions, total
treatment days) and total spending for the
elderly as a function of the LTCI spending for the
elderly beneficiaries.
Population Studied: We refer to the 245
administrative municipalities (Si, Goon, Gu) of
South Korea and their annual statistics on the
LTI and NHI program published by the NHIS for
2008-2011.
Principal Findings: We find the statistically
significant positive association of the LTCI
payment for the elderly with the NHI payment for
the elderly. The estimate (0.024, p<0.01)
indicates additional 2,602 million KRW spending
in the NHI account generated by the 1%
increase in the LTCI spending on the elderly.
Also the LTCI payment positively affects total
number of admissions (0.025, P<0.01) whereas
there finds no significant change in total number
of patients. This indicates that the LTCI program
does not induce non-users to new users of the
NHI service, but does motivate users to make
more frequent visits to the NHI-service
providers.
Conclusions: The LTCI program may motivate
more services use among the elderly and
accordingly lead to higher NHI spending for the
elderly. It is probable but yet empirically
ambiguous whether the NHI spending for acute
care among the elderly can be saved
by the provision of the LTC services with the
support of the LTCI program.
Implications for Policy, Delivery, or Practice:
Two lessons can be drawn for the directions of
health care reforms in Korea: first, any attempt
to enhance the generosity of the LTCI program
in regard to eligibility and benefit package
should take into account its cost-inducing effect
on the NHI program. Secondly, the coordinated
management of the LTC service and the NHI
service delivery should be installed to serve the
complex needs of the elderly for long-term care
and acute care in a cost-effective way.
Funding Source(s): Other KDI School
Poster Session and Number: B, #674
Racial Differences in Nursing Home
Residents’ Quality of Life
Tetyana Shippee, University of Minnesota;
Carrie Henning-Smith, University of Minnesota;
Greg Rhee, University of Minnesota; Robert
Held, Minnesota Department of Human
Services; Robert Kane, University of Minnesota
Presenter: Tetyana Shippee, Assistant
Professor, University of Minnesota
tshippee@umn.edu
Research Objective: Racial differences in longterm care have been traditionally studied in
terms of (1) accessibility and rate of admission
to nursing home (NH) and other long-term care
services, and (2) quality of care for specific
medical conditions and clinical procedures.
Despite the recent increase in the share of racial
and ethnic minority groups in skilled nursing
facilities, little is known about their quality of life
(QOL). This study examines racial differences
in nursing home residents’ QOL, when facilityand resident-level characteristics are controlled
for, by investigating the following: (1) Are there
racial differences in NH residents’ QOL?; (2) If
so, do these differences exist at the resident
and/or facility aggregate level?; and (3) Do these
differences persist across domains of QOL?
Study Design: QOL is assessed with a
multidimensional tool measuring six unique
domains (environment, personal attention, food,
engagement, and positive and negative mood),
along with a summary score. Due to sample size
constraints, race is constructed as a binary
white/non-white measure and we evaluate its
association with QOL on the individual and
facility level. We used cross-sectional
hierarchical linear modeling (HLM) to identify
significant facility- and resident-level predictors
for racial differences in each of the six QOL
domains (plus the summary score) and
predictive marginal effects to estimate QOL
scores by race.
Population Studied: The setting for this study
is Minnesota NHs. Data come from: (1)
Resident-reported QOL (n=10,929) collected in
a survey of residents in each NH; (2) Resident
clinical data from the Minimum Data Set; and (3)
Facility-level characteristics from the Minnesota
Department of Human Services (n=376).
Principal Findings: Race was a significant
predictor of most QOL domains on bivariate
level. Specifically, white residents had higher
satisfaction with personal attention, food
enjoyment, engagement, and the overall
summary score while non-white residents
reported higher satisfaction with environment. In
multivariate analyses, when we controlled for
health and other characteristics, racial
differences on the individual level remained only
for two domains: social engagement and food
enjoyment. On the facility-level, even when
controlling for other covariates, higher
percentage of white residents was associated
with better scores in all domains but
environment.
Conclusions: Racial differences in QOL exist
on both individual and facility levels. However,
individual-level racial differences may be mainly
explained by health status, pointing to the
cumulative disadvantages encountered by
minority elders. Proportion of white residents on
the facility level remained a significant predictor
of QOL across most domains, suggesting the
importance of examining structural policies,
community factors, and geographic inequality in
understanding QOL.
Implications for Policy, Delivery, or Practice:
Results may be used to inform interventions to
improve quality of life for the growing population
of non-white NH residents.
Funding Source(s): NIH
Poster Session and Number: B, #675
Healthy Aging, Patient-Centered Care and
Burden of Disease Considerations in Older
Adults with Hearing Loss (HL)
Annie Simpson, Medical University of South
Carolina; Kit Simpson, Medical University of
South Carolina; Judy Dubno, Medical University
of South Carolina
Presenter: Annie Simpson, Assistant Professor,
Medical University of South Carolina
simpsona@musc.edu
Research Objective: Recent focus on patientcentered outcomes research (PCOR), global
burden of disease (GBD), and an Institute of
Medicine (IOM) workshop on HLand healthy
aging shed light on important clinical and policy
implications when evaluating quality of life in
older adults. PCOR is a nationally focused effort
to provide evidence for interventions that
concentrate on what is important to patients.
The GBD study is an international collaboration
to examine and rank diseases based on their
impact on population health. The IOM workshop
brought attention to the broad health
implications of untreated HL in older adults. The
objective of this study is to explore the impact of
HL on quality of life on the US geriatric
population, separate from the length of survival,
using the 2000 Medical Expenditure Panel
Survey (MEPS) and 2010 Census data.
Study Design: Retrospective cohort utility and
population models.
Population Studied: Data included 12,542
subjects age 60-90 in the 2000 MEPS who
provided self-reported HL, HRQoL and
information on chronic conditions. The EQ5D
visual analog scale (VAS) transformation was
used to estimate marginal utility decrements for
5-year age categories and conditions. Utility
decrements were estimated using multivariable
regression including; five-year age group, HL,
hypertension, diabetes, angina, joint pain,
asthma, emphysema, or blindness. The
modeled decrements were applied to the US
census population by age group to estimate
annual QALYs lost from each condition for the
US 2010 population.
Principal Findings: Of the respondents 15.4%
had Mild HL and 1.1% had Moderate/Severe
hearing loss. The presence of other conditions
(utility decrement) were: Joint pain 53% (.0643),
hypertension 47.2% (.0292), diabetes 15.6%
(.0577), angina 9.8% (.0352), asthma 7.9%
(.0288), emphysema 4.5% (.1186), blindness
0.8% (.0836), and age decrement (.0033) per
year. When these decrements were applied to
the US population age 60-90 the QALY loss
ranged from a high of 821,918 for joint pain to a
low of 16,614 for blindness. The decrement from
hearing loss ranked 4th at 174,689 and was
outranked only by joint pain, hypertension and
diabetes.
Conclusions: HL may be expected to have a
substantial impact on healthy aging, which is not
obvious when quality of life decrements are
calculated to include survival, or when its effects
are diluted by including younger populations.
From a public health perspective HL ranks 31 in
the GBD, with much of the decrement being
conferred by HL in developing countries.
However, HL ranks 4th when the objective of the
estimate is to examine the impact on healthy
aging in the US geriatric population. These
exploratory findings warrant careful
consideration of interventions for age-related HL
in clinical practice, and suggest that further
research is needed to examine the effect of
hearing loss on the quality of life of otherwise
healthy older adults.
Implications for Policy, Delivery, or Practice:
It is important for primary care practitioners to
understand the impact of HL on their patients
and provide appropriate referrals and counseling
on amelioration. Our results raise critical issues
for health policy formation for the aging “Baby
Boomer” generation, and suggest that the
current lack of Medicare coverage for hearing
aids and related services should be reassessed.
Funding Source(s): NIH
Poster Session and Number: B, #676
Alignment of the Medicare Benefit Package
for People with Advanced Cancer with the
Preferences of Patients and their Caregivers
Donald Taylor, Duke University; Marion Danis,
National Institutes of Health; S. Yousuf Zafar,
Duke University Health System; Gregory
Samsa, Duke University Center for Learning
Health Care System, DCRI; Stevn Wolf, Duke
University Center for Learning Health Care
System, DCRI; Amy P. Abernethy, Duke
University Center for Learning Health Care
System, DCRI
Presenter: Donald Taylor, Associate Professor,
Duke University
don.taylor@duke.edu
Research Objective: To identify benefit
categories most important to Medicare
beneficiaries with cancer and their family
caregivers when faced with a resource
constraint that meant they could not choose
everything
Study Design: Patients seeking cancer care
from Duke University Health System and
community organizations were enrolled in a
facilitated, participatory decision making
exercise called Choosing Health Plans All
Together, that was held over a 2.5 hour period.
Caregivers were also enrolled. Participants
made benefit choices before and after
participating in discussions of priorities with one
another. Data were collected from August 2010March 2013. The resource constraint imposed
was substantial; participants had 50 stickers to
deploy, but 89 were required to pick the
maximum level of every benefit offered.
Population Studied: A total of 439 participants,
246 patients and 193 caregivers were enrolled in
the study. The four most common cancer types
were breast, lung, prostate and colon. Most of
the patients lived within a 2 hour drive of
Durham, N.C.
Principal Findings: Six of 15 benefit categories
were selected by more than 80% of participants:
cancer care, prescription drugs, primary care,
home care, palliative care, and nursing home
coverage. Only 12% of participants chose the
maximum level of cancer benefits offered, a
level of care commonly financed in the Medicare
program. Over 40% allocated some of their
scarce resources to a cash benefit that could be
used for any purpose, and a similar proportion
chose expanded long-term care to address the
limitations of frailty and advanced illness.
Conclusions: Under resource constraints,
participants chose benefits that differ from the
traditional Medicare benefit package in important
ways, including allocating scarce resources
away from cancer care toward more general
needs. While the priorities of cancer patients
elicited here are valuable and provocative, the
question remains how to best facilitate dialogue
and discussion about Medicare benefits in the
broader culture to inform policy.
Implications for Policy, Delivery, or Practice:
Refocusing the Medicare benefit package to
allow patient flexibility in choosing to forego
some traditional benefits in favor of others could
better align patient needs and preferences with
what is covered by the program. Our results
highlight the "holes" in the Medicare benefit
package related to Long Term Care, and ability
to receive palliative care concurrently with
curative treatments.
Funding Source(s): AHRQ
Poster Session and Number: B, #677
Integrating New Quality Provisions into
Nursing Homes
Gail Towsley, University of Utah; Katherine
Sward, University of Utah; Julius Kehinde,
University of Utah; Connie Madden, University of
Utah; Jia-Wen Guo, University of Utah
Presenter: Gail Towsley, Assistant Professor,
University of Utah
gail.towsley@nurs.utah.edu
Research Objective: Quality Assurance
Performance Improvement (QAPI) is a new
regulatory initiative by the Centers for Medicare
and Medicaid that requires nursing homes to
broaden their existing scope of Quality
Assurance and Assessment processes to
include components of quality improvement (QI)
and include widespread facility participation and
staff responsibility to the process. Historically,
regulatory provisions have focused on quality
assurance (QA)--a retrospective process to
determine if care processes meet quality and
regulatory standards whereas QI is a continuous
proactive data driven process to prevent or
decrease problems by identifying and resolving
persistent issues. The purpose of this study was
to understand QA and QI processes in rural
nursing homes with limited resources. Our aims
were to (1) describe the structure, processes,
and context in which measurement of quality
occurs, and (2) assess perceptions about staff
engagement in QI-supporting behaviors.
Study Design: We used a mixed methods
design to evaluate qualitative and quantitative
data. Qualitative data included interviews,
observations of the nursing home environment,
diagramming QA processes, and reviewing QA
data. Quantitative data included a modified
Agency for Healthcare Research and Quality
(AHRQ) Readiness Assessment Tool (RAT).
This instrument (57 items; 1= strongly disagree
to 5 = strongly agree) is designed to assess
nursing home readiness to undertake
organization-wide practice and culture changes
for QI.
Population Studied: We interviewed 18 key
personnel (including three corporate executives)
of four small (60 beds or less) rural nursing
homes located in Kansas and distributed RAT
surveys to all staff.
Principal Findings: Current QA efforts were
affected by organizational culture and behaviors
of inter-professional nursing home staff including
nursing, rehab, dietary, and other personnel. We
modeled the contextual and structural aspects of
QA processes to describe how, what, and
from/to whom information was collected, acted
upon, and reported. Our models demonstrated
task oriented processes led by Directors of
Nursing who also collected most of the QA data.
Eighty-four RAT surveys were returned; mean
scores for RAT subscales (management,
documentation/technology, education,
communication, workload/resources, and quality
improvement) ranged from a low of 2.92 (.74) for
workload/resources, to a high of 3.36 (.88) for
education. Staff voiced a willingness and
positive attitude toward organization changes
especially related to technology, if given
appropriate time and training.
Conclusions: Results from QA data collection
or QA meetings rarely were communicated to all
levels of staff, resolved as a team, or led to
system-wide or organizational changes. Many
staff reported QI was important and that they felt
they were not directly involved in QA processes
and decisions.
Implications for Policy, Delivery, or Practice:
Implications of our findings focus on ways to
incorporate QAPI approaches into the nursing
home setting. We model potential recommended
changes, such as integrating leadership
practices (e.g. shared vision) and conducting a
QAPI self-assessment of current QA processes.
Future research should include how integrating
all levels of inter-professional staff into QAPI
processes influences these processes and leads
to improved care delivery.
Funding Source(s): Other Deseret Health
Group
Poster Session and Number: B, #678
Comparing Home- and Community-Based
Care Models for Medicaid/Medicare Dual
Eligible Individuals
Janet Van Cleave, New York University; Sarah
Brosch, New York University Langone Medical
Center; Elizabeth Wirth, New York University
Langone Medical Center; Molly Lawson, Mount
Sinai Medical Center; Brian Egleston, Fox
Chase Cancer Center; Eileen Sullivan-Marx,
New York University College of Nursing; Mary
Naylor, University of Pennsylvania School of
Nursing;
Presenter: Janet Van Cleave, Assistant
Professor, New York University
janet.vancleave@nyu.edu
Research Objective: Medicare and Medicaid
dual eligible population consists of the nation’s
most poor and chronically ill individuals. The
population comprises 20% of Medicare and 15%
of Medicaid beneficiaries, yet account for 31% of
Medicare and 39% of Medicaid expenditures.
Approximately 1.5 million dual eligible
individuals receive home- and community-based
services (HCBS). However, research comparing
HCBS care models for dual eligible population is
limited due to separate and siloed agencies
serving these individuals. The purpose of this
study was to compare providers’ hours,
personnel costs, and patient hospitalizations of
two care models providing HCBS services for
dual eligible individuals.
Study Design: This study was a retrospective
analysis of medical record data. The dependent
variables for this analysis were providers’ hours
and personnel costs. Data were collected on
providers’ hours for 7 consecutive days per
patient at 1- 3- and 6-months after enrollment in
HCBS. Personnel costs were estimated using
Bureau of Labor Statistics 2012 Occupational
Employment and Wages. Hospitalizations or
short-term stays in skilled nursing facilities were
also collected over 7 consecutive day periods at
1-, 3-, and 6-months after enrollment in HCBS.
Data on study participants’ baseline functional
status and comorbidities were abstracted from
each organization’s enrollment assessment.
Generalized linear models (GLM) using
generalized estimating equations estimated
relative ratios of providers’ hours and personnel
costs of the two care models while controlling for
age, race, ethnicity, function, and comorbidity.
Population Studied: 49 dual eligible individuals
enrolled in Health Related Quality of Life: Elders
in Long Term Care Study (Mary D. Naylor, PI)
who received care from two capitated HCBS
programs in northeastern United States between
2007 and 2009. The two HCBS care models
were managed long-term care (MLTC) and
integrated care programs (MLTC n=31,
Integrated Care n=18). MLTC program
emphasizes coordinated care and in-home
services, whereas integrated care program
provides inter-professional primary care and
senior day care center services.
Principal Findings: The study population was
older (mean age = 79 years) and primarily Black
(56%). On average, MLTC patients received
greater number of care hours during the 21-day
data collection period than integrated care
patients (MLTC: 21 hours per patient; Integrated
Care: 4.7 hours per patient, p = .000). The GLM
analysis showed that, on average, MLTC
providers spent 4.0 times more hours per patient
than integrated care providers (p=.000). As a
result, MLTC program personnel were estimated
to cost 3.5 times more dollars per patient than
the integrated program (p=.000). During the data
collection period, MLTC participants were
hospitalized three times, whereas the integrated
care patients’ experienced one hospitalization
and two short–term stays.
Conclusions: Study findings suggest that
HCBS care models with integrated care and
senior centers may be more efficient in providing
care for dual eligible individuals than care
models emphasizing coordinated care and inhome services.
Implications for Policy, Delivery, or Practice:
More research is needed to better understand
which HCBS care models meet this vulnerable
population’s needs and desires while delivering
quality care.
Funding Source(s): NIH New York University
Research Challenge Fund Program
Poster Session and Number: B, #679
The Impact of Assistive Device Use on the
Social Participation of Older Adults
Judith Walsh, University of North Carolina at
Charlotte
Presenter: Judith Walsh, Research Assistant,
University of North Carolina at Charlotte
jwalsh23@uncc.edu
Research Objective: Literature shows that the
use of devices to assist with mobility and
sensory problems has become relatively more
prevalent in attempts to meet the needs of this
country's older population. While assistive
technology device (ATDs) use to counteract
these problems has been found to improve
functionality, their effectiveness in relation to
further outcomes has not been affirmed. One
such measure of success is participation in
social activity, which has been linked through
Social Capital Theory to both individual and
societal health benefits. The objective of this
research is to determine if the use of ATDs has
a positive impact on the level of social
involvement for older adults.
Study Design: Using a nationally representative
sample of older Americans, this paper examines
how these devices fit into the World Health
Organization's International Classification of
Functioning, Disability & Health Framework
(ICF) to maximize function and minimize
limitations for older people. The research uses
logistic regression to test hypotheses that an
individual's ATD use will have a positive impact
on participation in five separate types of social
activity. We control for other barriers that could
discourage participation such as health and
transportation issues, variables that measure
interest in the activities and a variety of
environmental and social support variables.
Marginal effects are used to show how the
probability of an individual participating in a
particular activity is predicted to change as each
of the independent variables changes ceteris
paribus.
Population Studied: The NHATS is a nationally
representative sample of individuals aged 65
and older, drawn from the Medicare enrollment
file. The hypotheses are premised on need and
the use of any assistive device by one who has
no potential for benefit would be irrelevant,
therefore we limit our sample to three groups
which are considered separately: a) older adults
with problems with mobility, b) older adults with
sensory problems and c) older adults
experiencing problems performing ADLs.
Principal Findings: For some activities, such
as visiting family and friends, we find that the
use of assistive devices does not have a
significant impact, but their use was more of a
factor in other types of participation. Contrary to
the hypotheses, initial results indicate that
devices to help individuals with mobility often
discourage rather than encourage social
involvement in activities such as attending
religious services and joining clubs. For those
with hearing problems wearing hearing aids
emerges as one of the most potent indicators of
participation in most activities.
Conclusions: The benefits of Social Capital
gained through participating in social activities
have been well documented. With limited
resources at both the personal and societal
level, it is important to inspect particular devices,
as some may have a larger impact on
participation than others.
Implications for Policy, Delivery, or Practice:
By improving one's capacity to perform, assistive
technology offers the person with limited
functionality the potential to acquire a sense of
autonomy and meaningful connection to the
community. If successful, it removes much of
the need for help from other persons, and
reduces the demands of disability care on both
families and public programs for older adults
with limitations.
Funding Source(s): No Funding
Poster Session and Number: B, #680
End of Life Care for a Non-Medicare Insured
Population
Elizabeth Wasilevich, Blue Cross Blue Shield of
Michigan
Presenter: Elizabeth Wasilevich,
Epidemiologist, Health Care Manager, Blue
Cross Blue Shield of Michigan
ewasilevich@bcbsm.com
Research Objective: End of life (EOL) care for
those with chronic diseases often falls short of
patient and family preferences. Delayed
communication of a poor prognosis may result in
delayed hospice admission, unwanted
treatments, and hospitalizations in the days
leading up to death. The objective of this project
was to describe EOL care and its affiliated costs
for a commercially insured population with high
risk chronic conditions, as little is known about
EOL care received by decedents less than 65
years.
Study Design: Non-traumatic deaths were
identified using medical claims with a discharge
disposition of ‘deceased’ and having zero
medical claims after the presumed date of
death. Deaths were restricted to those
decedents with a history of a chronic condition
that increases risk of death based on published
methods. The distribution of deaths by
demographic characteristics and location of
death were determined. EOL care was
measured by admission frequency and duration
of hospice care and inpatient stays, number and
type of health care practitioners seen, frequency
of life saving (e.g. feeding tube) and imaging
procedures, and cost during varying time
periods before death ranging from three days to
six months. Estimated proportions and
averages were accompanied by 95% confidence
intervals.
Population Studied: This study included nontraumatic deaths between 2010 and 2012
among Blue Cross Blue Shield of Michigan
(BCBSM) commercial members. Decedents
were restricted to Michigan residents 64 years or
less who had a high risk chronic condition and
continuous medical coverage during six months
prior to death.
Principal Findings: There were 5,232 nontraumatic deaths identified, of which 36.6%
occurred in the hospital. The average age at
death was 56 years. Over half of decedents
(64.0%) had one or more hospital admissions in
the 30 days prior to death. Forty percent of
decedents were never admitted to hospice while
10.8% were admitted to hospice within three
days of death. The average length of stay in
hospice during the six months prior to death was
48 days. A large majority of all decedents
(83.7%) saw 10 or more physicians during this
time period. During the last 30 days of life,
21.0% of decedents had a life saving procedure
and 5.4% had an imaging procedure (3.8% high
tech; 2.3% low tech). There was an average of
$76,000 in medical costs per decedent during
the last 6 months of life, conferring about $300
million in total costs.
Conclusions: While there may have been
incomplete and/or biased ascertainment of
deaths using claims data, evidence suggests
significant opportunity to improve EOL care for
this younger population with high risk chronic
diseases. Specifically, EOL care could be
improved by decreasing in-hospital deaths,
increasing earlier hospice admission, and
improving potential issues with care coordination
by enhancing communication between the many
practitioners serving the dying population.
Implications for Policy, Delivery, or Practice:
This work contributes to the understanding of
the EOL care for a younger population and
demonstrates that administrative claims data are
valuable in measuring EOL quality of care.
These results will be used to support innovative
provider incentive initiatives and inform
programs that promote advance care planning.
Funding Source(s): No Funding
Poster Session and Number: B, #681
A Personalized, Patient Centered, Report
Card for Nursing Homes – NHCPlus
Dana B. Mukamel, University of California,
Irvine; Derek Gustafson, University of California,
Irvine; Alpesh Amin, University of California;
David Weimer, University of Wisconsin-Madison;
Joseph Sharit, University of Miami; Dara Sorkin,
University of California, Irvine
Presenter: Dana B. Mukamel, Professor,
Department Of Medicine, University of
California, Irvine
dmukamel@uci.edu
Research Objective: The Nursing Home
Compare (NHC) report card published by the
CMS includes 19 clinical quality measures
(QMs), staffing, and citation information. These
measures are not correlated, making the choice
of a nursing-home a difficult task for consumers.
To help consumers, the report card also
provides a 5-star system, which combines the
individual QMs into a composite measure. The
composite is based on weights provided by an
expert panel. The objective of this study is to
develop an alternative: an iPad-based
application allowing consumers to create
composite scores for nursing homes based on
their personal health needs and preferences,
effectively creating a personalized, patient
centered report card.
Study Design: We developed an iPad
application designed to be used by patients and
their families at the hospital bedside. It includes
an educational module, explaining each QM and
its implications, a preference elicitation module,
guiding the patient through a value-clarification
exercise, and a personally-tailored solution
module that provides the patient and family with
information about the nursing homes in their
choice set, ranked on a composite measure
constructed from their personal preferences and
the NHC QMs, staffing, and citation information.
The application also allows patients to add
information about price and re-sort nursing
homes based on other criteria. Key informant
interviews were performed with patients and
family members who have tested the application
to obtain feedback about its usability,
usefulness, and applicability.
Population Studied: Twenty patient and family
members of patients who have recently been
discharged from the University of California
Irvine Medical Center to a nursing-home.
Principal Findings: All informants, without
exception, reported that an application like the
one they were experiencing would have been
extremely helpful at the time they had to choose
a nursing home. Several had accessed NHC,
but indicated that nonetheless the application
had provided an added value and would have
made their decision easier and better. All felt
that choosing a nursing home is an important
decision and said they would be willing to spend
a substantial amount of time (20-30 minutes or
more) with the application to help them make it.
Based on feedback, we changed the method of
preference elicitation, focusing on relative
ranking only. We also added several features,
such as different ways to view summary
preferences, presentation of information about
price and quality, and others.
Conclusions: Developing a personalized report
card based on the patient's preferences is
feasible with today’s mobile technology. Such
report cards are welcomed by patients and
families who feel a need for personalized
decision making.
In the era of personalized medicine and patientcentered care, report cards seem to be another
frontier. Like other decisions that patients and
families make about their care, choosing a
provider should be patient-centered and
personalized. The NHCPlus application we
developed and pilot tested offers a prototype
and framework for combining personalized
preferences with quality metrics for other report
cards, especially report cards with many QMs,
such as HEDIS, which are likely to challenge
users.
Funding Source(s): AHRQ
Poster Session and Number: B, #682
Behavioral Health
Hospital Costs for Persons with Mental
Health or Substance Use Disorders
Michael Abrams, The Hilltop Institute; Martiza
Webb, The Hilltop Institute; Julie Gielner, The
Hilltop Institute
Presenter: Michael Abrams, Senior Research
Analyst, The Hilltop Institute
mabrams@hilltop.umbc.edu
Research Objective: Mental health and
substance use disorders are explicit targets of
health care reform, and much research indicates
that this cluster of illnesses correlates with
exceptionally high overall medical expenditures.
This work uses a unique, multi-hospital
database to estimate the relative impact of
patient factors on hospital service (inpatient,
emergency, surgical) costs among those who
also have behavioral health diagnoses.
Study Design: Maryland’s all-payer (private and
public) hospital claims database was used to
identify adults with any behavioral health
diagnosis (mental health or substance abuse) in
their 2011-2012 records. The 2012 records were
further used to summarize expenditure, payer
(Medicare, Medicaid, private, uninsured),
demographic (age, race/ethinicity, region,
marital status), and diagnostic (269 flags
spanning most medical domains) information.
The isolated users were divided into quintiles by
rank-ordering their aggregate hospital
expenditures, and the top 20 percent of the
sample was marked as “high” utilizers with the
remainder marked as “low” utilizers. A
multivariate logistic regressions was constructed
with the “high”/”low” utilizer flag as the
dependent variable, and the other variables as
predictors of that expenditure outcome.
Population Studied: 513,370 adults (age >17
years) were found with at least one behavioral
health diagnosis and one hospital service in
2012. The top quintile of the population had
average expenditures of $42,400 versus only
$3,500 for all others.
Principal Findings: The logistic model was
highly significant, accounting for over 48 percent
of the variance. Nearly all effects were
significant (p<0.5). Medicare and Medicaid
coverage were both associated with decreased
adjusted-odds ratios (AORs) of high utilizer
status compared to privately covered persons
(AORs=0.76 and 0.78, and 95 percent
confidence intervals (CIs)=0.73-0.78 and 0.760.83, respectively). Blacks were less likely to be
high utilizers than Whites (AOR=0.86, CI=0.840.88), and Hispanics demonstrated an even
larger effect vs. Whites (AOR=0.63, CI=0.58-
0.69). Men had increased odds of being high
utilzers than women (AOR=1.16, CI=1.13-1.18),
and single persons were slightly less likely to be
high utilizers than those who were married
(AOR=0.95, 0.93-0.98). Baltimore City
correlated with significantly increased odds of
high utilizer status compared to 19 of the 23
separate county jurisdictions in Maryland
(across these 19 counties, AORs versus
Baltimore City ranged from 0.30-0.91, while the
other 3 counties were not significantly different
from Baltimore City).
Ten of 11 psychiatric diagnoses demonstrated
increased odds of high utilizer status. Substance
use disorders, family and social problems,
psychoses, and personality disorders all
increased the odds of high utilizer status (AORs
range=1.5-1.9) more than the other seven
psychiatric categories. Disease load outside the
behavioral health domain averaged 8.1 specific
disease or treatment flags (standard
deviation=7.3), and for each additional flag, the
adjusted-odds of being a high utilizer increased
by 26 percent (AOR=1.257, CI=1.255-1.259).
Conclusions: Several visible administrative
data variables are significant correlates of high
hospital costs. Included among these are
diagnostic markers for behavioral health
disorders.
Implications for Policy, Delivery, or Practice:
This work confirms and quantifies the
importance of behavioral health illness as
significant medical cost-driver at the same time
as it quantifies the relative impact of that
morbidity against several other visible factors
which also appear to drive high hospital costs.
Funding Source(s): Other
Poster Session and Number: B, #685
Methadone Maintenance Treatment Facilities
Are Not Associated with Increased Crime in
US Counties
Marcus Bachhuber, Philadelphia Veterans
Affairs Medical Center; Colleen Barry, Johns
Hopkins Bloomberg School of Public Health
Presenter: Marcus Bachhuber, Fellow,
Philadelphia Veterans Affairs Medical Center,
Robert Wood Johnson Foundation Clinical
Scholars Program
marcus.bachhuber@gmail.com
Research Objective: The number of persons
with opioid use disorder in the United States is
increasing rapidly, and methadone maintenance
is an effective treatment. New methadone
maintenance treatment facilities are often
opposed due to concerns over the potential for
increased crime. Therefore, we sought to
determine the association between methadone
maintenance treatment facilities and countylevel crime in the United States.
Study Design: We used yearly county-level
crime data (2002-2010) from the Federal Bureau
of Investigation Uniform Crime Reporting
Program and data on methadone maintenance
treatment facilities from the National Survey of
Substance Abuse Treatment Services. We
examined three main outcomes: (1) non-violent
property crimes (i.e., larceny, motor vehicle
thefts, burglary), (2) violent crimes (i.e., robbery,
aggravated assault, sexual assault, and
murder), and (3) drug crimes (i.e., possession,
sale, and manufacturing). We ran linear
regression models with fixed effects to examine
the association between the presence of
methadone maintenance treatment facilities and
crime-related outcomes. All models included a
county fixed effect to adjust for time-invariant
county-level factors, a year fixed effect to adjust
for factors affecting counties which differ over
time and a state*year interaction, as well as the
yearly county-level unemployment and poverty
rates. The main independent variable, presence
of a methadone maintenance treatment facility,
was coded in several different ways:
dichotomously (i.e., presence/absence of one or
more facilities), as a continuous variable (i.e.,
the number of methadone treatment facilities in
the county), geographic density (i.e., facilities
per 1,500 km2—the approximate median size of
all counties), and population density (i.e.,
number of facilities per 100,000 population).
Standard errors were calculated accounting for
the repeated-measures structure of crime data.
All analyses were performed with SAS 9.3 (SAS
Institute, Cary, NC, USA).
Population Studied: All counties in the United
States with complete crime data (n=2 956
representing 94% of counties).
Principal Findings: In the US in 2002, 293
counties contained one or more methadone
maintenance treatment facilities. Between 20032010, one or more facilities opened in 182
counties . For property and violent crimes, no
association was found between methadone
maintenance facilities and county-level crime
rate in any of the regression models. Significant,
but modest, decreases in drug crime rates were
associated with an incremental increase of one
facility (-1.3% [CI: -2.6%, -0.06%]), and an
increase in geographic density of one facility per
1,500 km2 (-0.9% [CI: -1.3%, -0.5%]). No
associations between facilities and drug crimes
were found in models characterizing the
presence of a methadone maintenance
treatment facility as dichotomous (i.e.,
presence/absence of one or more facilities) or
as a population density.
Conclusions: We found no evidence to suggest
methadone maintenance facilities are
associated with increases in property or violent
crimes at the county level, and some evidence
to suggest a modest association between
methadone maintenance treatment facilities and
decreased drug crime.
Implications for Policy, Delivery, or Practice:
Concerns that opening new facilities may lead to
increases in crime are not substantiated in this
analysis; in fact, opening of new facilities may be
linked to modestly decreased drug crime rates,
but further investigation is required.
Funding Source(s): VA
Poster Session and Number: B, #686
Do Health Shocks Improve Disease SelfManagement In Depressed Adults? Impact
Of Emergency Department Visit(s)
Neeraj Bhandari, Penn State; Yunfeng Shi, The
Pennsylvania State University; Larry Hearld,
University of Alabama at Birmingham; Megan
McHugh, Northwestern University, Feinberg
School of Medicine
Presenter: Neeraj Bhandari, Phd Student, Penn
State
nwb5090@psu.edu
Research Objective: The growing prevalence
of depression in the United States continues to
impose a significant burden on an already
strained mental health care delivery system.
Nonadherence to effective treatment (e.g., SSRI
antidepressants) remains a critically important
modifiable factor in reducing disease
prevalence. Emergency departments (ED) are
frequently the de-facto providers of treatment of
depression for many individuals with limited
access to regular treatment in outpatient
settings. Moreover, ED visits are viewed by
many as “teachable moments” that may prompt
changes in health behaviour (i.e., due to a
“health shock” effect), especially when there is a
comprehensive intervention program. Given
these facts, some researchers have proposed
using an ED visit for surveillance, screening, and
interventions designed to improve patient skills
in self-management of mental health conditions.
Understanding the impact of an ED visit on
adherence to depression treatment may help in
evaluating the “teachable” potential of these
settings for targeted interventions to improve
self-management of depression. In this study,
we examine the relationship between ED visit(s)
within the past year and self-reported adherence
to antidepressant medications and attendance of
physician-recommended counseling sessions
Study Design: Two-period panel data (2008-9
and 2011-12) from the Aligning Forces for
Quality Consumer Survey (AF4QCS) that was
administered as a part of the evaluation for
Aligning Forces for Quality (AF4Q), a national
program funded by the Robert Wood Johnson
Foundation (RWJF).Fixed effects logistic
regression was used to examine the relationship
between ED visit(s) and self-reported adherence
to antidepressant medications and attendance of
physician-recommended counseling sessions.
Population Studied: 850 adults with depression
that were surveyed both in 2008 and 2012.
Principal Findings: Between 2008 and 2012,
among those who reported an ED visit(s) in the
past year, the proportion of depressed adults
who were adherent to antidepressant
medications and reported attendance of
counselling sessions rose (69% to 71% and
29% to 37%, respectively). For those who did
not report an ED visit(s), the proportion who
were adherent to antidepressant medications fell
modestly (71% to 68%) and the proportion who
reported attendance of counselling sessions
rose slightly (27% to 30%). Compared to those
who did not have ED visit(s) in the past year, ED
visitors showed a significant increase in the
probability of adherence to medication for ED
visit(s) (26.17 percentage points, p<0.05) but no
significant change in the probability of
attendance of counselling (0.04 percentage
points, p=0.98) sessions.
Conclusions: We find that having an ED visit(s)
was associated with sizable and significant
increase in likelihood of self-reported adherence
to antidepressant medications in depressed
adults but no significant changes in self-reported
attendance of counseling sessions.
Implications for Policy, Delivery, or Practice:
Our results generally support the view that EDbased interventions may be an important
element of an overall strategy to improve
adherence in depressed patients, and such
settings may provide opportune moments for
positive behavioural change in a segment of the
population facing a host of access problems
stemming from capacity and managed care
constraints in mental health care provision.
Funding Source(s): N/A
Poster Session and Number: B, #687
Overcoming Barriers in Mental Health Care
Management Implementation
James Burgess, Boston University, School of
Public Health; Christopher Miller, VA Boston
Healthcare System; Jenniffer Leyson, VA
Boston Healthcare System; Justin Benzer, VA
Boston Healthcare System
Presenter: James Burgess, Professor, Boston
University, School of Public Health
jfburges@bu.edu
Research Objective: The primary objective of
this work was to assess how barriers and
facilitators affect implementation of Care
Management (CM) within a national Primary
Care/Mental Health Integration (PC/MHI)
program. The specific goals of this qualitative
research investigation were to identify (1)
barriers to the implementation of CM, (2) how
those barriers were overcome in successful
implementation, and (3) the degree to which
sites without formal CM programs do provide the
essential services of care management (i.e.,
monitoring and coordination).
Study Design: For sites with CM, interviews
identified how essential services of CM have
been implemented, the types of barriers
overcome during the implementation, and
organizational factors that may have facilitated
this process. For sites that have not yet
implemented CM, interviews identified the
barriers that have deterred implementation.
Interviews with these “no CM” sites also
evaluated whether and to what degree routine
practices already provide CM services.
Implementation was defined by the presence of
elements of CM whether or not the site formally
implemented CM as part of their PC/MHI
program. The conceptual framework for this
study is based on three key research-based
services of CM: (1) promoting patient education
and activation, (2) providing timely access to
mental health services, and (3) monitoring
mental health status over time. The barriers and
facilitators to the implementation of CM services
of patient education, activation, access, and
monitoring were identified with a semi-structured
interview guide aimed at identifying ideal states
and gaps in achieving those ideals. Qualitative
analyses involve three stages. First, memos are
being written to record interesting patterns and
relationships with theory. Second, coding is
being conducted based on the a priori and
emergent concepts. Third, thematic analyses
are drawing on the coded concepts to identify
factors that affect the implementation of CM
services of patient activation, access, and
monitoring.
Population Studied: A national survey was
used to select two Department of Veterans
Affairs Medical Centers (VAMCs) with strong
CM programs and two VAMCs that have not yet
implemented formal CM as part of their PC/MHI
programs. At each of the four sites, six to ten
individuals were recruited for interviews,
including leaders, care managers, mental health
providers co-located in the primary care setting,
psychiatrists on duty, and primary care
physicians and nurses.
Principal Findings: At the participant level of
analysis, leaders, mental health staff, and
primary care staff each have valuable and often
differing perspectives on implementation of CM.
Preliminary analysis suggests that successful
CM implementation requires collaboration
between primary care and mental health staff.
Primary Care providers need to understand the
value of CM to their practices and CM need to
be engaging as full partners in work design and
systems of care.
Conclusions: Evaluating complex
implementations in multi-disciplinary
environments requires careful attention to
provider-experienced organizational boundaries,
in this case between primary care and mental
health.
Implications for Policy, Delivery, or Practice:
The Affordable Care Act increases the
importance of behavioral health, but many
behavioral health interventions such as CM are
multidisciplinary. Complex interventions need to
be designed carefully and integrate perspectives
from all relevant stakeholders.
Funding Source(s): VA
Poster Session and Number: B, #688
Mental Health Treatment Utilization - The
Role of Health Insurance for Adults with
Serious Mental Health Disorders
Mason Burley, Washington State University Health Policy and Administration; Kenn Daratha,
Washington State University - College of
Nursing; Jae Kennedy, Washington State
University - Health Policy and Administration
Presenter: Mason Burley, Research Staff,
Washington State University - Health Policy and
Administration
mburley@wsu.edu
Research Objective: Serious mental illness
(SMI) affects over 11 million adults in the United
States (5 percent of the population). Mental
health disorders (depression, bipolar disorder,
and schizophrenia) account for three of the
leading ten causes of disability in developed
countries. While effective treatment alternatives
are available for individuals with these disorders,
the disease burden of SMI is increased as a
result of the underutilization of mental health
treatment. Only 50% of adults with SMI receive
mental health related treatment in a given year.
The objective of this study is to determine
factors associated with mental health treatment
utilization among adults with SMI. Specifically,
this research focuses on the role of health
insurance coverage on treatment utilization.
Access to insurance is particularly important for
this population given the recent passage of the
Mental Health Parity and Addiction Equity Act
(MHPAE), which took effect in 2010. The
MHPAE requires health insurance plans to
establish identical financial requirements (i.e.
copays) and treatment benefits (i.e. number of
visits) for both mental and primary health care.
Study Design: This observational cohort study
examines treatment utilization as reported by
adult respondents in the 2012 Behavioral Risk
Factor Surveillance System (BRFSS). A
multivariable logistical regression analysis
(accounting for the weighted stratified sample)
controls for a range of prognostic factors related
to treatment including demographics, physical
health condition, treatment attitudes, and
financial status (income, insurance).
Population Studied: In 2012, 16 states
administered the optional Mental Illness and
Stigma module to BRFSS. The un-weighted
study cohort includes 4,130 respondents
(representing 3,209,916 adults) with Serious
Mental Illness as identified by the Kessler-6 (K6) instrument. The K-6 instrument provides a
brief and valid assessment of SMI when
compared to structured clinical interviews using
DSM-IV criteria.
Principal Findings: Only 60% of adults with
SMI had health insurance coverage in the
previous year. After controlling for other
predisposing factors, adults with health
insurance coverage had higher odds of receiving
mental health treatment (OR 3.8, 95% CI 2.79-
5.31) for SMI disorders. In the fully adjusted
model, health insurance coverage represented
the factor with the highest odds associated with
mental health treatment.
Conclusions: After adjusting for differences in
age, race/ethnicity, education, income level and
presence of chronic health conditions, access to
health insurance was found to be highly
predictive of the decision to obtain mental health
treatment among adults with SMI. These
findings point to the importance of establishing
robust mental health benefits as required in the
MHPAE.
Implications for Policy, Delivery, or Practice:
Two implications for future policy and practice
are noted. First, the Affordable Care Act
provided for the expansion of Medicaid and
establishment of health insurance exchanges,
starting in 2014. Many adults previously
ineligible for coverage will now have access to
health insurance. States should utilize
community mental health providers and health
connectors to assist adults with SMI in finding
suitable insurance coverage. Second, we found
that 57% of adults with SMI saw a primary care
provider in the previous year. Collaborative care
arrangements and health homes may be an
effective means for increasing treatment
initiation and retention for adults with SMI
requiring mental health treatment.
Funding Source(s): No Funding
Poster Session and Number: B, #689
The Effects of a Public Health Insurance
Expansion on Behavioral Health Care Use
Marguerite Burns, University of Wisconsin;
Lindsey Leininger, University of Illinois- Chicago;
Laura Dague, Texas A&M University; Thomas
DeLeire, Georegetown University; Kristen
Voskuil, University of Wisconsin; Chris
Reynolds, University of Wisconsin; Donna
Friedsam, University of Wisconsin
Presenter: Marguerite Burns, Assistant
Professor, University of Wisconsin
meburns@wisc.edu
Research Objective: The effects of adult
Medicaid expansions on overall and preventive
health care use are increasingly well understood
as evidence emerges from early-expansion
states. The impact of these expansions on
behavioral health care use, including mental
health and substance use disorder (MHSUD)
services, is less clear. Limited evidence
supports the expectation that Medicaid
programs will provide more MHSUD services
after expanding eligibility to adults. Uncertainty
remains, however, about the magnitude of
increase and its composition across service
categories. This uncertainty contributes to
substantial budgetary and operational
challenges for Medicaid programs’ planning and
implementation activities as reported in a 2013
qualitative study of early-expansion states. This
paper reduces that uncertainty by providing the
first comprehensive assessment of the effect of
an adult Medicaid expansion on MHSUD
services across outpatient, emergency and
inpatient settings.
Study Design: The study population was
automatically enrolled in Wisconsin’s Medicaid
“Core Plan” on January 1, 2009. We exploited
this natural experiment to compare MHSUD care
use after this exogenous change (2009) relative
to prior use (2008) using fixed effects
regression. Data included encounter and claims
records from the Milwaukee County general
assistance medical program (GAMP) for the
uninsured period and the Wisconsin Medicaid
program for the insured period. We defined
MHSUD-related outpatient, emergency
department (ED), and inpatient use by
procedure and diagnostic codes using
established algorithms.
Population Studied: The cohort included 9,619
poor and uninsured childless adults that resided
in Milwaukee County, WI, and received any
health care in 2008 through the county indigent
program (i.e., GAMP).
Principal Findings: The percentage of the
cohort with any MHSUD visit or admission
increased from 30% in the 2008 pre-period to
38% post-Core enrollment. Post-Core, the
number of MHSUD outpatient visits/month
increased by 148% from a baseline rate of
0.045. The number of ED MHSUD visits/month
increased by 84% from 0.034 in 2008. No
significant change occurred in MHSUD inpatient
admissions.
Conclusions: States should anticipate MHSUD
care use among a large minority of poor,
childless adult enrollees. Increased MHSUD ED
use coupled with the increased rate of outpatient
visits raises questions regarding outpatient
access. Enrollment procedures that explicitly
assign a behavioral health provider to new
beneficiaries may facilitate care in the most
appropriate setting.
Implications for Policy, Delivery, or Practice:
The effect of Medicaid expansions on adults’
behavioral health service use will vary according
to the Benchmark plan that each state selects to
define benefits, the population’s health,
enrollment rates, and provider capacity. Thus, it
is critical to amass valid estimates from a variety
of settings that establish a reasonable range of
expected effects for state planning and
implementation purposes. This study provides
such estimates for a relatively poor and urban
population.
Funding Source(s): RWJF
Poster Session and Number: B, #690
Planning for Health Care Reform in an
Integrated Health System: Anticipated Impact
on Addiction Treatment and HIV Care
Cynthia Campbell, Kaiser Permanente; Derek
Satre, Department of Pyschiatry, University of
California, San Francisco; Andrea Altschuler,
Division of Research, Kaiser Permanente; Alison
Truman, Division of Research, Kaiser
Permanente; Sujaya Parthasarathy, Division of
Research, Kaiser Permanente
Presenter: Cynthia Campbell, Research
Scientist II, Kaiser Permanente
cynthia.i.campbell@kp.org
Research Objective: Health care reform has
important implications for patients with
substance use disorders (SUDs), including
those with HIV. This study examines how an
integrated health care delivery system has been
planning for health reform, particularly for
anticipated changes in membership and
services for these patient populations.
Study Design: The study is set in Kaiser
Permanente Northern California (KPNC). It uses
a mixed-methods approach, including qualitative
interviews with 25 clinical and operational
leaders, and quantitative analyses with
electronic health record and cost data. This
analysis presents the initial qualitative findings.
Population Studied: Qualitative interviews
were conducted with 25 clinical and operation
leaders with responsibility for implementing the
ACA, and with responsibility for behavioral
health services.
Principal Findings: Early initial findings
suggest that the delivery system anticipates
ACA-related changes in membership
composition but that many aspects remain
uncertain, including the prevalence of SUDs.
Interview themes include enhancing the ability to
respond quickly to changes in membership
needs including comorbid conditions, improving
“customer service” aspects of care in order to
compete in a rapidly changing marketplace,
modest increases in staffing for HIV and SUD
care; and changes in benefit structure. KPNC
has a range of internal and external strategies to
communicate with members and staff about key
aspects of ACA adaptation.
Conclusions: Qualitative interviews with KPNC
leaders reveal very active planning for ACArelated changes in member enrollment,
membership composition and benefit plans.
There is considerable uncertainty in the effect of
the ACA on membership, coming from the ACA
as well as other environmental sources, and
leaders emphasize the importance of flexibility in
the initial stages of implementation.
Implications for Policy, Delivery, or Practice:
The changes that this integrated delivery system
is in the process of addressing reflect the
uncertainty the nation has faced with the rollout
of health reform. Much is still unknown, and the
next two years will be critical in determining the
impact of health reform on SUD and HIV
populations.
Funding Source(s): NIH
Poster Session and Number: B, #691
Incremental Health Care Expenditures
related to Depression, Anxiety, Stress, and
Substance Abuse among Cancer Survivors
Jie Chen, University of Maryland at College
Park; Mir Ali, The Substance Abuse and Mental
Health Services Administration
Presenter: Jie Chen, Assistant Professor,
University of Maryland at College Park
jichen@umd.edu
Research Objective: There were approximately
13.7 million cancer survivors in the United
States in 2012. This number is expected to
increase to 18 million by 2020. Racial and ethnic
minority survivors continue to face worse health
care access and receive less timely health
treatment compared to the whites. Behavioral
health, such as psychological stress, and drug
abuse, are common comorbidities of cancer.
Unfortunately, racial and ethnic minorities also
have substantial barriers to mental health care
services. The delayed or neglected mental
health care among cancer survivors might
reduce the cost efficiency of cancer treatment
and increased overall health care cost,
especially among the racial and ethnic
minorities. The objective of this study is to
estimate the incremental health care
expenditures related to depression, anxiety,
stress, and substance abuse among cancer
survivors, with a focus on racial and ethnic
disparities.
Study Design: This study used national
representatively data set Medical Expenditure
Panel Survey (MEPS) from 2008-2011, the
linked data sets of Consolidated and Medical
Component Files of MEPS. ICD9 codes were
used to identify patients with cancer, depression,
anxiety, stress, substance abuse, and other up
to 10 chronic diseases, such as diabetes and
heart diseases. Health care costs were defined
as the total health care expenditures, including
the expenditures from the health insurance, outof-pocket payment, and/or government subsidy.
All the costs were adjusted using 2011
Consumer Price Index Medical Care
Component. The associations of behavioral
health disorders with health care expenditures
might be different along the distribution of the
expenditures. Hence, quantile multivariate
regressions were employed to measure the
different incremental health care expenditures
associated with behavioral health disorders.
Population Studied: Our final sample included
8,368 cancer survivors aged 18 and above.
Principal Findings: Having depression, anxiety,
and/or stress could significantly increase health
care expenditures among the cancer survivors,
and the incremental costs were more substantial
at the lower end of the health care expenditures.
Having substance abuse could also increase
health care expenditures, especially at the
higher end of the distribution. Among cancer
survivors, incremental costs related to disorders
of depression, anxiety, stress, and substance
abuse were the highest compared to other
cancer comorbidities, including diabetes, heart
diseases, etc. We also found that African
Americans had significantly higher health care
expenditures at the 75th-95th percentile of the
distribution, even after controlling all the social
economic and demographic characteristics.
Conclusions: This study showed evidence that
behavioral health disorders were the “most
expensive” comorbidities associated with
cancer. Compared to the whites, African
American cancer survivors had lower health
care expenditures at the lower end of the
distribution. However, they encounter
significantly higher health care expenditures
compared to other race and ethnicity, at the
higher end of the health care expenditures.
Implications for Policy, Delivery, or Practice:
Our results indicated the urgency to provide
timely behavioral health services among the
cancer survivors. Depression screening test is
free under the Affordable Care Act. This
screening test should be highly recommended
among the cancer survivors. Racial and ethnic
disparities in health care have been well
identified. Compared to the whites, minorities
usually have worse health care access, lower
health care utilization and expenditures. Our
study showed the disproportionate ratios of
racial disparities along the distribution of the
health care expenditures, which suggested that
the extremely high health care expenditures
among African American cancer survivors might
be caused due to the delaying or forgoing
effective treatment at the cancer prevention
period.
Funding Source(s): No Funding
Poster Session and Number: B, #692
Implementation of Behavioral Health
Interventions in Real World Scenarios: The
Problem of Social Inertia
Deborah Cohen, Oregon Health & Science
University; Larry Green, University of Colorado,
Denver; Bijal Balasubramanian, University of
Texas Health Science Center, Houston; Melinda
Davis, Oregon Health & Science University;
Rose Gunn, Oregon Health & Science
University; Benjamin Miller, University of
Colorado, Denver
Presenter: Deborah Cohen, Associate
Professor, Oregon Health & Science University
cohendj@ohsu.edu
Research Objective: Two decades ago the
Institute of Medicine recommended integrating
behavioral health and primary care, and recent
policy changes are stimulating action. We
examine the concept of social inertia in relation
to integrating care. Social inertia is peoples’
tendency to resist attitude and behavior change.
This study examined how social inertia slowed
integration efforts and describes strategies to
overcome it.
Study Design: A multiple site, comparative
case study using mixed methods, including site
observation and interviews, and care process
and outcomes measures.
Population Studied: Eleven primary care and
community mental health clinics located in
Colorado and participating in Advancing Care
Together (ACT), a demonstration project funded
by The Colorado Health Foundation to test
strategies for implementing evidence-based
integrated care.
Principal Findings: Social inertia at the
individual / interpersonal, organizational and
policy levels interacted to slow integration
efforts. At the organization level, leaders
resisted changing key aspects of organizational
culture to facilitate new models of integrated
care (e.g., policies, procedures, accountability)
and physical space constraints hampered new
professional collaborations. On the individual /
interpersonal, both primary care and behavioral
health clinicians resisted changing routine
behaviors related to intake assessments,
workflows, visit time, and clinical documentation.
Establishing collaborative processes – instances
when professionals worked together to develop
a diagnosis and treatment plan – were difficult to
implement because these new processes
required behavioral and attitudinal changes
among professionals. At the policy level, the
integrated care models being implemented by
ACT clinics were not supported with changes in
reimbursement. Clinic leaders were expressed
apprehension about making the substantial
changes required at individual and
organizational levels to integrate care for fear of
reversal, if not financially sustainable.
ACT clinics sought to overcome the status quo
reinforced by social inertia. We observed
effective strategies for doing this, including: (1)
leadership that set a vision for integration,
engaged practice members in this vision,
removed internal and external barriers to change
(including advocating for payment reform), and
gave on-the-ground staff time to innovate; (2)
training and orientation that fostered shadowing
and modeling of effective integrated care
behaviors, and the ability to openly discuss
critical barriers to collaboration across
professionals; (3) change agents / clinic
members with the authority to model and
encourage new behaviors and provide space for
reflecting and refining current practices; and (4)
ample opportunity and data to stimulate
reflection and assessment of what is working
and what is not.
Conclusions: Social inertia slowed integration
efforts in most clinics. A proactive approach that
includes facilitative leaders and engaging
behavioral health and primary care providers on
the ground is needed to overcome social inertia.
Implications for Policy, Delivery, or Practice:
Policy and practice leaders need to be aware of
the effects of social inertia and take steps to
overcome it in order to ensure that the vital
benefits associated with integrated care (e.g.,
reductions in cost, and improvement in quality
and patient experience) are achieved more
rapidly.
Funding Source(s): Other The Colorado Health
Foundation
Poster Session and Number: B, #693
Population Mental Health Outcomes before,
during, and after the Great Recession
Rada Dagher, University of Maryland; Jie Chen,
University of Maryland, College Park,
Presenter: Rada Dagher, Assistant Professor,
University of Maryland
rdagher1@umd.edu
Research Objective: The literature has shown
that economic recessions tend to increase the
risk of mental disorder. This study examines
population mental health outcomes during and
after the Great Recession compared to prerecession, by gender.
Study Design: We utilized 2005-2006, 20082009, and 2010-2011 data from the Medical
Expenditure Panel Survey, a nationally
representative survey of the U.S. noninstitutionalized civilian population. We
examined five mental health outcomes:
depression diagnosis, anxiety diagnosis, selfreported mental health, SF-12 MCS score, and
the K6 score (Kessler index).
Population Studied: The study weighted
sample included a total of 46,408 females and a
total of 34,905 males, aged 18 to 64 years old.
Principal Findings: Females had lower odds of
being diagnosed with depression during the
recession (OR=0.86) and post-recession
(OR=0.89) but higher odds of being diagnosed
with anxiety post-recession (OR=1.17),
compared to pre-recession. Females had better
self-reported mental health scores during the
recession (ß = 0.04) compared to pre-recession.
Males had lower odds of being diagnosed with
depression during (OR=0.79) and after
(OR=0.87) the recession compared to prerecession. Moreover, males had better selfreported mental health during (ß = 0.05) and
after (ß = 0.04) the recession and lower K6
scores post-recession (ß = -0.20) compared to
pre-recession. Findings from control variables
showed that among both males and females, the
unemployed, the less educated, and those in
lower Federal Poverty Level brackets had
consistently poorer mental health outcomes than
those with better socioeconomic conditions.
Conclusions: In this large nationally
representative study of the US population, we
found that females were more likely to be
diagnosed with anxiety post- recession
compared to pre-recession. Job insecurity is one
possible explanation for the increased anxiety
among females after the recession and future
research should ascertain this possibility.
Conversely, males and females had lower odds
of being diagnosed with depression and better
mental health scores during and after the
recession. One possibility is that during
economic downturns, there is more leisure time
to spend on family, friends, and exercise, which
may decrease the likelihood of depression.
Moreover, these associations could partly be
due to social protection programs in the U.S.
such as unemployment compensation and social
welfare which may have buffered the impact of
the recession on population mental health.
Implications for Policy, Delivery, or Practice:
While policymakers may interpret these results
as generally positive, our findings that the
unemployed, the less educated, and those in
lower income brackets had consistently poorer
mental health outcomes raises questions
regarding the impact of the recession on
disadvantaged groups. Policymakers should
invest in labor market programs that provide
group psychological support for the unemployed
and reintegrate workers in jobs and consider
other social policies such as debt relief
programs.
Funding Source(s): No Funding
Poster Session and Number: B, #694
New Measures of Community Mental Health
and Substance Use Burden and Access to
Care Using ED Utilization Data: The
Behavioral Health Emergency Department
Prevention Quality Indicators (ED-PQI)
Sheryl Davies, Stanford University School of
Medicine; Eric Schmidt, Stanford University
School of Medicine; Jose Maldonado, Stanford
University School of Medicine; Robert
Houchens, Truven Health Analytics; Carol
Stocks, Agency for Healthcare Research and
Quality; Kathryn McDonald, Stanford University
School of Medicine
Presenter: Sheryl Davies, Research Associate,
Stanford University School of Medicine
smdavies@stanford.edu
Research Objective: Mental health and
substance use contribute to a substantial
proportion of emergency department (ED) visits.
Beyond visits for toxicity or acute mental health
events, behavioral health disorders may
contribute to or complicate visits for physical
trauma and medical conditions. Individuals with
behavioral health diagnoses may have more
difficulty accessing health care for acute and
chronic illnesses and providing self-care,
potentially increasing ED use for general health
diagnoses. As such, the emergency department
data offers an unique lens into community
behavioral health. ED data has specific
strengths, including: 1) routinely collected, 2)
available for near real-time analysis, 3) relatively
consistent across the nation and 4) applicable to
various populations, including small areas. Our
objective was to develop multi-use measures of
community behavioral health usual nationally
representative ED datasets.
Study Design: Using structured Delphi
techniques and empirical validation, we
developed five measures of community
behavioral health, including ED visits for
substance use, ED revisits for substance use,
and three measures of ED visits for mental
health disorders (psychoses and bipolar
disorder; depression, anxiety and stress
disorders; revisits in individuals with mental
health conditions). Six additional indicators focus
on general health events, for which individuals
with substance use disorders are at high risk,
including ED visits for chronic disease
exacerbation and frequent ED visits for back
pain. We developed these indicators using the
combined Agency for Healthcare Research and
Quality Healthcare Cost and Utilization Project
(HCUP) State Inpatient Databases and State
Emergency Department Databases for 28
states, 2008-2010.
Population Studied: The HCUP databases
include 80,205,600 all-payer visits (28 states
and 1778 counties). The behavioral health
indicators include individuals 15 years and older.
Principal Findings: The indicators were rated
as useful in three separate review processes: 1)
a multispecialty internal review panel, 2) a
stakeholder panel of national experts and 3) two
Delphi panels. The ED visit rate for substance
use was 1.8 per 1000 population, and 24
percent returned to the ED with substance use
or trauma diagnoses. The ED visit rate for
depression, anxiety or stress reaction was 2.9
per 1000 and for psychoses or bipolar disorders,
1.1 per 1000 population. The all-cause revisit
rate was 39.7 percent for index visits with a
mental health diagnosis. Increasing area level
poverty, controlling for population age and
gender, were predictive of higher ED visit rates
for behavioral health (RR=1.5-1.8). Rates are
highly stable (year-to-year autoregressive
correlation = 0.88-0.99). In factor analysis,
behavioral health indicators were strongly
related to each other, with ED visit rates for
substance use and mental health being closely
related, and ED revisit rates for substance use,
mental health and repeat visits for back pain
being closely related. Behavioral health visit
rates were also moderately correlated with ED
visit rates for acute and chronic ambulatory care
sensitive conditions.
Conclusions: The metrics are reliable
indicators of community behavioral health at the
county level. ED revisit indicators capture a
distinct construct from utilization indicators.
Implications for Policy, Delivery, or Practice:
The metrics can be used to target opportunities
for health improvement, highlight disparities,
evaluate interventions, inform resource
decisions and provide public information on
community health.
Funding Source(s): N/A
Poster Session and Number: B, #695
Experiences with Patient-Centered Care for
Chronic Conditions – Relations with
Depression and Patient Activation
Daniel Fulford, Palo Alto Medical Foundation;
Shana Hughes, Palo Alto Medical Foundation
Research Institute; Laura Panattoni, Palo Alto
Medical Foundation Research Institute; Ming
Tai-Seale
Presenter: Daniel Fulford, Research
Psychologist, Palo Alto Medical Foundation
fulfordd@pamfri.org
Research Objective: Chronic conditions,
including hypertension (HTN) and diabetes
mellitus (DM), affect nearly 3 million Americans
and contribute to high health care costs and
poor quality of life (QoL). Depression is highly
prevalent among these patients and contributes
to poorer outcomes through diminished selfmanagement. Although patients with depression
generally report less favorable experiences with
care, little is known about the association
between experience with patient-centered care
and self-management for chronic conditions. We
examined the relationships among depression,
patient activation, and patient experiences with
primary care. Consistent with previous literature,
we hypothesized that depression would be
associated with diminished activation and
suboptimal experience with care. We also
predicted that experience with care would be
uniquely related to patient activation, above and
beyond the relationship with depression and
mental health related quality of life.
Study Design: Patients completed
questionnaires related to current symptoms of
depression (2-Item Patient Health
Questionnaire; PHQ-2), mental health related
QoL (Veterans Rand 12-Item Health Survey,
Mental Health Component Summary score; VR12 MCS), and self-management/activation
(Patient Activation Measure; PAM). They also
reported on their recent experiences with
primary care providers (Patient Assessment of
Care for Chronic Conditions; PACIC). We
examined the associations among depressive
symptoms, mental health related QoL (MQoL),
patient activation, and experience with care. We
also tested the unique association of experience
with care on patient activation using multiple
regression analysis, controlling for depression
and MQoL.
Population Studied: 526 patients with HTN
and/or DM receiving primary care in family
medicine or internal medicine departments in a
medical group practice in Northern California.
Principal Findings: In general, patients
endorsed moderate levels of depression and low
MQoL—roughly three-quarters of respondents
scored below population mean on the VR-12
MCS, and 12.2% met clinical cutoff for major
depression based on the PHQ-2 (= 3). In the
bivariate analyses, higher depression and lower
MQoL were related to lower patient activation
and poorer patient experiences with care.
Patients meeting clinical cutoff for depression (n
= 43) endorsed significantly lower activation than
those who did not (t = 4.07, p < .001), but did not
differ in experiences with care (t = 1.50, p > .05).
In the regression model, experiences with care
explained significant variance in patient
activation (R2 = .19), above and beyond that
explained by MQoL and depression (Adjusted
R2 = .10; a 1 SD increase in PACIC was
associated with a .32 SD increase in PAM).
Conclusions: Patients with chronic conditions
who reported depressive symptoms and poorer
MQoL were more likely to endorse diminished
activation. Furthermore, poorer patient
experiences with care contributed uniquely to
diminished activation, above and beyond their
association with depression and MQoL.
Implications for Policy, Delivery, or Practice:
Findings of the current study highlight the
potential impact of patients’ depression and
experiences with care for chronic conditions on
activation for self-management. When patients
with chronic conditions and depression present
to primary care, the depression is often not
treated. These findings provide further evidence
for the importance of addressing these
symptoms to improve patient-centered
experiences with care and activation.
Funding Source(s): Other Moore Foundation
Poster Session and Number: B, #696
Anxiety Disorder Diagnosis and Treatment
among Visits by Older Patients in the United
States
Jeffrey Harman, University of Florida
Presenter: Jeffrey Harman, Associate
Professor, University of Florida
jharman@phhp.ufl.edu
Research Objective: Previous research from
the 1990s demonstrated that anxiety disorders,
although highly prevalent, were often not
identified and treated among older patients in
the U.S. New estimates are needed to assess
current rates of identification and treatment of
anxiety disorders in this population to assess
whether the health care needs of this population
are being better met. The purpose of this study
is to assess rates of anxiety disorder diagnosis
during office visits made by patients ages 60+,
the type of physician seen, and rates of anxiety
treatment with medication and/or counseling and
psychotherapy.
Study Design: Data from the 2007 through
2010 National Ambulatory Medical Care Surveys
(NAMCS), a nationally representative survey of
physician office visits conducted annually by the
National Center for Health Statistics, are used.
The study uses a cross-sectional approach to
estimate rates of diagnosis of and treatment for
anxiety disorders during physician office visits.
All analyses use the survey procedures of Stata
to allow results to be nationally representative
and to produce confidence intervals that
correctly account for the complex sampling
strategy of the NAMCS.
Population Studied: Rates of anxiety disorder
diagnosis are assessed for all visits by patients
ages 60+ (N=21,832), while rates of anxiety
disorder treatment are assessed for all visits by
patients ages 60+ with a recorded anxiety
disorder diagnosis (N=295).
Principal Findings: Anxiety disorders were
diagnosed during only 1.2% of visits (95%CI:
1.0-1.4) made by older patients, with the
majority of these visits (56.0%; 95%CI: 46.8-
65.1) made to primary care providers. The next
most common types of provider seen were
psychiatrists and neurologists, which accounted
for 30.8% of visits (95%CI: 22.2-39.5). Some
form of anxiety treatment was offered during
78.9% of visits (95%CI: 71.3-86.6), although
anxiety treatment was less likely to be offered
during visits to primary care providers compared
to visits to psychiatrists or neurologists (73.1%
vs. 93.2%). Treatment with medications was the
most common form of treatment (71%; 95%CI:
63.1-79.1) for all types of physicians, with
benzodiazepines being the most common
pharmacologic treatment (48.6%, 95%CI: 39.857.3), followed by antidepressants (37.5%;
95%CI: 30.1-45.0). Behavioral treatment
(psychotherapy or mental health counseling)
was offered during 31.3% of all anxiety visits
(95%CI: 23.0-39.5).
Conclusions: Anxiety disorders remain underidentified during visits by older patients. When
anxiety disorders are diagnosed, treatment is
usually provided, although treatment is likely to
be with benzodiazepines, which increases risk of
dementia and falls among older patients.
Implications for Policy, Delivery, or Practice:
Interventions intended to increase awareness of
anxiety disorders and the importance of
treatment, and anxiety treatment options are
needed, and should be targeted to both primary
care physicians and older patients.
Funding Source(s): No Funding
Poster Session and Number: B, #697
Medicaid Utilization and Costs for Behavioral
Health Care Among Youth in Foster Care
Anika Hines, Truven Health Analytics; Tami
Mark, Truven Health Analytics; Dan Whalen,
Truven Health Analytics; Lauren Hughey,
Truven Health Analytics; Suzanne Fields,
Substance Abuse and Mental Health Services
Administration (SAMHSA)
Presenter: Anika Hines, Research Leader,
Truven Health Analytics
anika.hines@truvenhealth.com
Research Objective: Medicaid is a major
financer of healthcare for youth in foster care in
accordance with Title IV-E of the Social Security
Act. The purpose of this study was to provide
an overview of service use and costs for
physical health and behavioral health care
among youth who were enrolled in Medicaid
through the foster care system.
Study Design: We used data from the 2008
Medicaid Analytic Extract (MAX) file to tabulate
healthcare utilization and costs overall and for
behavioral health services among youth in foster
care. We include data from 49 States and the
District of Columbia. The state of Maine was
excluded due to missing data files. We
identified claims for overarching, non-capitated,
or fee-for-service, plans for inpatient, outpatient,
residential care, long-term care, durable medical
equipment, and prescription drug services. We
tabulated the number of claims, the total
payment for claims, and the number of unique
users represented in the claims for each service
type and for selected procedure codes. For
prescription drugs, we identified specific
medications that were used primarily to treat
mental and substance use disorders based on
their therapeutic classes. Claims, payments,
and unique users for prescription drugs were
tabulated similarly to other services.
Population Studied: Individuals enrolled in
Medicaid through foster care as indicated by the
MAX uniform eligibility code
Principal Findings: In 2008, Medicaid paid
approximately $5.8 billion dollars for the
healthcare of nearly one million children eligible
for Medicaid through the foster care system.
Approximately half of total Medicaid costs for
this population, or $2.6 billion dollars, were for
behavioral health services, including inpatient-,
outpatient-, residential-, and long term care as
well as prescription drugs. Approximately 39
percent of youth in foster care who were
covered by Medicaid had a behavioral health
diagnosis or used a psychiatric medication.
Among youth in foster care using behavioral
health care services, 84 percent used outpatient
behavioral health care and 57 percent used
psychiatric prescription medications. Although
only 7.3 percent of youth in foster care used
inpatient care, residential care, or long term
care, such as inpatient psychiatric facilities, for
behavioral health conditions, these services
contributed 25 percent of behavioral healthcare
costs. Medications for Attention Deficit
Hyperactivity Disorder (ADHD) were the most
commonly used behavioral health drugs within
this population.
Conclusions: Medicaid spent nearly $6 billion
on healthcare costs for foster care children in
2008; approximately half of these costs were for
behavioral health care.
Implications for Policy, Delivery, or Practice:
Behavioral healthcare is a critical part of the
overall healthcare of youth in foster care. More
research is needed on the specific types of
behavioral health services that children are
receiving under Medicaid to design effective and
efficient systems of delivery to this population.
Funding Source(s): Other Substance Abuse
and Mental Health Services Administration
(SAMHSA)
Poster Session and Number: B, #698
Projecting the Number of Active Duty Service
Members Returning from Deployment Who
Will Continue to Use Behavioral Health
Services in the Military Health System
Keith Hofmann, Kennell and Associates, Inc.;
Dave Kennell, Kennell and Associates, Inc.;
Marty Cohen, Kennell and Associates, Inc.; Geof
Hileman, Kennell and Associates, Inc.; Ron
Henke, TRICARE Management Activity
Presenter: Keith Hofmann, Data Analyst,
Kennell and Associates, Inc.
khofmann@kennellinc.com
Research Objective: To estimate the future
number of behavioral health (BH) users among
Active Duty Service Members (ADSMs)
returning from deployment. Estimating the future
demand among this population will help drive
budgets for BH services in the Military Health
System (MHS).
Study Design: ADSMs who deployed were
grouped into cohorts based on the fiscal year
they returned from their first deployment, and
were further stratified by whether they had one
or multiple deployments. Based on observed
trends in MHS eligibility, estimates of how many
members of each cohort will remain on active
duty over time were projected. ADSMs that had
at least one hospitalization or ambulatory
encounter within the MHS with an ICD-9-CM
diagnosis code indicating behavioral health are
defined as BH users. Future BH users per
eligible were projected for each cohort based on
historical trends in utilization. By combining
utilization trends with the eligibility projections,
an estimate of the number of future BH users
was derived.
Population Studied: Any ADSM who deployed
as part of a contingency operation from October
2002 to September 2012.
Principal Findings: Although the number of BH
users among ADSMs returning from
deployments has increased significantly since
2001, the number of users is beginning to
decline as the wars in Iraq and Afghanistan
come to a close. A number of factors are
responsible for the reversal of this trend. First,
as deployed troop levels are drawn down, fewer
ADSMs are deployed for the first time, so the
deployer population grows more slowly. Second,
some ADSMs that have already deployed leave
active duty and/or the MHS health system
entirely. Third, those ADSMs that remain MHS
eligible may stop using BH services. However,
several competing factors temper this decline in
utilization. First, the ADSMs that tend to remain
eligible in the MHS longer are also more likely to
utilize BH care while they are eligible.
Additionally, ADSMs who have multiple
deployments are both more likely to remain
eligible in the MHS and to use BH care than
those who have only deployed once. As a result
of all of these factors, the projected number of
BH users will slowly decline, and by 2020 the
projected estimates reach a level of about twothirds the peak level of users in 2012.
Conclusions: The demand for BH care among
ADSMs is greatly affected by the level of
deployments among that population. As
deployments slow down due to the end of the
wars in Iraq and Afghanistan, the number of BH
users among deployers slows down as well.
However, it will be many years until the demand
for BH care among this population completely
subsides.
Implications for Policy, Delivery, or Practice:
When considering funding for BH care for
ADSMs in the MHS, the conclusions of the wars
in Iraq and Afghanistan must be taken into
account. Our study projects that the demand will
decrease as the deployments decrease, but also
that a baseline amount of funding will be
necessary for those previously deployed ADSMs
who remain eligible in the MHS.
Funding Source(s): Other TRICARE
Management Activity (TMA)
Poster Session and Number: B, #699
Continuity of Outpatient Care among
Individuals with Newly Diagnosed
Schizophrenia under the NHI Program in
Taiwan: A Population-Based Cohort Study
Nicole Huang, National Yang-Ming University;
Department of Education and Research, Taipei
City Hospital; Hsin-Hui Huang, Institute of Public
Health, School of Medicine & Department of
Public Health, National Yang-Ming University;
Chuan-Yu Chen, Institute of Public Health,
School of Medicine & Department of Public
Health, National Yang-Ming University; YiingJenq Chou, Institute of Public Health, School of
Medicine & Department of Public Health,
National Yang-Ming University
Presenter: Nicole Huang, Professor, National
Yang-Ming University; Department of Education
and Research, Taipei City Hospital
syhuang@ym.edu.tw
Research Objective: Continuity of care has
been identified as an important service Principal
and performance measure of mental health
services. People with schizophrenia have
multiple and complex health care needs, and
continuity of care is considered essential to
assure positive outcomes for people with
schizophrenia. Despite the accumulating
literature on treatment adherence, transition of
care, and termination of contact, few studies
have assessed relationship continuity of care
among mentally ill patients in an outpatient
setting in Asia and the role of providers has
been rarely investigated. This study aimed to
assess relationship continuity of outpatient
mental care among individuals with newly
diagnosed schizophrenia under the National
Health Insurance (NHI) program in Taiwan, and
to determine patient and provider factors
associated with worse continuity of care.
Study Design: We conducted a retrospective
cohort study using the Longitudinal Health
Insurance Database released in 2000
(LHID2000) from the Taiwan. The LHID 2000
contains the longitudinal enrollment and
utilization data of 1,000,000 beneficiaries
randomly selected from all NHI beneficiaries.
Two continuity assessment indicators were
derived: the usual provider continuity (UPC)
index and the continuity of care index (COCI).
Both individual (demographics, socioeconomic
status, living arrangement, and health status)
and provider characteristics (demographics,
specialty, and characteristics of practice
locations) were analyzed. The generalized
estimating equations (GEE) statistical model
was used (first level: individuals; second level:
providers).
Population Studied: From the randomly
selected cohort, individuals with newly
diagnosed schizophrenia from 2000 to 2009
were identified. As the continuity of care index
becomes more reliable and robust as more visits
are included, we included only individuals with
schizophrenia who had been continuously
enrolled in the NHI program, and with three or
more mental health outpatient visits within one
year since their first diagnosis of schizophrenia.
Each individual was followed for one year since
receiving the first diagnosis of schizophrenia to
observe continuity of mental health outpatient
care.
Principal Findings: The final sample was
composed of 2,040 individuals with newly
diagnosed schizophrenia. The average scores of
the UPC and COCI were 0.80 and 0.69,
respectively, and were slightly higher than
continuity of care for common physical chronic
conditions in Taiwan. Other than those who
were unemployed (P=0.006) or those having
been hospitalized (P<0.001), individuals with
schizophrenia cared for by younger physicians,
non-psychiatrists (P<0.001), and treated at
mental health specialty institutions (P=0.018)
had significantly poorer usual provider
continuity. The results remained consistent for
the continuity of care index.
Conclusions: Our study is the first populationbased study to assess relationship continuity of
outpatient care among individuals with
schizophrenia in Asia. More importantly, in
addition to patient characteristics, we go further
by exploring the role of providers in continuity of
care among this vulnerable subpopulation.
Continuity of care for schizophrenia in Taiwan is
satisfactory, but significant variations observed
across some patient and provider groups
suggest that room for improvement exists.
Implications for Policy, Delivery, or Practice:
Our findings may help to advance the
understanding of mental care delivery in
outpatient settings and serve as an important
reference for policy makers in devising effective
health policies to improve continuity of care
among individuals with severe mental illnesses.
Funding Source(s): No Funding
Poster Session and Number: B, #700
Cost-sharing Exemption Program for
Mentally Ill Patients: Who Enrolls?
Nicole Huang, National Yang Ming University;
Department of Education and Research, Taipei
City Hospital; Hsin-Hui Huang, Institute of Public
Health, School of Medicine & Department of
Public Health, National Yang-Ming University;
Chuan-Yu Chen, Institute of Public Health,
School of Medicine & Department of Public
Health, National Yang-Ming University; YiingJenq Chou, Institute of Public Health, School of
Medicine & Department of Public Health,
National Yang-Ming University
Presenter: Nicole Huang, Professor, National
Yang Ming University; Department of Education
and Research, Taipei City Hospital
syhuang@ym.edu.tw
Research Objective: Mental health parity has
been a major challenge in numerous countries,
even in countries where universal insurance
coverage is provided. Cost-sharing obligations
particularly hinder access to care among people
with serious mental illnesses such as
schizophrenia. Therefore, governments
commonly institute welfare programs to exempt
this vulnerable subpopulation from cost-sharing
obligations to address the problem.
Paradoxically, a major concern for welfare
programs such as the disease-specific costsharing exemption program is that not every
eligible person enrolls. The enrollment process
requires efforts from both individuals and health
care providers. Unfortunately, the current
understanding of the association of providers
with enrollment is limited. This study aimed to
identify patient and provider characteristics
associated with the cost-sharing exemption
program enrollment among people newly
diagnosed with schizophrenia under the National
Health Insurance program in Taiwan.
Study Design: A retrospective cohort study was
conducted using the Longitudinal Health
Insurance Database 2000 (LHID2000) in
Taiwan. Under the NHI program, a person with
any of the major diseases or injuries listed can
apply for the cost-sharing exemption program.
Schizophrenia is on the list. Enrollees with
multiple diseases must apply separately for each
disease. Individuals can be exempted from the
cost-sharing requirement only for health services
associated with the diseases for which they
applied. Both individual (demographics,
socioeconomic status, living arrangement, and
health status) and provider characteristics
(demographics, specialty, and characteristics of
practice locations) were analyzed. The
generalized estimating equations (GEE)
statistical model was used.
Population Studied: The study population
comprised people who were newly diagnosed
with schizophrenia from January 1, 2000 to
December 12, 2007. Each people newly
diagnosed with schizophrenia was followed-up
for 1 year and 3 years from their first diagnosis
of schizophrenia to observe whether people with
schizophrenia enrolled in the cost-sharing
exemption program.
Principal Findings: The 1-year and 3-year
program enrollment rates were 52.4% and
58.1%, respectively. The five year trend shows
that the enrollment rate reached the plateau by
the fourth year since diagnosis (~60%). People
aged 35 years or older were 15%–24%
significantly more likely to enroll. Low-income
people and those who were hospitalized for
schizophrenia were significantly more likely to
enroll. Regarding provider characteristics,
patients cared for by psychiatrists (adjusted
odds ratio, AOR: 1.10, 95% confidence interval,
CI: 1.02–1.19) or by mental specialty institutions
(AOR: 1.10, 95% CI: 1.04–1.16) were
significantly more likely to enroll in the costsharing exemption program within the first year
of diagnosis.
Conclusions: This study demonstrates that
although people exhibiting a low SES and high
health needs enrolled, the overall enrollment
rate of the disease-specific cost sharing
exemption program by people newly diagnosed
with schizophrenia was unsatisfactorily low. The
trivial progress in the enrollment rate over time
may suggest that the reasons behind the low
enrollment rate persist over time. Future
research may help in this regard.
Implications for Policy, Delivery, or Practice:
The findings serve as important empirical
references to policy makers in addressing
mental health parity using welfare subsidies as a
major tool. The role of providers must not be
overlooked in increasing enrollment in diseasespecific welfare programs.
Funding Source(s): No Funding
Poster Session and Number: B, #701
The Use of the HIV Test: A Conflict Choice
Approach
Amir Khaliq, University of Oklahoma Health
Sciences Center; Robert Broyles, University of
Oklahoma Health Sciences Center, College of
Public Health; Ari Mwachofi, East Carolina
University, Brody School of Medicine,
Department of Public Health
Presenter: Amir Khaliq, Associate Professor,
University of Oklahoma Health Sciences Center
amir-khaliq@ouhsc.edu
Research Objective: The study introduces the
“Conflict-Choice model”(C-C) as an analytic
framework for studying consumer demand for
health and healthcare. The proposed approach
integrates the Theory of Consumer Behavior
(TCB), the Investment Theory of Demand (ITD),
and the Health Belief Model (HBM) into a single
model that might be applied to a wide spectrum
of health behavior and use of health services.
Study Design: Separating an episode of care
into two phases (patient initiated and physician
dominated), the proposed Conflic-Choice model
focuses on the first phase and is limited to the
individual’s decision to seek service. This phase
is dominated by two conflicting and undesirable
outcomes that the patient seeks to avoid. The
first is discomfort or dis-utility that accompanies
the use of care. The second is the discomfort of
illness and a reduced ability to perform social
and economic roles, an outcome that may result
in a potential decline in income. In this conflictchoice situation, the interrelation between two
undesirable conditions and related avoidance
gradients result in a behavioral equilibrium. The
study applied this framework to the use or nonuse of HIV tests.
The theoretical framework of the conflict-choice
model was applied to the data derived from the
responses of 196,081 individuals who
participated in the Behavioral Risk Factor
Surveillance System (BRFSS) survey in 2003.
BRFSS is a well-known population survey
conducted annually on risk factors and health
conditions in the general population in the U. S.
Data on variables that may influence the use of
HIV test were available in the 2003 survey but
unavailable in more recent years. To test the
validity of the conflict-choice model that
integrates three separate theoretical
perspectives into a single modality, a logistic
regression analysis was carried out using HIV
tests as the dependent variable. As such, the
use or non-use of HIV tests was treated as a
function of the integrated conflict-choice model.
Population Studied: 196,081 participating
individuals in the Behavioral Risk Factor
Surveillance System (BRFSS) of 2003. BRFSS
is a well-known population survey conducted
annually on risk factors and health conditions in
the general population in the U. S.
Principal Findings: The results of logistic
regression analysis correctly identified 64.2
percent of the respondents who used HIV test
with all of the coefficients being statistically
significant at P = .01. The analysis uniformly and
accurately predicted influence of barriers to
care, reported susceptibility, and the perceived
benefits of early detection on the use of the HIV
test. The respondents who reported that it was
important to know their health status and those
who consulted with a provider of care regarding
preventive measures were among the most
likely to use the HIV test.
Conclusions: The analyses supported the
expectations based on the newly developed
conflict-choice theoretical framework and
support the adoption of policies that reduce the
tendency to avoid care while increasing the
avoidance of undesirable health outcomes.
Implications for Policy, Delivery, or Practice:
Funding Source(s): N/A
Poster Session and Number: B, #702
External Effects of a State Psychiatric
Hospital Waitlist Policy on Emergency
Department Utilization
Elizabeth La, RTI Health Solutions; Joseph
Morrissey, University of North Carolina at
Chapel Hill; Marisa Domino, University of North
Carolina at Chapel Hill; Kristen Hassmiller Lich,
University of North Carolina at Chapel Hill; Anna
Waller, University of North Carolina at Chapel
Hill; Julie Seibert, Truven Health Analytics
Presenter: Elizabeth La, Senior Research
Health Economist, RTI Health Solutions
eholdsw@live.unc.edu
Research Objective: Nationally, people in
psychiatric crisis can experience long delays
before being admitted to a state psychiatric
hospital. These delays are driven by state
hospital bed shortages, which have prompted
many states to implement waitlist policies in an
effort to avoid operating overcrowded treatment
units. The objective of this study was to
estimate the external effects of these waitlist
policies on the frequency and length of stay of
general hospital emergency department (ED)
visits.
Study Design: North Carolina Medicaid data
(2004-2009) were used to evaluate post-waitlist
changes in overall and behavioral health-related
ED utilization by people with severe mental
illness (SMI). Descriptive and time trend
analyses were used to compare pre- and postwaitlist differences in the proportion of people
with any ED visits in each month, the mean
number of ED visits in each month, and ED
length of stay. These outcomes were also
examined using a difference-in-difference
approach, testing the hypothesis that waitlists
were associated with more frequent and longer
ED visits as people experiencing psychiatric
crises were forced to wait in communities for
admission to state hospitals with nowhere else
to go.
Population Studied: The study’s sample
included North Carolina Medicaid enrollees aged
18-64 years with a diagnosis of SMI
(n=160,143). Elderly enrollees aged 65-74
years who were not in a skilled nursing facility
were used as a control group in difference-indifference analyses.
Principal Findings: For adults with SMI on
Medicaid, the unadjusted percent of personmonth observations with any behavioral healthrelated ED visits increased from 5.2% prewaitlist implementation to 6.2% post-waitlist
(18.6% relative increase, p<0.001). Results
from difference-in-difference analyses indicated
that state hospital waitlists were associated with
small increases in the frequency and length of
stay of ED visits by Medicaid enrollees with SMI.
Specifically, waitlists were associated with a
0.9% increase in the probability of having any
ED visits in a given month (overall or behavioral
health-related, both at p<0.001), as well as 0.12
day and 0.27 day increases in ED length of stay
for overall and behavioral health-related ED
visits, respectively (both at p<0.001). Postwaitlist increases in the number of ED visits in a
given month were not clinically meaningful.
Conclusions: Findings from the current study
provide preliminary evidence that North Carolina
EDs were not overburdened by Medicaid
enrollees with SMI during the post-waitlist era.
Implications for Policy, Delivery, or Practice:
Despite limited effects of the waitlist policy on
ED utilization, additional research is needed to
ensure that people previously served in state
hospitals are not increasingly ending up in other
sub-optimal locations, such as jails and prisons.
Further research is also needed to determine
whether effects of the waitlist vary by state
psychiatric hospital region, as well as whether
results from the current study, which are specific
to Medicaid enrollees with SMI, extend to people
who are uninsured or diagnosed with other
mental health and substance abuse disorders.
Funding Source(s): AHRQ
Poster Session and Number: B, #703
Medicaid Coverage and Financing of
Medications to Treat Substance Use
Disorders
Tami Mark, Truven Health Analytics; John
Richardson, University of Michigan; Hollis Lin,
Truven Health Analytics; Mady Chalk, Treatment
Research Institute
Presenter: Tami Mark, Vice President,
Behavioral Health and Quality Research, Truven
Health Analytics
tami.mark@truvenhealth.com
Research Objective: Medicaid programs are an
important source of coverage for individuals with
substance use disorders and are one of the
largest payers of medications for treating
substance use disorders. This study reviews the
evidence regarding coverage of medications for
the treatment of substance use disorders
(specifically alcohol and opioid use disorders).
Study Design: We reviewed benefit information
regarding coverage of medications to treat
alcohol and opioid use disorders in all 51
Medicaid programs. We searched for
information on Medicaid programs that were
employing innovative substance abuse
treatment delivery and financing models.
Population Studied:
Principal Findings: Only 16 Medicaid programs
include all available medications for the
treatment of alcohol and opioid disorders on
their Preferred Drug Lists (PDLs). Disulfiram,
buprenorphine, buprenorphine-naloxone, and
oral naltrexone were available on the PDL of all
51 Medicaid programs. However, Campral,
Vivitrol, and Methadone were each only
available in approximately 30 state PDLs.
Many Medicaid programs are using benefit
management techniques, such as prior
authorization and quality limits. Buprenorphinenaloxone most commonly required prior
authorization (48 out of 51 Medicaid programs
require prior authorization). Among all Medicaid
programs, 11 have a lifetime limit on the use of
buprenorphine-naloxone. Lifetime limits are
rarely used for any other type of schedule III
medication.
Three examples of Medicaid programs that are
using innovative financing and delivery
approaches are Massachusetts, Vermont, and
Maryland. Common aspects of these models
include leveraging non-physicians to increase
the number of patients who can be treated and
coordinating care with primary care and other
non-specialty substance abuse providers.
Conclusions: There is significant variation
among Medicaid programs in their coverage of
substance abuse medications and benefit
design techniques to contain costs and
encourage appropriate use of substance abuse
medications.
Implications for Policy, Delivery, or Practice:
Medicaid serves over 60 million adults and
children annually and over 1 in 10 adult
Medicaid beneficiaries have a substance use
disorder. Medicaid pharmaceutical and
therapeutics committees may need more
information about the chronic nature of
substance abuse conditions and the
effectiveness of substance abuse medications.
There may also be opportunities for Medicaid
programs to learn from each other about
innovative financing and delivery approaches.
Funding Source(s): Other Substance Abuse
and Mental Health Services Administration
Poster Session and Number: B, #704
Reporting Male Military Sexual Assault:
Challenges as Viewed by Male Enlisted and
Officer Personnel in the Reserves and
National Guard
Michael McClain, Iowa City VAMC; Michelle
Mengeling, Iowa City VA; Brenda Booth,
Department of Psychiatry, University of Arkansa;
James Torner, College of Public Health,
University of Iowa
Presenter: Michael McClain, Couple & Family
Therapist, Iowa City VAMC
michael.mcclain@va.gov
Research Objective: To better understand the
lack of reporting by male victims of military
sexual assault (MST) from an ecological
perspective and the degree of endorsement
about male rape myths by men from the unique
culture of the Reserves and National Guard.
Study Design: Seven focus groups were held in
Iowa, Nebraska, Illinois, Missouri and Kanas.
with informants identified through the Defense
Manpower Data Center listing. A coding team
transcribed and analyzed transcripts utilizing
NVivo 8.0.
Population Studied: Twenty-nine informants
that included both enlisted and officer personnel
from the Army and Air Force Reserve and
National Guard serving during
OEF/OIFparticipated.
Principal Findings: Despite the efforts of the
military and the Sexual Assault Prevention and
Response Office, informants had little
knowledge of reporting options for male victims
of MST. Participants had little to no knowledge
that males are sexually victimized in the military
and difficulty accepting men were assaulted and
that services apply to male victims (“The fact
that I don’t know about this stuff, men bein’
raped, means 9 times out of 10, if I go back to
my social group today at the-at the end of our
day meeting today, none of us know about men
being raped. That’s stuff that happens in prison.
Not our military.” “The army’s had a program in
place where you had to go through training on,
ya know, EEO and sexual and this-and-that.
That is geared towards really ma-male-female
relations… I mean, it’s geared-it’s geared
towards the man’s always gonna be assaulting
the woman… not the man’s assaulting another
man”). When given information on restricted and
non-restricted reporting, most informants were
highly skeptical of the confidentiality and
effectiveness of reporting protections (“We have
a commander, we-we use that chain, I don’t –
but I don’t think it’s necessarily put out that
they’re—what you’re sayin’ is protected, ya
know?” “In order to prosecute somebody in U…
UCMJ, you must have evidence. And the
evidence is a complaint, and the best complaint
is from the individual. And that’s the only way
you’re gonna get that, and that’s why people
won’t come forward, unless there’s going…
unless they know, or, unless they’ve been
pushed”). Officers were somewhat more
confident investigations would be conducted
diligently and procedurally rigorous than were
enlisted personnel (“I suppose uh, you know if
you’re truly following your-your leadership, the
requirements and-and uh… you should take it
no differently. Um… you should investigate”).
The culture of the Reserve and National Guard
are significant challenges to feeling confident in
confidentiality (Administratively, it would be
similar… but sure it would be different because
you’re… in Active Duty, it’s… people from all
over the United States, you get reassigned all
the time, and - whereas in a reserve unit you’re
kind of stuck to, you know, you’re from this
region, and if you’re gonna stay in the Guard or
Reserve, everybody—everybody knows you.
And there’s no way you’re really gonna… be
anonymous anymore”). The unique culture of
the Reserves and National Guard, coupled with
the endorsement of many male rape myths (“In
the military, we have this image of our military,
that a male can handle himself in any situation.
Alright? And to tell me a man got raped – first
thing that’s gonna go through my mind is, first of
all, how did you allow someone to rape you
without killin ‘em?”) indicated powerful barriers
to male victims likely stepping forward to report
assaults from the perspectives of informants.
Conclusions: In spite of great efforts to be
inclusive of male victims of MST, male members
of the Reserve and National Guard seem to little
knowledge of options available and little
confidence in the confidentiality of the process
because of the nature of these types of units. In
addition, acceptance of male rape myths
provides a powerful disincentive to the victim to
report assaults.
Implications for Policy, Delivery, or Practice:
Improved awareness efforts of male MST
occurrences, male rape myths, and reporting
options are important to disseminate. The
confrontation to challenge male rape myths is
sufficiently lacking such that beliefs are for
support of the victim rather than blaming the
victim. It is important for clinicians to understand
the cultural experience of male victims to
provide sensitive care as it is unlikely victims will
report assaults under the current mores of the
Reserve and National Guard cultures.
Funding Source(s): VA
Poster Session and Number: B, #705
Managing and Improving Behavioral Health
Care Quality in Private Health Plans
Elizabeth Merrick, Brandeis University;
Constance Horgan, Brandeis University;
Deborah Garnick, Brandeis University; Sharon
Reif, Brandeis University; Maureen Stewart,
Brandeis University; Candi Ramos, Brandeis
University; Dominic Hodgkin, Brandeis
University; Amity Quinn, Brandeis University
Presenter: Elizabeth Merrick, Senior Scientist,
Brandeis University
merrick@brandeis.edu
Research Objective: Health plans play a key
role in promoting behavioral health care quality.
The study aim was to determine national
estimates of private health plans’ activities
related to behavioral health care quality,
including incentives and recognition programs.
Study Design: This is a cross-sectional analysis
of survey data. We conducted product-level
univariate and bivariate analyses (n=925
products) with weighted data.
Population Studied: Data are from the 2010
Brandeis Health Plan Survey on Alcohol, Drug
and Mental Health Services, a nationally
representative survey of private health plans
regarding behavioral health care. We sampled
private health plans in 60 market areas to obtain
data on each market-specific plan’s top 3
commercial products, reporting here on 385
plans (88% response) completing the quality
improvement module.
Principal Findings: Over 96% of health plan
products conducted patient surveys regarding
behavioral health services. Most reported
aggregated results to providers and purchasers.
Tracking the National Committee on Quality
Assurance’s HEDIS behavioral health measures
was also nearly universal (97%), with results
commonly reported to providers, purchasers and
enrollees. About one-third had provider
incentives or recognition programs tied to
performance measures. About one-third of
products required providers to use standardized
instruments to assess clinical outcomes for
behavioral health patients with 72% of those
reporting results to individual providers and 65%
reporting aggregated results to both enrollees
and purchasers.
Conclusions: Nearly all health plan products
conduct patient surveys and track performance
measures relating to behavioral health. A
minority require providers to do standardized
outcomes assessment. Most do not use
provider incentives for these behavioral health
quality activities.
Implications for Policy, Delivery, or Practice:
Results provide an important picture of private
health plans’ activities regarding behavioral
health care quality. Health plans’ involvement in
quality-related activities in behavioral health can
vary by specific type of activity. Furthermore,
many plans have begun using incentives and
recognition programs in keeping with general
medical care trends, but there is room to grow in
use of these approaches. Plans are clearly
active in the quality area and some are using
financial and non-financial methods to impact
service quality.
Funding Source(s): N/A National Insitute on
Drug Abuse; National Institute on Alcohol Abuse
and Alcoholism
Poster Session and Number: B, #706
Patterns of Initiation of Second Generation
Antipsychotics for Bipolar Disorder
Christopher Miller, VA Boston Healthcare
System; Mingfei Li, Edith Nourse Rogers
Memorial VA Medical Center; Rob Penfold,
Group Health; Austin Lee, Edith Nourse Rogers
Memorial VA Medical Center; Eric Smith, Edith
Nourse Rogers Memorial VA Medical Center;
David Osser, VA Boston Healthcare System;
Laura Bajor, VA Boston Healthcare System;
Mark Bauer, VA Boston Healthcare System
Presenter: Christopher Miller, Investigator, VA
Boston Healthcare System
Christopher.Miller8@va.gov
Research Objective: Second generation
antipsychotics (SGAs) received FDA approval
for bipolar disorder in the 2000s. Although they
have demonstrated efficacy, they are also costly
and may cause significant side effects. This
paper aims to explore the factors associated
with SGA prescriptions within the Veterans
Health Administration, as well as clinical
outcomes that follow in the year after treatment
initiation. Results will inform our understanding
of clinical decision-making for this disorder and
provide insights into the mechanisms of
adoption of medical innovations more broadly.
Study Design: Using a large national database,
generalized estimating equations identified
demographic, clinical, and comorbidity variables
associated with initiation of an SGA prescription
on a month-by-month basis. Separate analyses
used multinomial logistic regression to
investigate factors associated with individual
SGAs.
Population Studied: We gathered
administrative data from the Department of
Veterans Affairs on 170,713 patients with bipolar
disorder between fiscal years 2003-2010,
treated by 47,293 unique providers. Of these
patients, 113,510 qualified as potential SGA
initiators.
Principal Findings: The number of patients
with bipolar disorder using SGAs nearly doubled
between 2003 and 2010 (31,779 in 2003; 61,697
in 2010; p<.0001 for annual increase).
Correspondingly, SGA use grew (compared to
growth of the bipolar population) at a rate of
about 7% per year, although this growth
disappeared when controlling for other patientlevel demographic and clinical factors. Providers
were more likely to initiate SGA treatment for
patients with prior psychiatric hospitalizations,
psychotic features, and a sleep disorder
diagnosis (odds ratios between 1.5 and 2.6; all
p<.0001). Over time, however, SGA initiation
became more common for patients with milder
symptoms. Providers located in the southern
region of the country were more likely to initiate
SGA treatment than those in other regions (odds
ratio > 1.2, p<.0001). Most medical
comorbidities were only modestly associated
with SGA initiation as a whole, although
significant findings emerged for individual SGAs.
Conclusions: The number of veterans with
bipolar disorder receiving SGA prescriptions is
rising due to a steady influx of new initiators into
a growing population. Although illness severity
predicts SGA initiation, SGA use is also
spreading to less severely ill individuals.
Additionally, regional variation is prominent
without clear clinical basis. Thus month-bymonth analyses reveal dynamic patterns of
adoption of SGAs both temporally and spatially.
We hypothesize that these patterns may be
driven by spatiotemporal clustering of early
adopters, and propose a line of research to
assess this potential mechanism.
Implications for Policy, Delivery, or Practice:
Better knowledge of the factors affecting
providers’ decisions to prescribe SGAs can help
in crafting interventions to encourage best
practices in medication prescribing.
Spatiotemporal clustering analyses may also
prove useful in identifying "hot spots" from which
innovative prescribing practices originate and
spread.
Funding Source(s): VA
Poster Session and Number: B, #707
Preferences and Barriers to Care Following
Psychiatric Hospitalization in the Veterans
Health Administration
Paul Pfeiffer, Veterans Health Administration &
University of Michigan; Marcia Valenstein,
Veterans Health Administration & University of
Michigan; Nicholas Bowersox, Veterans Health
Administration & University of Michigan; Dara
Ganoczy, Veterans Health Administration;
Jennifer Burgess, Veterans Health
Administration
Presenter: Paul Pfeiffer, Psychiatrist, Veterans
Health Administration & University of Michigan
ppfeiffe@umich.edu
Research Objective: The period following
psychiatric hospitalization is one of high risk for
suicide and readmission. Improving the quality
of care following psychiatric discharge may
reduce these risks. The purpose of this study
was to assess patients’ preferences for services
and barriers to high quality care posthospitalization.
Study Design: Surveys were mailed to patients
2 to 4 weeks following discharge from the
inpatient psychiatric units of two Midwestern VA
medical centers in 2012-2013. The survey asked
patients to indicate unmet needs for a variety of
mental health-related services and to rate the
perceived helpfulness of potential new services.
Barriers to engaging in psychotherapy were
assessed utilizing items from the Mohr
Perceived Barriers to Psychological Treatment
scale.
Population Studied: A total of 766 patients
were mailed surveys, 288 (38%) were returned;
242 of these patients also granted a waiver to
access their electronic medical records for
demographic and diagnosis data. Participants
were 89% male, 64% White, 14% AfricanAmerican, 21% other or unknown race, and had
a mean age of 51. The most common diagnosis
was a unipolar depressive disorder (31%)
followed by bipolar disorder (21%), substance
use disorder (13%), posttraumatic stress
disorder (12%) and psychotic disorders (11%).
Principal Findings: The most commonly
reported unmet need for services was for
individual counseling (24%), followed by housing
assistance (16%), employment assistance
(16%), group counseling or support group
(14%), family or couples counseling (14%),
medication review (11%), and intensive
outpatient treatment (9%). Participants
substantially preferred in-person counseling
(77%) over telephone counseling (13%) or
counseling via internet video (2%).
Potential new services that participants
perceived would be at least moderately helpful
were increasing the support received from family
or friends (70%) and one-to-one support from
another Veteran (63%), while home visits from a
nurse or mental health clinician (39%), internetbased support groups (22%), and internet selfhelp programs (21%) were less frequently
perceived as potentially helpful.
The most frequently endorsed barriers to
attending counseling or psychotherapy after
hospitalization were problems with
transportation (44%), talking about upsetting
issues (36%), and lack of energy or motivation
(36%). Patients were less likely to report
barriers due to prior negative experiences with
counseling (24%), concerns related to stigma
(24%), or having counseling documented in the
medical record (24%).
Conclusions: Individual psychotherapy was the
most common gap in services identified by
patients who had been recently discharged from
a psychiatric hospitalization. Patients reported
that transportation was an important barrier to
engaging in therapy, yet expressed little interest
in one potential alternative, tele-psychotherapy.
Patients did express considerable interest in
programs to increase involvement and support
from family and friends and Veteran peers posthospitalization.
Implications for Policy, Delivery, or Practice:
Novel methods for reducing travel barriers to
psychotherapy that do not rely on telehealth
technologies should be explored. Further
investigation of the feasibility and efficacy of
family and peer support interventions for
increasing post-hospital treatment engagement
and reducing adverse outcomes is warranted.
Funding Source(s): VA
Poster Session and Number: B, #708
Have Physician Practices Improved
Engagement of Patients in Tobacco
Cessation Efforts? Results from a
Longitudinal Study
Patricia Ramsay, University of California,
Berkeley; Stephen Shortell, University of
California, Berkeley; Lawrence Casalino, Weill
Cornell Medical College
Presenter: Patricia Ramsay, Research
Specialist, University of California, Berkeley
pramsay@berkeley.edu
Research Objective: Based on a wide body of
literature, the 2008 US Dept. of Health and
Human Services Clinical Practice Guidelines
recommend specific interventions and support of
tobacco-dependent patients in clinical practice.
We report unique longitudinal data from three
national surveys conducted between 2006-2013
demonstrating the extent to which physician
practices of all sizes provide such support for
patients who use tobacco, the organizational
factors and external incentives associated with
this support, and how these factors have
changed over time.
Study Design: A numeric score summarizing
support activities for patients who use tobacco
was calculated for all physician practices. The
score included implementation of: 1) a formal
system to identify tobacco users, 2) guidelinebased reminders to physicians at point of care
for tobacco cessation screening, 3) routine
referral to tobacco counseling for a majority of
patients, and 4) feedback to physicians on
quality of care for tobacco users. We examined
the following organizational characteristics to
determine their association with these support
activities: physician practice size; ownership
(physician-owned, hospital or health system
owned, or non-profit/community health center);
practice type (primary care only or multispecialty); and percentage of annual revenues
from Medicaid (>=25% vs. <25%). We also
examined the effect of external financial
incentives (income received specifically as a
result of scores on HEDIS tobacco-related
measures).
Population Studied: Data are included from
three national studies of physician practices: the
National Study of Physician Organizations III,
conducted from 2012-2013 (n=1403, response
rate: 47.6%), the National Study of Small and
Medium-Sized Physician Practices, conducted
from 2007-2009 (n=1745, response rate: 63.2),
and the National Study of Physician
Organizations II, conducted from 2006-2007
(n=538, response rate: 60%). Although specialty
practices were not excluded from these studies,
the current analyses were restricted to practices
with a primary care component (at least 33%
primary care physicians).
Principal Findings: In 2006, physician
practices with more than 20 physicians reported
implementing an average of 1.9 out of the 4
support activities measured (n=274). In 2013,
the mean score for physician groups in this size
range had increased by 37% to2.6 (n=191,
p<0.0001). Increases were observed across all
subgroups, with the largest statistically
significant increases found among solely primary
care practices, practices with 20-99 physicians,
and physician-owned practices. Among
practices with fewer than 20 physicians, the
mean tobacco support index score was 2.2 in
2008 (n=1330) and 2.3 in 2012-13 (n=885).
However, there were statistically significant
increases for multi-specialty groups and
practices with 3-7 physicians.
Conclusions: While larger practices made the
most improvement in their tobacco cessation
efforts, all practices are providing significantly
less than the four recommended tobacco
cessation activities, suggesting the need for all
practices to give greater emphasis to engaging
patients with regard to tobacco cessation efforts.
Implications for Policy, Delivery, or Practice:
As practices move toward becoming patientcentered medical homes and participate in
accountable care organizations, greater
emphasis and support may be placed on
engaging in these recommended prevention
activities. This could also be encouraged by the
incorporation of tobacco cessation metrics into
quality improvement payment incentives on the
part of third party payers.
Funding Source(s): RWJF
Poster Session and Number: B, #709
Clinician Beliefs and Practices Regarding
Screening and Brief Intervention for Drug
Use of their Community Health Center
Patients
Anjani Reddy, University of California Los
Angeles; Lillian Gelberg, University of California
Los Angeles; Ron Andersen, University of
California Los Angeles
Presenter: Anjani Reddy, NRSA Fellow,
University of California Los Angeles
areddy@mednet.ucla.edu
Research Objective: Integration of behavioral
health including substance use problems into
primary care is an essential benefit that federally
qualified health centers (FQHCs) will offer as
part of the Affordable Care Act (ACA). It is
important for primary care clinicians to approach
their patients’ drug use disorders with
confidence and responsibility. This study
explores FQHC primary care clinicians’ beliefs
and practices in integrating illicit drug use
assessment and treatment into their practices
with a screening, brief intervention and referral
to treatment (SBIRT) protocol.
Study Design: We administered a 10-minute
questionnaire to 68 primary care clinicians of 5
FQHCs in Los Angeles.
Population Studied: Primary care clinicians of
5 FQHCs in Los Angeles.
Principal Findings: Clinicians expressed
limited confidence in their ability to deal with
illicit drug use of their patients, scoring on
average 3.31 on a five point Likert scale. Twothirds reported that they assess for drug use
routinely ‘at every visit’ and/or ‘at annual visits’.
When asked how often they counsel regarding
drug use (on a five point Likert Scale from
‘Never’ to ‘Always’), the median response was 4
(‘Usually’). Regarding their perspectives on the
best practical resource for addressing drug use
in their clinics, 45.6% named primary care
clinicians. A minority (29.4%) of clinicians had
completed a clinical rotation dealing with
substance use, and 27.2% reported that more
than 10 hours of their training was devoted to
substance use problems. Having a substance
use rotation was associated with greater
confidence in SBIRT (p<0.01). More hours of
substance use training was associated with
greater confidence (p=0.01) and routinely
addressing substance use in their patients
(p=0.04).
Conclusions: Our findings suggest that clinician
confidence and practices in substance use care
are not optimal, but are associated with
increased substance use education.
Implications for Policy, Delivery, or Practice:
Further work should examine whether improving
clinicians’ education/training at community
health centers, via the establishment of an
SBIRT protocol improves substance use care
practices and facilitates the ACA’s mandate to
integrate substance use into routine primary
care of FQHCs.
Funding Source(s): NIH
Poster Session and Number: B, #710
Satisfaction with Insurance Plans and
Providers Among Persons with Mental
Illness: Findings from a panel data analysis
Kathleen Rowan, Minnesota Population Center
Presenter: Kathleen Rowan, Ihis Graduate
Research Assistant, Minnesota Population
Center
kath.rowan@gmail.com
Research Objective: Health insurance
coverage increases access to health care;
however, insurance plans can impose
administrative problems for care-seekers while
poor interactions with providers can reduce
continuity in treatment and adherence. These
negative experiences can be more difficult to
navigate for persons with mental illness. Few
studies have examined the association of mental
health status and satisfaction with insurance
plans and providers. Given recent changes in
the insurance market and health care reform, it
is timely to examine the prevalence and
variation of experiences with insurance plans
and providers for people with mental illness.
Here we address three research questions: 1)
Do persons with mental illness experience
greater problems with insurance plans
compared to those without mental illness? 2)
Do persons with mental illness experience
greater problems with providers compared to
those without mental illness? 3) How do these
experiences vary by public and private coverage
for persons with mental illness? We assess
problems by levels of severity of mental illness,
a distinction which enables stakeholders to
identify problems among persons with the
greatest need for treatment.
Study Design: Data come from seven panels
(2004 to 2011) of the Medical Expenditure Panel
Survey and the Integrated Health Interview
Survey. We measure mental health status at
multiple time points which we use to assess
severity of problems. Mental illness is defined
as serious psychological distress (SPD)
measured by the Kessler-6 scale. Satisfaction
with insurance plans is measured in the same
manner for public and private plans and includes
five problems: finding a doctor; getting approval
for treatment; finding information; customer
service; and paperwork. Indicators of
satisfaction with providers include: whether the
usual provider seeks the person’s advice when
choosing treatments; asks about medications
from other doctors; asks the person to help
make treatment decisions; and explains
treatment options. Responses are dichotomized
into "any problem" in contrast to no problem.
We report descriptive statistics on the
prevalence of problems, use multivariate logistic
regression to examine the association of mental
health status and outcomes, and contrasts to
compare problems between public and private
plans.
Population Studied: Adults age 18 to 64 with
any health care coverage.
Principal Findings: In both the public and
private sector, persons with mental illness have
more problems finding a doctor compared to
persons without mental illness. In the private
sector, persons with mental illness have more
problems getting approval for treatment
compared to persons without mental illness and
are less likely to be asked to help make
treatment decisions and provided options for
treatment. We found no differences in
experiences with providers by mental health
status in the public sector.
Conclusions: Insurance plans should examine
ways to expand access to providers and
providers should look for ways to improve the
quality of care through inclusion of patients in
treatment decisions.
Implications for Policy, Delivery, or Practice:
The Affordable Care Act (ACA) promises to
increase access to insurance to millions of
American. As access to care increases, it
important to continue to monitor adverse
experiences with plans and providers for those
with mental illness, as mental health affects
overall functioning in many domains in life.
Funding Source(s): N/A
Poster Session and Number: B, #711
Incentives in Public Alcohol and Drug
Treatment Payment Systems: Intended and
Unintended Effects
Maureen Stewart, Brandeis University; Sharon
Reif, Heller School for Social Policy and
Management, Brandeis University; Constance
Horgan, Heller School for Social Policy and
Management, Brandeis University; Beth Mohr,
Heller School for Social Policy and
Management, Brandeis University
Presenter: Maureen Stewart, Senior Research
Associate, Brandeis University
mstewart@brandeis.edu
Research Objective: Innovative ways are
needed to improve the quality of treatment for
substance use disorders (SUDs). Performancebased contracting (PBC) aims to align program
incentives and purchaser goals, yet is
uncommon in drug abuse treatment systems. In
2007, Maine implemented a second-generation
PBC system with financial incentives, for
outpatient programs, addressing problems
identified with its previous PBC. This study
aims to 1) Determine whether rewarded
measures and outcomes changed under the
PBC; 2) Examine whether there was client
selection due to the PBC; 3) Explore possible
unintended effects of the PBC.
Study Design: We examined a natural
experiment using state administrative data from
a period two years before the PBC through five
years after the PBC was introduced (20052012). Multilevel modeling techniques with a
difference-in-difference approach and a nonPBC comparison group were employed.
Population Studied: 25,105 admissions to
publicly funded addiction treatment agencies in
Maine between 2005 and 2012.
Principal Findings: Preliminary analyses
(N=25,105 admissions) indicated that the
probability of receiving four or more outpatient
treatment sessions did not change significantly
between the 2005 – 2007 pre-period and the
2008 – 2010 post-period (OR 0.9, p =.16) and
was not significantly different in the PBC
agencies from the non-PBC agencies. Similarly
there was no significant change in the probability
of remaining in treatment for 90 days or more.
The most significant predictors of retention were
client characteristics (primary substance,
previous treatment, criminal justice involvement
and homelessness). Client selection problems
were not identified in an analysis of admission
rates for individuals with diagnosed mental
disorders. We continue to refine these models
and examine the impact of the PBC on other
factors and client outcomes.
Conclusions: Introduction of a PBC is a
significant change in payment design that may
affect how addiction treatment services are
delivered, thus increasing our understanding of
its effects is critical. In preliminary analyses we
find the overall net effect of incentivized contract
very small and not significant.
Implications for Policy, Delivery, or Practice:
Payment systems may require additional
refinement in order to improve quality of
substance abuse treatment. It is likely that
implementation, both at the system level and
within individual agencies receiving incentives,
requires more focus in order to drive changes.
Further analyses will examine these aspects as
well as the possibility of differential effects over
specific time periods after the PBC was in place.
The fact that no selection bias was identified
indicates that thoughtful design of a PBC allows
it to be put in place without negative
consequences.
Funding Source(s): NIH
Poster Session and Number: B, #712
Wellness Initiatives: National Findings from
Private Health Plans
Sharon Reif, Heller School for Social Policy and
Management, Brandeis University; Constance
Horgan, Heller School for Social Policy and
Management, Brandeis University; Elizabeth
Merrick, Heller School for Social Policy and
Management, Brandeis University; Ann-Marie
Matteucci, Heller School for Social Policy and
Management, Brandeis University; Maureen
Stewart, Heller School for Social Policy and
Management, Brandeis University
Presenter: Maureen Stewart, Senior Research
Associate, Brandeis University
mstewart@brandeis.edu
Research Objective: Wellness programs
offered by employers are becoming ubiquitous,
and can be seen as a part of the preventiontreatment continuum for a variety of potential
health problems, both behavioral and physical.
Health plans frequently offer wellness activities
or more formal programs as part of their benefits
packages for employers, yet little is known about
how these are structured, and the role of
incentives in encouraging enrollees to
participate. Further, it is unclear to what extent
tracking of alcohol use and other substance use
is part of these wellness activities. Our objective
is to examine the what activities are included in
health plans’ wellness programs, determine the
degree to which mental health and substance
use are considered, and determine if and how
incentives are used to drive participation.
Study Design: This is a cross-sectional analysis
of nationally representative survey data. We
conducted product-level univariate and bivariate
analyses (n=939 products) with weighted data.
Population Studied: Data are from the 2010
Brandeis Health Plan Survey on Alcohol, Drug
and Mental Health Services, a nationally
representative survey of private health plans
regarding behavioral health care. We sampled
private health plans in 60 market areas to obtain
data on each market-specific plan’s top 3
commercial products, reporting here on 389
plans (89% response).
Principal Findings: The vast majority of health
plan products offer a health risk assessment
(HRA) (96%), online self-assessment tools
(93%) and health coaching (99%). Alcohol use
and drug use are frequently included in the
HRAs (87% and 78% respectively) and selfassessments are nearly always available for
alcohol use and drug use (98% and 85%
respectively). Enrollees are usually (82%)
offered incentives to complete HRAs, in the form
of premium reductions (82% of those with
incentives), premium increases if not completed
(44%), reduced cost-sharing (49%), other
monetary incentives (65%) or chance to win a
prize (81%).
Conclusions: Financial incentives are used by
health plans and may be viewed as a tool to
encourage enrollees to evaluate their own
wellness, which is expected to in turn have an
impact on healthy behaviors. Further, health
plans are taking a broad approach to wellness
going beyond the typical diet and exercise
approaches to include concerns about risky
drinking and substance use.
Implications for Policy, Delivery, or Practice:
Health plans have adopted a variety of wellness
initiatives which are fairly broad in the topics that
they address. A focus on prevention and
wellness is highlighted in the Affordable Care
Act, and these findings suggest that health plans
are well situated to address wellness and
prevention among their enrollees. Moving
forward, it would be valuable to have a more indepth understanding of the more active wellness
approaches (i.e., health coaching) and to
determine if these wellness initiatives translate
to better health for enrollees.
Funding Source(s): NIH
Poster Session and Number: B, #713
Adjunctive Antipsychotic Use for
Depression: Cost Implications for Medicaid
Yan Tang, University of Pittsburgh; Walid F.
Gellad, VA Pittsburgh Health Care System,
RAND, and University of Pittsburgh; Marcela
Horvitz-Lennon, RAND, University of Pittsburgh;
Julie M. Donohue, University of Pittsburgh
Presenter: Yan Tang, PhD Candidate,
University of Pittsburgh
yat11@pitt.edu
Research Objective: Major depressive disorder
(MDD) is a leading cause of disability in the
United States. The mainstay of pharmacological
treatment is antidepressant medication.
However, in recent years, some antipsychotics
have received FDA approval for adjunctive
treatment of depression. Little is known about
the prevalence of, or economic consequences
of, antipsychotic use in those with depression in
spite of the high costs of antipsychotics and their
high risk of side effects. We examined these
issues in a large Medicaid program due to the
important role Medicaid plays in financing care
for people with MDD.
Study Design: We obtained data for all 1 million
adult enrollees in Pennsylvania Medicaid from
2007-2011. We identified patients with >1
inpatient or >2 outpatient claims with a diagnosis
of MDD between 1/ 1/2008 and 12/31/2010. The
index event was the first diagnosis of MDD after
a period of >12 months with no diagnosis of
MDD or prescription for antidepressant or
antipsychotic. We included individuals
continuously enrolled for this 12-months preindex period and 12 months after the index
diagnosis, with at least one antidepressant or
antipsychotic prescription fill following the index
diagnosis. We excluded enrollees with
diagnoses of schizophrenia, bipolar disorder,
other psychoses, or autism as these may have
been the primary indications for an
antipsychotic. We defined adjunctive
antipsychotic treatment as an overlap of >30
days supply of any antidepressant and any
antipsychotic. Enrollees with MDD were then
classified into three groups: those with
antidepressants use only (no antipsychotic),
those with adjunctive antipsychotic use, and
those with non-adjunctive antipsychotic use (no
30-day overlap). We used a generalized linear
model (GLM) with gamma distribution and log
link to compare post-index period total
prescription drug spending for these three
groups. Spending was adjusted for differences
in sex, age, race/ethnicity, eligibility category,
insurance type, and health status (using
Elixhauser comorbidity index).
Population Studied: Non-dual eligible Medicaid
enrollees aged 18-64 with MDD.
Principal Findings: Of the 9,529 enrollees with
MDD approximately half (48%) used
antidepressants only (no antipsychotics). Fully
38% of the sample used adjunctive treatment
with antipsychotics and 14% used antipsychotics
without a 30 day overlap with their
antidepressant. Patients in adjunctive treatment
group were more likely to be SSI eligible (62%
vs. 45%, p<.0001) and had higher Elixhauser
comorbidity index (2.77 vs. 2.17, p<.0001) than
those in antidepressants only group. Unadjusted
mean total prescription drug spending was more
than 2-fold higher for those with adjunctive
treatment than those on antidepressants only
($5,422 vs. $2,540). Unadjusted drug spending
was $3,444 for the non-adjunctive group.
Adjusted mean prescription drug spending was
$3,019 higher (p<.0001) for the adjunctive
treatment group and $1,095 higher (p<.0001) for
the non-adjunctive group compared to the
antidepressants only group.
Conclusions: Antipsychotic drugs were used by
more than half of enrollees with major
depression in this large state Medicaid program.
Adjunctive use of antipsychotic drugs was
associated with significantly higher total
prescription drug spending.
Implications for Policy, Delivery, or Practice:
Managed care organization and Medicaid
directors may consider monitoring the treatment
outcomes and costs associated with adjunctive
antipsychotic treatment to improve the
appropriateness and value of this treatment
strategy.
Funding Source(s): Other CTSI - RAND
University of Pittsburgh Health Initiative Pilot
Grant, NIMH (MH087488)
Poster Session and Number: B, #714
How Policy Effectiveness was Achieved
while Spending Inequities Remained in
California's Realigned Public Mental Health
System
Megan Vanneman, VA Palo Alto Health Care
System + Stanford University School of
Medicine; Lonnie Snowden, University of
California, Berkeley; William Dow, University of
California, Berkeley
Presenter: Megan Vanneman, Postdoctoral
Fellow, VA Palo Alto Health Care System +
Stanford University School of Medicine
mev@stanford.edu
Research Objective: To evaluate whether the
legislative intent for a greater focus on
community-based care was met after the
realignment of California’s public mental health
system, and the extent to which it was achieved
in different counties.
Study Design: We use fixed-effect regression
models to test for spending differences between
counties with more or less racially/ethnically
diverse populations as well as historically lower
or higher spending levels per beneficiary.
Population Studied: All California Medicaid
specialty mental health claims for full-scope 0-25
year old beneficiaries (on average 156,000
individuals per year) and provider of service files
from fiscal years 1993-94 to 2003-04.
Principal Findings: Over time, counties
contributed a greater portion of their budgets to
community-based care. However, counties with
relatively low spending per beneficiary continued
to spend less than those counties with
historically higher spending levels. These
disparities are most noticeable in counties with
larger racial/ethnic minority populations that also
have lower spending levels per beneficiary. To
illustrate this point, it is useful to compare the
community-based expenditures of counties with
lower racial/ethnic diversity and high baseline
spending to that of counties with larger
racial/ethnic diversity and low baseline
spending. At the beginning of the study their
expenditures were quite different. Less diverse
counties with high baseline spending dedicated
$3,606 per beneficiary per year to communitybased care, while more diverse counties with
low baseline spending dedicated $1,890 per
beneficiary per year. This disparity persisted
over time, as each additional year after
realignment is associated with approximately
$196 more spending per beneficiary in both of
these subsets of counties.
Conclusions: Although the evidence suggests
that this realignment policy was effectively
implemented, future efforts that place more
attention on system-level disparities will be
better suited to increase equity.
Implications for Policy, Delivery, or Practice:
With respect to policy effectiveness and equity,
the findings of this research are informative for
California and other states with decentralized
systems as well as those considering or
implementing further decentralization policies. In
the case of California, perhaps the State could
monitor performance differences between
county public mental health systems, and aid
those systems that are struggling. Because the
State maintains some oversight authority for
county-based public mental health systems, it
could provide assistance to those counties that
are not taking full advantage of opportunities.
The State could also further examine and
address funding inequities between counties for
children, youth, and young adults served by the
public mental health system. A greater focus on
racial/ethnic minority populations and transition
age youth in California’s Mental Health Services
Act of 2004 may help to address disparities for
the child, youth, and young adult populations
studied here. Additionally, as of 2006, cultural
competence reports had to be submitted by
each county. However, these efforts could be
compromised if differences are not addressed
as additional services benefitting this population
have recently been devolved to county-based
public mental health systems.
Funding Source(s): Other University of
California, Berkeley
Poster Session and Number: B, #715
Linking the Legislative Process to the
Consequences of Realigning California’s
Public Mental Health System
Megan Vanneman, Center for Innovation to
Implementation (Ci2i) + Center for Primary Care
and Outcomes Research (PCOR); Lonnie
Snowden, University of California, Berkeley
Presenter: Megan Vanneman, Postdoctoral
Fellow, Center for Innovation to Implementation
(Ci2i) + Center for Primary Care and Outcomes
Research (PCOR)
mev@stanford.edu
Research Objective: To understand the
motivations behind California’s 1991 realignment
– the transfer of responsibility, resources, and
accountability for social, health, and mental
health services from the state to the counties –
and the impact this devolution has had on the
county-based public mental health system.
Study Design: An iterative mixed-methods
approach was used, involving analysis of 11
years of cross-sectional panel data for Medicaid
specialty mental health beneficiaries in all 58
California counties as well as 10 key informant
interviews with public mental health experts from
the State and counties (with varying levels of
historical spending per capita and racial/ethnic
diversity). We conducted purposeful sampling
for historical key informants and maximum
variation sampling for administrative key
informants, along with preliminary analysis and
fixed effect regression models using claims and
provider data from the California Department of
Mental Health. Key informant interviews were
used for expansion purposes to better frame and
interpret the statistical results.
Population Studied: Ten key informants with
extensive clinical, policy, and administrative
expertise, in conjunction with all California
Medicaid specialty mental health claims for fullscope 0-25 year old beneficiaries (on average
156,000 individuals per year) and provider of
service files from fiscal years 1993-94 to 200304.
Principal Findings: The realignment legislation
provided the incentives and flexibility for
counties to move toward an even more
community-based model of care, and we see
this trend in the quantitative data over time. The
decision to base Realignment allocations off
counties’ historical spending levels along with
minimal under-equity payments to address these
differences helped to institutionalize the
spending disparities observed in the quantitative
data over time.
Conclusions: The collaborative development of
Realignment legislation helped with the creation
of goals in which county-based systems
believed, and ultimately with the overall success
that counties achieved in creating a more
community-based system for public mental
health. However, the legislation also reinforced
some structural differences between county
systems that contribute to continued financial
disparities.
Implications for Policy, Delivery, or Practice:
A similar legislative process could be used in
sectors beyond public mental health to ensure
that expectations for a policy’s consequences
are made clear when resources, authority, and
responsibility are devolved. Given inevitable
financial and administrative limitations,
anticipated changes in effectiveness, efficiency,
and equity should be addressed and explicitly
prioritized.
Funding Source(s): Other University of
California, Berkeley
Poster Session and Number: B, #716
Mental and Substance Use Disorder
Spending Projections
Tracy Yee, Truven Health Analytics; Katharine
Levit, Truven Health Analytics; John Richardson,
University of Michigan, Ann Arbor; Sasha
Frankel, Results for Development Institute;
Lauren Hughey, Truven Health Analytics; Tami
Mark, Truven Health Analytics
Presenter: Tracy Yee, Research Leader,
Truven Health Analytics
tracy.yee@truvenhealth.com
Research Objective: The goal of this study is to
project spending on mental and substance use
disorder (M/SUD) treatment through 2020,
overall, by payer and provider, and relative to all
health care spending. We project spending on
M/SUD treatment as a result of enrollment
expansions from the Affordable Care Act (ACA).
Study Design: The projection methodology is
analogous to that used by CMS in estimating
future national health expenditures. We
employed a five-factor model that allocates
spending growth to five determinants (i.e.,
changes in population, prices, utilization, general
inflation, and residual influences), and a
production model, which estimates input costs
used to produce services. We modeled the
impacts of the ACA by estimating the effect of
the insurance enrollment expansion and
legislative effects on spending across payers
and providers. We estimated insurance
enrollment effects by determining the size of the
eligible population, likely take-up rates, user
rates for M/SUD treatment, the elasticity of
demand with respect to insurance, and spending
per user for M/SUD services. Projections of
prescription drug spending were developed by
evaluating the timing of expirations on drug
patents and their impact on prices.
Population Studied:
Principal Findings: We project that spending
on mental and substance use disorder treatment
will grow more slowly than all health spending,
resulting in a share of all health spending that
falls from 7.3 percent in 2009 to 6.5 percent in
2020. The main reason for the slowdown is the
projected reduction in spending on mental health
drugs resulting from patent expirations and new
game-changing drugs entering the market. In
addition, the lingering effects of the 2007-2009
recession on treatment spending in state
psychiatric hospitals will also slow mental health
spending growth. Treatment spending for
substance use disorders is expected to remain
only about 1 percent of all health spending in
2020.
The slowdown comes despite additions from the
ACA beginning in 2014 that are expected to
amount to 2.7 percent increase to M/SUD
spending in 2020. The ACA is expected to
increase spending on substance use disorder
treatment by 7 percent in 2020, largely because
many of the new users are expected to be
young adults who have higher prevalence of
substance use disorders.
The ACA is also projected to alter mental health
financing, primarily with increased spending by
Medicaid ($7.6 billion) and private insurers ($2.6
billion), and a reduction in spending on out-ofpocket (by $1.0 billion). Nevertheless, mental
health spending by Medicaid is expected to fall
as share of Medicaid budgets, as is mental
health spending as a share of private insurance.
Conclusions: Overall, spending on mental and
substance use disorders is likely to remain a
falling share of all health expenditures through
2020, even for major payers such as Medicaid
and private insurance.
Implications for Policy, Delivery, or Practice:
In 2014, the ACA will provide mental health and
substance abuse insurance coverage to millions
of individuals who are currently uninsured. The
costs of this increased access will be largely offset by the declining price of prescription
medications. These two factors should combine
to make treatment more affordable and reduce
unmet need.
Funding Source(s): Other Substance Abuse
and Mental Health Services Administration
(SAMHSA)
Poster Session and Number: B, #718
Disruptive Behavior Disorders in Medicaidenrolled Children: Disease Prevalence and
Medication Expenditures
Lirong Zhao, Centers for Medicare & Medicaid
Services; Vetisha McClair, Centers for Medicare
& Medicaid Services; Caitlin Cross-Barnet,
Centers for Medicare & Medicaid Services;
Sophia Chan, Centers for Medicare & Medicaid
Services; William D. Clark, Centers for Medicare
& Medicaid Services
Presenter: Lirong Zhao, Social Science
Research Analyst, Centers for Medicare &
Medicaid Services
lirong.zhao@cms.hhs.gov
Research Objective: Disruptive Behavior
Disorders (DBDs), including Conduct Disorder,
Oppositional Defiant Disorder and Attention
Deficit Hyperactivity Disorder, are among the
most common mental health conditions among
children under 20. DBD exert burdens on the
patients, their families, and the healthcare
system. Few studies have examined Medicaid
national-level spending on psychotherapeutic
drugs. Understanding Medicaid drug
expenditures and trends can help foster policies
that encourage medically appropriate and
financially prudent care to vulnerable children.
This study utilizes claims data to assess DBD
prevalence and prescription expenditures for
Medicaid beneficiaries under age 20.
Study Design: Children with a DBD diagnosis
were identified from Medicaid Analytic eXtract
(MAX), a national Medicaid administrative
research database. Unique beneficiaries in 49
states and DC, number of Medicaid covered
prescription drug fills (PDFN), and Medicaid paid
amount (MPA) were calculated for each year
from 2006-2009. The proportion of
psychotherapeutic drugs to all drug payments
was calculated. PDFN and MPA were
aggregated per beneficiary per year (PBPY) for
comparisons across years. The outcomes were
examined by gender, age, and race categories
and were compared by dividing the study cohort
by service delivery type: Fee-For-Service (FFS)
or Managed Care Organization (MCO).
Population Studied: Medicaid beneficiaries
under age 20 with at least one DBD diagnosis in
Medicaid services claims from 2006 through
2009. MCO encounter data may be incomplete,
which could constrain population selection.
Principal Findings: In 2006, 1.31 million (or
3.95%) of the Medicaid children who had claims
information were diagnosed with at least one
DBD. Among them, almost 90% of DBDdiagnosed children had at least one prescription
filled each year. While total PDFN and MPA
increased from 2006 to 2009 (27.2% for PDFN
and 28.8% for MPA), aggregated per beneficiary
(PBPY) PDPN and MPA remained relatively
stable (less than 3%). The total MPA for
prescribed drugs in 2006 was 1.62 billion, 54.6%
of which was for psychotherapeutic drugs. The
proportion of psychotherapeutic drugs increased
across years, ending at 57.7% in 2009. Analysis
by age, gender, and race showed an increase in
PDFN, MPA, and psychiatric drugs proportion
across years. Beneficiaries age 19-20 and
males had higher PBPY of MPA than other
subgroups. Increase in PBPY of MPA and
psychiatric drugs proportion were evident across
service delivery types.
Conclusions: Increases in DBD diagnoses in
Medicaid children who had claims information
indicate an increasing need for comprehensive
pediatric psychiatric and psychological care.
Psychotherapeutic drugs accounted for a major
portion of Medicaid DBD children’s medication
costs. More research is needed to investigate
classes of psychotherapeutic drugs and
potential factors associated with the high
proportion of psychiatric drug prescriptions and
trends across years. Data sources that include
all Medicaid MCO DBD beneficiaries and that
disaggregate diagnosis and medication
information are preferable for further study.
Implications for Policy, Delivery, or Practice:
While more children with DBDs are receiving
care, the appropriateness of the care provided
may vary. Medicaid, as the largest insurance
provider for children in the US, faces challenges
in providing effective and appropriate care for
this vulnerable population. Medicaid could
potentially use its market leverage to foster
reforms in the treatment of children with DBDs.
Funding Source(s): No Funding
Poster Session and Number: B, #719
Projected Spending on Psychotropic
Medications for 2013-2020
Tracy Yee, Truven Health Analytics; Dominic
Hodgkin, Brandeis University; Cindy Thomas,
Brandeis University; Margaret O'Brien, Brandeis
University; Katharine Levit, Truven Health
Analytics; John Richardson, University of
Michigan; Tami L Mark, Truven Health Analytics;
Kevin Malone, Substance Abuse and Mental
Health Services Administration
Presenter: Tracy Yee, Research Leader,
Truven Health Analytics
tracy.yee@truvenhealth.com
Research Objective: Spending on psychotropic
medications has grown rapidly in recent
decades, and by 2009 medications accounted
for 25 percent of all US spending in the
treatment of mental illness and substance use
disorders. However, many brand psychotropic
drugs will lose patent protection in the next few
years, and there are few new medications in
development. Changes in the trajectory of
spending on mental health and substance abuse
medications should be important to patients,
providers, insurers, and policymakers who are
trying to balance the goals of improving patient
health while maintaining or reducing costs. We
present projections of drug spending over the
period 2012-2020 that take into account patent
expirations and other factors.
Study Design: We use the National Prescription
Audit database compiled by IMS Health, which
are based on a sample of electronic pharmacy
claims transactions. Data was by drug class and
subclass over the period 2002 through 2012. In
addition, we spoke with experts and reviewed
the literature to determine possible changes in
the new drug pipeline, and about possible
reasons for utilization to change. We determined
which medications were going off patent in order
to simulate the effect of this on utilization and
prices. Projections of price and quantity were
carried out at the product level for five specific
drug classes of particular importance, and at the
class level for all other psychotropic drug
classes. Based on expert opinion, literature
review, and observations in the data, we project
that generic drugs that will be replacing
medications going off patent will experience a
drop in price to 30% of the branded drug price
within two years of patent expiration and that
most consumers will switch to these lower priced
generic versions.
Population Studied: Data summarize utilization
for the US population as a whole.
Principal Findings: We project that growth in
spending on psychotropic medications will slow
down over the period 2012–2020. The average
annual increase is projected to be just 2.7
percent per year, continuing the steady
deceleration in annual psychotropic medication
spending growth that had peaked at 25.6
percent in 1998. The projected growth for
psychotropic medications is considerably slower
than what is expected for prescription drugs
overall.
Conclusions: The main drivers of this expected
deceleration include the slowdown in
development of new drugs, upcoming patent
expirations which will lower prices, and the
growing ability of payers to manage utilization
and promote generic use.
Implications for Policy, Delivery, or Practice:
The slowdown in spending on psychotropic drug
spending will relieve some cost pressures on
payers, particularly Medicare and Medicaid,
which between them fund more than half of all
spending on antipsychotic drugs.
Funding Source(s): Other Substance Abuse
and Mental Health Services Administration
Poster Session and Number: B, #720
Value Associated with Insourced versus
Outsourced Depression Collaborative Care
in Rural Federally Qualified Health Centers
Jeffrey Pyne, VA HSR&D Center for Mental
Healthcare and Outcomes Research; John
Fortney, UAMS Psychiatric Research Institute;
Sip Mouden, Community Health Centers of
Arkansas Inc.; Liya Lu, UAMS Psychiatric
Research Institute; Teresa Hudson, VA HSR&D
Center for Mental Healthcare and Outcomes
Research; Dinesh Mittal, VA HSR&D Center for
Mental Healthcare and Outcomes Research
Presenter: Jeffrey Pyne, Research
Scientist/staff Physician, VA HSR&D Center for
Mental Healthcare and Outcomes Research
jmpyne@uams.edu
Research Objective: Collaborative care for
depression has been found to be effective and
cost-effective in primary care settings. However,
collaborative care for depression has been
implemented using on-site and off-site models.
This study examined the cost-effectiveness of
on-site practice-based collaborative care
(PBCC) versus off-site telemedicine-based
collaborative care (TBCC) for depression in
Federally Qualified Health Centers (FQHCs).
Study Design: Multi-site randomized pragmatic
comparative cost-effectiveness trial. Subjects
were enrolled 2007-2009 from five participating
Arkansas FQHCs, each serving 5,362-13,050
unique primary care (PC) patients. Participating
FQHCs did not have on-site mental health
specialists. Subjects randomized to PBCC
received evidence-based care from an on-site
PC provider and nurse depression care
manager (DCM). Those randomized to TBCC
received evidence-based care from an on-site
PC provider and off-site telephone DCM,
telephone pharmacist, tele-psychologist and
tele-psychiatrist. All DCMs had access to a
web-based patient registry and decision support
system (https://www.netdss.net/). Base case
analysis costs used FQHC healthcare utilization
costs and secondary analysis used national cost
estimates. Effectiveness measures were
depression-free days (DFDs) and qualityadjusted life years (QALYs) derived from DFD,
Medical Outcomes Study SF-12, and Quality of
Well Being scale (QWB). Telephone interview
data were collected at baseline, 6-, 12, 18months.
Population Studied: 19,285 patients were
screened for depression, 14.8% (n=2,863)
screened positive (PHQ9 =10) and 364 were
enrolled. Patients with serious mental illness
and acute suicide ideation were excluded. The
population-based sample was randomized
(stratified by clinic) to TBCC or PBCC.
Principal Findings: Mean base case
incremental cost-effectiveness ratio (ICER)
using DFDs and bootstrapped sample was
$10.78/DFD. Mean base case ICER using SF12 QALYs was $33,464/QALY. Sensitivity
QALY analyses ranged from $14,754/QALY
(DFD QALY4) to $37,261/QALY (QWB QALY).
Secondary national ICER was $25,753/QALY
(SF-12 QALY). Sensitivity analyses ranged from
$11,532/QALY (DFD QALY4) to $29,234/QALY
(QWB QALY).
Conclusions: This study supports the costeffectiveness of the TBCC intervention in
medically underserved primary care settings
according to commonly used thresholds for costeffectiveness.
Implications for Policy, Delivery, or Practice:
These results inform decisions about whether to
insource or outsource depression care
management. This decision is particularly
relevant within the current context of PatientCentered Medical Home recognition, valuebased purchasing, and potential bundled
payments for depression care.
Funding Source(s): NIH
Poster Session and Number: B, #721
Intimate Partner Violence-Related Traumatic
Brain Injury in a Sample of Female Veterans
Katherine Iverson, VA Boston Healthcare
System; Terri Pogoda, VA Boston Healthcare
System, Center for Healthcare Organization and
Implementation Research
Presenter: Katherine Iverson, Clinical Research
Psychologist, VA Boston Healthcare System
katherine.iverson@va.gov
Research Objective: Intimate partner violence
(IPV) is a significant population health problem.
Relative to civilians, female veterans are at
higher risk for lifetime IPV. Traumatic brain injury
(TBI), a condition defined as a blow or jolt to the
head that disrupts brain function, is a “signature
injury” of recent Veterans. Although IPV is
prevalent among female veterans and is known
to increase women’s risk for TBI, the occurrence
of probable IPV-related TBI has not been
examined in the female veteran population. The
objective of this study was to identify the
occurrence of IPV-related TBI in a sample of
female veterans, and to examine the
associations of IPV-related TBI with
posttraumatic stress disorder (PTSD) and
depressive symptoms.
Study Design: Cross-sectional study of female
veterans who participated in a larger survey
between February and April 2013. IPV-related
TBI was assessed with a modified version of the
Veterans Affairs (VA) TBI screening tool.
Women were considered to have a probable IPV
act to the head if they reported experiencing at
least one of six events by an intimate partner
(e.g., having their head pushed or shoved into a
wall, car, furniture, or object; being
strangled/choked). Consistent with VA’s
assessment for probable TBI, women were then
considered to have probable IPV-related TBI
history if they self-reported that IPV events were
associated with loss of consciousness, altered
consciousness (e.g., being dazed or confused),
posttraumatic amnesia, concussion, or head
injury. The survey also included validated
measures of PTSD (PTSD Checklist; PCL) and
depressive (Center for Epidemiologic Studies
Depression Scale; CES-D) symptoms.
Population Studied: The sample from the
larger study comprised 198 female patients of
VA hospitals in New England. The response rate
for the survey was 79.8%. The current study
included 175 women who responded to
questions concerning IPV-related TBI.
Principal Findings: Among the 175 participants
in this study, the percentage of women who
reported experiencing at least one act of IPV to
the head was 32% (n = 56). Of these 56 women,
37.5% (n = 21) screened positive for probable
IPV-related TBI history. Women who
experienced probable IPV-related TBI reported
significantly more severe PTSD (mean PCL
scores: 57.8 versus 36.7, p<.0001) and
depressive (mean CES-D scores: 29.3 versus
21.3, p<.0001) symptoms than those who
experienced IPV-related acts without probable
TBI.
Conclusions: These findings suggest that IPVrelated TBI is a common health issue for female
veterans, with at least one in ten women in the
current sample meeting criteria for probable IPVrelated TBI history. In addition, IPV-related TBI
history is strongly associated with current mental
health symptoms in this sample of female
veterans.
Implications for Policy, Delivery, or Practice:
The current findings can inform VA’s efforts to
implement national guidelines with respect to
IPV detection and care for female veterans. VA
has established TBI evaluation and treatment
services, and with additional IPV training for
providers, this healthcare system could be wellpositioned to identify IPV-related TBI and to
provide more comprehensive health care and
preventative services.
Funding Source(s): VA
Poster Session and Number: B, #722
Complex Chronic Conditions
Preventing Stroke in Youths with Sickle Cell
Disease
Michael Abrams, The Hilltop Institute; Jennifer
Smith, The Hilltop Institute; Carl Mueller, The
Hilltop Institute; Marlene Miller, Johns Hopkins
University Schools of Medicine and Public
Health; James Casella, Johns Hopkins
University School of Medicine; David Bundy,
Medical University of South Carolina
Presenter: Michael Abrams, Senior Research
Analyst, The Hilltop Institute
mabrams@hilltop.umbc.edu
Research Objective: Youths (children and
adolescents) with sickle cell disease (SCD) are
at heightened risk for stroke. Transcranial
Doppler (TCD) imaging is a non-invasive way to
detect those at especially high risk. When such
risk is detected, ongoing blood transfusion is
indicated as an established method to reduce
subsequent stroke by more than 90 percent.
Accordingly, annual TCD screening is
recommended for youths with SCD, though such
screening often does not occur. This research
tests if a mailing outreach to guardians and
primary care providers (PCPs) of youths with
SCD can increase the use of TCD to screen for
elevated stroke risk.
Study Design: Administrative claims data from
Maryland were used to isolate SCD cases
enrolled in Medicaid during the period spanning
November 2010 to October 2011 (base year).
Medicaid data were further used to isolate base
year and follow-up period (November 2011 to
July 2012) data that summarized demographics
(age, race, gender, regional density), Medicaid
enrollment, TCD use, crisis service (inpatient
and emergency department (ED)) use, and
preventative/specialty (well or hematology)
visits. Multivariate logistic regressions were
constructed using receipt of TCD as the
dependent variable and all other information as
explanatory variables, including whether or not a
letter was sent to an individual’s guardian and
PCPs.
Population Studied: A total of 550 youths (age
2-16 years) with SCD were identified for this
study- none of which had Medicaid records
revealing TCD in the 12 months prior to the
intervention or during a 1.5 month mailing
interval period. Letters were sent to the
guardians and PCPs of 130 of these youths by
their managed care organization. Comparison of
the independent variables studied between
these 130 intervened subjects and the 420
remaining “controls” reveal no major imbalances
between those two groupings.
Principal Findings: The logistic model was
significant and predicted 20 percent of the
variance; however, the mailing variable was not
a significant predictor of TCD use (adjustedodds ratio (AOR)=0.86, 95 percent confidence
interval (CI)= 0.34-2.0). Other effects were nonsignificant (p>0.05), with two notable exceptions:
1) those classified as disabled by Medicaid
enrollment criteria had somewhat increased
odds of TCD screening (AOR=2.6; CI=1.1-6.2);
and 2) those visiting a hematologist during the
follow-up period had substantially increased
odds of receiving a TCD (AOR=8.8; CI=3.7-21).
Conclusions: A single mailing intervention does
not appear to be effective at increasing use of a
non-invasive though important screening
technique for youths at risk for SCD-related
stroke. However, disability status and specialist
visits were correlated with recommended
screening.
Implications for Policy, Delivery, or Practice:
The disability status correlate suggests that
those pre-identified by Medicaid with high
morbidity have increased access to screening
for stroke, perhaps because this medically
needy population is already the focus of much
clinical attention to manage the illness (or
constellation of illnesses) that made them qualify
for disability status. The hematologist visit
correlate, the strongest finding for this
investigation, suggests that health
communication and other interventions to
encourage use of secondary preventative
services might be more successful if they target
general (e.g., “see an expert”) rather than
specific (e.g., “seek this treatment”) messaging
and outcomes.
Funding Source(s): RWJF
Poster Session and Number: C, #1022
Stroke Incidence Following Traumatic Brain
Injury in Older Adults
Jennifer Albrecht, University of Maryland School
of Pharmacy; Xinggang Liu, University of
Maryland School of Pharmacy; Gordon Smith,
University of Maryland School of Medicine;
Mona Baumgarten, University of Maryland
School of Medicine; Gail Rattinger, Fairleigh
Dickinson University School of Pharmacy;
Steven Gambert, University of Maryland School
of Medicine; Patricia Langenberg, University of
Maryland School of Medicine; Ilene Zuckerman,
University of Maryland School of Pharmacy
Presenter: Jennifer Albrecht, Postdoctoral
Fellow, University of Maryland School of
Pharmacy
jalbrecht@rx.umaryland.edu
Research Objective: Traumatic brain injury
(TBI) is a significant health problem among older
adults that results in cognitive and functional
disability. Stroke may be a cause of disability
post-TBI. Older adults with TBI have multiple
chronic conditions, and are at increased risk of
both intracranial hemorrhage and
thromboembolic events during hospitalization for
TBI. However; it is unclear whether increased
risk continues following hospital discharge. The
objective of this study was to estimate incidence
rates of hemorrhagic and ischemic stroke
following hospital discharge for TBI among
Medicare beneficiaries aged =65 and compare
them with pre-TBI rates.
Study Design: This was a retrospective
analysis of Medicare claims data. The primary
exposure was TBI, defined on inpatient claims
by ICD-9 codes 800.xx, 801.xx, 803.xx, 804.xx,
850.xx- 854.1x, 950.1-950.3, 959.01. The
primary outcomes were hemorrhagic or
thrombotic stroke. Hemorrhagic stroke was
defined on inpatient claims by ICD-9 codes
430.xx-432.xx. Ischemic stroke was defined on
inpatient claims by ICD-9 codes 433.xx, 434.xx,
435.xx, 437.0x, 437.1x. Beneficiaries contributed
follow-up time to our study if they were enrolled
in Medicare Parts A and B, with no Part C
enrollment, at any time during the study period
(2006-2009). We counted stroke events before
and after TBI and used follow-up time as the
denominator for our incidence calculations. To
determine if TBI increased the risk of stroke, we
conducted a time to event analysis with TBI as a
time-varying exposure.
Population Studied: The study population was
selected from a 5% random sample of Medicare
beneficiaries =65 years hospitalized for
traumatic brain injury during 2006-2009 who
survived to hospital discharge. (N=16,936)
Principal Findings: Medicare beneficiaries with
traumatic brain injury were predominantly female
(62%) and white (90%), with an average age of
81 (standard deviation, 7.9) years. Heart failure
(46%) and history of depression (41%) were the
most common comorbidities. The annualized
incidence rate per 1,000 for hemorrhagic stroke
following TBI was 23.9 (95% confidence interval
(CI) 21.8, 26.3). There was a six-fold higher rate
of hemorrhagic stroke post-TBI compared to
pre-TBI (adjusted rate ratio (RR) 6.5; 95% CI
5.3, 7.8). The annualized incidence rate per
1,000 for ischemic stroke following TBI was 84.4
(95% CI 80.3, 88.6). The rate of ischemic stroke
was significantly higher post-TBI than pre-TBI
(adjusted RR 1.3; 95% CI 1.2, 1.4). Incidence
rates for both hemorrhagic and ischemic stroke
were highest in the month after discharge from
TBI hospitalization and decreased over time. At
1 year post-hospital discharge for TBI, the
incidence rate of hemorrhagic stroke remained
double the baseline rate.
Conclusions: Following TBI, there is an
elevated risk of stroke among older adults.
Implications for Policy, Delivery, or Practice:
The increased risk of both hemorrhagic and
ischemic stroke following TBI among patients
aged =65 raises concerns because prophylactic
treatments for ischemic stroke include
anticoagulation therapy which could increase
hemorrhagic stroke risk and may be
contraindicated post-TBI.
Funding Source(s): NIH
Poster Session and Number: C, #1023
Benefits and Risks of Anticoagulation
Following Traumatic Brain Injury
Jennifer Albrecht, University of Maryland School
of Pharmacy; Xinggang Liu, University of
Maryland School of Pharmacy; Stephen
Gottlieb, University of Maryland School of
Medicine; Mona Baumgarten, University of
Maryland School of Medicine; Gail Rattinger,
Fairleigh Dickinson University School of
Pharmacy; Steven Gambert, University of
Maryland School of Medicine; Patricia
Langenberg, University of Maryland School of
Medicine; Ilene Zuckerman, University of
Maryland School of Pharmacy
Presenter: Jennifer Albrecht, Postdoctoral
Fellow, University of Maryland School of
Pharmacy
jalbrecht@rx.umaryland.edu
Research Objective: Risk of venous
thromboembolism and stroke increases
substantially following TBI. Treatment with
anticoagulant therapy can reduce risk of
thrombotic events post-TBI, but this benefit must
be balanced against the potential for a higher
risk of bleeding, particularly intracranial
hemorrhage. The increased risk of hemorrhage
associated with anticoagulant therapy following
traumatic brain injury creates a serious dilemma
for medical management of multimorbid older
patients: should anticoagulant therapy be
resumed after traumatic brain injury and if so,
when? The objective of this study was to
estimate the risk of thrombotic and hemorrhagic
events associated with warfarin therapy
resumption following traumatic brain injury.
Study Design: This was a retrospective
analysis of Medicare claims data. The primary
exposure was warfarin utilization in each 30 day
period following hospital discharge for traumatic
brain injury. Monthly warfarin utilization was
determined from Medicare Part D prescription
drug event files. The primary outcomes were
hemorrhagic and thrombotic events following
hospital discharge for traumatic brain injury. A
hemorrhagic event was defined on inpatient
claims by the following ICD-9 codes:
hemorrhagic stroke (430.xx-432.xx); upper
gastrointestinal bleeding (531.xx, 532.xx,
533.xx, 534.xx, 578.xx); adrenal hemorrhage
(772.5x). Thrombotic events included ischemic
stroke (433.xx, 434.xx, 435.xx, 437.0x, 437.1x)
and pulmonary embolism (415.1x). A composite
of hemorrhagic or ischemic stroke was a
secondary outcome.
Population Studied: All Medicare beneficiaries
=65 years hospitalized for traumatic brain injury
(defined by ICD-9 codes 800.xx, 801.xx, 803.xx,
804.xx, 850.xx- 854.1x, 950.1-950.3, 959.01)
during 2006-2009 who received warfarin in the
month prior to injury and survived to hospital
discharge (n=12,215).
Principal Findings: Medicare beneficiaries with
traumatic brain injury were predominantly female
(64%) and white (93%), with an average age of
81 (standard deviation, 7) years and 82% had
atrial fibrillation. Over the 12 months following
hospital discharge, 53% received warfarin during
at least one 30-day period. We examined the
lagged effect of warfarin on outcomes in the
following period. Warfarin utilization was
associated with a decreased risk of thrombotic
events (relative risk (RR) 0.75; 95% confidence
interval (CI) 0.64, 0.88) in the following 30-day
period but an increased risk of hemorrhagic
events (RR 1.33; 95% CI 1.11, 1.60) in the
following 30-day period. Warfarin was
associated with decreased risk of hemorrhagic
or ischemic stroke (RR 0.80; 95% CI 0.69, 0.92)
in the following period.
Conclusions: Following traumatic brain injury, a
large percentage of Medicare beneficiaries do
not resume anticoagulation therapy with
warfarin. Resumption of warfarin therapy
following TBI is associated with an increased
risk of hemorrhagic events and a decreased risk
of thrombotic events.
Implications for Policy, Delivery, or Practice:
Results from this study suggest that despite
increased risk of hemorrhage, there is a net
benefit for most anticoagulated patients, in terms
of a reduction in risk of thrombosis, from
warfarin therapy resumption following hospital
discharge from TBI.
Funding Source(s): N/A
Poster Session and Number: C, #1024
Linking Patients with Complex Diabetes to
Self-Care Programs
Sarah Krein, VA Ann Arbor Healthcare System;
Ann Annis Emeott, VA Ann Arbor Healthcare
System; Bree Holtz, Michigan State University;
Wendy Morrish, VA Ann Arbor Healthcare
System; Jennifer Davis, VA Ann Arbor
Healthcare System
Presenter: Sarah Krein, Health Science
Research Specialist, VA Ann Arbor Healthcare
System
skrein@umich.edu
Research Objective: Research has
demonstrated that patients with chronic
conditions can improve their health through
participation in self-care programs. However,
getting patients who are likely to benefit to enroll
in these programs can be challenging. The
purpose of this study is to identify key patient
factors that promote or impede efforts by nurses
in connecting patients with complex diabetes to
appropriate health services.
Study Design: This is a retrospective analysis
of data collected as part of a computer-based,
nurse-led intervention (called the Navigator
program) delivered within a large VA primary
care clinic.
Population Studied: Patients were identified
via a population-based registry as having
diabetes and meeting criteria for being at highrisk for complications, and were referred to the
Navigator program. This study includes patients
with an assessment completed by the Navigator
nurse between December 2010 and January
2013.
Principal Findings: The majority of patients
were male (97%), with an average age of 65
years. Among 397 patients with a completed
initial assessment, 68% were referred by the
nurse to a variety of VA programs or services. In
total, 270 patients generated 501 referrals.
Approximately 37% of referrals were for selfcare programs, such as in-person and
telephonic exercise intervention programs, home
telehealth, and a care management program
involving patients’ informal caregivers. However,
almost one-fifth (19%) of patients declined to
accept any referrals. A higher percentage of
patients who declined all referrals reported that
they understood their health condition (100% vs.
91%, p=.04), were not interested in working on
their health condition (73% vs. 46%, p<.01), and
were not interested in traveling for programs
(77% vs. 61%, p=.01), as compared to those
who did not decline referral. Additionally, those
declining referrals had higher mean confidence
levels for completing health tasks (9.2 vs 8.6,
p=.03), and managing their health (9.0 vs 8.4,
p=.02). Compared to patients who declined all
referrals, patients who did not decline had a
higher mean number of primary care visits in the
previous year (5.01 vs. 4.16, p=.02), higher
mean pain scores reported for the week prior to
the assessment (4.7 vs. 3.4, p<.01), and higher
mean pain scores representing the highest level
of pain experienced in the previous four weeks
(5.5 vs. 4.2, p<.01). There were no differences in
reported general health and self-efficacy in
managing their condition between the two
groups.
Conclusions: Among our study population of
high-risk diabetic patients, a willingness to work
on their health condition and travel for services,
knowledge of their health condition, confidence
in managing their health, pain level, and
previous primary care visits seem to predict
patients’ receptivity to accepting referrals to
health services and self-care programs.
Implications for Policy, Delivery, or Practice:
Understanding what factors influence patients’
decisions to consider and participate in self-care
programs has important implications for program
design and the development of effective
strategies to encourage the use of these
programs. This information can also inform
outreach efforts by care providers and program
staff to identify and engage patients who are
likely to benefit from these self-care activities.
Funding Source(s): VA
Poster Session and Number: C, #1025
Quality Compliance Score – A New
Methodology for Identification the Potential
for Future Savings Based on the Adherence
to Evidence-Based Care Guidelines
Ogi Asparouhov, LexisNexis/MEDai; Anton
Berisha, LexisNexis/MEDai
Presenter: Ogi Asparouhov, Chief Scientist,
LexisNexis/MEDai
oasparouhov@medai.com
Research Objective: The importance of proper
selection of target population for disease and
case management, preventive, patient
engagement and motivation efforts has been
established as key factor toward improved
clinical and financial performance as well as
patient satisfaction.
Study Design: The Quality Compliance Score
(QCS) is a member-level indicator which
identifies the potential to maximize future
savings over the next 24 months by adherence
to evidence-based care guidelines. The QCS
model utilizes 142 guidelines: 108 diseaserelated and 34 preventive care guidelines
covering 21 chronic diseases and five groups of
preventive services.
Population Studied: The model was trained on
three years of longitudinal claims data,
representing 2,070,669 member lives. Benefit
plans included Commercial, Medicaid and
Medicare.
Principal Findings: 1. The QCS increases as
the number of major disease groups or the
number of guidelines increases.
2. Lower QCS are seen when preventive care
guidelines predominate.
3. Chronic diseases such as schizophrenia,
CAD and heart failure rank higher for
impactability in the QCS model than patients
with other illness.
4. Guidelines appearing in the top 20 and
bottom 20 by average QCS were judged
clinically appropriate.
5. Patients with very high severity (catastrophic
patients) received a severity adjusted QCS
reflected in lower values indicative of lower
savings potential.
6. As guideline non-compliance rate per person
increased, the QCS increased. Patients 100%
non-compliant with guidelines have on average
higher QCS than patients 100% compliant.
7. Patients 100% compliant with guidelines
received a QCS due to eligibility for guidelines
and the need to remain compliant over the next
24 month period. The model assumes some
savings potential exists.
Conclusions: The developed Quality
Compliance Score could be used for filtering out
the most impactable individuals from both cost
savings potential and clinical improvements
perspective, realized by achieving but more
importantly by maintaining compliance to
Evidence Based Medicine Guidelines.
Implications for Policy, Delivery, or Practice:
Implications for Practice:
The Quality Compliance Score model provides
the means to manage the right people at the
right time with the right interventions using the
right tools; and achieve the following desired
goals: best clinical outcomes and maximization
of the Return on Investment (ROI).
Funding Source(s): Other MEDai
Poster Session and Number: C, #1026
The Changing Face of Polypharmacy in
Australia. Did a Large Price Increase Make a
Difference?
Peter Balram, Department of Human Services
Presenter: Peter Balram, Pharmacentical
Adviser, Department of Human Services
peter.balram@humanservices.gov.au
Research Objective: The Australian
government subsidizes more than three quarters
of the cost of medications. However, one
unintended effect of this is stockpiling by
patients and the subsequent waste of
medication and taxpayer money. In order to
overcome this patients are required to pay a
copayment for their medication, which is
effective in reducing waste of medication but
leaves some patients on multiple medicines
vulnerable to simultaneous increases in the cost
of all their prescriptions. Some patients respond
to the increased price by reducing or ceasing
their medication. This study investigates the
impact of a twenty percent increase in
copayment for patients who take five or more
medications per month (polypharmacy).
Study Design: Data for this study were supplied
by Medicare Australia which administers the
Pharmaceutical Benefits Scheme (PBS) for the
Australian Government. A comparative analysis
was conducted of medications purchased the
year before the copayment rise came into effect
and medication purchases the following year.
Population Studied: A total of 13,523 patient
records comprising details of 540, 411
prescriptions were compared over 2004 and
2005.
Principal Findings: About seven percent of
patients used at least five medicines, and one
fifth used 2-4 medications per month (minor
polypharmacy). Just over two thirds of patients
aged over 64 demonstrated some form of
polypharmacy.
More than half of patients on cardiovascular
medications were subject to either minor
polypharmacy or polypharmacy. Multiple
medication use was also common in patients
taking diabetes, musculoskeletal and nervous
system medications.
Compared to 2004, over half of the
polypharmacy and minor polypharmacy patients
purchased significantly fewer medications
following the copayment increase. The biggest
reductions in medications were for
cardiovascular medications, notably ACEInhibitors. Minor polypharmacy patients
increased their purchase of alimentary
tract/diabetes medications, while those in the
polypharmacy category significantly reduced
these medications in 2005.
Conclusions: Polypharmacy in Australia is
increasing, particularly amongst younger adults
aged 25-44. This creep of polypharmacy into
the younger age groups is most likely driven by
factors such as early detection, and aggressive
treatment of chronic diseases such as high
blood pressure, diabetes, and depression.
However, increasing polypharmacy also leads to
potential financial hardship. People especially
vulnerable to hardship are the elderly and the
chronically ill, who both significantly reduced
their use of medication for hypertension,
diabetes and asthma following a major cost
increase. This may ultimately increase use of
other health care services and overall
expenditure.
Implications for Policy, Delivery, or Practice:
A major contributor of polypharmacy is
medicines being prescribed by multiple
prescribers at different times, a scenario
encountered when medications are commenced
in hospital and not revaluated on discharge.
Monitoring of patients’ medications on discharge
would eliminate hospitalisation as a risk factor to
polypharmacy.
Initiatives including clinical guidelines for
deprescribing and medicines reviews need to be
expanded to reduce unnecessary polypharmacy.
Phasing in large copayment changes for specific
medication groups rather than introducing an
across the board copayment increase should
protect financially vulnerable patients with
chronic illness. Finally, the change in paradigm
to early detection and aggressive therapy of
chronic diseases requires regular monitoring of
the potential for polypharmacy.
Funding Source(s): Other Medicare Australia
Poster Session and Number: C, #1027
Medication Effectiveness with the Use of
Tumor Necrosis Factor Inhibitors among
Texas Medicaid Patients Diagnosed with
Rheumatoid Arthritis
Jamie Barner, University of Texas at Austin;
Abiola Oladapo, Baxter; Karen Rascati,
University of Texas at Austin; Kristin Richards,
University of Texas at Austin; Kenneth Lawson,
University of Texas at Austin; Suzanne Novak,
Austin Outcomes Research; David Harrison,
Amgen
Presenter: Jamie Barner, Professor, University
of Texas at Austin
jbarner@austin.utexas.edu
Research Objective: Although adalimumab
(ADA), etanercept (ETN), and infliximab (IFX)
have not been directly compared in clinical trials,
results from the majority of indirect treatment
comparisons suggest comparable efficacy and
safety profiles. However, these tumor necrosis
factor (TNF) inhibitors differ in administration
method and dosing flexibility, which may result
in differences in medication use profiles. The
objective was to determine the percentage of
rheumatoid arthritis (RA) patients on ADA, ETN
and IFX effectively treated using a validated
comparative effectiveness algorithm designed
for large retrospective databases.
Study Design: This was a retrospective
database study. Based on an RA medication
effectiveness algorithm (Curtis et al. 2011), a
medication was considered effective if all of the
following six criteria were met: 1) high
medication adherence (medication possession
ratio (MPR) =80%); 2) no switching or adding of
biologics; 3) no additions of new non-biologic
disease modifying antirheumatic drugs
(DMARDS); 4) no increase in dose of biologics
or frequency of administration of biologics; 5) no
more than 1 glucocorticoid (GC) joint injection;
and 6) no increase in dose of oral GC.
Propensity score (PS) matching was employed
and paired tests (i.e. McNemar’s) and
multivariate conditional logistic regression
analysis were used. Demographic (age, gender,
race) and clinical (pre-index DMARDs, pain and
glucocorticoid use, RA-related and non-RA
related visits, number of non-study medications,
total utilization cost, and comorbidity index)
characteristics served as covariates in the PS
matching.
Population Studied: Adult (18-63 years) Texas
Medicaid patients diagnosed with RA (ICD-9 CM
code 714.0x) and on ADA, ETN or IFX were
included. The index date was the first
prescription for ADA, ETN or IFX with no
prescription for a biologic in the 6-month preindex period (i.e., biologic naïve). The study
timeframe was July 1, 2003 to Aug 31, 2011.
Patients were grouped based on their index
medication; prescription and medical claims
were analyzed over an 18-month period (6months pre- and 12-months post-index).
Principal Findings: After PS matching, 822
patients were included. The majority (69.2%)
was between 45-63 years, female (88%) and
Hispanic (53.7%). The 3 study groups were
significantly different on only 2 of the 6
effectiveness criteria; adherence and biologic
dose increase. A significantly higher proportion
of IFX users (38.3%) were adherent compared
to ETN (p<0.0001) users (16.4%) and a
significantly higher proportion of ETN users
(98.2%) did not dose escalate compared to ADA
(88.7%; p<0.0001) and IFX (80.3%; p<0.0001)
users. The multivariate analysis indicated no
significant differences in overall effectiveness
among the TNF inhibitors, nor any differences
among the covariates.
Conclusions: Although IFX users were more
likely to be adherent and ETN users were less
likely to dose escalate, when assessed using the
RA medication effectiveness algorithm, the
results showed comparable effectiveness
among ADA, ETN and IFX.
Implications for Policy, Delivery, or Practice:
In absence of direct clinical trial comparisons,
the Curtis algorithm may be a useful tool in
assessing RA medication effectiveness using a
retrospective database. Medicaid should
consider how adherence and dose escalation
impact other outcomes such as health care
costs.
Funding Source(s): Other Amgen
Poster Session and Number: C, #1028
Continuity of Medication Management and
Continuity of Care: Conceptual and
Operational Considerations
Chris Beadles, Department of Veteran Affairs
Medical Center, Durham NC; Matthew
Maciejewski, Department of Veteran Affairs
Medical Center, Durham NC; Matthew Crowley,
Department of Veteran Affairs Medical Center,
Durham NC; Joel Farley, Eshelman School of
Pharmacy, University of North Carolina at
Chapel Hill; Corrine Voils, Department of
Veteran Affairs Medical Center, Durham NC
Presenter: Chris Beadles, Post-doctoral Fellow,
Department of Veteran Affairs Medical Center,
Durham NC
beadles@email.unc.edu
Research Objective: Continuity of care (COC)
is a well-established concept considered a key
element in delivering high quality care and a
centerpiece of Accountable Care Organizations
(ACO) and Patient Centered Medical Homes
(PCMH). Prior research indicates that better
COC, defined as receiving care from a single
provider (or team), is associated with better care
experiences, fewer emergency room visits, and
fewer hospitalizations. Conversely,
discontinuous or fragmented care has been
associated with duplication of services,
medications, diagnostic tests, and procedures.
ACOs and PCMHs have been tasked with
improving care quality and reducing health
expenditures by better coordinating care.
Aggregate improvements may be concentrated
in patients with multiple chronic conditions
(MCC) who often see multiple providers, incur a
disproportionate share of health expenditures,
and require a greater number of medications
than patients with single chronic conditions.
Increasing number of providers and chronic
medications is associated with greater risk of
adverse drug events and drug-drug interactions.
While COC may be useful for characterizing
care quality in general populations, it may be
insufficient for patients with multiple chronic
conditions (MCC) whose interactions with
providers involve management of complex
medication regimens. Continuity of care may be
more relevant to MCC patients if focused on the
subset of providers who prescribe medications
that are essential for managing their chronic
illnesses, which we refer to as continuity of
medication management (COMM). We outline
the conceptual underpinnings of COC and
COMM and summarize the similarities and
differences between COC and COMM. We also
assess whether COC and COMM are empirically
interchangeable in a convenience sample of
patients with cardiometabolic conditions,
summarize the limited empirical evidence to
support COMM, and discuss needed future
research.
Study Design: We utilized a simple
retrospective cohort from a single VA medical
center to compare number of prescribers in
fiscal year to number of providers in a fiscal
year.
Population Studied: Veteran enrollees with one
or more cardiometabolic conditions:
hypertension, diabetes mellitus, hyperlipidemia,
congestive heart failure who had at least one
outpatient encounter with an affiliated provider
and a medication filled for one or more
cardiometabolic conditions.
Principal Findings: Evidence against a 1:1
provider-prescriber relationship was
demonstrated with significantly greater mean
number of providers compared to prescribers.
Conclusions: COC and COMM are
conceptually and empirically distinct constructs
in primary care. COMM may represent a
superior measure of continuity among patients
with MCC requiring medication management.
Implications for Policy, Delivery, or Practice:
COC and COMM non-exchangeability has
implications for future research, policy and
clinical care. Research evaluating associations
between continuity and patient outcomes should
carefully select COC or COMM based on the
population and outcomes studied. For policy,
attribution rules to attribute patients to providers
for purposes of performance assessment and
quality reporting may vary between COC and
COMM based algorithms. While intuitive and
potentially more relevant for MCC patients,
COMM as a measure in care quality requires
future research refining the operational definition
and strength of association with health
outcomes.
Funding Source(s): VA
Poster Session and Number: C, #1029
Technical Assistance Needs to Support Care
Delivery Patients with Bleeding Disorders:
Findings from a National Survey of
Hemophilia Treatment Centers
Jessica Bellinger, University of South Carolina;
Medha Iyer, University of South Carolina
Presenter: Jessica Bellinger, Postdoctoral
Fellow, University of South Carolina
bellingj@mailbox.sc.edu
Research Objective: Bleeding disorders,
including hemophilia A and B, Von Willebrand
disease (VWD), and other genetic or acquired
bleeding disorders affect at least 36,600 people
in the United States. The present care system
for patients with bleeding disorders is
comprehensive and multidisciplinary, particularly
comprehensive care delivered in hemophilia
treatment centers (HTC). Collaborations
between HTCs, Regional Hemophilia Networks
(RHNs), the Maternal and Child Health Bureau
(MCHB), and national consumer organizations
have greatly contributed to the current care
model.
The National Hemophilia Program Coordinating
Center (NHPCC) was created in 2012 to
facilitate and coordinate activities carried out by
the RHNs to optimize the health of this special
population. The research objective was to
identify technical assistance priorities for the
HTCs to support NHPCC implementation.
Study Design: In 2013, we conducted an
electronic survey of HTC staff in all eight RHN
regions in the United States. The electronic
survey included items related to HTC staffing,
active patient population size, treatment
specialties or services directly provided during
comprehensive care clinic at the HTC (or at the
institution of free-standing HTCs), types and
sources of technical assistance received in the
past, and anticipated technical assistance needs
in the future, and preferences for technical
assistance delivery.
Population Studied: The survey was
electronically distributed to staff in 143 HTCs in
the United States. Of the 647 survey attempts,
119 (18.4%) were not completed and not
included in the analysis. Of the 528 complete
surveys, 99.24% of respondents (n=524)
consented to take part in the survey.
Individual-level analyses were conducted in
STATA 11. Chi-square tests were used to
identify difference.
Principal Findings: Survey respondents were
represented in each of the eight HRSA regions.
The primary HTC staff role of the respondents
was examined to understand the context of their
work in caring for patients with hemophilia and
blood disorders. Nurses including RNs, LPNs,
and NPs (n=173; 33.02%), hematologists (n=91;
17.37%), and social workers (n=86; 16.41%)
were the top three survey respondents. Over a
third of respondents (34.16%) were staff in
hemophilia treatment centers with 101-250
patients.
RHNs were cited as a source of technical
assistance (TA) by 75.38% of respondents. The
top five TA needs identified by the respondents
were: health record data management (n=315),
staff development (n=314), 340B pharmacy
(n=307), lifespan transition (n=261), and
geographic access and 340B pharmacy (n=255).
Conclusions: HTC staff were receptive to
receiving technical assistance from the NHPCC
and their RHNs to support coordinated care
delivery to patients with bleeding disorders.
Priorities were identified on the patient-level
(electronic health records) and the practice-level
(staff development).
Implications for Policy, Delivery, or Practice:
The findings of the survey will help facilitate the
evolution of the eight HRSA-funded regional
networks into a cohesive and high performing
national network that works in concert with
stakeholders to improve access to care, quality
of care delivery, and clinical and patient reported
care outcomes.
Funding Source(s): HRSA
Poster Session and Number: C, #1030
When Complex Care Goes Complementary:
Closing the Loop on Integrated Care for
Children with Special Health Care Needs
Christina Bethell, Oregon Health and Science
University; Narangel Gombojav, Child and
Adolescent Health Measurement Initiative,
Oregon Health and Science University; Stephen
J. Blumberg, Center for Disease Control and
Prevention, National Center for Health Statistics;
Adam Carle, University of Cincinnati School of
Medicine; John Neff, Children's Hospital and
Regional Medical Center; Paul Newacheck,
Institute of Health Policy Studies, University of
California at San Francisco; Thomas K. Koch,
Division of Pediatric Neurology, Oregon Health
& Science University
Presenter: Christina Bethell, Professor, Oregon
Health and Science University
bethellc@ohsu.edu
Research Objective: Children with complex
health problems, particularly children with
special health care needs (CSHCN), have
higher medical expenditures, experience more
barriers to care and gaps in quality of medical
care than other children. Less is known about
population-based associations between
complementary and alternative medicine (CAM)
use and conventional medical care (CMC)
experiences for CSHCN. This study assesses
patterns of CAM use for children with chronic
health conditions and CSHCN and associations
with conventional medical care utilization,
expenditures, access and quality of care.
Study Design: Data from the most recent
available national data on CAM use and CMC
expenditures and quality of care among CSHCN
are used. This data file links the 2007 National
Health Interview Survey (NHIS) and 2008
Medical Expenditure Panel Survey data at the
child level. The most recent national survey
data on CAM (2012 NHIS) is also used. The
Bivariate and logistic and two-part regression
analyses were employed.
Population Studied: Children age 0-17 years
included in the National Health Interview Survey
and the Medical Expenditure Panel Survey. All
estimates are adjusted for non-response bias
and weighted to represent the noninstitutionalized population age 0-17.
Principal Findings: CAM use was more
prevalent among CSHCN (AOR 1.89, 24.7%)
and children with multiple chronic conditions (2
CC: AOR 2.98, 24.9; 3+ CC, AOR 5.42, 37.9%).
CAM use is higher among more complex
CSHCN who meet criteria for the Affordable
Care Act Diagnostic Condition List (ACA Dx,
AOR 4.18, 42.2%) or experience difficulties in
daily activities or functioning (AOR 4.19, 36.4%).
Regardless of clinical condition group CAM use
was found to be significantly higher among
children with anxiety/stress (Rate Ratio ranging
from 1.44 to 2.67 across condition groups) and
children who missed two or more weeks of
school (Rate Ratio ranging from 1.58 to 3.26).
Conclusions: CAM use is associated with the
complexity and intensity of children’s health
conditions and service needs, difficulties in
accessing and poorer quality of CMC.
Implications for Policy, Delivery, or Practice:
Children with complex health problems receive
multiple forms of conventional, complementary
and alternative care, emphasizing the need for
well integrated and coordinated pediatric care
systems within the context of a medical home.
Funding Source(s): NIH
Poster Session and Number: C, #1031
Value-Based Insurance Design: Falling Short
for Those with Multiple Chronic Conditions
Christine Buttorff, Johns Hopkins Bloomberg
School of Public Health
Presenter: Christine Buttorff, Phd Candidate,
Johns Hopkins Bloomberg School of Public
Health
cbuttorf@jhsph.edu
Research Objective: Value-based insurance
designs change the level of cost sharing to
promote services perceived to be high value
from the insurer or policy maker’s perspective.
However, it is unclear how people with multiple
chronic conditions react to value based
insurance design because they may not be
willing or able to switch to lower cost
alternatives. The behavior of these individuals is
important because they are the heaviest
consumers of medical services and may be
more susceptible to short term complications
from poorly managed conditions. The objective
of this paper evaluates how adults with groups
of multiple chronic conditions respond to a
value-based insurance design change that
increased copayments on brand name drugs
while decreasing copayments on generics.
Study Design: A regression discontinuity
design exploits a 2010 co-pay change that
raised copayments on preferred brands, nonpreferred brands and specialty medications
while decreasing generic copayments.
Outcomes included total medical and drug
spending, as well as the use of outpatient,
emergency department and inpatient services.
Population Studied: Data consists of drug and
medical claims from Maryland’s high-risk pool
for the years 2007-2011. High-risk pools offer
insurance to those with preexisting conditions
who were denied coverage on the individual
market and who do not have access to
employer-based insurance. The sample was
restricted to those aged 18-64 and who were
continuously enrolled for one-year period before
and after the policy change.
Principal Findings: The copayment policy
change has a statistically significant impact on
those with increasing numbers of chronic
conditions, but the magnitudes are small. The
use of both brand and generic drugs increased
less than 3% across all numbers of chronic
conditions after the policy change, while the
overall number of fills decreased less than 1%.
This varies little across the different numbers of
chronic conditions despite the significantly
higher copays paid by those with multiple
chronic conditions.
The copay change does appear to have
impacted the use of other services: outpatient
visits dropped by 0.17 visits per month while the
probability of an ED visit increased 65% for
those with 10+ chronic conditions. However,
these could also reflect temporary changes in
service utilization at the start of a new plan year.
Inpatient visits, including those for ambulatory
care sensitive conditions and 30-day
readmissions, decreased after the policy
change. While prescription drug costs saw
barely any change, medical costs, and therefore
total spending, decreased 30% in the period
after the policy change for those with 10+
chronic conditions, due primarily to the change
in inpatient admissions.
Conclusions: This study finds little impact on
the use of prescription drugs after a value-based
insurance design initiative on those with multiple
morbidities. More research is needed on
working-aged adults with multiple chronic
conditions and how they respond to cost sharing
in order to confirm results.
Implications for Policy, Delivery, or Practice:
Other behavioral incentives than cost sharing
may be needed to manage the health of those
with multiple chronic conditions.
Funding Source(s): Other Jayne Koskinas Ted
Giovanis Foundation for Health and Policy
Poster Session and Number: C, #1032
Health Risk Behaviors of Medicaid
Recipients Diagnosed with Chronic Mental
and Physical Illness
Kim Case, University of Florida; Kimberly Case,
University of Florida, Institute for Child Health
Policy; Dena Stoner, Texas Department of State
Health Services (DSHS); Keith Muller, University
of Florida, Institute for Child Health Policy;
Martin Wegman, University of Florida, Institute
for Child Health Policy; Jill Herndon, University
of Florida, Institute for Child Health Policy;
Elizabeth Shenkman, University of Florida,
Institute for Child Health Policy
Presenter: Kim Case, Research Scientist,
University of Florida
kim.case@ufl.edu
Research Objective: Medicaid enrollees with
co-occurring physical and mental health
conditions often engage in risky health
behaviors, including tobacco use, poor nutrition,
and sedentary behavior. Risk behaviors tend to
occur in clusters and can contribute to poor
health status and increased costs. The aim of
this study was to examine how health risk
behaviors cluster among individuals in Medicaid
with co-occurring physical and behavioral health
conditions.
Study Design: As part of a longitudinal
randomized trial examining the benefits of health
navigation and financial support on health risk
behaviors, baseline data from 1,569 Medicaid
enrollees with co-occurring physical and mental
health diagnoses was examined. All enrollees
participated in an extensive survey on current
health risk behaviors, health status (SF-12-V2R)
and subjective cognitive status (MOS-CognitiveR). Health risks included excessive alcohol use,
concerns about weight, tobacco use, sedentary
behaviors, emotional stress, poor eating habits,
and pain.
Population Studied: Medicaid claims and
enrollment data were utilized to group
participants into three diagnostic categories:
Chronic Behavioral health illness + Chronic
Physical health illness (BH+PH; n=533), Serious
mental illness (SMI; n=470), and a group
comprising all three diagnoses (BH+PH+SMI;
n=566). All enrollees were between the ages of
21 to 61 and enrolled in the STAR+PLUS
Medicaid program in Texas. Enrollees with
intellectual or cognitive diagnoses were
excluded.
Principal Findings: Preliminary analyses of the
full sample found that emotional stress (96%),
poor eating habits (79%), sedentary behavior
(75%), and smoking (55%) were the most
common health risks reported. Excessive
alcohol use was reported by 11% of the sample.
BH+PH+SMI enrollees had significantly higher
health risk rates in weight, smoking and
sedentary behaviors as compared to the other
two diagnostic groups. Pain was reported
significantly more in enrollees with chronic
physical health diagnoses. Most enrollees
reported 3-5 (72%) or 6-7 (25%) health risks.
Enrollees who reported a higher number of
health risks also reported lower subjective
cognitive functioning scores. Older participants
reported significantly more health risks
compared to the younger participants, with no
significant differences between genders. An
exploratory factor analysis, using a promax
rotation to account for correlations among health
risks, supported two factors. The first factor
included sedentary behaviors, emotional stress,
and pain (psycho-social health risks). The
second factor included smoking and excessive
drinking.
Conclusions: Enrollees report multiple health
risk behaviors which can have a long-term
impact on severity and prevalence of chronic
disease. Enrollees presenting with all three
diagnosis categories had the highest number of
health risks.
Implications for Policy, Delivery, or Practice:
The health risk clusters identified can assist
policy makers and providers in structuring
interventions to incorporate several different
health topics within one intervention program.
For example, programs designed to increase
physical activity in enrollees could also provide
information on techniques for pain and stress
reduction. By tailoring the intervention to the
population, there could be a larger impact
across several areas of health. Both clinical and
public health practitioners can utilize this
information to guide efforts in addressing
multiple risk behaviors in the context of chronic
physical and mental illness.
Funding Source(s): CMS
Poster Session and Number: C, #1033
Improving Management of Chronic
Conditions Using Telehealth: A Mixed
Method Evaluation of a Multisite Intervention
in New York
Linda Weiss, The New York Academy of
Medicine; Tongtan Chantarat, The New York
Academy of Medicine; Ebele Benjamin-Gardner,
The New York Academy of Medicine; Brenda
Bartock, Visiting Nurse Service of Rochester
and Monroe Co., Inc.; Victoria Hines, Visiting
Nurse Service of Rochester and Monroe Co.,
Inc.; Sharon Legette-Sobers, Greater Rochester
Health Foundation
Presenter: Tongtan Chantarat, Research
Analyst, The New York Academy of Medicine
tchantarat@nyam.org
Research Objective: Telehealth services offer
promise for patients with chronic conditions who
require enhanced monitoring and support.
Although telehealth has been associated with
improved disease self-management, improved
health, and reduced healthcare use, more
information is needed to optimize
implementation and outcomes.
Study Design: We conducted a mixed-method
evaluation of a 3-arm telehealth intervention
from 2011-2013. Eligibility criteria and service
model differed somewhat by arm. A
convenience sample of patients completed
baseline, discharge and follow-up surveys to
assess health status, service utilization, disease
self-management, and perceptions of the
program. A subset of patients and providers
participated in qualitative interviews that probed
more deeply into selected topics. Claims data
were analyzed to compare healthcare utilization
and cost before, during and after the program,
and to compare the intervention group to
matched controls. Data from the telehealth
devices were studied to identify trends in
frequency of out-of-range alerts.
Population Studied: Patients with
hypertension, diabetes, congestive heart failure,
or chronic obstructive pulmonary disease were
eligible for telehealth services. Patient selection
procedures and eligibility criteria differed by arm
and included high health care use, low health
literacy, poor disease self-management, and/or
being homebound.
Principal Findings: The majority of participants
were female (52%), high school educated or
less (69%), and had hypertension (90%).
Seventy-four percent had more than one chronic
condition. In the 6 months prior to enrollment,
71% had ER visits and 59% had
hospitalizations. Survey, claims and interview
data indicated improved disease management
and some reductions in healthcare use during
telehealth enrollment. However, claims data
suggest a rebound in healthcare use postdischarge. Findings also suggested significant
differences in outcomes by patient
characteristics, including age and illness.
Targeting the service to the highest risk patients
based on prior service use yielded more
immediate reductions in health service utilization
and costs. However, our data also suggested
that utilizing telehealth as an early intervention
may result in better disease management and
more healthy behaviors, with greater long term
implications. Telehealth device data over a 12week period showed a significant reduction in
the number of out-of-range alerts per week, with
the most consistent reductions among
hypertension patients. High satisfaction with
telehealth was reported and many wanted the
service to last longer. Satisfaction among
providers was higher in the arm with an on-site
telehealth nurse, because the nurse was more
familiar with patient needs and communication
between the patient, provider and nurse was
more consistent.
Conclusions: Our findings indicate that
telehealth results in improved disease selfmanagement, positive behavior changes, and
reduced health care use and cost during the
program period. Patients had very positive
perceptions of the service.
Implications for Policy, Delivery, or Practice:
Our data suggest the utility of telehealth, as well
as the need for continued efforts to identify
patients that benefit most from the service;
optimal duration of the intervention; and most
appropriate model for service delivery.
Funding Source(s): Other Greater Rochester
Health Foundation
Poster Session and Number: C, #1034
“Sugar” by Any Other Name: How Patients
and Providers Talk and Fail to Talk about
Chronic Conditions
Ellis Dillon, Palo Alto Medical Foundation;
Shana D. Hughes, Palo Alto Medical Foundation
Research Institute; Mary Carol Mazza, Palo Alto
Medical Foundation Research Institute; Ming
Tai-Seale, Palo Alto Medical Foundation
Research Institute
Presenter: Ellis Dillon, Post-doctoral Fellow,
Palo Alto Medical Foundation
ellis.c.dillon@gmail.com
Research Objective: This research examines
how providers and patients with diabetes and/or
hypertension talk about the condition for which
patients are seeking treatment. We explore
language use concerning diagnosis and
prognosis, which has the potential to influence
patient understanding and decision-making.
Study Design: Findings are drawn from the
qualitative component of a mixed methods study
of primary care for patients with chronic
conditions including diabetes and hypertension.
The study was conducted at a large, non-profit,
multispecialty group practice in northern
California which serves four counties and more
than 850,000 patients.
Population Studied: Observation of 12 primary
care appointments, including 12 patients with
chronic conditions and 9 physicians.
Principal Findings: Patients and providers
frequently use colloquial names like “sugars” or
“pressure” instead of biomedical terms for
chronic conditions. In addition, talk about
prognosis and risk often is absent, limited, or
downplays possible adverse outcomes. Further,
providers sometimes use language in ways that
seem intended to build rapport, (e.g., “We need
a colonoscopy”) rather than prioritizing the
accurate and detailed exchange of clinical
information. These linguistic patterns occur even
in appointments where one might expect more
direct and comprehensive discussion.
Conclusions: Certain patients and providers
talk about diagnosis and prognosis in colloquial
terms and focus on immediate consequences
(e.g. changes in medications and lab values)
rather than long-term health goals and possible
outcomes. Further research is needed to
understand the reasons for these linguistic
patterns and their effects on patients.
Implications for Policy, Delivery, or Practice:
Our findings highlight particular situations for
which physicians might benefit from instruction
on communication strategies that accommodate
both rapport-building and clinical precision.
Funding Source(s): Other Gordon and Betty
Moore Foundation
Poster Session and Number: C, #1035
Effects of Newly Diagnosed Cancer on
Medication Management of Prevalent
Diabetes
Mujde Erten, University of Vermont; Bruce
Stuart, University of Maryland Baltimore, School
of Pharmacy; Amy Davidoff, Agency for
Healthcare Research and Quality
Presenter: Mujde Erten, Assistant Professor,
University of Vermont
merten@med.uvm.edu
Research Objective: A new cancer diagnosis
often initiates a cascade of healthcare decisions
related to provider, cancer treatment, and site of
care, that have important consequences for
management of other chronic conditions such as
diabetes. Medication adherence may decline
after a cancer diagnosis if the prognosis is poor,
or due to the diversion of focus, time, and
financial resources away from management of
comorbid conditions. The objective of this study
is to evaluate drug therapy for diabetes before
and after a new cancer diagnosis among
Medicare beneficiaries, testing for effects of poor
prognosis and financial crowdout.
Study Design: This retrospective study
examined changes in adherence to three drug
classes (oral hypoglycemic agents, RAASinhibitors, and statins) among Medicare
beneficiaries with diabetes, comparing those
with and without a new cancer diagnosis who
were observed for at least 13 months.
Adherence was measured as the percentage of
days covered (PDC) for each drug class, within
6-month pre-index and post-index periods
(POST), with index date defined as the month of
cancer diagnosis (cancer cohort) or an artificial
index date assigned to the comparison group.
Good prognosis was assigned for beneficiaries
who survived at least 6 additional months
beyond the observation period (GOODPROG).
Low income subsidy (LIS) receipt was used as a
proxy for out-of-pocket price of cancer treatment
and other medication therapy. We estimated
difference-in-difference-in-differences (DDD)
equations, with interactions between POST,
CANCER, and either GOODPROG or LIS, with
controls for demographics and selected chronic
conditions.
Population Studied: Medicare beneficiaries
with diabetes were identified from the 2007-2008
Chronic Condition Warehouse (CCW) 5%
enrollment sample, using a CCW “diabetes
ever”-flag. Beneficiaries with ICD-9 diagnosis
codes of cancer (one inpatient or two outpatient
claims) during January through December 2007
after a prior 12-months “washout” period were
identified as newly diagnosed cancer patients
(N=4,457). The comparison group represented
beneficiaries with diabetes who had no evidence
of cancer during the study period (N=29,187).
Our sample was restricted to beneficiaries
continuously enrolled in stand-alone prescription
drug plans (PDPs) and beneficiaries who
survived at least 6-months after their index
dates.
Principal Findings: Both unadjusted and
adjusted estimates indicate PDC declines prePOST for all three medication classes, with
substantially larger declines for CANCER. The
DDD results suggest that GOODPROG was
associated with smaller PDC declines, with
larger interaction effects for CANCER. More
specifically, in all three drug classes we found
small declines in medication adherence for
cancer survivors relative to beneficiaries without
a cancer diagnosis (3-5 percentage points; for
all three p<0.001), whereas longer term cancer
survivors had much better adherence to all three
drug classes (10-12 percentage points higher;
for all three p<0.001) relative to beneficiaries
with cancer who had a poor prognosis. No
effects were found for LIS.
Conclusions: A diagnosis of cancer among
Medicare beneficiaries with diabetes significantly
reduces adherence with evidence-based
medications recommended in diabetes
treatment guidelines. Poor cancer prognosis is
an important factor, but does not account for the
full decline.
Implications for Policy, Delivery, or Practice:
Improved care coordination for Medicare
beneficiaries diagnosed with cancer will likely
improve adherence to medications for preexisting chronic conditions.
Funding Source(s): Other American Cancer
Society
Poster Session and Number: C, #1036
Primary Care Physicians’ Use of Behavior
Change Counseling with Chronic Condition
Patients
Daniel Fulford, Palo Alto Medical Foundation;
Ellis Dillon, Palo Alto Medical Foundation
Research Institute; Shana Hughes, Palo Alto
Medical Foundation Research Institute; Mary
Carol Mazza, Palo Alto Medical Foundation
Research Institute; Laura Panattoni, Palo Alto
Medical Foundation Research Institute; Ming
Tai-Seale, Palo Alto Medical Foundation
Research Institute
Presenter: Daniel Fulford, Research
Psychologist, Palo Alto Medical Foundation
fulfordd@pamfri.org
Research Objective: The long-term course of
chronic conditions can be vastly improved by
self-management behaviors, including
adherence to treatment and making sustained
lifestyle changes. Increasingly, primary care
physicians (PCPs) are expected to engage
patients with chronic conditions and activate
them for self-management during office visits.
Behavior Change Counseling (BCC), an
adaptation of Motivational Interviewing suitable
for primary care settings, encourages physicians
to explore the patient’s thoughts, feelings, and
plans for change. Although there is evidence
that patients of physicians trained in BCC show
improved health outcomes, little is known about
the use of BCC among physicians not
specifically trained in this technique. We
hypothesized that PCPs would show varied BCC
skills during visits for chronic conditions, which
could be related to both provider differences in
communication style, as well as applicability of
specific visits to behavior change topics.
Study Design: Observational study of audiorecorded primary care office visits. We scored
recordings of physicians’ discussion of behavior
change during primary care visits using the
Behavior Change Counseling Index (BECCI), a
widely used measure in healthcare settings.
Scores range from 0 (“not at all”) to 4 (“a great
extent”). An average score for each visit is
calculated based on 11 items. Four doctorallevel health services researchers achieved
consensus on three recordings, with the
remaining scored by two of these raters.
Population Studied: Nine physicians in Family
Medicine and Internal Medicine departments in a
group medical practice in Northern California.
Principal Findings: Mean BECCI scores
ranged from 0.50 (“not at all” to “minimal” use of
BCC) to 1.73 (“minimal” to “to some extent”),
with a mean (SD) score across physicians of
1.10 (0.36), indicating “minimal” use of behavior
change counseling.
Conclusions: PCPs exhibited varied BCC skills.
Overall, scores were low, which could indicate
either less developed skills in behavior change
counseling, or could simply indicate less
opportunity for behavior change discussion in
these appointments (e.g., the patient in the
recorded visit was doing well in selfmanagement and thus behavior change was not
a topic of importance, or there were acute issues
to be addressed).
Implications for Policy, Delivery, or Practice:
The need to communicate effectively with
patients about lifestyle change is a growing
pressure for practitioners. The BECCI may be
useful in identifying PCPs’ use of BCC in their
care of chronic conditions; however, the scale’s
applicability in this context is limited due to the
variation in content of visits. As chronic
conditions requiring lifestyle changes are
becoming more common, it is crucial to explore
how PCPs can motivate and assist patients to
make important changes. Training in the use of
BCC may be a valuable tool for improving
patient activation and health outcomes.
Funding Source(s): Other Moore Foundation
Poster Session and Number: C, #1037
Incorporating the Use of Functional Health
Status Within a Diagnosis Driven Clinical
Risk Adjustment Model
Richard Fuller, 3M HIS; Norbert Goldfield, 3M
HIS; Jack Hughes, Yale University School of
Medicine
Presenter: Richard Fuller, Medical Director, 3M
HIS
rfuller@mmm.com
Research Objective: Measures of functional
status may be at least as useful for predicting
costs of care as clinical variables such as
diagnosis codes, but have not been routinely
incorporated within measures of clinical risk
adjustment.
Study Design: Claims based analysis
Population Studied: Using a 5% development
sample of Medicare fee-for -service (FFS)
claims from 2006 and 2007 that included
functional status assessments, we created nine
functional groups defined by the interaction of
self-care, mobility, incontinence and cognitive
impairment. We then applied the nine functional
groups to a validation set of 100% Medicare
FFS data for 2010 and 2011. We specified a
regression model linking Clinical Risk Groups
(CRGs), a diagnosis-based patient classification
model, to an augmented model including the
functional groups.
Principal Findings: We made the linkage
between the two classifications by computing a
baseline CRG adjusted resource prediction for
enrollees without assessment data and
introduced this prediction into the augmented
classification for enrollees with assessment
data. Coefficients obtained from the resulting
regression model demonstrated significant
effects associated with functional status, with
similar coefficients obtained from both the
development and validation data.
Conclusions: Use of the functional groups
improved fit, measured by the R2 statistic, by 5
percent across all Medicare enrollees. The
individual domains of functional health status
demonstrate a non-linear relationship with
increasing resource use.
Implications for Policy, Delivery, or Practice:
The interaction of functional domains can be
used to create stable adjustments suitable for
integration with the underlying risk adjustment
model while continuing to provide classification
and aligned payment for enrollees that did not
require a functional assessment.Funding
Source(s): Other MedPAC
Poster Session and Number: C, #1038
Population Based Study of Use of Opioid
Analgesics by Persons with Diabetes
mellitus and Differences by Gender
Santosh Gautam, University of Texas Health
Science Center at Houston; Luisa Franzini,
Divison of Management, Policy and Community
Health, School of Public Health, University of
Texas Health Science Center at Houston;
Barbara Turner, REACH Center, University of
Texas Health Science Center at San Antonio
Presenter: Santosh Gautam, Graduate
Assistant, University of Texas Health Science
Center at Houston
santosh.gautam@uth.tmc.edu
Research Objective: To examine the use of
opioid analgesics (OAs) therapy in a population
based cohort of privately insured persons with
diabetes mellitus and to examine characteristics
of men and women taking these medications.
Study Design: The study used Blue Cross Blue
Shield of Texas (BCBSTX) claims database
from 1/ 2008 to 12/2011. We initially identified
278,348 individuals diagnosed with diabetes in
at least one inpatient or several outpatient
encounters. Of these, we identified persons age
18-64 and enrolled for 12 or more continuous
months in a PPO/PPO+ plan including drug
coverage. Exclusion criteria included: non-basal
cell cancer diagnosis or drug-dependence
diagnosis. We created variables for
demographics and clinical indicators for 5
categories of chronic non-cancer pain (CNCP),
mental health conditions, and alcohol abuse. We
created an indicator for OA treated persons
prescribed Schedule II/III OAs in a pill form. For
the entire study cohort, we also created
indicators for any prescription for each of three
categories of other drugs used for CNCP
including: benzodiazepines, zolpidem, and
antidepressants. Among OA users, we
examined OA therapy and other drugs used for
CNCP in the first 6 months following the first
filled OA prescription. For OA therapy, we
aggregated the number of opioid pills received
into less than 120 or 120 or more short acting
OAs (4/day x 30 days) as short or longer
duration. We also created a flag for receipt of
any long-acting OAs. Differences between the
groups were assessed using the chi-square test
for categorical variables and t-test for continuous
variable.
Population Studied: A retrospective cohort of
privately insured individuals in Texas diagnosed
with diabetes (N=51,656), divided into two
groups of OA-users (n=21,515) and non OAusers (n=30,141).
Principal Findings: Among persons with
diabetes, OA users were significantly more likely
to be women (47.2 vs 39.1 percent of non OAusers) and the proportions of OA users with
other clinical conditions were usually twice as
high as in non-OA users including: five CNCP
conditions, depression, anxiety, PTSD,
psychotic disorder, and alcohol abuse. Use of
other drugs for CNCP was over three times
higher in the OA-users than non-OA users.
Among the subset of OA-users, women were
younger (49.6 vs 51.1 yrs), and more likely
(P<0.01) to be diagnosed with a CNCP condition
and mental health disorder but less likely to
have an alcohol abuse diagnosis. In both
genders, 15 percent of OA-user received long
duration prescriptions and 1 percent received
long-acting opioids but women were more likely
to receive other drugs for CNCP.
Conclusions: In a cohort with diabetes, women
were more likely to be prescribed OAs than men
and persons treated with OAs had more CNCP,
mental health, and alcohol abuse comorbidities
as well as treatment with other drugs for CNCP.
However, among OA users, women had similar
OA treatment patterns but significantly more
comorbidities and more use of other drugs for
CNCP.
Implications for Policy, Delivery, or Practice:
Women with diabetes appear to be more likely
to be prescribed OAs and among users, more
likely to receive other potentially addictive drugs
than men.
Funding Source(s): N/A
Poster Session and Number: C, #1039
Chronic Disease Self-Management Programs
May Reduce Hospitalization Costs in
Medicare Beneficiaries
Ekta Ghimire, Acumen LLC; Daniella Perlroth,
Acumen LLC; Grecia Marrufo, The Lewin Group;
Emil Rusev, Acumen LLC; Benjamin Howell,
CMS; Erin Murphy, CMS; Catherine Lewis, The
SPHERE Institute; Michael Packard,
Georgetown University
Presenter: Ekta Ghimire, Research Analyst,
Acumen LLC
eghimire@acumenllc.com
Research Objective: Community-based
programs that help individuals manage their
chronic conditions may also help curb Medicare
spending. We used Medicare claims to evaluate
the effects of participating in the nationally
disseminated Chronic Disease SelfManagement Program (CDSMP) on health
service use and costs of Medicare beneficiaries.
Prior cost-effectiveness studies of CDSMP
showed promising results but used self-reported
data on a relatively small number of individuals,
not all of whom were Medicare beneficiaries.
Study Design: We conducted a retrospective
matched cohort study. Applying a difference-indifferences estimation method, we estimated the
differences in outcome changes between
CDSMP participants and matched controls in the
one-year period following each participant's
program start date, relative to the one-year
period prior. We also conducted sensitivity
analyses based on participation levels and
survival status in the year after enrollment.
Population Studied: Cohort included 13,329
individuals (average age: 74.4, 80-percent
female) located in various US states who
enrolled in CDSMP anytime between 2010 and
2011, and were continuously receiving Medicare
fee-for-service benefits throughout the study
period. Individuals who were long-term
institutionalized or receiving hospice and end
stage renal disease care were excluded.
Principal Findings: CDSMP enrollment was
associated with unplanned hospitalization cost
savings of 245 USD per person (95-percent CI:
52, 437) at 1 year but not with statistically
significant reductions in average number of
hospitalizations. CDSMP enrollment was
associated with an increase in ER costs of 27
USD per person (95-percent CI: 9, 46), a
corresponding increase of 0.03 ER visits per
person (95-percent CI: 0.01, 0.06); and an
increase of 0.41 physician visits per person (95percent CI: 0.31, 0.51). Total cost savings were
not detected.
We did not find hospital cost savings in the
subpopulation analysis on survivors. CDSMP
enrollment was however associated with a
statistically significant increase in physician and
non-institutional service costs of 87 USD per
person (95-percent CI: 7, 167) at 1 year for
survivors. Other results for survivors were
similar to those in the main analysis.
Completion of a CDSMP workshop (attendance
of all 6 sessions) was associated with total cost
savings of 944 USD per person (95-percent CI:
383, 1506) at 1 year. This effect was sustained
but smaller for survivors.
Conclusions: CDSMP participation could
reduce hospital costs for Medicare beneficiaries.
While we did not find evidence of total cost
savings within the first year, we found that
workshop completion may be particularly
important in achieving net savings. The
program may also activate participants to seek
medical attention for previously unmet medical
needs in physician offices or the ER.
Our retrospective study is subject to selection
bias. Participants and controls are likely to have
health or behavioral differences that influenced
their enrollment in CDSMP and also
independently influenced outcomes. Our
matching variables based on Medicare data may
not have adequately captured such differences.
Implications for Policy, Delivery, or Practice:
Further research is needed to assess whether
observed near-term effects on hospital costs
could translate to net Medicare savings beyond
the first year. A prospective study could help
disentangle selection bias by allowing the
identification of a refined set of variables to
model program enrollment.
Funding Source(s): CMS
Poster Session and Number: C, #1040
A Qualitative Assessment of Clinical Review
To Improve Risk Stratification: Why and How
Do Clinicians Risk Select
Vivian Haime, Partners HealthCare; Clemens
Hong, Massachusetts General Hospital; Laura
Mandel, Partners HealthCare; Jessica
Moschella, Partners HealthCare; Namita Mohta,
Partners HealthCare; Christine Vogeli, Harvard
Medical School
Presenter: Vivian Haime, Project Manager,
Partners HealthCare
vhaime@partners.org
Research Objective: As one of thirteen
successful year-one Pioneer ACOs, the Partners
Healthcare delivery system relied in large part
on its Integrated Care Management Program
(iCMP) to better coordinate care for its Medicare
and commercial complex patients. This primary
care-integrated complex care management
program uses a hybrid approach to select
patients for intervention. Primary care
physicians (PCPs) and their respective care
managers (CMs) are asked to review a list of
patients identified by a claims-based algorithm in
order to identify a final set of patients to target
for intervention. We sought to understand the
criteria PCPs/CMs used when selecting iCMP
patients to assess the added value of provider
input.
Study Design: We used a semi-structured
interview guide designed to obtain information
on patient selection processes and criteria.
Interview responses were open coded by two
independent reviewers through an iterative
process, and analyzed using NVivo. The
Andersen Behavioral Model of Health Services
Use provided a theoretical base for the coding
scheme.
Population Studied: A convenience sample of
20 PCP and CM dyads stratified by practice size
and region.
Principal Findings: Interviewees considered
issues in three main areas when selecting
complex patients for the iCMP: health system,
social/environmental, and patient level factors.
PCPs and CMs assessed how well they felt the
existing health system met the patients’ care
needs, and whether the iCMP had sufficient
resources and capacity to meet patients’ unmet
needs. Interviewees noted that they might
exclude patients whose specialty care teams
(e.g., psychiatry, oncology or transplant)
comprehensively met the needs of the patients
and their families. Interviewees also looked for
factors that predispose patients to inappropriate
healthcare utilization and limit patients’ abilities
to manage their conditions. These included
patient predisposing characteristics (e.g.
difficulties with health literacy or navigation;
vulnerability related to older age, frailty, and
mobility issues; or personality traits or cognitive
factors that decrease executive function) and
social and environmental enabling factors (e.g.,
absence of social supports, issues at home, or
barriers to care). Interviewees also considered
the patients’ medical need (e.g. severity of
disease, unstable disease, and medical
complexity as defined by multiple comorbidities
and complicated medication regimens).
Interviewees noted that for many patients, it is
often the interplay between underlying mental
health issues and medical need that complicates
medical treatment and management. Often, a
combination of these factors were thought to
contribute to problematic health behaviors such
as frequent or inappropriate utilization (including
both acute care utilization and practice-level
utilization, such as frequent visits or calls) and
poor adherence to treatment regimens and/or
scheduled follow-up visits.
Conclusions: In selecting patients for a
complex care management program, primary
care physicians and care managers assessed
areas of patient complexity that are not easily
captured through administrative data, most
notably the impact of underlying mental health
components and the outside social environment
on patients’ ability to manage their health
conditions.
Implications for Policy, Delivery, or Practice:
A better understanding of the role and value of
provider input in selecting the most appropriate
and actionable complex patients for care
management will increase the opportunity to
improve care and reduce cost for our most
vulnerable patients.
Funding Source(s): No Funding
Poster Session and Number: C, #1041
The Influence of Chronic Conditions on SelfReported Health Status in the Elderly
Barbara Guerard, Peoples Health; Vincent
Omachonu, University of Miami; Raymond
Harvey, Peoples Health; Bisakha Sen,
University of Alabama Birmingham
Presenter: Barbara Guerard, Peoples Health
Barbara.Guerard@peopleshealth.com
Research Objective: The impact chronic
conditions have on the Medicare Advantage
plan populations’ perception of their health is an
important area for continued research. As the
population ages and survey data are relied upon
as key elements in policy design, Medicare
Advantage plans are keenly aware that self-
reported health status on surveys significantly
impacts an organizations’ Star quality rating as
well their major strategic business decisions and
care models.
This study explores the relationship between
participant’s survey responses to questions
adapted from the Consumer Assessment of
Health Providers and Systems Survey (CAHPS)
and Health Outcomes Survey (HOS); reflective
of their overall physical and mental health and
the following chronic conditions: diabetes,
decubitus, congestive heart failure, coronary
artery disease, chronic kidney disease,
cerebrovascular disease, chronic obstructive
pulmonary disease, arthritis and dementia. The
purpose is to demonstrate the influences of
chronic conditions on self-reported mental health
and physical well-being within a Medicare
Advantage population.
Study Design: A simple random sample of plan
participants from provider practices with 200
hundred or more plan participants was used to
mail a total of five thousand surveys to mutually
exclusive members in September 2011, October
2011, and January 2012.
Multinomial regression analysis was used to
explore the relationship between self reported
physical and mental health and specific chronic
conditions adjusting also for age, gender, and
race.
Population Studied: A Medicare Advantage
plan in southeastern Louisiana.
Principal Findings: Thirty-two percent of the
surveys were returned of which 51 did not meet
eligibility criteria or had responses out of range,
yielding a final sample of 4,325 eligible
respondents. Six chronic conditions; dementia,
diabetes, congestive heart failure,
cerebrovascular disease, coronary artery
disease, and rheumatoid arthritis were
significant predictors of the likelihood of a selfreported status of worsening overall physical
health. Four chronic conditions; dementia,
diabetes, coronary artery disease, and
decubitus, were predictors for the likelihood of a
self-reported status of worsening overall mental
health. Results also demonstrated that it was
less likely that non-minority respondents, as
compared to minority respondents, would selfreport their health as better versus about the
same while respondents over the age of 65 were
more likely to report their physical health as
worsening.
Conclusions: Findings from this study provide
insight into the influence of chronic conditions on
the assessment and rating of one’s perceived
physical health and mental well-being in a
Medicare Advantage plan population. Multiple
studies have indicated that there are a number
of factors that relate to one’s self-perception of
health and functional status. Individuals with
chronic conditions have a long period of time to
develop a conceptual framework of their
conditions through their own experiences as well
as the experiences of others with similar
conditions.
Implications for Policy, Delivery, or Practice:
Enrollees with multiple chronic conditions, as
well as health plans, may benefit from a more
structured integration of members into the health
plan to better identify the future need for
resources. This could facilitate the design of
specific models of care and business policies
aimed at improving the overall health outcomes
for health plan members.
Funding Source(s): No Funding
Poster Session and Number: C, #1042
Examining Telehealth Applications for
Evaluation and Treatment of Veterans with
possible Mild TBI
Rachel Martinez, Edward Hines, Jr. VA
Hospital;Timothy Hogan, Edith Nourse Rogers
Memorial VA Medical Center; Keshonna Lones,
Edward Hines, Jr. VA Hospital; Salva Balbale,
Edward Hines, Jr. VA Hospital; Joel Scholten,
Washington DC VA Medical Center; Douglas
Bidelspach, Lebanon VA Medical Center;
Bridget Smith, Edward Hines, Jr. VA Hospital
Presenter: Rachel Martinez, Social Science
Analyst, Edward Hines, Jr. VA Hospital
rachael.n.martinez@va.gov
Research Objective: The objective of this
project was to explore perspectives of Veterans
Health Administration (VHA) healthcare
providers (TBI specialists and telehealth clinical
technicians (TCTs)) in implementing clinical
video telehealth (CVT) for assessment and
treatment of mild traumatic brain injury (mTBI)
among Veterans of Operation Enduring
Freedom, Operation Iraqi Freedom, and
Operation New Dawn.
Study Design: Semi-structured interviews were
conducted with TBI specialists and TCTs from
15 Veterans Affairs Medical Centers (VAMC)
and Community Based Outpatient Clinics
(CBOC). Interview topics included the step-bystep protocol for administering comprehensive
traumatic brain injury evaluations (CTBIE);
advantages and disadvantages of conducting
CTBIE in person versus CVT; barriers to
implementing telehealth applications for TBI
assessment and management; strategies and
facilitators for enhancing CVT implementation;
additional resources needs; and perceptions of
current training needs. Interviews lasted
approximately 30 minutes, were audio-recorded,
and transcribed verbatim. Transcribed interviews
were coded using a qualitative content analysis.
In addition, providers were asked to fill out a
short background questionnaire prior to the
interview.
Population Studied: The study population
included providers participating in VHA
Rehabilitation and Prosthetic Services (RPS)
TBI Teleconsultation Pilot Project administering
CTBIE over CVT. Subjects were recruited from a
complete list of providers involved in the pilot. A
total of 26 (N) TBI specialists and TCTs agreed
to participate.
Principal Findings: The most formidable
barriers to implementing CVT for TBI evaluation
and management included challenges with
scheduling (e.g., coordinating the schedules
between two different sites), setting up the clinic
(e.g., equipment issues), and conducting a
physical exam over a virtual modality (i.e., the
provider must rely on the TCT to be their hands).
Subjects also discussed strategies to enhance
implementation of CVT for TBI evaluation and
treatment. Some of the most common strategies
included establishing good relationships and
communication with staff, making a personal
connection and establishing rapport with
patients over CVT, and providing accessible
resources to both patients and providers.
Moreover, responses indicated that there are far
more advantages to utilizing CVT—including
travel convenience, cost-effectiveness, and
patient satisfaction—than there are
disadvantages (e.g., limitations in assessing
comorbid conditions besides TBI).
Conclusions: Findings from this study suggest
that CVT is generally perceived by providers as
an effective technology through which to assess
patients for mTBI. Our interviews provide
information about best practices in facilitating
CVT implementation and improving CTBIE
access and treatment plan development.
Implications for Policy, Delivery, or Practice:
Telehealth can potentially increase access to
healthcare for Veterans remotely located from
VA facilities. Subsequent work will focus on
facilitating communication over CVT to further
benefit Veterans with TBI, spinal cord injury
(SCI/D), and other conditions.
Funding Source(s): VA
Poster Session and Number: C, #1043
Positive Deviance in Obesity: Characteristics
of counties with relatively low obesity rates
Jessica Holzer, Yale University School of Public
Health; Maureen Canavan, Yale University
School of Public Health; Elizabeth Bradley, Yale
University School of Public Health
Presenter: Jessica Holzer, Post-doctoral
Fellow, Yale University School of Public Health
jessica.holzer@yale.edu
Research Objective: The objectives of this
exploratory project were to identify counties that
experienced relatively low rates of obesity
despite being located in states with top quintile
rates of obesity and to identify characteristics
about those counties that might account for their
relatively low rates of obesity.
Study Design: Using data from the Robert
Wood Johnson County Health Rankings and
Ratings database, we identified counties whose
rates of obesity were in the lowest quintile
nationally but were located in states with rates of
obesity in the highest quintile nationally; we
termed these counties “positive deviants.” We
examined county characteristics including:
population characteristics and demographics;
the healthcare system, including per capita rates
of doctors and dentists, numbers and types of
hospitals, number of beds per capita, the county
health department, county health clinics, and
availability of health education programs across
the counties; information about recreational
resources and physical activity among county
citizens; rates of education and types of higher
educational institutions; and median income and
poverty levels within the county to understand
how positive deviants counties may differ from
other counties.
Population Studied: 3,141 counties across the
United States, including Alaska and Hawaii.
Principal Findings: We identified 12 positive
deviant counties. Compared with other
counties, positive deviant counties have on
average higher per capita rates of dentists and
physicians, higher rates of college education,
and larger populations than the national
average. The median income for the counties
was roughly the same as the national average,
but the counties had on average lower rates of
children in poverty, lower percent of the
workforce unemployed, lower percent of adults
reporting no leisure time physical activity, and
lower rates of adult uninsurance.
Conclusions: Positive deviant counties are
different than the population average on many
policy-relevant characteristics. The relative
similarity in median income between the positive
deviant counties and the national average
suggests that the deviance is not merely driven
by higher incomes. However, differences in
resource availability denoted by the higher rates
of dentists and physicians, higher rates of
college education, and lower rates of child
poverty, unemployment, physical inactivity, and
uninsurance may indicate significantly different
contexts for the citizens of positive deviant
counties compared to other counties nationally.
Additional work to understand the nature of the
different contexts in positive deviant counties is
ongoing.
Implications for Policy, Delivery, or Practice:
Characteristics of positive deviant counties are
targets for further investigation and may present
policy levers for counties and states to address
obesity trends.
Funding Source(s): AHRQ
Poster Session and Number: C, #1044
Intermediate Clinical Outcomes of Diabetes
under Different Pay-for-Performance
Incentive Designs: A Nationwide PopulationBased Study
Hui-min Hsieh, Kaohsiung Medical University,
Taiwan; Herng-Chia Chiu, Department of
Healthcare Administration and Medical
Informatics, Kaohsiung Medical University; ShyiJang Shin, Graduate Institute of Medical
Genetics, College of Medicine, Kaohsiung
Medical University; Division of Endocrinology
and Metabolism, Kaohsiung Medical University
Hospital
Presenter: Hui-min Hsieh, Assistant Professor,
Kaohsiung Medical University, Taiwan
hsiehhm@gmail.com
Research Objective: Since 2001, Taiwan has
implemented a pay-for-performance (P4P)
program under National Health Insurance that
rewards doctors based on quality of care
provided for their diabetes mellitus (DM)
patients. There are two periods of evolutions for
the incentive methods. In the first period (20012006), the program aimed at pay for
participation, which was to provide financial
incentives only for “processed-based services”.
Beginning at 2007, in addition to process
outcomes, the program paid extra bonuses for
two intermediate outcomes (i.e. pay for
decreasing rate of HbA1c>9.5 and LDL>130).
The purpose of this study is to examine whether
different financial incentive designs under P4P
program affect quality of type 2 diabetes.
Study Design: This study used a longitudinal
quasi-experimental design to evaluate the longterm effects of different P4P incentive methods
on patient quality of care. Diabetes patients who
were newly enrolled in the P4P program during
two periods 2002-2003 (DM1) and 2007-2008
(DM2) were identified. Each patient was
followed for three years and censored if they
dropped out or were died. The first enrolled date
was defined as index date. Both process (i.e.,
executive rates of examining AC, HbA1c, lipid
profile, etc.) and intermediate outcomes were
examined. We hypothesized that quality
outcomes may have greater changes when
intermediate outcomes were directly
incentivized. We used generalized estimating
equation (GEE) models with repeated measures
and first auto-regressive adjustments for
statistical analysis.
Population Studied: Data sources were
derived from two population-based nationwide
databases. One is national P4P database, and
the other is national health insurance
administrative claim data. Diabetes were
included if: (1) they have at least two primary
diagnosis codes in outpatient visits or any
diagnosis codes with ICD-9-CM codes 250
during DM1 and DM2; (2) they were newly
enrolled into the P4P program; (3) they were
age older than 18 years old. Total numbers of
DM1 and DM2 were 75,962 and 78,191,
respectively.
Principal Findings: In DM1 and DM2, patients
average follow-up visits were 5.83 and 6.78
visits; 47.31% and 50.03% of patients were
male; mean ages were 60.07 and 59.12 years
old. Most of the executive rates of each process
outcome had similar trends in DM1 and DM2.
With respect to the specific rewarded
intermediate outcomes, compared to the
baseline data, rate of HbA1c>9.5 decreased
51% in DM1 and 64.44% in DM2; and rate of
LDL>130 decreased 12.3% in DM1 and 48.7%
in DM2.
Conclusions: The study findings indicated there
was an association between patients’ process or
intermediate outcomes and P4P incentive
designs. When the intermediate quality
outcomes were rewarded in DM2, these
indicators seem to have better improvements in
DM2 compared to DM1.On the contrary, process
outcomes were similar when those process
indicators were incentivized in both periods.
Implications for Policy, Delivery, or Practice:
The public and policy makers may concern
whether financial incentive design under P4P
program may affect patient quality. Our empirical
findings suggest that healthcare providers
respond to the rewards and provide care given
the target of incentive designs. Therefore,
financial incentive methods under P4P program
need to be designed carefully in order to provide
better care.
Funding Source(s): Other DOH Grant
DOH101-NH-9018: “Evaluation of Pay-forPerformance and Establishment of Prospective
Payment for Integrated Care”.
Poster Session and Number: C, #1045
Are Patients Satisfied with Chronic Care
Delivery under the National Health Insurance
Program in Taiwan?
Guan-Ting Huang, National Yang Ming
University; Nicole Huang, Institute of Hospital
and Health Care Administration, School of
Medicine, National Yang-Ming University,
Department of Education and Research, Taipei
City Hospital; Hsiao-Yun Hu, Institute of Public
Health & Department of Public Health, School of
Medicine, National Yang-Ming University;
Department of Education and Research, Taipei
City Hospital; Yiing-Jenq Chou, Institute of
Public Health & Department of Public Health,
School of Medicine, National Yang-Ming
University
Presenter: Guan-Ting Huang, Student, National
Yang Ming University, Taipei, Taiwan
pondin_21@hotmail.com
Research Objective: Due to the rising
prevalence of chronic conditions, the delivery of
effective and high-quality chronic care has
emerged as a major focus in health care
systems. In contrast to enormous emphasis
placed on patient-centered quality assessments
in many western countries, very limited
information is available in understanding quality
of chronic care delivery from patient’s
perspective in Asian countries. This study aimed
to assess patients’ experience with chronic care
delivery under the National Health Insurance
program in Taiwan and to identify patient and
provider characteristics associated with negative
patient experiences.
Study Design: We conducted a cross-sectional
study. Socio-demographics characteristics,
socioeconomic status, nature and type of
chronic conditions, and self-rated health status
were collected by face-to-face interview. Among
those with one or more chronic conditions, we
also collected additional information on the
characteristics of their usual source of care and
the number of chronic conditions managed by
their usual providers. Individuals’ perception of
quality of care was also collected using the
Patient Assessment of Chronic Illness Care
(PACIC) instrument. Multiple linear regression
and tobit regression were applied to analyze the
data.
Population Studied: This study surveyed 2,635
adult participants of a government funded health
examination program in 3 hospitals in Taipei
from April to September 2013. Of these
participants, we targeted 830 elderly patients
with hypertension, diabetes and hyperlipidemia.
The three chronic conditions were selected
because of their high prevalence in Taiwan.
Principal Findings: The average overall PACIC
score was 1.85 of a possible 5, which was far
lower than the findings reported in other
countries. Of the five subscales, delivery
system/practice design subscale had the highest
score (3.25) and the remaining four subscales
ranged from 1.37 to 1.96. After adjusting for
other characteristics, significant differences in
overall scores were associated with marital
status and self-rated health. Patients who were
divorced or in widowhood had significantly lower
overall score (P=0.013) and lower scores in all
the subscales except for delivery
system/practice design. In contrast, patients with
poorer self-rated health were particularly
unsatisfied with system/practice design of their
chronic care delivery (P=0.004). Patients who
were usually cared by male physicians,
physicians who practice in a public institution, or
in clinics had considerably lower PACIC scores,
but the differences did not reach the level of
statistical significance. More interestingly,
patients whose chronic conditions were
managed by a single provider had substantially
lower PACIC score than those whose chronic
conditions were managed by multiple providers.
Conclusions: From the patient’s perspective,
the quality of chronic care for the three common
chronic conditions (hypertension, diabetes, and
hyperlipidemia) was poor in Taiwan. The overall
PACIC score and all five subscales did not vary
dramatically across patient and provider
characteristics. The negative patient experience
was prevalent among older individuals and
across different physician groups.
Implications for Policy, Delivery, or Practice:
Despite extensive efforts in improving quality of
chronic care delivery in Taiwan, patients’
experience with chronic care delivery is far from
satisfactory and suggests much room for
improvement. The analyses of overall scores
and subscales of PACIC may help to identify
areas for improvement in chronic care delivery
or practices.
Funding Source(s): No Funding
Poster Session and Number: C, #1046
Using Vignettes to Explore Missed
Opportunities in Primary Care for Patients
with Chronic Conditions
Shana D. Hughes, Palo Alto Medical
Foundation; Ellis Dillon, Palo Alto Medical
Foundation Research Institute; Mary Carol
Mazza, Palo Alto Medical Foundation Research
Institute; Ming Tai-Seale, Palo Alto Medical
Foundation Research Institute
Presenter: Shana D. Hughes, Qualitative
Analyst/Research Assoc., Palo Alto Medical
Foundation
hughess@pamfri.org
Research Objective: Missed opportunities are
well documented in clinical communication, but
there is less understanding of their causes.
Patients with diabetes and hypertension have
many health concerns that if not fully addressed
may become magnified and potentially more
debilitating, difficult to treat, and costly over time.
This study asks how and when physicians and
medical assistants respond to patient concerns
in primary care appointments, and uses
vignettes embedded in interviews to explore
care teams’ perceptions of these instances.
Study Design: Findings are drawn from
qualitative observations and interviews which
are part of a mixed methods study of primary
care for patients with chronic conditions. The
research was conducted at a large, non-profit,
multispecialty group practice in northern
California which serves four counties and more
than 850,000 patients. The observational
component included primary care visits with
patients with chronic conditions. Subsequent
interviews with physicians and medical
assistants probed professional experience with
this type of care and included standardized
vignettes. The vignettes invited interviewees to
react to realistic examples of clinical
communication, permitting triangulation between
observed behavior and interview responses.
Population Studied: Appointment observations
involved 12 patients with diabetes and/or
hypertension, 9 physicians, and 9 medical
assistants. Semi-structured interviews were
conducted with 3 physicians and 3 medical
assistants.
Principal Findings: Qualitative analysis of
appointment observations revealed that
physicians and medical assistants sometimes
miss critical opportunities to discuss important
health concerns or decisions. We developed an
emergent taxonomy of these “missed
opportunities:” (1) missed cues and (2)
presumed responses. A missed cue occurs
when a patient indicates a concern (e.g., “I’m
trying to get used to salads…”) but is met with
no response, a cursory response, or a deflection
of the concern. A presumed response occurs
when a physician or medical assistant does not
engage the patient in discussion. For example,
rather than asking if a patient is amenable to
HIV testing, the provider states, “And you don’t
want an HIV test.” Preliminary findings from
interviews indicate that missed opportunities are
most commonly attributed to time constraints in
primary care and the goal of “getting [the visit]
done.” In response to the vignettes, physicians
and medical assistants readily identified
strategies to improve communication, even
though our observation demonstrated
inconsistent application of these techniques in
practice.
Conclusions: Providers sometimes miss
opportunities to fully engage with patients
regarding their health concerns or care
decisions. These missed opportunities represent
an important area for improvement in primary
care for patients with chronic conditions.
Although the vignette technique is not frequently
used in health services research, in this study it
exposed a divergence between observed
behavior and beliefs about professional best
practices.
Implications for Policy, Delivery, or Practice:
Further research is needed to explore
meaningful strategies to reduce missed
opportunities. This knowledge can translate to
user-centered design for behavioral or
organizational interventions intended to improve
the experience and quality of care for patients
with chronic conditions.
Funding Source(s): Other Gordon and Betty
Moore Foundation
Poster Session and Number: C, #1047
Super-Utilizers: Population Dynamics and
Trends
Tracy Johnson, Denver Health and Hospital
Authority; Deborah Rinehart, Denver Health and
Hospital Authority; Josh Durfee, Denver Health
and Hospital Authority; Dan Brewer, Denver
Health and Hospital Authority; Holly Batal,
Denver Health and Hospital Authority; Joshua
Blum, Denver Health and Hospital Authority;
Kathy Thompson, Denver Health and Hospital
Authority; Paul Melinkovich, Denver Health and
Hospital Authority
Presenter: Tracy Johnson, Director, Health
Care Reform Initiatives, Denver Health and
Hospital Authority
tracy.johnson@dhha.org
Research Objective: To provide a detailed,
longitudinal, program-/policy-relevant portrait of
individuals with high levels of potentially
avoidable utilization. Specifically,
- Describe their burden of chronic disease and
social determinants of health
- Assess population dynamics and stability of
trends over time
- Assess longitudinally their natural history of
utilization/cost, across settings and payers
- Assess explicit/implicit super-utilizer program
design assumptions
Study Design: A retrospective, secondary
analysis of administrative data, using
longitudinal and cross-sectional descriptive
methods.
Population Studied: The study population
included n=4774 publicly-insured and uninsured
adult patients at an urban safety net integrated
delivery system that met “super-utilizer” criteria
one or more months during the study period
(May 1, 2011- April 1, 2013.) Super-utilizers are
patients with 3 or more hospitalizations in the
previous year as well as patients with a serious
mental health diagnosis and 2 or more
hospitalizations in the previous year.
Principal Findings: Over a two-year period,
• Consistently 3 percent of adult patients at an
urban safety net integrated delivery system
qualified as “super-utilizers”, accounting for 30
percent of adult institutional charges;
• Super-utilizers had stable, overall group
characteristics in terms of chronic disease,
payer, social determinants of health, and costs;
• Super-utilizers were nonetheless
heterogeneous with identifiable patient
subgroups (e.g., cancer, emergency Medicaid
dialysis, trauma, orthopedic complications, and
co-morbid chronic and mental health conditions);
• Super-utilizers were NOT stable at the
individual level, with significant monthly cycling
in and out of the “super-utilizer” status (of the
super-utilizers on 5/1/12, only 20 percent still
met the super-utilizer definition one year later);
• 43.3 percent of super-utilizers did not have an
established medical home;
• More than one-quarter of super-utilizers are
managed care patients for whom it is possible to
assess utilization outside the study and 85
percent of these patients had such charges (20
percent of charges);
• Significant regression to the mean (44 percent)
in per person charges was observed in the
qualifying year compared to the subsequent
year, absent any intervention.
Conclusions: The mere fact a small percentage
of the population drives a significant portion of
overall costs is not sufficient for program design.
On the one hand, our longitudinal analysis
documents the stability of the super-utilizer
demographic profile and costs over time, which
suggests that focused attention on superutilizers could yield a positive return on
investment. On the other hand, the
heterogeneity within the super-utilizer population
argues against a one-size-fits-all approach and
in favor of multiple, tailored, intervention
strategies. Furthermore, the fact that patients
cycle in and out of super-utilizer status and that
many use multiple hospitals and lack primary
care attachments have implications for when,
where, and how providers can optimally
intervene. Finally, our data indicate that superutilizer program evaluations need to account for
significant, expected regression to the mean or
program impact will be overstated.
Implications for Policy, Delivery, or Practice:
Gwande’s 2011 New Yorker article on “hot
spotters” ignited and the ACA has fueled
national interest in super-utilizers or “triple fail”
populations. However, despite the original
article’s emphasis on data-driven program
design, super-utilizer program development has
outpaced the descriptive literature. Many of the
published studies of super-utilizers to date are
cross-sectional, point-in-time snapshots and are
limited to a single payer or hospital setting.
Funding Source(s): CMS
Poster Session and Number: C, #1048
The Hospital and Geographic Concentration
of Care for High-Cost Patients
Karen Joynt, Harvard School of Public Health; E.
John Orav, Harvard School of Public Health;
Ashish Jha, Harvard School of Public Health
Presenter: Karen Joynt, Instructor, Harvard
School of Public Health
kjoynt@hsph.harvard.edu
Research Objective: A small proportion of
patients is responsible for the majority of
healthcare spending in the Medicare program.
However, we know little about the concentration
of care for these high-cost patients: namely,
whether they are evenly distributed across
hospitals and geographical areas, or whether
some hospitals or regions are disproportionately
likely to care for these patients. Understanding
more about the concentration of care would
allow policymakers and clinical leaders to more
effectively target interventions to reduce costs
and improve outcomes in this important patient
population.
Study Design: We used national Medicare
claims data from 2010 and assigned a standard
cost to each claim based on Medicare fee
schedules. Using standardized costs allowed us
to identify patients who were high utilizers of
healthcare services, regardless of local costs of
living. We summed these costs to create an
annual cost of care for each beneficiary, and
defined “high-cost patients” as those in the
highest decile. We then calculated, for each
hospital in the country, the proportion of their
admissions that were for high-cost patients, and
defined “high-cost patient hospitals” (HCP
hospitals) as those in the highest decile. We
compared the characteristics of these HCP
hospitals to other U.S. hospitals. Finally, we
calculated, for each hospital referral region
(HRR) in the country, the proportion of their
patients that were high cost, and defined “highcost patient HRRs” (HCP-HRRs) as those in the
highest decile. We compared HRR
characteristics between HCP-HRRs and all
other U.S. HRRs.
Population Studied: 1,236,797 Fee-for-service
Medicare beneficiaries, 123,679 of whom were
considered high-cost.
Principal Findings: High-cost patients
accounted for almost 60% of hospitalizations
overall; however, only 8.6% of hospitalizations
for high-cost patients occurred in HCP hospitals.
The 462 HCP hospitals had, on average, 72.6%
of their claims accounted for by high-cost
patients, compared with 56.0% at non-HCP
hospitals. HCP hospitals were smaller, more
often rural, and more often public; over half were
located in the South. They had marginally
worse performance on hospital quality alliance
processes of care metrics (95% versus 96%,
p=0.01) and similar mortality rates as non-highcost patient hospitals. The HCP hospitals had
higher readmission rates. The 30 HCP-HRRs
had, on average, 13.1% of their Medicare
beneficiaries labeled as high-cost, compared
with 9.7% in non-HCP-HRRs. HCP-HRRs had
higher proportions of black patients (17.9%
versus 10.1%) but similar proportions of patients
in poverty (14.3% versus 12.4%, p=0.12). HCPHRRs performed worse on many quality metrics,
including screening and preventive care,
avoidable emergency department visits,
preventable hospitalizations, and preventable
mortality (108.7 versus 94.6 deaths per 100,000
population, p=0.004).
Conclusions: Care for high-cost patients is
relatively dispersed at both the hospital and
HRR level. Hospitals serving the highest
proportions of high-cost patients appear to have
somewhat worse care and higher readmission
rates. HRRs with a high proportion of high-cost
patients have worse performance on numerous
quality measures.
Implications for Policy, Delivery, or Practice:
Targeting policy and clinical interventions at
hospitals and regions serving a disproportionate
number of high-cost patients may have the
potential to both reduce costs and improve
quality of care, both at the hospital and the
population level.
Funding Source(s): Other Rx Foundation
Poster Session and Number: C, #1049
Opioid Use and Walking Among Patients
with Chronic Low Back Pain
Sarah Krein, VA Ann Arbor Center for Clinical
Management Research; Amy Bohnert, VA Ann
Arbor Center for Clinical Management
Research; Hyungjin Kim, VA Ann Arbor Center
for Clinical Management Research; Meredith
Harris, University of Cincinnati College of
Medicine; Caroline Richardson, VA Ann Arbor
Center for Clinical Management Research
Presenter: Sarah Krein, Research Health
Scientist, VA Ann Arbor Center for Clinical
Management Research
skrein@umich.edu
Research Objective: Opioids are commonly
used to help treat back pain, despite rising
concerns and recommendations to use
alternative strategies. The purpose of this study
was to examine the interaction between selfreported opioid use and a walking intervention
on step counts in Veterans with chronic back
pain.
Study Design: Retrospective, longitudinal
analysis of data collected as part of a
randomized trial of an internet mediated walking
program for managing back pain. Step counts at
baseline, 6 and 12 months were collected via an
uploading pedometer. Self-reported opioid use
was collected at baseline along with
demographic and other pain-related information.
A linear mixed-effects model was used to
assess relationships between baseline opioid
use and step counts for those in the intervention
versus control group over time.
Population Studied: Veterans who participated
in a randomized trial of a walking intervention to
reduce back pain-related disability (n = 229, 118
control and 111 intervention).
Principal Findings: Over 40% (n = 99) of
participants reported using opioid medications at
baseline. While there were no differences in
age or percent randomized to the intervention
group, opioid users compared to non-users
reported higher pain levels (6.3 vs. 5.8, p = .02),
had higher disability scores (10.5 vs. 8.6, p =
.02) and lower average daily step counts (4005
vs. 4811, p = .02) at baseline. Unadjusted
results showed relatively large increases in daily
steps (more than 1,000 steps) at both 6 and 12
months for opioid users assigned to the
intervention group, while changes among nonopioid users and opioid users assigned to the
control group appeared more modest (less than
500 steps) or even negative. After adjustment,
this finding was even more pronounced, with a
time averaged predicted mean increase of more
than 1200 daily steps for opioid users assigned
to the intervention. This change was significantly
different from the predicted mean reduction of
nearly 400 steps for opioid users in the control
group (p = .004) and slight reduction (about 15
steps) for non-opioid users in the intervention (p
= .02). However, it was not significantly different
when compared with the predicted mean
increase of about 660 steps among non-opioid
users in the control group (p = .34).
Conclusions: Our findings revealed a notable
increase in objectively monitored step counts at
both 6 and 12 months among study participants
using opioids at baseline and who were
assigned to the intervention group. This
included an increase of more than 1000 steps
(nearly 1/2 mile) on average each day
throughout the 12 month study period.
Implications for Policy, Delivery, or Practice:
These findings may suggest that opioid use
facilitates participation in walking as a form of
exercise for patients with chronic back pain.
More importantly, however, these data show that
patients receiving opioids are both willing and
able to engage in walking to help manage their
back pain when provided with additional support
to do so. Our results emphasize the importance
of supporting the use of alternative pain
management strategies for patients with chronic
back pain who are receiving opioids.
Funding Source(s): VA
Poster Session and Number: C, #1050
Does Socioeconomic Status Moderate the
Effect of Increasing Chronic Disease Burden
on Three-Year Survival in a PopulationBased Cohort?
Natasha Lane, University of Toronto; Andrea
Gruneir, Women's College Research Institute;
Colleen J. Maxwell, University of Waterloo
School of Pharmacy; Susan E. Bronskill,
Institute for Clinical Evaluative Sciences; Walter
P. Wodchis, University of Toronto Institute of
Health Policy, Management and Evaluation
Presenter: Natasha Lane, MD-PhD Student,
University of Toronto
natasha.lane@mail.utoronto.ca
Research Objective: Lower socioeconomic
status (SES) is associated with an increased risk
of multimorbidity. Other research has shown that
increases in the degree of multimorbidity are
associated with poorer survival. The direct link
between the SES gradient in multimorbidity and
survival, however, has not been well described.
The goal of this study was to determine whether
the effect of increasing multimorbidity on patient
survival is moderated by SES.
Study Design: This retrospective cohort study
used linked administrative data from April 1,
2009 to March 31, 2012. A multivariable
proportional hazards regression model was
constructed to examine correlates of individuals’
survival over a three-year time period.
Covariates included the number and type of
chronic conditions, as well as sociodemographic
and health service utilization characteristics.
Interactions between neighbourhood income
quintile (SES proxy) and the number of chronic
conditions were examined.
Population Studied: 6,639,089 Ontarians up to
age 105 years as of April 1, 2009 with at least
one of the following 16 conditions were included:
cardiac arrhythmia, acute myocardial infarction,
hypertension, chronic coronary syndrome,
congestive heart failure, stroke, asthma, chronic
obstructive pulmonary disorder, diabetes,
osteoporosis, rheumatoid arthritis, osteo- and
other arthritis, depression, dementia, cancer, or
renal failure.
Principal Findings: Individuals with higher
multimorbidity had significantly poorer survival
than those with fewer conditions. After
controlling for demographics and health services
utilization, higher neighbourhood income quintile
was associated with longer survival. However,
there was no significant interaction between high
disease burden and income quintiles in
predicting survival.
Conclusions: Although increasing
multimorbidity did reduce survival in this
population-based cohort, the relationship
between high disease burden and survival was
not moderated by SES, as measured by
neighborhood income quintiles.
Implications for Policy, Delivery, or Practice:
The number of chronic conditions individuals
have is highly predictive of their survival and
should be considered when targeting
interventions to improve health outcomes in
complex patients. Although there is an SES
gradient in survival among multimorbid patients,
the impact of high disease burden on survival is
consistent across SES groups.
Funding Source(s): Other Health System
Performance Research Network (HSPRN),
sponsored by the Ontario Ministry of Health and
Long-Term Care
Poster Session and Number: C, #1051
Unintended Benefits? The Potential
Economic Impact of Addressing Risk Factors
for Alzheimer’s Prevention
Pei-jung Lin, Tufts Medical Center; Zhou Yang,
Department of Health Policy and Management,
Rollins School of Public Health, Emory
University; Howard Fillit, Alzheimer’s Drug
Discovery Foundation; Joshua Cohen, Center
for the Evaluation of Value and Risk in Health,
Institute for Clinical Research and Health Policy
Studies, Tufts Medical Center; Peter Neumann,
Center for the Evaluation of Value and Risk in
Health, Institute for Clinical Research and Health
Policy Studies, Tufts Medical Center
Presenter: Pei-jung Lin, Assistant Professor,
Tufts Medical Center
plin@tuftsmedicalcenter.org
Research Objective: Some chronic conditions
such as diabetes and cardiovascular diseases
(CVDs) have been identified as potentially
modifiable risk factors for Alzheimer’s disease
and related dementias (ADRD). This study
examines the potential health and economic
ADRD-related impacts of addressing those risk
factors. This is a critical question given the lack
of effective disease-modifying interventions for
ADRD, the potentially high-price of ADRD
therapy, and the pressure from payers,
employers, and governments to restrain costs.
Study Design: We used logistic regression to
estimate the probability of ADRD in a certain
year, and developed a two-part model to predict
a person’s annual Medicare and Medicaid
expenditures by ADRD status. We constructed
a Medicare cohort-based Dynamic Aging
Process simulation model to examine the
relationship between the reduction of possible
ADRD risk factors and ADRD onset and
duration, and costs to Medicare and Medicaid.
We explored four hypothetical scenarios: a 10%
prevalence reduction in diabetes, hypertension
and CVD (each considered separately), and a
10% reduction in BMI among overweight or
obese older adults.
Population Studied: Medicare beneficiaries
age 65 and older with ADRD identified from the
1997-2005 Medicare Current Beneficiary Survey
Cost and Use Files.
Principal Findings: Our simulation predicts
that: at baseline, 14.5% of the 65-year-old
cohort would develop ADRD, that the average
onset age would be 80.7 years, and that those
who developed ADRD would have the condition
for an average of 5.03 years. Reducing the
prevalence rate of CVD by 10% would decrease
the ADRD risk by 0.6% (from 14.5% to 13.9%),
delay onset by 0.1 years (from age 80.7 to age
80.8), and reduce time spent living with ADRD
by 0.03 years (from 5.03 years to 5.00 years).
This would translate into eliminating U.S.
population time spent with ADRD from 663 to
635 million person-months and yield substantial
savings for Medicare (reducing costs from $136
to $116 billion) and Medicaid (from $122 to $105
billion) as a result of lower ADRD costs. At the
population level, we estimate that reducing BMI
by 10% among overweight or obese
beneficiaries would save $6 billion for Medicare
and $35 billion for Medicaid as a result of lower
ADRD costs. Reducing diabetes by 10% would
save $7 billion for Medicare and $1 billion for
Medicaid, and reducing hypertension by 10%
would save $12 billion for Medicare and $12
billion for Medicaid as a result of lower ADRD
spending.
Conclusions: Our simulation suggests that
reducing the prevalence rates of certain chronic
conditions, especially CVDs, may yield
“unintended benefits”, including lower ADRD
risk, delayed onset, reduced disease duration,
and substantial savings in ADRD costs.
Implications for Policy, Delivery, or Practice:
Anticipating the possible consequences of
addressing modifiable ADRD risk factors could
help Medicare and Medicaid officials, the
Congressional Budget Office, and other
stakeholders to assess future costs and health
care needs and to conduct public health
planning for ADRD. While more research is
needed to better establish these effects, our
study suggests that addressing certain
modifiable risk factors could achieve a
“compression of morbidity" and increase ADRDfree life expectancy, which may ultimately
improve population well-being and reduce costs.
Funding Source(s): Other Alzheimer’s Drug
Discovery Foundation
Poster Session and Number: C, #1052
Technology Use, Chronic Conditions, and
Healthcare Utilization among VA Primary
Care Patients
Sara Locatelli, Department of Veterans Affairs;
Sherri LaVela, Department of Veterans Affairs;
Feinberg School of Medicine, Northwestern
University
Presenter: Sara Locatelli, Post-doctoral
Research Fellow, Department of Veterans
Affairs
sara.locatelli@va.gov
Research Objective: Examine relationships
between technology self-efficacy, anxiety, and
usage, and chronic comorbid conditions, and
healthcare utilization among primary care
Veteran patients receiving care at a large VA
hospital.
Study Design: Semi-structured interviews
assessed participant demographics (e.g., age,
gender, employment status), current technology
use (any versus none; frequency), along with
their confidence (self-efficacy) and anxiety with
using technology for healthcare needs. With
participant consent, VA electronic record data
were used to obtain chronic comorbid
conditions, healthcare utilization (number of
inpatient days and outpatient visits over the
previous year), and distance from the facility (in
miles, computed by participant zip code).
Population Studied: 121 Veterans recruited
through primary care clinics at a large VA
hospital
Principal Findings: Individuals with liver
disease had lower technology self-efficacy
(p=0.04) and higher technology anxiety (p=0.04)
than individuals without this diagnosis. Though
measures of technology self-efficacy and anxiety
did not correlate with utilization, three individual
scale items related to inpatient utilization.
Confidence in viewing online personal health
records (PHRs) (p=0.02) related to lower
inpatient utilization; confusion with using email
for healthcare needs (p=0.04), and confusion
with viewing medical record online (p=0.02)
related to higher utilization. Lower inpatient
utilization was observed among individuals who
reported any use of computers (p=0.01), the
internet (p=0.04), email (p=0.01), and health
information seeking (p=0.02), and individuals
who reported any use of email (p=0.04) had
fewer outpatient appointments, during the
previous year. Individuals who reported weekly
or greater use of computers (p=0.05) had
significantly fewer inpatient days, and individuals
who reported weekly or greater use of
smartphones (p=0.04) had significantly more
outpatient appointments. Individuals with liver
disease were less likely to report weekly or
greater use of computers (p=0.03), the internet
(p=0.02), and email (p=0.02). Individuals with
depression were less likely to report weekly or
greater health information seeking (p=0.04).
Additionally, individuals with alcohol use
disorders were less likely to report using
computers (p=0.05), the internet (p=0.03), and
Twitter (p=0.04), and individuals with substance
abuse disorders were less likely to report weekly
or greater use of tablet computers (p=0.05), the
internet (p=0.02), and email (p=0.03). However,
individuals with substance abuse disorders were
more likely to report weekly or greater use of
text messages (p=0.02).
Conclusions: Our findings suggest that Veteran
primary care patients who use technology tend
to utilize less face-to-face healthcare with the
exception of frequent smartphone users having
higher outpatient utilization. Confidence with
viewing PHRs predicted lower inpatient
utilization, and confusion with using email and
PHRs predicted higher inpatient utilization. In
contrast, greater confidence with
receiving/viewing text messages from one’s
physician predicted higher outpatient utilization;
it is possible individuals more comfortable with
this communication method also desire
immediate answers, resulting in more outpatient
visits. Greater use of these communication
methods may reduce outpatient visits for routine
questions.
Implications for Policy, Delivery, or Practice:
Providing resources to support and encourage
technology use may increase patient access,
and potentially decrease in-person utilization.
This may be especially beneficial for individuals
with comorbidities and greater utilization
patterns.
Funding Source(s): VA
Poster Session and Number: C, #1053
The Cross-Section of Preventable
Hospitalizations and Healthcare Associated
Infections
Andrea Lorden, Texas A&M Health Science
Center
Presenter: Andrea Lorden, Graduate Assistant
Researcher, Texas A&M Health Science Center
alorden@msn.com
Research Objective: This research had two
objectives. First, identify and quantify
characteristics of the individuals who experience
a potentially preventable hospitalization and
acquire a healthcare associated infection during
the same admission. Second, build upon and
improve the methodologies for identifying
healthcare associated infections from hospital
administrative data for use in policy research.
Study Design: This is a retrospective
observational study using a cross-sectional
design.
Population Studied: From the Texas inpatient
discharge summary for 2011, individuals with a
potentially preventable hospitalization or a
healthcare associated infection were included.
Potentially preventable hospitalizations were
identified through the Agency for Healthcare
Research and Quality Prevention Quality
Indicators. Healthcare associated infections
included central-line associated blood stream
infections, catheter associated urinary tract
infection, ventilator associated pneumonia,
Clostridium difficile infection, and surgical site
infections for coronary artery bypass, hip
replacement and knee replacement.
Principal Findings: Over 900 individuals were
identified with both a preventable hospitalization
and a healthcare associated infection during the
same inpatient stay during 2011. Primarily 45
years or older, more than 65 percent of
individuals with both potentially preventable
hospitalization and a healthcare associated
infection identified Medicare as their primary
payer. An additional 26 percent identified
Medicaid or private insurance as their primary
payer. Cost analyses are anticipated to reflect
differences between payer groups for those with
a preventable hospitalization and those with
both a preventable hospitalization and a
healthcare associated infection.
Conclusions: Potentially preventable
hospitalizations have long been associated with
barriers in access to preventive care due to
uninsurance. However, given that 92 percent of
individuals with both events identified Medicare,
Medicaid, or private insurance as the primary
payer, policy that focuses on quality preventive
care could be justified by the substantial savings
anticipated through the prevention of
hospitalizations and resulting reduced exposure
to and acquisition of healthcare associated
infections.
Implications for Policy, Delivery, or Practice:
While the number of individuals in the crosssection of potentially preventable hospitalization
and healthcare associated infection is small, the
increased healthcare costs and the large
proportion of Medicare and Medicaid
beneficiaries may require policy initiatives and
program interventions to do more to create a
change in this population.
Funding Source(s): No Funding
Poster Session and Number: C, #1054
The Impact of Most Traumatic Life Event on
Post-traumatic Stress Disorder (PTSD)
Symptoms in Women Veterans
Betsy McGee, VA Health Care System &
University of Iowa Department of Psychiatry;
Brenda M. Booth, University of Arkansas for
Medical Sciences; James C. Torner, College of
Public Health & Carver College of Medicine,
University of Iowa; Anne Sadler, Iowa City VA
Health Care System & University of Iowa
Department of Psychiatry
Presenter: Anne Sadler, Physician, Iowa City
VAMC
anne.sadler@va.gov
Research Objective: To examine posttraumatic stress disorder (PTSD) diagnoses and
severity in relation to type (self-oriented vs.
other-oriented) of most traumatic life event
Study Design: Retrospective cohort study
design using a computer-assisted telephone
interview to collect data on socio-demographic
and military characteristics, lifetime trauma
exposure, physical and mental health.
Population Studied: 1004 US service women
(< 52 years of age; mean=38 years) enrolled in
two Midwestern Veterans Affairs (VA) Health
Care Systems or outlying clinics within the five
years preceding study interview (63% response
rate). Military combat (29%), the sudden death
of a close friend or relative (75%) and one of
more lifetime (SA, 62%; 32% experienced SA
during military service)), not mutually exclusive.
Participants were asked to identify their most
traumatic life event and respectively, to rate the
presence and severity of PTSD symptoms
(Posttraumatic Symptom Scale). Lifetime
trauma exposure was measured by traumatic
events associated with PTSD (i.e., combat, SA,
other), and was further classified into selforiented (i.e., events directed toward self: SA,
personal illness) and other-oriented (i.e., events
directed at another person: witnessing
injury/death).
Principal Findings: 772 (77%) endorsed at
least one of the queried traumas as their most
traumatic event. 23% (n=178) met criteria for
current diagnosis of PTSD. Most frequently
endorsed traumatic events were grouped into
three categories: Combat (8% n=66; e.g.
incoming artillery during combat), SA (33%,
n=253; attempted or completed rape), and other
(59% n=453; e.g. the sudden death of a close
friend or relative). Participants whose most
traumatic event was a combat trauma or SA
were more likely to have current PTSD (OR
3.04, CI (1.7, 5.4) and OR 3.5, CI (2.4, 5.0),
respectively). When trauma events were
classified into self-oriented (50% n=388) vs.
other-oriented (50% n=384), participants who
endorsed a self-oriented trauma as the most
distressing were more likely to have PTSD than
those endorsing other-oriented trauma ((50%
n=384), OR 2.4 CI(1.7, 3.4)). Among
servicewomen who experienced both combat
and in-military sexual assault, 55% identified
sexual assault as the most traumatic.
Conclusions: Although experiencing a
traumatic event was prevalent in these women
Veterans, the type of trauma was a more
specific predictor of PTSD than trauma exposure
alone. Specifically, women who reported
combat trauma or SA as their most traumatic
event were three times more likely to report
PTSD symptoms than those who reported other
trauma events. Likewise, compared to otheroriented traumas, those traumatic events of a
self-oriented nature posed greater risk for PTSD
symptoms.
Implications for Policy, Delivery, or Practice:
Further understanding is needed regarding the
heterogeneity of trauma events and their impact
on the development and severity of PTSD in
military women, a population well-recognized as
being at elevated risk for exposure to traumatic
events over their lifetime. These findings
highlight the importance of trauma-informed
clinical practice and policies guiding care
delivery in women Veterans.
Funding Source(s): VA
Poster Session and Number: C, #1055
"It Doesn't Work Like That Here" A
Structured Approach to Contextualising
Interventions to Manage Chronic Conditions
in International Settings
Martin McKee, London School of Hygiene and
Tropical Medicine; Dina Balabanova, LSHTM;
Isabelle Risso-Gill, LSHTM; Charlotte
Kuhlbrandt, LSHTM; Helena Legido-Quigley,
LSHTM; Khalid Yusoff, Universiti Teknologi
MARA, Kuala Lumpur; Patricio Lopez Jaramillo,
FOSCAL, Colombia; Salim Yusuf, Population
Health Research Institute, McMaster University,
Canada
Presenter: Martin McKee, Professor of
European Public Health, London School of
Hygiene and Tropical Medicine
martin.mckee@lshtm.ac.uk
Research Objective: To develop a contextually
appropriate model of care to improve
management of hypertension that will be
evaluated in a cluster randomized controlled trial
in low and middle income countries
Study Design: Multi-method health system
assessment. The operation of the health system
is assessed through the eyes of the population
and front line professionals. Data are collected
from documentary review, observation, semistructured interviews with patients, providers
and policy-makers, and focus groups by trained
field workers, using manuals containing
definitions, sampling guides, interview guides
etc. Analysis draws on the concept of realist
evaluation and soft systems theory.
Population Studied: Development was
undertaken in two urban and two rural
communities each, in Colombia and Malaysia,
participating in the Prospective Urban and Rural
Epidemiology Study (21 country study of
cardiovascular disease)
Principal Findings: The assessment enabled
the initial concept of the intervention (task
shifting to mid level workers, combination
therapy, patient education, simplified guidelines)
has had to be refined considerably. In Malaysia,
it was necessary to address a) health beliefs
(Malay, Traditional Chinese, Ayurvedic) that
conceived as illness as a self-limiting disorder,
rejecting long term treatment for an
asymptomatic condition; b) complex treatment
pathways where patients moved between
western and traditional health workers;
diagnostic guidelines recognizing low chance of
patient returning for confirmation. In Colombia
design addressed:drug supply problems; high
costs of drugs; confusion about insurance
coverage; fragmentation of system so that
patients get lost.
Conclusions: Models of care for chronic
diseases cannot simply be transferred into new
settings. The design of interventions in two
middle income countries was refined
substantially by first undertaking a detailed
assessment of the health systems they were to
be implemented within. the issues in each
country were quite different.
Implications for Policy, Delivery, or Practice:
Originally developed as a structured approach to
using tracer conditions to assess health system
performance from the patient perspective, our
assessment tool offers a means of ensuring that
interventions in different countries (and
potentially ethnic groups within a country) take
full account of context. A failure to do so would
have resulted in evaluating interventions that
missed the major barriers to appropriate care
in each setting
Funding Source(s): Other Canadian Institutes
for Health Research
Poster Session and Number: C, #1056
Determinants of Drug Compliance and Blood
Pressure Control in Hypertensive Patients at
the Obafemi Awolowo University Teaching
Hospital, Ile- Ife
Akinyemi Oluwafunmi, Obafemi Awolowo
University, Ile- Ife, Nigeria; Adedeji Onayade,
Obafemi Awolowo University; Anthony
Akintomide, Obafemi Awolowo University;
Margaret Afolabi, Obafemi Awolowo University;
Gbola Olayiwola, Obafemi Awolowo University;
Wilson Erhun, Obafemi Awolowo University
Presenter: Akinyemi Oluwafunmi, Assistant
lecturer, Obafemi Awolowo University, Ile- Ife,
Nigeria
moskolad@yahoo.com
Research Objective: The objectives of the
study were to assess the knowledge of
hypertension among the patients attending the
cardiology outpatient clinic at the Obafemi
Awolowo University Teaching Hospitals
Complex (OAUTHC) Ile- Ife, to determine the
patients’ compliance with antihypertensive
medication through self- reporting, to correlate
the blood pressure readings of the patients with
their self- reported drug compliance; and
determine the factors influencing patients’
compliance with antihypertensive drugs
Study Design: The study was a cross sectional
descriptive survey. A questionnaire administered
through interview was used for the study. It
consisted of 6 sections addressing sociodemographic data, standardized questions on
high blood pressure to determine respondents’
knowledge of hypertension, self- reported
compliance and hypothesized factors affecting
compliance to antihypertensive medications. It
also consisted of standardized questions to
determine if respondents had social support and
satisfaction with care. Compliance was
assessed using a previously validated selfreport tool. Compliance rate was calculated as
pills taken over a specified period divided by pills
prescribed for that period and expressed as a
percentage
Population Studied: Three hundred and thirty
adult patients with essential hypertension
attending outpatient cardiology clinic at the
OAUTHC were studied. The inclusion criteria
was that patients must have been on
antihypertensive drugs for at least 6 months
while patients with diabetes and other
complications were excluded from the study
Principal Findings: Nearly ninety percent of
respondents were compliant with their
antihypertensive. This corresponds with a good
number of respondents (over sixty percent)
having controlled blood pressure. Eighty- five
percent of respondents have a good knowledge
of hypertension. Possible determinants of noncompliance identified include: running out of
supply (nineteen percent), feeling well (fourteen
percent), work schedule (nearly ten percent),
missing clinic (approximately six percent),
Travelling (about eight percent), forgetfulness
(twelve percent), side effects (ten percent), faith
belief (about four percent), drug not available at
pharmacy (approximately six percent), fasting
for religious purposes (ten percent) and high
cost of drugs (nearly seven percent).
Conclusions: Compliance with antihypertensive
in the study population was found to be very
high and this corresponds with a higher blood
pressure control among the compliant group.
Hypertension knowledge and drug regimen were
found to be the strongest factors influencing
compliance.
Implications for Policy, Delivery, or Practice:
Hypertensive patients need to be educated on
their disease condition, the aim of their
treatment as well as the risk factors for
hypertension. Also, patients with poor economic
status would benefit from the provision of health
insurance scheme in order to reduce the burden
of high cost of medication. Patients need to be
educated on the adverse effects of their
medications and how to manage such effects in
order to enhance compliance.
Furthermore, monitoring by health professionals
would go a long way in improving compliance
with anti-hypertensive
Funding Source(s): Other personal
Poster Session and Number: C, #1059
Improving Efficacy of Treatment Plans for
Hospitalized Patients with Uncontrolled
Diabetes during Transitions of Care
Michael Oravec, Summa Health System; Lynn
Clough, Summa Health System; James Salem,
Summa Health System; Jason Kunz, Summa
Health System; Michelle Cudnik, Summa Health
System; Megan Elavsky, Northeast Ohio
Medical University; Robert Woods, Northeast
Ohio Medical University
Presenter: Michael Oravec, Research
Associate, Summa Health System
oravecm@summahealth.org
Research Objective: Medical care during
transitions from inpatient to outpatient provides
opportunities for assessing the efficacy and
intensification of current treatment plans for
patients with uncontrolled diabetes. The
resource-intense inpatient setting offers
opportunities to understand the impact of
comorbidities on optimal management of
diabetes. The post-hospital follow up visit in the
outpatient setting offers opportunities to address
the impact of comorbidities on self-management.
However, clinical inertia may inhibit providers
from realizing these opportunities. The purpose
of this study is to assess the impact of
hospitalization on patients with diabetes and
factors associated with medication adjustment
and glycemic control during the follow up
outpatient care.
Study Design: We conducted a retrospective
cohort study of 300 hospitalized and 333
nonhospitalized patients with diabetes. Patients
had baseline and outcome A1c taken
approximately 6-12 months apart, during regular
outpatient care over the same period.
Hospitalized patients received both inpatient and
outpatient treatment between the two A1c
measures. Multivariate linear regression was
used to model predictors of A1c change from
baseline to outcome. Multivariate logistic
regression was used to model predictors of
medication adjustment (in patients with baseline
A1c less than 7 percent) or intensification (in
patients with baseline A1c 7 percent or greater)
between baseline and outcome clinic visits. The
multivariate analyses were adjusted for the
presence of comorbidities by use of a scale
comprised of the sum of nine costly comorbid
conditions. In addition, depression was included
as an independent predictor due to its
documented influence on diabetes treatment.
Population Studied: Our study population
consisted of patients with diabetes in a hospitalbased internal medicine residency continuity
clinic.
Principal Findings: Hospitalization was not a
significant predictor of A1c change. Hospitalized
patients with baseline A1c less than 7 percent
were more likely to have therapy adjusted (OR
3.048, p=.0042), but this trend did not extend to
intensification in patients with baseline A1c 7
percent or greater (OR 0.975, p=.2487). A
significant predictor of medication intensification
was having a specialized Chronic Care Modelbased outpatient diabetic planned visit (DPV)
(OR 1.633, p=.0202). Depression was not a
significant predictor for medication therapy
change in well-controlled diabetics, but was
actually associated with a lower likelihood for
medication intensification in poorly-controlled
diabetics (OR 0.496, p=.0040).
Conclusions: This study supports previous
research in that encounters related to transitions
of care from inpatient to outpatient setting may
be missed opportunities to improve the long
term outcomes for patients with uncontrolled
diabetes. These critical opportunities can be
used to improve overall efficacy of treatment
plans and treat comorbid depression. Based on
our findings, hospitalized patients may benefit
from a standard screening for depression
symptoms during inpatient stays with outpatient
follow up visits focused on treatment
intensification that also addresses depression.
Implications for Policy, Delivery, or Practice:
Given the evidence of the effectiveness of the
DPV in our clinic, a next step for research is to
redesign the DPV proximate to discharge to
include treatment of depression and assess
efficacy for A1c reduction.
Funding Source(s): No Funding
Poster Session and Number: C, #1060
Trends in the Number of Manufacturers for
Critical Chemotherapy Drugs: Implications
for Future Drug Shortages and Treatment
Helen Parsons, University of Texas Health
Science Center at San Antonio; Susanne
Schmidt, The University of Texas Health
Science Center at San Antonio; Mary Jo Pugh,
The University of Texas Health Science Center
at San Antonio; Anand Karnad, The University of
Texas Health Science Center at San Antonio
Presenter: Helen Parsons, Assistant Professor,
University of Texas Health Science Center at
San Antonio
parsonsh@uthscsa.edu
Research Objective: Congress has identified
the critical need to evaluate contributors to and
solutions for chemotherapy drug shortages. In
2010, the U.S. Food and Drug Administration
identified 178 reported drug shortages, 132 of
which involved sterile injectable drugs, such as
cancer drugs. These shortages may force
physicians to prioritize patients, improvise
standard treatment regimens and potentially
choose unproven treatment options for patients
with curable diseases. Therefore, the objective
of this research is to evaluate trends in the
number of distinct manufacturers for critical
chemotherapy drugs over time.
Study Design: We conducted a retrospective
observational study using information obtained
from the 2003-2009 Redbook: Pharmacy’s
Fundamental Reference. As the pre-eminent
resource for clinical pharmacists and physicians,
the Redbook publishes comprehensive
manufacturing information for all FDA-approved
drugs in the US each year. We abstracted
complete information on the number of distinct
manufacturers for each of the FDA-approved
cancer drugs for first-line treatment of colon,
lung and breast cancer by year. Exploratory
analysis was used to quantify trends in
manufacturing by cancer site and drug over
time.
Population Studied: Drug manufacturers for
FDA-approved chemotherapy drugs.
Principal Findings: There were 16 drugs
approved to treat first-line colon, breast and lung
cancer in 2003, which increased to 31drugs in
2009. The median number of manufactures for
colon cancer drugs was 7 (range:1-8) in 2003
vs. 1 in 2009 (range:1-16); 3 for breast cancer
drugs in 2003 (range:1-11) vs. 3 in 2009
(range:1-16); and 5.5 (range:1-11) for lung
cancer drugs in 2003 vs. 3 (range:1-14) in 2009.
While there were only 3 FDA-approved colon
cancer drugs in 2003, one of them had only one
manufacturer as a result of a patent. By 2009, 5
out of 8 FDA-approved colon cancer drugs had
only one manufacturer—but the two mainstay
therapies, fluorouracil and leucovorin, had 7-16
manufacturers across all years. Out of 7 FDAapproved drugs for breast cancer in 2003, 2 had
only one manufacturer. By 2009, 6 out of 13 had
only one manufacturer- but again many of the
components of common therapeutic regimens
(cyclophosphamide, doxorubicin, flourouricil,
and methotrexate) had 3-16 manufacturers
across all years. Finally, for lung cancer, there
were 6 FDA-approved drugs in 2003, 2 of which
had only one manufacturer. In 2009, 5 out of 10
drugs had only one manufacturer, but the
common therapeutic regimens (carboplatin,
cisplatin, and etoposide) all had between 1-11
manufacturers.
Conclusions: Our data indicate that the
majority of mainstay chemotherapy regimens for
colon, lung and breast cancer have multiple
registered manufacturers each year, which can
fluctuate greatly. However, the number of
manufacturers does not necessarily correlate
with the available supply for a given drug
considering the large number of reported drug
shortages for chemotherapy drugs that continue
throughout the US.
Implications for Policy, Delivery, or Practice:
Despite the large number of registered
manufacturers for mainstay chemotherapy drugs
in the US, drug shortages remain common.
Future policies should promote the
dissemination of clear information regarding total
drug supply and current manufacturers to
evaluate contributors to and predictors of
shortages in the oncology community.
Funding Source(s): NIH
Poster Session and Number: C, #1061
The Underutilization of Rehabilitation
Services for Older Adults with Cancer
Mackenzi Pergolotti, University of North
Carolina, Chapel Hill; Allison Deal, Lineberger
Comprehensive Cancer Center, Univeristy of
North Carolina at Chapel Hill; Phyo Htoo,
Univeristy of North Carolina at Chapel Hill;
Hyman Muss, Lineberger Comprehensive
Cancer Center, Univeristy of North Carolina at
Chapel Hill
Presenter: Mackenzi Pergolotti, Post-Doctoral
Fellow, University of North Carolina, Chapel Hill
pergolot@email.unc.edu
Research Objective: Older adults are at greater
risk of cancer and suffering adverse
consequences of that cancer and its associated
treatments. Cancer rehabilitation services seek
to reduce morbidity, mortality and improve the
quality of life of individuals, however little is
known about the needs and use of cancer
rehabilitation for older adults. The objectives of
this study are to: (1) identify the needs of older
adults with cancer and (2) determine the
predictors of use of rehabilitation services by this
group.
Study Design: This study will analyze data from
an institution-based registry of older adults 65 +
with cancer. Variables include cancer type and
status, functional status, co-morbidities, falls,
timed up and go (TUG), social status, emotional
and tangible support scores, demographics, and
cognition. Descriptive and regression analyses
will be performed.
Population Studied: 533 individuals with
cancer over the age of 65 were examined.
Principal Findings: Preliminary findings
suggest out of the 533 patients with cancer, 69%
have at least one functional deficit and 50%
have at least two functional deficits requiring
supportive services. Of the patients with
functional deficits defined, 40% have difficulty
with Activities of daily Living (ADL); 35% with
Instrumental Activities of Daily Living (IADL) and
25% have fallen at least once in the last 6
months.It is hypothesized that Caucasian
American patients with breast cancer are most
likely to receive rehabilitation services.
Conclusions: This project is significant because
it is the first to outline and define predictors of
functional need for cancer rehabilitation in an
older population. At least 50 % of the individuals
assessed needed cancer rehabilitation. These
results outline the great need for intervention
and referrals to supportive care for older adults
with cancer.
Implications for Policy, Delivery, or Practice:
There is severe underutilization of cancer
rehabilitation for older adults with cancer. Future
policy should be directed towards identifying and
breaking down barriers to receipt of
rehabilitation services by older cancer patients
to decrease morbidity, and improve quality of
life.
Funding Source(s): NIH
Poster Session and Number: C, #1062
Associations between Traumatic Brain
Injury, Suspected Psychiatric Conditions,
Terri Pogoda, VA Boston Healthcare System; , ;
Kelly Stolzmann, VA Boston Healthcare System,
Center for Healthcare Organization and
Implementation Research; Katherine Iverson,
VA Boston Healthcare System, Center for
Healthcare Organization and Implementation
Research, National Center for Posttraumatic
Stress Disorder; Boston University School of
Medicine; Errol Baker, VA Boston Healthcare
System, Center for Healthcare Organization and
Implementation Research; Maxine Krengel, VA
Boston Healthcare System; Boston University
School of Medicine; Henry Lew, Defense and
Veterans Brain Injury Center (DVBIC), Virginia
Commonwealth University, Richmond, VA; John
A. Burns School of Medicine, University of
Hawaii at Manoa; Mark Meterko, VA Boston
Healthcare System, Center for Healthcare
Organization and Implementation Research;
Boston University School of Public Health; ,
Presenter: Terri Pogoda, Research Health
Scientist, VA Boston Healthcare System
terri.pogoda@va.gov
Research Objective: To examine the
relationship between sociodemographic
characteristics, traumatic brain injury (TBI)
history, suspected psychiatric conditions, current
health symptoms, and employment status in
Veterans evaluated for TBI in the Department of
Veterans Affairs (VA).
Study Design: Retrospective cross-sectional
database review of TBI evaluations documented
between October, 2007 and June, 2009.
Population Studied: Operation Enduring
Freedom/Operation Iraqi Freedom Veterans (n =
11,683) who completed a comprehensive VA
TBI evaluation.
Principal Findings: The main analysis was a
multinomial logistic regression. Adjusted odds
ratios (aOR) and 95% confidence intervals (CIs)
are described below. Relative to Veterans who
were employed/students, those who were
unemployed/not looking for work were
significantly more likely to have: 1) completed
high school or less, relative to a Bachelor’s
degree (aOR = 1.74, 1.34-2.26); 2) deploymentrelated moderate (aOR = 1.61, 1.28-2.03) or
severe (aOR = 1.49, 1.23-1.79) TBI history than
mild TBI history; 3) suspected PTSD (aOR =
1.38, 1.18-1.60), depression (aOR = 1.20, 1.061.36), and drug abuse/dependence (aOR = 2.88,
2.04-4.07); and 4) more severe self-reported
affective (aOR = 1.24, 1.12-1.38), cognitive
(aOR = 1.26, 1.16-1.38), and vestibular
symptoms (aOR = 1.22, 1.11-1.34). A
comparison with Veterans who were
unemployed/looking for work yielded similar, but
less robust, findings.
Conclusions: Veterans who have moderate or
severe TBI, suspected psychiatric conditions,
and who self-report more severe affective,
cognitive, and vestibular symptoms are at
greater risk for being unemployed/not looking for
work, relative to those who are
employed/students. Focused outreach and
intervention for the unique health and
employment needs of OEF/OIF Veterans may
help this cohort achieve their academic and
vocational potential.
Implications for Policy, Delivery, or Practice:
Including vocational rehabilitation as part of an
inter-disciplinary TBI evaluation process may
facilitate OEF/OIF Veterans in their readjustment
to civilian life.
Funding Source(s): VA
Poster Session and Number: C, #1063
Impact of Diagnosis and Treatment of
Colorectal Cancer on Health-Related Quality
of Life Among Older Americans
Caroleen Quach, University of North Carolina at
Chapel Hill; Hanna K. Sanoff, University of North
Carolina at Chapel Hill; Grant R. Williams,
University of North Carolina at Chapel Hill;
Jessica C. Lyons, University of North Carolina at
Chapel Hill; Bryce B. Reeve, University of North
Carolina at Chapel Hill
Presenter: Caroleen Quach, Graduate
Research Assistant, University of North Carolina
at Chapel Hill
cquach@unc.edu
Research Objective: Limited research on the
impact of diagnosis and treatment of colorectal
cancer (CRC) on health-related quality of life
(HRQoL) among older Americans exists. Study
objectives were to: 1) estimate change in
HRQoL from before to after CRC diagnosis and
2) evaluate whether HRQoL declines among
CRC patients across disease stages are greater
compared to matched controls without cancer.
Study Design: This population-based study
used the Surveillance, Epidemiology, and End
Results-Medicare Health Outcomes Survey
(SEER-MHOS) dataset. CRC patients (n=349)
were matched to non-cancer controls (n=1,745)
using propensity scores. Analysis of covariance
models estimated HRQoL change—as
measured by the Medical Outcomes Study Short
Form-36 (SF-36)/Veterans RAND 12-item
Health Survey—from before to after CRC
diagnosis. Covariates included sociodemographic, clinical and survey characteristics.
The SF-36 has 8 subscales (physical functioning
[PF], role limitation due to physical health [RP],
bodily pain [BP], general health [GH], mental
health [MH], role limitation due to emotional
health [RE], social functioning [SF], vitality [VT]),
and two summary scores, Mental Component
Summary (MCS) and Physical Component
Summary (PCS). Higher scores indicate better
HRQoL.
Population Studied: Medicare managed care
beneficiaries aged>64 years diagnosed with
CRC between completion of baseline and followup MHOS and controls who completed both
MHOS.
Principal Findings: Mean time from diagnosis
to MHOS follow-up was 12.3 (SD 9.8) months
for CRC patients (n=103 Stage I, 122 Stage II,
95 Stage III, 29 Stage IV). CRC patients
experienced statistically and clinically significant
declines in MCS (-8.25; p<.01) and PCS (-9.16;
p<.01). Higher disease stage was associated
with poorer MCS and PCS scores at follow-up.
Stage I patients reported decreases in 6
subscales (all except MH and RE; each p<.05)
at follow-up, while Stages II, III and IV patients
experienced decrements in all domains except
MH (each p<.05). Compared to controls, CRC
patients had lower scores in all subscales
except MH, with the greatest declines observed
in the RP and RE domains. When comparing
HRQoL change among patients across disease
stages with controls, differences in PCS decline
were significant for all disease stages; while
differences in MCS change were only significant
for Stage III and Stage IV patients. Declines in
the GH and VT subscales among patients within
each disease stage were significant compared
to change among controls. Declines in HRQoL
within each disease stage were significantly
worse (each p<.05) compared to the change
observed among controls for the following
subscales: Stage I (GH, VT); Stage II (RP, GH,
VT); Stage III (RP, BP, GH, SF, VT); Stage IV
(RP, GH, RE, SF, VT).
Conclusions: CRC diagnosis and treatment
have significant adverse effects on HRQoL,
particularly fatigue and general health. Overall
physical health declined across all disease
stages. HRQoL decrements among Stage III
and IV patients are consistently greater in most
subscales compared to controls.
Implications for Policy, Delivery, or Practice:
A better understanding of which HRQoL
domains are most affected by CRC is crucial for
effective disease management among older
Americans. Further, increased debility among
later-stage patients should be considered when
weighing the risks and benefits of treatment.
Funding Source(s): NIH
Poster Session and Number: C, #1064
Art Therapy Among Military Service
Members and Veterans with Post-Traumatic
Stress Disorder: A Systematic Review
Jeremy Ramirez, California State University,
Long Beach; Mercedes Guilliaum, California
State University, Long Beach; Erlyana Erlyana,
California State University, Long Beach
Presenter: Jeremy Ramirez, Student, California
State University, Long Beach
jeremy.c.ramirez@gmail.com
Research Objective: The objective of this
systematic review is to investigate the
effectiveness of art therapy as a supplemental
treatment for current military service members
and veterans, of both genders, diagnosed with
Post-Traumatic Stress Disorder (PTSD). The
preferred method of treatment for patients
enrolled in the Veterans Health Administration
(VHA), who are diagnosed with PTSD, is
Cognitive-Behavioral Therapy (CBT). CBT,
however, is effective in treating only two of the
three hallmark symptom clusters associated with
PTSD, i.e., re-experiencing and hyper-vigilance,
but not nearly as effective in treating the third
symptom cluster, avoidance/emotional numbing.
As a result, patients do not receive adequate
treatment for the symptoms related to
avoidance/emotional numbing, which include:
social detachment, memory loss, and
depression.
Study Design: This systematic review was
guided using the Preferred Reporting Items for
Systematic Reviews and Meta-Analysis
(PRISMA) statement. An electronic literature
search of two databases was conducted using
CINAHL and an enhanced version of PubMed.
All articles containing the terms “art therapy” and
“post-traumatic stress disorder” as subject
headings were identified. Articles were included
in the review if they met the following inclusion
criteria: discussed the treatment of art therapy
among persons with post-traumatic stress
disorder. Case studies, recommendations and
reviews were excluded as well as articles where
art therapy was not included as a principal
treatment of the study. English-only articles were
selected and there were no restrictions on
publication date or status.
Population Studied: The target population for
this review is military service members
diagnosed with PTSD.
Principal Findings: The literature search
identified 132 total references, 12 of which met
the inclusion and exclusion criteria and were
selected for review. Each of the articles’
objectives, target population, methods, and
principal findings were examined. The majority
of these articles suggest that through the
practice of art therapy, patients with PTSD
experienced at least three significant outcomes:
(1) the ability to express thoughts which could
not previously be verbalized, (2) improved social
relationships which led to reduced social
detachment, and (3) a general reduction in reexperiencing, hyper-vigilance, and
avoidance/emotional numbing symptom clusters
with notable improvements in experiencing less
anxiety, being able to control intrusive thoughts,
and feeling less emotionally numb.
Conclusions: Military service members with
combat exposure and military sexual trauma
(MST) are prone to developing PTSD. Given the
effectiveness art therapy has in treating the
avoidance/emotional numbing symptom cluster,
it is not meant to replace CBT, but rather, it is
meant to be offered in addition to CBT in order
to produce a greater comprehensive care
package offered to veterans diagnosed with
PTSD who are enrolled in the VHA network.
One of the barriers for why it is not yet widely
implemented is due to the highly masculinized
culture of the military. Male service members,
who are more prone to developing PTSD than
females, commonly view PTSD as
emasculating.
Implications for Policy, Delivery, or Practice:
Veterans diagnosed with PTSD
disproportionately suffer from a myriad of health
and socio-economic disparities. As a result of
service members participating in art therapy
workshops, a reduction in veteran
unemployment, substance abuse,
homelessness, incarceration, and suicide could
be realized.
Funding Source(s): No Funding
Poster Session and Number: C, #1065
The Association between Disability, Health
and Multiple Chronic Conditions among Dual
Eligibles
Amanda Reichard, University of New
Hampshire; Michael Fox, Division of Human
Development and Disability / Centers for
Disease Control and Prevention
Presenter: Amanda Reichard, Assistant
Professor, University of New Hampshire
Amanda.Reichard@unh.edu
Research Objective: The objective of this study
is to better understand the relationship between
functional disabilities and multiple chronic
conditions within a high cost segment of the U.S.
publicly-insured population, the dual eligibles.
Two research questions were posed: 1. What is
the prevalence of multiple chronic conditions for
discrete socio-economic and health
characteristics among dual eligibles in the
United States overall and stratified by age
category and type of disability limitation?
2. What are the odds of having multiple chronic
conditions among dual eligibles adjusted for
socio-demographic and health characteristics,
stratified by age category and type of disability
limitation?
Study Design: Medical Expenditure Panel
Survey (MEPS) data for 2005 through 2010
were stratified by ages 18 to 64 and 65 or older
to account for unique subsets of dual eligibles.
Prevalence of MCC was calculated for those
with physical disabilities, physical plus cognitive
disabilities, and all others, accounting for
sociodemographic and health-related factors.
Adjusted odds for having MCC were calculated
by using logistic regression.
Population Studied: Noninstitutionalized adults
aged 18 years or older who were identified as
dual eligibles if they had “coverage at any time”
for both Medicare and Medicaid in years 2005 2010.
Principal Findings: We found that 53% of dual
eligibles 18-64 had MCC compared to 73.5% of
65+. For dual eligibles with MCC, mean age and
total annual expenditures were 51 and $18,137
(18-64), and 75 and $14,364 (65+), respectively.
Sixty-five percent of all dual eligibles had 2+
chronic conditions and among those 65+ with
physical disabilities and cognitive limitations,
36% had four or more, with hypertension and
arthritis the leading conditions. Dual eligibles
18-64 having a usual source of care were 127%
more likely to have MCC than those without a
usual source of care, but among those 65+,
these odds rose to 202%. Women aged 18-64
were 24% more likely to have a MCC than men.
Conclusions: Within the dual eligible
population, disability subgroups have unique
associations with multiple chronic conditions.
Understanding disability better among dual
eligibles allows us to better understand the
relationship between health and chronic
conditions for dual eligible populations and other
segments of our society with complex health and
medical needs.
Implications for Policy, Delivery, or Practice:
Understanding the relationship between health
and disability better among dual eligibles can
allow policy makers to more effectively design
cost-containment strategies, case managers to
better navigate consumer-directed health plans,
and dual eligibles to participate in practices that
could lead to improved health at reduced costs
to themselves and newly emerging health
systems.
Funding Source(s): CDC
Poster Session and Number: C, #1066
Cancer and Medical Debt: Evidence from the
Panel Study of Income Dynamics
Patrick Richard, Uniformed Services University
Presenter: Patrick Richard, Assistant Professor,
Uniformed Services University
patrick.richard@usuhs.edu
Research Objective: High treatment costs of
cancer in the United States may impose
substantial financial hardships on patients and
their families through the accrual of medical
debt. This paper investigates the impact of
cancer on medical debt and examines whether
the Medicaid expansion in the Affordable Care
Act (ACA) will reduce medical debt associated
with cancer.
Study Design: From a theoretical standpoint,
this study starts with Becker’s household
production model to hypothesize that household
members are altruistic and jointly maximize
utility given the household’s preferences,
income, assets, budget set and price of care.
Faced with an idiosyncratic health shock such
as cancer, households produce health in a
manner that is consistent with the Grossman
model by using a combination of market and
non-market inputs. Depending on insurance
coverage and treatment costs, the household
may face high out-of-pocket costs (the actual
price of receiving care). Based on the
permanent income theory, households that
experience high out-of-pocket costs may have to
borrow money, hence accumulating medical
debt, to smooth consumption. This paper
considers three potential mechanisms through
which cancer may have an effect on medical
debt: financial resources (employment, income
and wealth), insurance and out-of-pocket costs.
Population Studied: This study uses data from
the 2011 Panel Study of Income Dynamics
(PSID), the first year in which unsecured debt
was divided into several categories such as
credit card debt, student loans, medical debt,
legal bills and debt from relatives. The sample
was restricted to 5,918 households with heads
of household who were between 18 and 65
years old. An indicator of household cancer
status was created if the head of the household
or his/her spouse reported being diagnosed with
cancer by a health professional.
Principal Findings: I use several OLS and tobit
model specifications and find a positive and
significant association between cancer and
medical debt. For instance, the (semi) elasticity
of cancer on medical debt ranges from 0.79 to
0.42 (p<0.001) for tobit models (expected log of
medical debt at the means, given that debt has
not been censored). Subsample analyses,
conditional on households with positive debt
(secured and unsecured), show similar results
(0.86-0.48, p<0.001). Additionally, inverse
probability weighting propensity score models
show elasticities of cancer on medical debt
ranging from 0.40 to 0.32 (p<0.001). These
models account for systematic differences in
observables between households reporting a
diagnosis of cancer compared to those with no
cancer. Although it is unlikely that medical debt
would cause cancer, this study plans to use
instrumental variables to address potential
omitted variables bias. Analyses stratified by
income quartiles and insurance types will be
conducted as well to address heterogeneity in
findings. Finally, counterfactual policy simulation
of the ACA Medicaid expansion in reducing
medical debt associated with cancer will be
conducted.
Conclusions: Findings will contribute to the
growing literature on medical debt and the
broader literature of health and socioeconomic
status (SES) while accounting for issues of
endogeneity.
Implications for Policy, Delivery, or Practice:
These findings will help to educate and guide
policymakers, patients, physicians, and public
health professionals about the potential financial
burden of cancer treatment.
Funding Source(s): Other
Poster Session and Number: C, #1067
The Burden of Medical Debt Faced by
Households with Chronic Health Conditions
in the United States
Patrick Richard, Uniformed Services University;
Laura Burke, Uniformed Services University
Presenter: Patrick Richard, Assistant Professor,
Uniformed Services University
patrick.richard@usuhs.edu
Research Objective: Several studies have
documented that patients with chronic health
conditions face substantial financial hardships
such as high medical bills compared to those
without chronic health conditions, due to high
out-of-pocket costs. This study examines the
effect of chronic health conditions on medical
debt while accounting for potential endogeneity
issues.
Study Design: From a theoretical standpoint,
we use Becker’s household production model as
a starting point. We hypothesize that household
members maximize utility by jointly choosing
consumption, savings, and borrowing levels
given the household’s preferences, income,
assets, budget set and price of care. Faced with
a chronic health condition, households produce
health in a manner that is consistent with the
Grossman model by using a combination of
market and non-market inputs. Depending on
insurance coverage and treatment costs, the
household may face high out-of-pocket costs
(the actual price of receiving care). Based on the
permanent income theory, households that
experience high out-of-pocket costs may have to
borrow money, hence accumulating medical
debt, to smooth consumption. This paper
considers three potential mechanisms through
which chronic health conditions may have an
effect on medical debt: financial resources
(employment, income and wealth), insurance
and out-of-pocket costs.
Population Studied: This study uses data from
the 2011 Panel Study of Income Dynamics
(PSID), the first year in which unsecured debt
was divided into several categories such as
credit card debt, student loans, medical debt,
legal bills and debt from relatives. The sample
was restricted to 4,758 households with heads
of household who were between 18 and 65
years old. The Healthcare Cost and Utilization
Project Chronic Condition Indicator was used to
measure chronic conditions reported by either
the head of the household or his/her spouse.
Principal Findings: We use several logtransformed OLS and tobit model specifications
and find a positive and significant association
between the number of chronic health conditions
and medical debt. For instance, the elasticity of
chronic health conditions on medical debt
ranges from 0.85 to 1.42 (p<0.001) for tobit
models (expected log of medical debt at the
means, conditional on censoring). Subsample
analyses, conditional on households with
positive debt (secured and unsecured), show
similar results. Additionally, propensity score
models that account for systematic differences
in observables between households reporting
chronic health conditions and those with no
chronic health conditions show similar and
robust findings. For instance, elasticities of
inverse probability weighting propensity score
models range from 0.97 to 0.70 (p<0.001).
Additionally, this study plans to use instrumental
variables to address potential omitted variables
bias and reverse causality. Analyses stratified by
income quartiles and insurance types will be
conducted as well to address heterogeneity in
findings. Finally, counterfactual policy simulation
of the ACA Medicaid expansion in reducing
medical debt associated with chronic health
conditions will be conducted.
Conclusions: Findings will contribute to the
growing literature on medical debt and the
broader literature of health and socioeconomic
status (SES), while accounting for issues of
endogeneity.
Implications for Policy, Delivery, or Practice:
This analysis will help to educate and guide
policymakers, patients, physicians, and public
health professionals about the potential financial
burden of the treatment of chronic health
conditions.
Funding Source(s): Other
Poster Session and Number: C, #1068
The Multi-Disciplinary Team Improves the
Ability of a Practice to Provide
Comprehensive Care to Patients with
Chronic and Complex Conditions in Ontario,
Canada and New Zealand
Juliet Rumball-Smith, University of Toronto;
Walter Wodchis, University of Toronto; Toni
Ashton, University of Auckland; Tim Kenealy,
University of Auckland; Jan Barnsley, University
of Toronto
Presenter: Juliet Rumball-Smith, Research
Associate, University of Toronto
juliet@rumballsmith.co.nz
Research Objective: To investigate the
association between the multi-disciplinary team
and the delivery of comprehensive care to adults
with chronic complex conditions within primary
care in Ontario, Canada and New Zealand.
Study Design: General practices in Ontario and
New Zealand were surveyed using the Quality
and Costs of Primary Care in Europe
(QUALICOPC) questionnaire. Regression
models were used to calculate estimates of the
association between the number of disciplines
working within each center and indicators of the
comprehensiveness of care. The models
included terms to control for the age and sex of
the general practitioner, practice roster size,
demographic characteristics of patients, and
country.
Population Studied: 352 general practitioners
throughout New Zealand (n=169) and Ontario
(n=183) completed the physician questionnaire.
After exclusion of those with missing answers,
the data from approximately 300 participants
were available for analysis.
Principal Findings: There were positive
associations between the number of disciplines
on the primary care team and a number of
indicators of the comprehensiveness of care.
These outcomes included: the extent of
independent nurse service provision (including
health promotion activities, immunization) (beta
0.11; 95% CI 0.06, 0.16) the provision of special
sessions for the elderly and patients with
diabetes or hypertension (0.14; 95% CI 0.07,
0.21); participation in disease management
programs for asthma, chronic obstructive
pulmonary disease and diabetes (0.15; 95% CI
0.05, 0.24); and the comprehensiveness of the
equipment available at the practice (0.74; 95%
CI 0.60 – 0.88).
Conclusions: The number of disciplines on the
primary care team is associated with the
comprehensiveness of care available for
patients with complex and chronic conditions.
Implications for Policy, Delivery, or Practice:
There is published evidence that the multidisciplinary team may provide more coordinated
and effective clinical care for some groups of
patients. Our findings suggest that this strategy
is also associated with the ability to provide
more comprehensive care for patients with
chronic complex conditions within the patientcentered medical home.
Funding Source(s): N/A Canadian Institutes of
Health Research
Poster Session and Number: C, #1069
Effect of Electronic Communication During
Deployment on PTSD and Deployment
Adjustment in OEF/OIF Reserve and National
Guard Servicewomen
Anne Sadler, Iowa City VAMC; Michelle
Mengeling, Iowa City VAHCS, Dept of Internal
Medicine, University of Iowa; James Torner,
University of Iowa College of Public Health;
Brenda Booth, Department of Psychiatry,
University of Arkansas for Medical Sciences
Presenter: Anne Sadler, Researcher, Deputy
Director- Iowa City Vamc Mental Health Service
Line, Iowa City VAMC
anne.sadler@va.gov
Research Objective: To investigate differences
in PTSD between OEF/OIF Reserve and
National Guard (RNG)servicewomen who did or
did not use electronic communication (EC)
during deployment. It also described the
association between EC and self-reported
deployment adjustment.
Study Design: This was a cross sectional
retrospective study with participants stratified by
deployment experience.
Population Studied: A community sample of
665 OEF/OIF RNG servicewomen returning
from deployment within three years prior to
study participation participated (70% response
rate).
Principal Findings: Median ge age of
participants was 37 years. Most servicewomen
(93%) used some form of electronic
communication with almost half (47%) using
daily and 40% weekly. Traumatic exposures
during deployment were common with most
frequently reported exposures: fear of being
killed (61%), seeing a dead body (49%), having
an IED explode nearby (26%), interacting with
enemy POWs (28%), and being responsible for
body searches (28%). No difference was found
in PTSD diagnoses (65%, PSS-I) by EC use,
although differences were found in
servicewomen’s EC perceptions by PTSD
diagnoses. Most participants (88%) reported
being better able to cope during deployment
after EC. Women with PTSD were more likely
to: limit their EC time to focus on their
deployment duties (59%v36%, p <.001); worry
more about personal safety after EC (15%v6%,
p<.006); learn things during EC they wish they
hadn’t (36%v18%, p.002) and pretend things
were ok during EC (78%v43%,p<.0001).
Servicewomen without PTSD reported they
worried less about their family (81%v68%,p<.01)
or their significant other (59%v38% p<.006) after
EC.
Conclusions: Deployed RNG servicewomen
have frequent and severe traumatic exposures
during deployment. EC was reported by most to
help them cope better during deployment,
women with PTSD had more negative stressors
associated with EC, such as having to pretend
things were ok and more worry. EC use was not
found to associated with PTSD.
Implications for Policy, Delivery, or Practice:
While EC was not a protective factor for postdeployment PTSD, it had both positive and
negative associations for deployed RNG
servicewomen.
Funding Source(s): VA
Poster Session and Number: C, #1070
Variations in Health Care Use by Children
with Chronic Illness: A Population-wide
Analysis
Lee Sanders, Stanford University; Vandana
Sundaram, Stanford University; Lisa
Chamberlain, Stanford University; Ewen Wang,
Stanford University; Paul Wise, Stanford
University
Presenter: Lee Sanders, Stanford University
leesanders@stanford.edu
Research Objective: To assess populationwide variation in the use and quality of care
received by children with serious chronic illness.
Study Design: Cross-sectional study of
administrative data.
Population Studied: From paid claims for
323,922 children enrolled > 6 month in a row
(2007-2012) in California's Title V program -- we
described the mean, median and interquartile
ranges for per child annualized use of care in 8
domains (hospital, primary, subspecialty,
emergency, diagnostic, home health, residential,
dental, pharmacy). Non-parametric tests were
used to compare differences by child age,
parent language, county, disease complexity,
primary diagnostic category, and most common
primary diagnoses. Heirarchical-clustering
analysis was applied by diagnostic category.
Intra-class coefficients assessed betweencounty variation in outpatient-to-hospital ratios.
Principal Findings: Most common frequencies
of care use were in pharmacy (mean 18.44, SD
30.38), home health (14.79, 21.46), inpatient
(13.7, 28.4), diagnostic (14.31, 22.16), and
subspecialty (5.36, SD 6.22). Children < 1 year
had high overall and hospital use; with no agerelated variations from 2-18 years. Children
living in Spanish-speaking households and in
rural regions had a lower use of home health.
Three dominant care-use patterns were
hospital+pharmacy (neonatology, cardiology,
pulmonology, oncology, orthopedics), homehealth+outpatient (neurology, ENT), and
outpatient+pharmacy (endocrinology,
hematology, gastroenterology). Significant
regional variation was observed for most quality
metrics -- including outpatient MD visit during
the 30 days prior to hospitalization (mean
57.1%) and during the 7 days after
hospitalization (mean 21.1%). For the most
common diagnostic category
(neurodevelopment), significant regional
variations were observed in the use of outpatient
care generally and pre- and post-hospitalization.
Conclusions: Use of care by children with
serious chronic illness varies by disease
complexity, parent language and county -suggesting policy opportunities for improved
efficiency and cost reduction.
Implications for Policy, Delivery, or Practice:
Regionalized delivery systems may respond
best by developing learning collaboratives
across institutions to improve the value of
outpatient care services for children with chronic
illness.
Funding Source(s): Other California Health
Care Foundation
Poster Session and Number: C, #1071
Are There Symptom Groupings Associated
with Mild TBI? A Cluster Analysis of
Neurobehavioral Symptom Data among
Operation Enduring Freedom/Operation Iraqi
Freedom Veterans
Mark Meterko, VA Boston Healthcare System,
Center for Healthcare Organization and
Implementation Research; Errol Baker, VA
Boston Healthcare System, Center for
Healthcare Organization and Implementation
Research; Terri Pogoda, VA Boston Healthcare
System, Center for Healthcare Organization and
Implementation Research; Kelly Stolzmann, VA
Boston Healthcare System, Center for
Healthcare Organization and Implementation
Research; Katherine Iverson, VA Boston
Healthcare System, National Center for
Posttraumatic Stress Disorder; Maxine Krengel,
VA Boston Healthcare System; Boston
University School of Medicine; Marjorie Nealon
Seibert, VA Boston Healthcare System, Center
for Healthcare Organization and Implementation
Research; Nina Sayer, Minneapolis VA
Healthcare System; University of Minnesota
Presenter: Kelly Stolzmann, Research Analyst,
VA Boston Healthcare System (152M)
kelly.stolzmann@va.gov
Research Objective: To attempt to identify
consistent patterns of neurobehavioral
symptoms among Veterans judged to have
experienced a mild traumatic brain injury (TBI)
based on an evaluation conducted in the
Department of Veterans Affairs (VA).
Study Design: Retrospective secondary
analysis of a national database of demographic,
military background, injury exposure history, and
clinical information collected during
comprehensive TBI evaluations (CTBIE)
conducted in the VA between October 2007 and
June 2009. Self reported symptom severity was
measured by the Neurobehavioral Symptom
Inventory (NSI), a component of the CTBIE.
Based on previous factor analyses of the NSI,
four summary scales (Affective, Cognitive,
Somato-sensory and Vestibular) were computed
by averaging the scores across relevant items.
Cluster analysis of the NSI scales were
conducted on a derivation sample and replicated
in a confirmation sample.
Population Studied: Operation Enduring
Freedom/Operation Iraqi Freedom Veterans who
completed a CTBIE and who, based on that
examination, both: (1) met VA/Department of
Defense criteria for mild TBI with regard to loss
of consciousness, post-traumatic amnesia,
and/or alteration of consciousness; and (2) were
judged by the clinician conducting the CTBIE to
have experienced a mild TBI (n = 6871). This
total subject pool was then randomly split into
derivation and confirmation samples. Success
of randomization was confirmed by comparing
the two samples with regard to demographics
and symptom severity.
Principal Findings: Examination of the
dendrogram describing the successive
combination of clusters into hierarchical groups
in the derivation sample suggested four, five and
seven cluster solutions as plausible. To decide
among these alternatives, we generated
symptom profile plots for each of the cluster
models. Clusters were characterized by those
NSI dimensions on which the average score of
cluster members was above the mid-point of the
5-point severity scale, indicative of “moderate”
severity. The five-cluster model was the best
compromise between collapsing groups with
substantive differences and distinguishing
between groups with relatively minor
differences. The five clusters were: (1) High
across all dimensions, (2) High on the AffectiveCognitive-Vestibular, (3) High on AffectiveCognitive, (4) High on Affective only, and (5)
Low across all dimensions. Results in the
confirmation sample substantially replicated
these findings, but suggest that the AffectiveCognitive-Vestibular and Affective (only) clusters
may not be robust.
Conclusions: We were able to demonstrate,
and partially replicate, a five cluster model of
symptom patterns in a large sample of Veterans
with a history of mild TBI.
Implications for Policy, Delivery, or Practice:
Symptom clusters may provide clinically
meaningful insights into the trajectory of
symptom progression or difficulties regarding
post-deployment reintegration into civilian life for
Veterans with mild TBI and thus help guide
treatment and support.
Funding Source(s): VA
Poster Session and Number: C, #1073
Blood Pressure Control at Primary Care
Office Visits by Adults with Hypertension
and Comorbid Diabetes or Chronic Kidney
Disease, United States, 2005 and 2010
Anjali Talwalkar, Centers for Disease Control
and Prevention; Sayeedha Uddin, Centers for
Disease Control and Prevention
Presenter: Anjali Talwalkar, Senior Service
Fellow, Centers for Disease Control and
Prevention
atalwalkar@cdc.gov
Research Objective: To examine blood
pressure control at primary care office visits by
adults with hypertension over time and to assess
differences in control by presence of diabetes or
chronic kidney disease
Study Design: Data from the 2005 and 2010
National Ambulatory Medical Care Survey
(NAMCS) were analyzed. NAMCS is an annual,
cross sectional survey of visits to nonfederal
physician offices in the United States. Blood
pressure control at visits by patients 18 years
and older with documented hypertension was
examined. Visits to physicians in primary care
specialties (internal medicine, geriatrics, family
practice, obstetrics and gynecology, pediatrics)
with blood pressure recorded were included.
Visits by pregnant and postpartum patients were
excluded. Controlled blood pressure was
defined per JNC VII guidelines as a systolic
blood pressure (SBP) less than 140 mmHg and
diastolic blood pressure (DBP) less than 90
mmHg for patients without diabetes and chronic
kidney disease (CKD) and SBP less than 130
mmHg and DBP less than 80 mmHg for patients
with diabetes or CKD. Control was alternatively
defined as an SBP less than 140 mmHg and
DBP less than 90 mmHg for all patients as a
comparison. Differences were evaluated with
two tailed t tests and a significance level less
than 0.05.
Population Studied: Visits to nonfederal, office
based physicians in primary care specialties by
nonpregnant, nonpostpartum patients aged 18
years and older with documented hypertension
Principal Findings: Between 2005 and 2010,
the percentage of visits with controlled blood
pressure by patients with hypertension and
without diabetes or CKD increased from 58
percent to 63 percent (p less than 0.05). There
was no change in the percentage of visits with
controlled blood pressure by patients with
hypertension and with diabetes or CKD or at
overall visits by patients with hypertension. The
percentage of visits with controlled blood
pressure by patients with diabetes or CKD was
significantly lower; it was controlled at 28
percent of visits in 2005 and at 34 percent of
visits in 2010. Applying the 140 over 90 target
to all visits, the percentage of overall visits by
patients with hypertension with controlled blood
pressure did significantly increase from 57
percent in 2005 to 63 percent in 2010, and there
was no difference in control between visits by
patients with and without diabetes or CKD in
either year.
Conclusions: Blood pressure control improved
between 2005 and 2010 at visits by patients with
hypertension and without diabetes or CKD but
not at visits by patients with one of these
comorbidities based on JNC VII blood pressure
targets. Blood pressure was also controlled at
fewer visits by these patients compared to visits
by patients without diabetes or CKD. However,
when using the same blood pressure target for
all patients, control at overall visits by patients
with hypertension has improved, and the level of
control is similar at visits by patients with and
without diabetes or CKD.
Implications for Policy, Delivery, or Practice:
Monitoring quality indicators for chronic
conditions is critical to clinical improvement
efforts. The new JNC VIII guidelines will impact
reported measures of blood pressure control at
physician office visits.
Funding Source(s): CDC
Poster Session and Number: C, #1074
Treatment Targets and Cardiovascular
Morbidity and Mortality in Patients with
Hypertension
Alexander Turchin, Brigham and Women's
Hospital; Wenxin Xu, Harvard Medical School;
Saveli Goldberg, Massachusetts General
Hospital; Maria Shubina, Brigham and Women's
Hospital
Presenter: Alexander Turchin, Assistant
Professor Of Medicine, Brigham and Women's
Hospital
aturchin@partners.org
Research Objective: Hypertension is the most
common risk factor for cardiovascular events
worldwide. The optimal systolic intensification
threshold, time to medication intensification after
the first elevated blood pressure measurement,
and time to blood pressure follow-up after
medication intensification in the management of
hypertension are unknown. We sought to
establish the systolic intensification threshold,
time-to-intensification and time-to-follow-up
associated with the lowest risk of cardiovascular
events or death among adults with hypertension.
Study Design: We performed a retrospective
cohort study of patients with hypertension in The
Health Improvement Network (THIN) database.
Hypertension was defined as the presence of a
hypertension-related diagnosis code in the
medical record. The primary composite outcome
was defined as death or acute cardiovascular
event (myocardial infarction, stroke, congestive
heart failure, or peripheral vascular disease). We
analyzed the relationships between the systolic
intensification threshold, time-to-intensification of
anti-hypertensive medications (from the first
blood pressure measurement higher than the
systolic intensification threshold) and time-tofollow-up after an anti-hypertensive
intensification over the course of a 10-year
treatment strategy assessment period, and time
to primary outcome. Systolic intensification
threshold was defined as the lowest blood
pressure above which anti-hypertensive
medications were intensified. The Cox
regression model was adjusted for age, sex,
smoking status, socioeconomic deprivation,
history of diabetes or cardiovascular disease,
Charlson Comorbidity Index, body mass index,
medication possession ratio, and blood pressure
during the treatment strategy assessment
period.
Population Studied: We identified 81,178 adult
patients with previously diagnosed hypertension
who had been followed in primary care practices
for at least 10 years. Mean patient age was 58.1
years, and 59.7 percent of patients were female.
Median follow-up time after the treatment
assessment period was 37.7 months; 8,362
patients (10.3%) died or had an acute
cardiovascular event during the follow-up period.
Preexisting cardiovascular history (myocardial
infarction, peripheral vascular disease,
cerebrovascular accident, congestive heart
failure or peripheral vascular disease) was
present in 9.5% of patients.
Principal Findings: No difference in outcome
risk was seen between systolic intensification
thresholds of 130-150 mmHg, while systolic
intensification thresholds greater than 150
mmHg were associated with progressively
greater risk. Outcome risk increased
progressively from the lowest (0-1.4 months) to
the highest quintile of time to medication
intensification. The highest quintile of time tofollow-up (>2.7 months) was also associated
with increased outcome risk.
Conclusions: For patients with hypertension,
intensification of anti-hypertensive medications
within 1.4 months of the first measurement of
systolic blood pressure = 150 mm Hg followed
by re-measurement within 2.7 months was
associated with the lowest rate of cardiovascular
events and death of any cause.
Implications for Policy, Delivery, or Practice:
In patients with hypertension, both appropriate
blood pressure targets and timely treatment and
follow-up are important for achieving optimal
outcomes.
Funding Source(s): Other Harvard Medical
School
Poster Session and Number: C, #1075
Inequalities in Healthcare for Patients with
Sickle Cell Disease: Due to Racialization of
the Disease, Stigma, and History
Ashley Valentine, Impaq International LLC
Presenter: Ashley Valentine, Research Analyst,
Impaq International LLC
avalentine@impaqint.com
Research Objective: The objective of the study
is to explore the relationship between race and
the lack of research available about Sickle Cell
Disease (SCD). This lack of research limits the
accessible treatment options available for
patients. Given the history of racism in the U.S.
and U.K., Sickle Cell Disease was labeled as a
“black disease.” This racism gave power to
stereotypes and also racialized the illness. As a
result, racialization of the illness created
“cultural” symptoms, of the disease rather than
allowing the disease to be recognized as a
biological disorder.
In America, there are about 70,000 people who
have Sickle Cell Disease, and in the UK, about
15,000 people. People with SCD have sickle
hemoglobin (HbS), which is different from the
healthy hemoglobin (HbA). This process
produces severe episodes of pain and ultimately
can damage the tissues and vital organs and
lead to other serious medical problems, which
can result in fatality. Secondary symptoms,
which are often overlooked, include depression,
growth delay, social exclusion, lack of
progression in school, loss of job, stress on the
caregivers and family . There is no cure for
Sickle Cell Disease; however, the most widely
used treatments are opioid medication to treat
the symptoms and blood transfusions to give the
body healthy oxygen filled blood cells when the
sickled cells die. Racialization of the disease,
along with misconceptions of race versus
ethnicity and stigma all combine in limiting
access to healthcare for people with SCD along
with decreasing their life chances and quality of
life. This study looks to explore the relationship
between race and healthcare from the
perspective of patients and their caretakers.
Study Design: The study uses an
ethnographical approach with eight semi-formal
interviews. Group members agreed to
participate in 30-45 minute semi-formal
interviews throughout the study. The
participants were asked a series of questions in
regards to experiences they had with denial of
pain medication, patient perspective of the
quality of care they were receiving, and the
impacts of sickle cell on their lives. All
participants were over the age of 18.
Population Studied: The population includes
Sickle Cell patients, Sickle Cell caretakers and
family members of those with SCD. The
participants are members of a sickle cell support
group in South East London.
Principal Findings: Inequalities in healthcare
were clearly reported by the participants. The
majority of the patients reported denial of
medication and neglect by the hospital staff
while in the hospital. Respondents expressed
the idea that often times medical professionals
lack sympathy, respect, and concern for Sickle
Cell patients due to negligence and lack of
concern for their quality of care as a
consequence of race.
Another major theme was a lack of consistent
education and knowledge about SCD in the
medical field. The respondents reported
numerous occasions of being both patients and
educators simultaneously. Likewise, patients'
treatment were dependent on medical
professionals' familiarity the disease, as well as
their subjective idea of how the disease affects
the patients.
Conclusions: Physicians that have a committed
interest to Sickle Cell Disease and who are
knowledgeable about symptoms and care, lead
to less racialized medicine and opens access to
healthcare for thousands of people who are
currently have limited access to healthcare
services.
Implications for Policy, Delivery, or Practice:
This study could raise public awareness about
Sickle Cell Disease, racialized medicine and the
health implications for the patients. This shift in
public concern has the potential to further
research and continuity for Sickle Cell
education. Ultimately, this could improve patient
outcomes, not only their physical health but with
their overall quality of life.
Funding Source(s): No Funding
Poster Session and Number: C, #1076
Understanding the Characteristics and Types
of Adult and Pediatric Cancers Treated with
Proton Beam Radiotherapy in a Population of
Commercially Insured Patients
Holly Van Houten, Mayo Clinic; Robert Miller,
Mayo Clinic; Robert Foote, Mayo Clinic; Sameer
Keole, Mayo Clinic; Steven Schild, Mayo Clinic;
Nadia Laack, Mayo Clinic; Thomas Daniels,
Mayo Clinic; William Crown, Optum Labs; Nilay
Shah, Mayo Clinic
Presenter: Holly Van Houten, Senior Health
Services Analyst, Mayo Clinic
vanhouten.holly@mayo.edu
Research Objective: The aims of our study
were to advance the understanding of (i) the
characteristics of patients treated with proton
beam radiotherapy (PBRT), (ii) identify the most
common cancers treated with PBRT and (iii)
report the number of patients treated with PBRT
overall and by diagnosis In a commercially
insured population.
Study Design: Retrospective descriptive
analysis was performed. SAS statistical software
(SAS version 9.3 for Windows; SAS Institute
Inc., Cary, North Carolina) was used to perform
all statistical analyses.
Population Studied: We used administrative
claims data in this study including medical
claims, pharmacy claims and eligibility
information from a large, national US health
plan. The individuals covered by this health plan,
about 96 million unique patients, in years 1994
to 2012, are geographically diverse across the
US, with greatest representation in the South
and Midwest US census regions. The plan
provides fully insured coverage for professional
(e.g., physician), facility (e.g., hospital), and
outpatient prescription medication services.
Medical (professional, facility) claims include
International Classification of Diseases, 9th
Revision, Clinical Modification (ICD-9-CM)
diagnosis codes, ICD-9 procedure codes,
Current Procedural Terminology, Version 4
(CPT-4) procedure codes, Healthcare Common
Procedure Coding System (HCPCS) procedure
codes, site of service codes, provider specialty
codes, and health plan and patient costs.
Principal Findings: We identified 1,544 unique
patients treated with PRT from 2002 to 2012.
The number of patients treated increased from
14 in 2002 to 386 in 2012. Overall, most of the
patients were male (80%) with a mean age of 56
years. Adult patients outnumbered pediatric
patients from 2002 (93%) to 2012 (91%). A
variety of cancers were treated with PBRT. Our
data shows that in 2002 eye cancer was the top
diagnosis, followed by bone/soft tissue and
benign tumors. By 2006 prostate cancer was at
the top of the list (41%) and continues to be the
top cancer treated with PBRT in 2012 (55%).
Top diagnoses treated with PBRT in 2012 for
adults were prostate (61%,) head/neck (9%),
lung (8%), central nervous system (CNS)/brain
cancer (5%) and bone/soft tissue (5%) and for
pediatrics in 2012 were CNS/brain (69%,)
bone/soft tissue (17%), and eye (6%). Overall
for years 2002-2012, the top diagnosis treated
among adults is prostate cancer and among
pediatric patients is CNS/brain cancer. In
contrast to the 1,544 patients treated with PBRT,
473,000 patients were treated with conventional
radiotherapy other than protons. Of patients
treated with radiotherapy, PBRT made up <
0.5% of cases.
Conclusions: Although the use of PBRT to treat
prostate cancer in adults has grown, the
absolute number of cancer patients treated with
PBRT remains extremely low in relation to the
overall number of patients treated with
conventional radiotherapy.
Implications for Policy, Delivery, or Practice:
The very low utilization rate of PBRT means that
it is unlikely to substantially impact the level of
expenditure on cancer care on a national basis
for commercially insured patients in the current
environment. Changes in equipment cost,
coverage decisions, clinical evidence, and
reimbursement rates could alter the utilization
rate.
Funding Source(s): No Funding
Poster Session and Number: C, #1077
Access to Preventive Health Services for
People with Diabetes: Do Long Working
Hours Impede Access?
Xiaoxi Yao, Ohio State University; Allard
Dembe, The Ohio State University College of
Public Health; Thomas Wickizer, The Ohio State
University College of Public Health; Bo Lu, The
Ohio State University College of Public Health
Presenter: Xiaoxi Yao, PhD Candidate, Ohio
State University
yaoxx.03@gmail.com
Research Objective: It is critical for workers
with diabetes to adhere to recommended
preventive health services, to prevent severe
complications and avoid loss of productivity and
employment. This study measures key
indicators of care received by full-time diabetic
workers. Specifically, we analyze whether
working long-hour schedules impedes their
ability to obtain needed diabetic and routine care
services.
Study Design: Data from the Medical
Expenditure Panel Survey for 2002 through
2010 were used to determine average weekly
work hours for full-time employees’ having
private health insurance (predominantly
employer-sponsored insurance). Among those
individuals, we measured the use of three
diabetes-related preventive care services
recommended by the American Diabetes
Association: A1C tests, dilated eye
examinations, and foot examinations.
Additionally, we assessed their utilization of four
common preventive health services: flu
vaccinations, routine check-ups, cholesterol
screening tests, and dental check-ups.
Multivariate logistic regression analyses tested
the association between working long hours
(e.g., more than 60 hours per week) and not
obtaining a particular preventive care service
(i.e. odds ratios greater than 1.0 indicate less
likelihood of obtaining care than those working
shorter hours).
Population Studied: The study population
consisted of 2,938 full-time privately insured
employees who reported having been diagnosed
with diabetes. We confined the study population
to workers covered by private health insurance,
to control for the significant variation in the
quality of care between insured and uninsured
individuals, and among private and public
insurers.
Principal Findings: Among employees with
diabetes, 76 percent received A1C tests at least
twice a year, 59 percent received an annual
dilated eye exam, and 69 percent obtained an
annual foot exam. The use of preventive health
services was also suboptimal, especially for flu
vaccinations and dental check-ups. Only 50
percent of workers received an annual flu
vaccination and 49 percent received dental
check-ups twice a year. Regression analyses
found that working over 60 hours per week was
a significant barrier to obtaining routine checkups (OR=1.80, p<0.05), cholesterol check-ups
(OR=2.49, p<0.01), and dental check-ups
(OR=1.56, p<0.05). The association with flu
vaccinations (OR=1.17) was in the expected
direction but was not statically significant. No
statistically significant association was found
between working long hours and obtaining
dilated eye examinations, foot examinations, or
A1C tests.
Conclusions: These findings suggest the
possibility that people with diabetes give most of
their attention to controlling their diabetes, which
is desirable. However, paradoxically, this may
imply that given the limited time available
because of long working hours, they are less
likely to fulfill other routine and preventive health
care needs.
Implications for Policy, Delivery, or Practice:
The Affordable Care Act expands health
insurance coverage and eliminates copayments
for many preventive health services. Therefore,
it is even more important to identify non-financial
barriers to care. Some individuals may be able
to work long hours and still fulfill their chronic
care (e.g., diabetes) needs. However,
employers and policy makers should be aware
that these kind of long-hour work demands may
consequently make it more difficult for the
workers to obtain routine preventive care
services.
Funding Source(s): No Funding
Poster Session and Number: C, #1078
Relationship between Primary Care Home
Visits and Hospitalization and Emergency
Department Utilization among Complex
Medicare Beneficiaries
Julia Zucco, Centers for Medicare and Medicaid
Services; Marsha Davenport, Centers for
Medicare and Medicaid Services
Presenter: Julia Zucco, Health Insurance
Specialist, Centers for Medicare and Medicaid
Services
julia.zucco@cms.hhs.gov
Research Objective: Medicare beneficiaries
with multiple chronic conditions and functional
dependencies have extremely high health care
costs. This population may encounter
challenges in accessing medical services,
making provider home visits a practical care
delivery option. Although the concept of home
visits is not new, there is limited research on this
type of service. It is possible that home visits
offer improved access and more comprehensive
health care, leading to reductions in other types
of utilization such as hospitalizations and
emergency department (ED) visits. The purpose
of this study is to assess the relationship
between primary care home visits and hospital
and ED utilization.
Study Design: A retrospective observational
study was conducted using 2009 Medicare Feefor-Service claims in order to review the most
current data prior to the enactment of the 2010
Affordable Care Act (ACA). A 20% national
random sample of Medicare beneficiaries with at
least two chronic conditions, at least two
activities of daily living (ADL) limitations, and a
history of hospitalization and post-acute care
services comprised the population of interest.
Beneficiaries in Medicare Advantage plans or
long-term-care institutions were excluded.
Hospitalization and ED utilization were
compared between Medicare beneficiaries
receiving 3 or more primary care home visits
(n=20,430) and those receiving less than 3 visits
(n=231,985). We used logistic regression to
adjust for demographic characteristics, urban
residence, dual eligibility (Medicare and
Medicaid), and chronic conditions.
Population Studied: The Medicare
beneficiaries had an average age of 78.23
(SD=6.55), and 3.74 chronic conditions
(SD=2.10). Males comprised 38.15% of the
sample. Other characteristics of the study
population included 31.44% dual eligible,
84.11% Caucasian, and 79.47% of the
beneficiaries resided in an urban environment.
Principal Findings: Among those receiving
home visit services, beneficiaries received an
average of 8.23 home visits in 2009 (SD = 5.83,
range = 3.00-139.00). Those who received
primary care home services utilized inpatient
services an average of 1.43 times (SD=1.81 vs.
1.15 times, SD=1.66) and the ED an average of
2.72 times (SD=3.48, vs. 2.13 times, SD=3.22).
Compared to those who received less than 3
home visit services, beneficiaries who received
3 or more home visits had a higher odds of
hospitalization in that year (adjusted odds ratio =
1.03, 95% CI 1.00-1.07). However, these
beneficiaries had a lower odds of visiting the ED
(adjusted odds ratio = 0.83, 95% CI 0.80-0.86).
Conclusions: This observational study
suggests that, among complex Medicare
beneficiaries, primary care home visits are
associated with slightly more hospitalizations,
but less ED visits.
Implications for Policy, Delivery, or Practice:
A more detailed examination of home visits in a
similar population as studied here will be critical
to understanding the full impact on
hospitalizations and ED visits. Section 3024 of
the ACA directs the Centers for Medicare &
Medicaid Services (CMS) to test the effects of
payment incentives to practices that provide
primary care home visits in the demonstration,
entitled “Independence at Home” (IAH). The
Evaluation of IAH will provide further evidence
on whether such visits lead to lower Medicare
utilization and costs, without reduction in quality
of care.
Funding Source(s): CMS
Poster Session and Number: C, #1079
Pathways to Recovery as Understood from
High-Achieving Individuals with
Schizophrenia
Amy Cohen, Greater Los Angeles VA
Healthcare Center; Alison Hamilton, VA HSR&D
Center for the Study of Healthcare Innovation,
Implementation & Policy, UCLA; Elyn Saks,
University of Southern California Gould School
of Law; Stephen Marder, VA Desert Pacific
Mental Illness Research, Education, and Clinical
Center (MIRECC), UCLA
Presenter: Amy Cohen, Psychologist, Greater
Los Angeles VA Healthcare Center
ancohen@ucla.edu
Research Objective: Many individuals with
schizophrenia experience profound problems in
functioning. However, better outcomes are
possible, especially with increasing availability of
recovery-oriented services in mental health
clinics. The President’s New Freedom
Commission states, “Recovery...is the process
in which people are able to live, work, learn, and
participate fully in their communities.” There is
much to be learned about the experience of
recovery, especially from those who achieve
academic, occupational and social success
despite active psychotic symptoms. Study
objectives were to understand: 1) characteristics
of “high-achieving” individuals with
schizophrenia, and 2) strategies facilitating their
current functioning level.
Study Design: This study was an in-depth,
mixed methods study with high-achieving
individuals with schizophrenia. We used
purposive sampling to identify eligible individuals
in collaboration with community mental
healthcare practitioners. Subjects were
interviewed to confirm diagnosis and assess
current symptoms and functioning using goldstandard measures. Then, each individual
participated in an intensive 4-hour personcentered interview. Interview data were
analyzed using the constant comparison
method.
Population Studied: Twenty individuals with
schizophrenia who were currently employed in a
professional, technical, or managerial job per the
Dictionary of Occupational Titles, or who were
full-time students or caretakers.
Principal Findings: The sample was comprised
equally of men and women, average age of 40
years, and just over half had never married. The
sample was ethnically diverse. The majority had
master’s or doctoral degrees. Three held
professional jobs, 7 held technical jobs, 5 held
managerial jobs, 4 were full-time students, and 1
was a caretaker of elderly parents. Only two
were making over $50,000/year. In terms of
their psychiatric history and current symptoms,
the majority started experiencing psychiatric
symptoms around 18, and by the time of the
interview had experienced an average 21 years
with schizophrenia. The majority had a history
of 2-5 psychiatric hospitalizations; three had
never been hospitalized. Many reported current
mild to moderate levels of delusions,
hallucinations, social withdrawal, blunted affect,
and unusual thought content. Across the
sample, several compensatory mechanisms
were identified as helpful, including cognitive
techniques, exercise, control of their
environment (including noise and light), and
checking their symptoms with others. Other
important aspects of their self-care were
medication adherence, staying physically
healthy, engaging spirituality, and, for some,
engaging in the mental health recovery
movement. Across the sample, common
avoidances included illicit drugs and alcohol,
travel, crowds, and isolation.
Conclusions: This is one of the first in-depth
studies examining the life trajectories of highachieving individuals with schizophrenia. There
were many common elements between the
stories of this sample and the recovery literature,
including the importance of hope, a focus on
strengths (as opposed to deficits), the
importance of redefining self, participating in
meaningful activities, building a feeling of
empowerment, assuming control over one’s
illness, and overcoming stigma.
Implications for Policy, Delivery, or Practice:
The research and clinical communities could
benefit from the knowledge gained from these
individuals, and clinicians and families would
benefit from the hope their stories provide. Our
groundwork could potentially lead to improved,
more consumer-driven and consumer-informed
services for individuals with schizophrenia.
Funding Source(s): Other Greenwall
Foundation and Larson Foundation
Poster Session and Number: C, #1080
Consumer Choice and Behavioral
Economics
Medication Adherence among Commercially
Insured Individuals: Role of Cost Sharing
and Access Factors
Ibrahim Abbass, University of Texas School of
Public Health; Ajit Appari, University of Texas
School of Public Health; Jordan Mitchell,
University of Houston Clear Lake; Lee Revere,
University of Texas School of Public Health
Presenter: Ibrahim Abbass, Graduate Research
Assistant, University of Texas School of Public
Health
ibrahim.m.abbass@uth,tmc.edu
Research Objective: The objective of this study
is to examine the association of medication
adherence, among commercially insured
individuals, with cost sharing burden and factors
influencing access to medication including
number of pharmacies, and poverty level.
Study Design: We conducted a retrospective
cohort study of commercially insured individuals
on statin therapy between January 2008 and
December 2012 in the state of Texas. The
primary data source is all claims data including
pharmacy claims obtained from the BlueCross
BlueShield of Texas. Additionally, data on
socioeconomic status and pharmacies comes
from the Research Triangle Institute’s Spatial
Impact Factor database, and the Texas State
Board of Pharmacy.
Medication adherence for statin was measured
using medication possession ratio (MPR) where
MPR < 0.8 was considered non-adherence and
coded as 1, with 0 otherwise. Factors influencing
access to medication were measured by count
of active pharmacies, and proportion of
household with public assistance at ZCTA (zip
code tabulation area) level. The analysis was
performed using clustered weighted logistic
regression, adjusting for individuals’
characteristics and residence neighborhood
characteristics. These covariates included age,
gender, residence urbanity, cardiovascular risks,
comorbid conditions (diabetes, and mental
disorders), patients’ level of risk, whether the
prescribing physician was a cardiologist, and,
and ZCTA level distribution of race/ethnicity and
educational level. The total medical care
utilization six months prior to starting statin
medication was used as a proxy of patients’
level of risk. The model was weighted using the
length of duration statin therapy and errors were
clustered at the ZCTA level to account for
intragroup correlation. Data extraction and
management were performed using SAS 9.3,
and the Stata 13.1 was used for statistical
analysis.
Population Studied: 50,687 commercially
insured individuals in the state of Texas.
Principal Findings: Slightly less than half of the
sample population were non-adherent (49.8%)
during the study period. Monthly cost sharing of
statin medication was significantly associated
with more non-adherence (adjusted odds ratio
[adj. OR], 1.067, 95% confidence interval [1.056
– 1.078]). Individuals living in areas receiving
more public assistant are more likely to be non–
adherent (adj. OR: 1.139, CI [1.066 – 1.22]; adj.
OR: 1.30, CI [1.20 – 1.41]) for the second and
third quartiles respectively. The count of
pharmacies was in patients’ residential areas
was not associated with medication adherence
[p-value = 0.423. Other factors that were
significantly associated with less adherence
included patients being low utilizer of medical
care versus median utilizer (OR=1.14, CI [1.081.120]), having diabetes mellitus (OR=1.13, CI
[1.07 – 1.18]), mental disorders (OR==1.09, CI
[1.04-1.34]). Patients living in areas dominated
by black or Latinos were more likely to be nonadherent compared to areas dominated by white
or non-Latinos [OR=1.76, CI [1.41-2.19]),
(OR=1.47, CI [1.30, 1.66]).
Conclusions: The increased cost sharing of
statin drugs and living in poor areas are
associated with less medication adherence.
Implications for Policy, Delivery, or Practice:
Patients’ copay policies should be re-evaluated
to facilitate access to medication and hence,
produce the desired effect on patients. This
evaluation policy could also take into
consideration patients’ level of income.
Funding Source(s): No Funding
Poster Session and Number: A, #174
Financial Vs Non-Financial Incentives in
Improving the Quality of Health Production –
A Healthcare Provider Perspective
Olubiyi Aworunse, University of Texas Health
Science Center at Houston
Presenter: Olubiyi Aworunse, Graduate
Research Assitant, University of Texas Health
Science Center at Houston
biyisky@yahoo.com
Research Objective: The aim of enhancing
quality of health production is to improve the
process of delivery and health outcomes by
rendering the best evidenced services to
consumers while at the same time aiming to
curtail costs. An incentive is any factor (financial
or non-financial) that provides motivation for a
particular course of action, or counts as a
reason for preferring one choice compared to
alternatives. Intrinsic sources of motivation for
clinicians include the likelihood that patients’
health will improve as a result of a course of
action, and motivation from performing a task
well. Other sources of motivation include social
and peer group norms, where a particular choice
is regarded by others as the right thing to do, as
particularly admirable, or where the failure to act
in a certain way is condemned.
Study Design: Systematic review of studies
assessing the effect incentives in improving the
production of consumer’s health from the health
care provider’s perspective.
Population Studied: Data Source - Englishlanguage literature (1 January 2003 – 30
October 2013) from searches in PubMed, Ovid
Medline, Google Scholar, and reference lists of
retrieved articles.
Study Selection – Empirical studies of the
relationship between incentives designed to
improve the quality of health production in
consumers and a quantitative measure of the
healthcare quality. The incentives are directed
solely on the healthcare providers in order to be
included in the study for the review. In addition
excluded were articles which were editorials,
opinion pieces or reports.
Principal Findings: A total of 22 articles were
reviewed for this study. Out of the 22 articles, 5
were systematic reviews, 3 were randomized
clinical trials, 3 were quasi experiments. 1
retrospective study, 1 case study, 1 interrupted
time series, 5 observational study, 2 cross
sectional and 1 case control study. 17 of the
studies reviewed used financial incentives as the
primary form of intervention, 3 used both
financial and non-financial incentives as primary
form of intervention and 2 used non-financial
incentives.
Conclusions: Despite the mixed results on the
use of incentives in healthcare, it still serves as
a potent tool in improving process of care even
though measured outcomes are not met in some
cases. The most used form of incentive from the
review is financial incentives while non-financial
although not frequently used still ties to a form of
reimbursement of some sort for the providers
Implications for Policy, Delivery, or Practice:
In order for there to be a comprehensive
improvement in the quality of health production,
policy makers should always align both types of
incentives as a tool for intervention in the
delivery of healthcare
Funding Source(s): No Funding
Poster Session and Number: A, #175
Where’s My Smokes? Outcomes of Plain
Packaging Legislation of Tobacco in
Australia
Peter Balram, Department of Human Services;
Kathy Ahern
Presenter: Peter Balram, Pharmaceutical
Advisor, Department of Human Services
peter.balram@humanservices.gov.au
Research Objective: Legislation to remove
brands from tobacco packaging was introduced
in January 2013 with the intent of removing
positive signifiers such as glamor or success.
Since then tobacco packages are uniformly
“poo” brown with the only brand information
being the name of the cigarette written in plain
font. The purpose of this research was twofold:
investigate the attitudes of smokers to the
proposed legislation before it became law, and
the effects of the legislation after it came into
effect.
Study Design: This qualitative study consisted
of two rounds of interviews of smokers three
months before plain packaging occurred and
three months after. Transcripts were
thematically analyzed.
Population Studied: Prior to the enactment of
the legislation, 11 smokers aged 18 - 25 were
approached in public designated smoking areas
and interviewed about their smoking habits and
the upcoming legislation. Three months after
the legislation came into effect participants were
re-interviewed.
Principal Findings: Prior to the legislation, all
participants predicted that plain packaging would
have no effect because they had established
their favorite brands and smoking habits.
Participants were resentful at what they
perceived as government interference in their
personal lives, but recognized that the legislation
was naively well intentioned. Post legislation
interviews confirmed that smokers did not
change their smoking habits or brands, but that
an extra layer of annoyance now existed. Plain
packaging led to regular purchase of the “wrong”
brand and accidently picking up someone else’s
tobacco during socializing. However, even
when they were sold or picked up the “wrong”
cigarettes in error, participants would smoke
them, even if the cigarettes were stronger than
their usual ones. The only time they would
return the “wrong” packet or challenge someone
who had mistakenly picked up their cigarettes
was if their brand was the more expensive
brand.
Many more changes were brought in postlegislation than expected. Packs now included
very disturbing images of diseased people on
both sides of the pack, the packaging was
noticeably flimsier, and individual cigarettes had
longer filters and no brand identification.
Conclusions: Post legislation smokers tended
to be annoyed but complacent about the
packaging changes. As in the pre legislation
interviews, participants post legislation indicated
that price had the greatest influence on whether
or not they would reduce or quit smoking.
However, all participants felt the replacement of
glamorous images of smoking with new
signifiers of cheap ugliness would probably deter
people from starting smoking.
Implications for Policy, Delivery, or Practice:
This legislation led to increased resentment at
what was perceived to be government
interference in smokers’ lives, although this
resentment was short lived. Significantly, no
participant focused their resentment at the
political party which passed the legislation. This
suggests that in Australia at least, politicians
need not fear adverse political repercussions for
bringing in anti-smoking legislation such as plain
packaging. Secondly, plain packaging appears
to be effective in removing the ‘glamour’ image,
which may well reduce the number of new
smokers. Research of people who commenced
smoking after January 1, 2013 is required to test
this hypothesis.
Funding Source(s): Other Univesity of
Queensland
Poster Session and Number: A, #176
The Impact of Reference Pricing on
Expenditures and Quality Outcomes for
Arthroscopic Surgery, Cataract Surgery, and
Colonoscopy
Timothy Brown, University of California,
Berkeley; James Robinson, University of
California, Berkeley; Kevin Bozic, University of
California, San Francisco; Emily Finlayson,
University of California, San Francisco;
Christopher Whaley, University of California,
Berkeley
Presenter: Timothy Brown, Assistant Professor,
University of California, Berkeley
timothy.brown@berkeley.edu
Research Objective: To determine the impact
of reference pricing applied to arthroscopic
surgery, cataract surgery and colonoscopy on
insurance expenditures, consumer expenditures,
and quality outcomes.
Study Design: We employed a quasiexperimental design comparing a treatment and
comparison group, both covered by PPO
insurance issued by the same insurance carrier.
Our treatment group was subject to reference
pricing limits if they received services at a
hospital, but was not subject to reference pricing
limits if they received services from an
ambulatory surgery center. Ambulatory surgery
centers have been found to generally charge
prices at or below the reference prices.
Reference prices were $6,000 per arthroscopy
procedure, $2,000 per cataract surgery, and
$1,500 per colonoscopy. The comparison group
was not subject to reference pricing and each of
these categories of procedures was covered in
the usual manner regardless of where received.
Difference-in-differences models are used to
model each procedure category separately. The
data analyzed include claims data for three
years prior to the implementation of referencing
pricing and for two years after the
implementation of reference pricing. Financial
outcomes examined include price (allowed
charge), insurer expenditures, and consumer
expenditures. Quality outcomes examined
include complications attributable to each
category of procedure as determined by medical
experts.
Population Studied: Enrollees in the PPO
products of the California Public Employees'
Retirement System (CalPERS), aged 18-64 and
who received treatment in California, made up
the treatment group. Enrollees in the PPO
product outside of CalPERS, but issued by the
same insurance carrier, aged 18-64 and who
received treatment in California, made up the
comparison group. The data covered the years
2009 to 2013.
Principal Findings: Findings are scheduled to
be complete by May 2014. We expect a
decrease in prices, insurer expenditures and
consumer expenditures due to reference pricing
and no changes in quality. These expectations
are based on our previously published research
on knee/hip replacement using a similar study
design and population.
Conclusions: Reference pricing applied to
arthroscopic surgery, cataract surgery, and
colonoscopy results in a reduction in
expenditures without a change in the rate of
complications.
Implications for Policy, Delivery, or Practice:
Reference pricing should be expanded
nationally to appropriate procedures, procedures
that are standardized and for which there is little
variation in outcomes.
Funding Source(s): Other California Public
Employees' Retirement System
Poster Session and Number: A, #179
Spouses vs. Parents: Who is Providing
Dependent Coverage to Young Adults Under
State-Level Reforms?
James Burgdorf, University of California - San
Diego
Presenter: James Burgdorf, Staff Research
Associate, University of California - San Diego
jburgdorf@ucsd.edu
Research Objective: To examine the
effectiveness of state-level dependent coverage
expansions using detailed private insurance
outcomes that specify coverage source.
Study Design: Difference-in-difference models
are used to estimate the effect of state-level
dependent coverage expansions on finely
detailed categories of coverage. This study
uses the 2001-2009 files of the Current
Population Survey’s Annual Social and
Economic Supplement, covering calendar years
2000-2008.
Population Studied: This study considers
effects among young adults ages 19 through 29.
Subpopulations of age, gender, race/ethnicity,
and marital status are considered.
Principal Findings: Certain published results
on state-level parental coverage expansions are
flawed, as the reported increases are driven by
changes in the marital rate and spousal
coverage. Other published results appear to be
valid, but these effects are most concentrated
near ages at which one would have lost
coverage without reform.
Conclusions: This study shows evidence that
one study’s results on “dependent” coverage are
in fact driven by changes in marriage and rates
of spousal coverage, indicating a problem with
the model itself. A second study seems to have
produced valid results, but these apply most
strongly to individuals at ages at which one
would typically have lost parental coverage
before reform.
Implications for Policy, Delivery, or Practice:
Adult dependent coverage expansions seem to
have modestly increased parental coverage
among young adults who would have otherwise
aged out of parental plans, consistent with a
“passive” effect rather than an “active” effect that
enrolls previously uninsured individuals. Future
state-level increases in limiting ages may be
expected to have similar effects around the
ACA-mandated cutoff point of 26.
Funding Source(s): No Funding
Poster Session and Number: A, #180
The Relation of Social Networks and Health
Service Demand
Nathan Dong, Columbia University
Presenter: Nathan Dong, Assistant Professor,
Columbia University
gd2243@columbia.edu
Research Objective: Amid increasing interest
in how social relationships play an important role
in health and health behavior, it remains unclear
whether social interaction benefits health literacy
and in turn affects individuals’ healthcare
consumption. More specifically, this article
proposes a research hypothesis to address the
question: Do individuals who are strongly tied to
other individuals within the social networks
become more health conscious or literate and
hence use more health services?
Study Design: We pool the two survey results
to conduct OLS, IV and median regressions to
assess the relationship between social networks
and health service utilization. In the first set of
analysis, the dependent variable is the total
number of visits to healthcare providers. To
avoid the double-counting problem in health
service utilization, in the second set of analysis
we break down the service demands to three
categories: the number of visits to: 1) hospitals,
2) ERs, and 3) doctors. The main predictor
variables are social networks, income, marriage
status, ethnicity, age, gender, insurance
coverage, and working hours. Other model
covariates include family size, education,
geographic location, whether living in a
metropolitan area, and whether owing a
business. We use two dependent variables to
proxy for an individual’s social networks.
Population Studied: We use the Health
Tracking Household Survey 2007 and 2010
which is a successor of the Community Tracking
Study Household Survey. This U.S. householdrepresentative, cross-sectional survey of civilian
and non-institutionalized individuals contains
information on health insurance coverage,
access to care, perceptions of care delivery and
quality of care, use of health services, health
status, consumer engagement, use of health
care information, and demographic information.
Principal Findings: People who are socially
"active" became more health-conscious or
literate and hence use more health care services
in hospitals, ERs, and doctor clinics. It also finds
that people of younger (age effect), male
(gender effect), having higher income (wealth
effect), being married (family effect), of white
race (ethnicity effect), having longer working
hours (time-availability effect), owning a
business (risk-taking effect), having more years
of education (literacy effect), and having no
insurance coverage (disincentive effect) tend to
avoid seeking health services.
Conclusions: Given that the prior research only
study the impact of social networks on health,
health literacy, and health behavior, this article
seeks to present a contribution by focusing on
the relation between social networks and health
service demand, and what other variables
besides social networks determine a consumer’s
utilization of health services and what kind of
health service. Such an update in the literature
is critical to understanding how contributing
factors of health service demand have changed
over the past decade. We provide new evidence
on quantitatively understanding how different
aspects of personal characteristics, time and
geographic locations influence individual
decisions to use health service.
Implications for Policy, Delivery, or Practice:
These findings can help policy makers to
increase the effectiveness of public policies that
aims at increasing utilization of healthcare or
businesses and entrepreneurs to target the
underutilized groups or regions for health
services by advertising on internet social media,
local newspapers, or community groups.
Funding Source(s): No Funding
Poster Session and Number: A, #181
Preventive Care Utilization among the
Employed
Teresa Gibson, Truven Health Analytics /
Harvard Medical School; Emily Ehrlich, Truven
Health Analytics; William Marder, Truven Health
Analytics
Presenter: Teresa Gibson, Vice
President/Lecturer, Truven Health Analytics /
Harvard Medical School
tbgibson1@gmail.com
Research Objective: Nearly half of adults in the
US do not use commonly recommended
preventive services (Centers for Disease Control
and Prevention, 2012). The Affordable Care Act
(ACA) offers opportunities to increase
preventives services. This includes providing
services such as breast cancer screening for
women over 40 and colorectal screening for
adults over 50, at no cost to the consumer.
However, many of the changes associated with
the ACA are designed to improve access to care
for those who previously had no usual source of
health care or means of payment for these
services. The purpose of this study is to assess
the varying levels of preventive service use
among an insured population to further
understand consumer choices related to
prevention of chronic disease.
Study Design: A panel data set was created
with employees as the cross sectional unit and
calendar quarter as the unit of time. We
estimated the likelihood of receipt of the
following preventive services: colon cancer
screening, lipid screening, flu shot. The
likelihood of breast cancer screening and
cervical cancer screening were estimated for
women. Annual wages for each employee were
appended, and five categories of wages were
created (0-10th percentile <$43,197, 10-25th
percentile $43,197-<$61,850, 25-50th percentile
$61,850-<$86,254, 50th-75th percentile
$86,254-<$119,081 and over the 75th percentile
$119,081+). The likelihood of a preventive
service was regressed on wage category, age
group, gender, health status (comorbidity index
and psychiatric diagnostic groupings), and other
sociodemographic characteristics. We adjusted
for clustering over time by employee and
included fixed effects for employer/health plan.
Population Studied: Employees of six large
firms who were enrolled from January 2010
through June 2012. (n=27,119).
Principal Findings: Employees in the lowest
wage category were 33% less likely (p<0.01) to
receive colon cancer screening than employees
in the highest wage category, with the likelihood
of cancer screening increasing along a gradient
with wage category. Wage category was not
associated with cholesterol screening, although
employees win the lowest wage category were
half as likely (49.6% less likely, p<0.01) to
receive a flu shot as those in the highest wage
category. Among women, the likelihood of
breast cancer screening increased with each
wage category, with employees in the lowest
wage category 36% less likely (p<0.01) to
receive screening than employees in the highest
wage category. Women in the two wage
categories falling below the 25th percentile of
wages were less likely to receive cervical cancer
screening than those in the highest wage
category, 24.2% less likely and 9.5% less likely
(p<0.05) respectively. However, women
employees with wages between the 25th and
75th percentile had cervical cancer screening
rates similar to those in the highest wage
category (p>0.05).
Conclusions: This study finds that wage levels
are associated with use of preventive services.
These results suggest that additional strategies,
beyond financial incentives, may impact
consumer health behaviors with respect to use
of preventive health care services.
Implications for Policy, Delivery, or Practice:
The results of this study indicate a need to
continually monitor the use of adult preventive
services to inform additional strategies and
interventions to increase the use of these
services across the broad US population.
Funding Source(s): No Funding
Poster Session and Number: A, #183
Early Indications of the Public’s Perceptions
of Comparative Effectiveness Research
Mindy Hu, Mathematica Policy Research;
Derekh Cornwell, Mathematica Policy Research
Presenter: Mindy Hu, Survey Researcher,
Mathematica Policy Research
MHu@mathematica-mpr.com
Research Objective: To provide insights
regarding the general public’s knowledge,
attitudes, and behaviors relevant to comparative
effectiveness research (CER).
Study Design: We conducted six two-hour
focus groups in three metropolitan areas
(Boston, Massachusetts; Chicago, Illinois; and
San Francisco, California) between November
and December 2011. Participants were
segmented based on whether they resided in an
urban or non-urban location and whether they
took an “active” or “passive” approach to making
health care decisions. Active vs. passive
participants were identified based on responses
to screening questions about self-confidence in
their ability to identify when they needed medical
care and the extent to which they were proactive
about preparing/asking questions during doctor
visits. All focus group sessions occurred at
professional focus group facilities and were
conducted by moderators using a standardized
protocol that included questions and vignettes to
gauge participants’ understanding and reaction
to key elements of CER, such as the use of
alternative evidence-based treatment options as
part of the health care decision-making process.
Participants discussed scenarios that captured
this key element of CER.
Population Studied: We selected focus group
participants from lists compiled by the focus
group facilities. A recruitment script was
provided to each facility to convey the purpose
of the research. In total, the participants across
all three cities consisted of 58 adults ages 21 to
77 with a mix of education levels, gender, and
race to ensure broad representation of the
general public.
Principal Findings: Key themes that emerged
from the focus groups included:
Knowledge-Awareness-Understanding. Levels
of knowledge and awareness of CER were
limited. Most participants were unfamiliar with
the term CER and with related research
initiatives. The knowledge gap was common
across active and passive health care
consumers, though, once the concept was
explained, participants expressed favorable
opinions.
Attitudes-Opinions. Most respondents expressed
positive attitudes towards the potential benefits
of CER, though many expressed an
understanding of certain concepts that differ
from those of policymakers and subject matter
experts. For example, participants expressed a
more encompassing definition of medical care to
include time spent with their physician and did
not restrict it to specific medical services.
Behaviors-Experiences. Participants’ opinions
on the level of familiarity they expected their
doctors to have regarding alternative treatment
options were mixed. Some participants expected
specialists to be more familiar than primary care
providers. Active consumers were more likely
than passive consumers to view health care
decision-making with their doctor as a
collaborative process.
Conclusions: In this small, but broadly
representative sample of the general public,
levels of knowledge and understanding of CER
were low but there was a generally positive view
of the concepts underlying CER. Public
perceptions of key elements of CER were
generally consistent with those of policymakers.
Overall, these findings support an underlying
preference among the general public for more
information to help inform health care decision
making. However, the findings also indicate that
more efforts will be needed to clearly
communicate key aspects of CER to the general
public.
Implications for Policy, Delivery, or Practice:
The results underscore the need to improve
understanding of the concept and practice of
CER among the general public.
Funding Source(s): Other ASPE
Poster Session and Number: A, #184
How Does Exchanging Health Information
Affect Hospital Market Shares?
Peiyin Hung, University of Minnesota Division of
Health Policy; Jeffrey McCullough, University of
Minnesota Division of Health Policy and
Management; Ira Moscovice, University of
Minnesota Division of Health Policy and
Management; Michelle Casey, University of
Minnesota Rural Health Research Center
Presenter: Peiyin Hung, Ph.D Student,
University of Minnesota Division of Health Policy
payinin@gmail.com
Research Objective: In the wake of health care
reform, health information exchange (HIE) has
gained significant momentum. The existing
literature explores its vast potential, but does not
examine this technology's demand-side effects.
We aims to evaluate the impact of HIE on
hospital choice.
Study Design: Using patient-level hospital
choice data, we employed difference-indifference estimation with two discrete choice
models – conditional choice and nested logit,
which control for patient age, gender, race,
insurance type, driving distance, as well as
hospital and market characteristics, hospital
fixed effects, and year dummies. A hospital is
included in a given patient’s “choice set” if 1) it is
within 50 miles of the patient’s location
coordinates, or 2) at least 10% of the patients in
the same ZIP code chose the hospital in
question. The maximum number of hospitals
included in one’s choice set is 20.
Population Studied: This study includes all
patients admitted to any Pennsylvania hospital
from 2005-2011 for congestive heart failure
(CHF) (N=14,527; Diagnosis Related Group
(DRG)=291,292,293) or hip/knee replacement
surgery (N=39,870; DRG=466,467,468). We
linked data from the Healthcare Cost and
Utilization Project’s Pennsylvania State Inpatient
Database, the American Hospital Association
(AHA) Annual Survey, the 2007 AHA Healthcare
IT Database, and Area Health Resource Files.
Principal Findings: Both CHF and hip/knee
replacement patients in Pennsylvania were
significantly more likely to choose a hospital that
exchanges health information (HIE hospital)
than one which does not (non-HIE hospital).
This trend increased more rapidly for hip/knee
replacement patients than CHF patients. We
found that rural HIE-hospitals had a decreasing
trend of average admissions for hip/knee
replacement, while the number of admissions in
urban HIE-hospitals increased significantly
(p<0.014). The odds of patients choosing HIEhospitals did not vary by the scope of HIE (only
used within a hospital system, only used outside
a system, or both). Younger patients and selfpaid patients were more likely than their
counterparts to gravitate toward HIE-hospitals.
Conclusions: HIE is a significant factor in
hospital choice for the two conditions examined.
However, the magnitude varies by hospital
characteristics, patient age, and insurance type.
Implications for Policy, Delivery, or Practice:
Policies encouraging HIE should go beyond
supply-side effects and consider market effects .
Certain hospital characteristics such as rurality,
service volume, and range of services, may
influence the degree to which HIE use affects its
market share. Hospital management and
policymakers should incorporate and be aware
of patient flow and changes in patients’ hospital
choices when adopting HIE policies for hospital
care.
Funding Source(s): No Funding
Poster Session and Number: A, #185
How Employee Earnings are Associated with
CDHP Plan Choice
David Jordan, Slippery Rock University; David
Jordan, Slippery Rock University
Presenter: David Jordan, Associate Professor,
Slippery Rock University
dwj12000@yahoo.com
Research Objective: The focus of this paper is
the association between employee earnings and
plan choice. Although research examines the
association between earnings and CDHP
choice, it assumes a simple linear relationship,
which may not be the case.
Study Design: A cross sectional nonexperimental ex post facto design was used to
examine data from a single large employer in
four regions of the United States. The
dependent variable is plan choice. Analysis is at
the contract level. Multinomial logit regression is
used because discrete nominal dependent
variable responses are analyzed simultaneously.
Population Studied: This study is based on a
single large self-insured employer’s ESI enrollee
population. The data source is a regulated
publicly traded holding company with assets of
approximately forty billion dollars. It employs
about 20,000 persons in East North Central,
South Atlantic, East South Central, and West
South Central United States. The workforce is
comprised of salaried and hourly (exempt and
non-exempt) positions including administrative,
technical, skilled trades and non-skilled laborers
(union and non-union), various levels of
management, and professional generalists.
Principal Findings: Where prior research finds
a positive linear association between earnings
and CDHP choice, this research does not, in the
broader sense. Taken in total, results find lowest
earners are associated with choosing one form
of CDHP (the HRA), and highest earners are
associated with choosing a different type of
CDHP (the HSA eligible HDHP). Furthermore,
when CDHP choice is examined for enrollees in
the middle 80 percent earner group, no
statistically significant association is found
between earnings relative to CDHPs versus
Managed Care plan choice. Prior research treats
all CDHPs as a homogenous choice among
plans in their plan choice estimations, while in
fact studies also acknowledge the many forms of
CDHPs with distinctly differing cost structures
and characteristics in the literature.
Conclusions: Where prior research finds a
positive linear association between earnings and
CDHP choice, this research does not, in the
broader sense. Taken in total, results find lowest
earners are associated with choosing one form
of CDHP (the HRA), and highest earners are
associated with choosing a different type of
CDHP (the HSA eligible HDHP). Furthermore,
when CDHP choice is examined for enrollees in
the middle 80 percent earner group, no
statistically significant association is found
between earnings relative to CDHPs versus
Managed Care plan choice. Prior research treats
all CDHPs as a homogenous choice among
plans in their plan choice estimations, while in
fact studies also acknowledge the many forms of
CDHPs with distinctly differing cost structures
and characteristics in the literature.
Implications for Policy, Delivery, or Practice:
The distribution of earners across plans may
affect the risk pool and employers need to be
vigilant regarding premium risk adjustment.
Lower earning enrollees may not want to risk the
larger initial cost sharing that they perceive the
HSA eligible HDHP to have due to the high
deductible. They may also perceive their need
for health care to be minimal and take the
chance their costs, if any, will be Results may
suggest that low-earning employees seek the
lower premium cost of the HRA versus the PPO
Managed Care plan. Albeit, the HSA eligible
HDHP has the lowest premium cost to enrollees,
but also has a high deductible that must be
funded entirely by the enrollee(s) if they incur
medical costs, whereas the HRA’s high
deductible is partly offset by the employer
funded spending account. This suggests
enrollees may not want to risk the larger initial
cost sharing that they perceive the HSA eligible
HDHP to have due to the high deductible. They
may also perceive their need for health care to
be minimal and take the chance their costs, if
any, will be covered by the employer-funded
HRA account. Premium contributions are
markedly higher for the HRA than the HDHP, but
the risk and uncertainty of incurring costs under
a high deductible HDHP may be unattractive.
Furthermore enrollees may lack the disposable
income to self-fund the HSA to help offset costs
in the HDHP. Qualitative findings by Green, et
al. (2006) support these possible explanations.
As in this study, Green, et al.’s (2006) research
included a similar choice between lower and
higher deductible CDHPs. They found enrollees
who chose the higher deductible CDHP did not
expect to need care and preferred a plan with
low premiums. Those who chose the lower
deductible HRA did so because its premiums
were cheaper than the Managed Care option,
but they expected the employer funded account
to greatly assist with minimizing their risk of outof-pocket costs (Green, et al. 2006).
Results for those in the middle-earners group do
not support prior research that finds a positive
association between earnings and CDHP choice
(Barry et al., 2008; Lo Sasso et al., 2004;
Parente et al., 2004a, 2004b, 2008; Tollen et al.,
2004; U.S. Department of Health & Human
Services, 2009; U.S. Government Accountability
Office, 2006). However, the one study with
similar choices across plan characteristics and
study population finds earnings and CDHC
choice not significant as in this study (Green et
al., 2006). This study suggests the individual
plan characteristics are more critical than the
moniker of CDHP assigned to HRAs and HSA
eligible HDHPs relative to the association
between earnings and plan choice.
Funding Source(s): No Funding personal
Poster Session and Number: A, #186
The Association between Enrollee Prior Total
Cost Sharing and Health Status: Not quite
the Same across CDHP Versus Managed
Care Plan Choice
David Jordan, Slippery Rock University
Presenter: David Jordan, Associate Professor,
Hcam Program Coordinator, Slippery Rock
University
david.jordan@sru.edu
Research Objective: To examine the
association of Total Cost Sharing (TCS) and
Health Status/Relative Risk Score (RRS) with
Consumer Directed Health Plan choice, and if
these factors are similarly associated as they
relate to health care access and use through
plan choice. The analysis is at the household
level when a Managed Care Preferred Provider
Organization (PPO), HRA, and HSA eligible
HDHP are offered concurrently in an ESI
program.
Study Design: This study employs a cross
sectional non-experimental ex post facto design
that examines data from a single large employer
in multiple regions of the United States. The unit
of analysis is the enrollee household. First,
descriptive statistics are used to describe the
enrollee population relative to available plans.
Then, multivariate analysis is used to examine
hypothesis developed using a theoretical model
relative to household out-of-pocket costs and
health status.
Population Studied: This study examines
census data from a single large employer’s
enrollee population. Enrollee and plan data are
collected from the employer’s human resources
information system (HRIS), and a data
management vendor contracted by the
employer’s broker for managing their ESI claims
data.1 The data incorporates health plan
enrollment, claims, socio-demographic, and plan
data related to the ESI program for 2005 and
2006. The data include employees and
household members eligible for benefits who
were continuously enrolled from January 1, 2005
to December 31, 2009, and under 60 years of
age.
Principal Findings: This study lends additional
support to the literature that CDHPs enjoy both
favorable selection and a positive association
between lower premium cost and plan choice
(Barry et al., 2008; Green, et al., 2006; Lo Sasso
et al., 2004; Parente et al., 2004a; Parente et al.,
2004b; Parente et al., 2008; Tollen et al., 2004).
However, findings differ slightly between health
status (measured via a Relative Risk Score) and
household total cost sharing (household out-ofpocket costs plus premium contributions) with
CDHP choice.
Findings suggest enrollment relative to TCS
appears to be inversely hierarchal for enrollee
premiums and plan generosity. As prior total
cost sharing (TCS) decreases, enrollees are
more likely to choose a plan with a greater
emphasis on lower premium cost than greater
benefit generosity. Alternatively, as TCS
increases, the importance of plan generosity
increases.
Enrollees in the “lowest” total cost sharing
(bottom ten percent) are most likely to choose
the lowest cost highest initial cost sharing (least
generous) CDHP. Enrollees with “lower” total
cost sharing (the bottom two quintiles), are more
likely to choose the plan deemed to represent
“middle” premiums and initial cost sharing
(middle generosity) verses the highest premium
and lowest initial cost sharing (most generous)
Managed Care plan.
As with TCS, Relative Risk Score (RRS) and
CDHP plan choice also suggests favorable
selection for CDHP enrollment. Findings indicate
that enrollees with the best household health
status (in the “lowest” forty percent or bottom
quintile for RRS), are more likely to choose
either CDHP, but by a minimal extent are most
likely to choose the lowest generosity plan
versus the most generous Managed Care plan.
Choice of either CDHP by the lowest earners is
nearly equal in likelihood. Enrollees in the
“lower” forty percent (bottom two quintiles of
RRS) are more likely to choose the “middle”
generosity plan versus the PPO Managed Care
plan. Choice of the lowest generosity CDHP and
membership in the lower forty percent of RRS is
not statistically significant.
Conclusions: Of interest is that for households
with the lowest Relative Risk Score (healthiest
twenty percent), either CDHP is more likely to be
chosen over the Managed Care plan. The
coefficients and exponents (B) are very similar
for the both CDHPs versus the PPO Managed
Care plan. However, for enrollees in the lower
Relative Risk Score (RRS) group (the healthiest
forty percent), only middle generosity plan
choice is significant. One explanation could be
that when enrollees fall into the “healthiest”
group, there is less concern for high initial outof-pocket costs associated with the lowest
generosity CDHP, but for enrollees in the group
that has a broader range of RRS (forty percent
versus twenty percent), the least generous
CDHP becomes less attractive than the middle
generosity CDHP. The middle generosity CDHP
in this study may represent a “middle ground”
option between the high-risk high cost CDHP
and the low-risk high cost PPO Managed Care
plan.
Taking the findings in total, an empirical
measure of health status (RRS) appears to have
a slightly different association with CDHP choice
than total cost sharing (TCS). Enrollees’ out-ofpocket costs (TCS) may affect their perceived
need for health care (and related financial
exposure to related costs) across plan choices
differently than their actual health status (RRS).
One possible explanation may be different
health care use behaviors relative to need, or
that some individuals are more likely to seek
care than others. Alternatively, it may suggest
that out-of-pocket dollars are more apparent to
the household than the complex and often
uncertain status of their health, or exposure to
financial uncertainty related to health care is
driven primarily by what was spent previously
regardless of health status.
Implications for Policy, Delivery, or Practice:
The balance between plan premium cost and
benefits generosity must be considered for ESI
programs relative to whom are the high users of
health care (including families verses single
subscribers), and those with diseases and with
chronic conditions. This can impact the viability
of the program and the health of enrollees. If the
cost and benefit generosity of the ESI program
are not conducive to employee needs, enrollees
could defer necessary care due to cost, miss
early detection of more serious problems
through avoidance of routine preventive care, or
even choose to go without insurance coverage
at all. Furthermore, enrollee costs may be more
pertinent to employer sponsored insurance
choice sets than empirical surveys of population
health measures.
Funding Source(s): No Funding
Poster Session and Number: A, #187
Health Care Utilization and Financial Burden
of Health Care among Adults with High
Deductible Health Plans with and without
Health Savings Accounts
Whitney Kirzinger, National Center for Health
Statistics; Robin Cohen, National Center for
Health Statistics
Presenter: Whitney Kirzinger, Health
Statistician, National Center for Health Statistics
wkirzinger@cdc.gov
Research Objective: Previous research has
found that the percentage of privately insured
persons under 65 years of age covered by a
high deductible health plan (HDHP) has
increased from 19.2% in 2008 to 33.4% in the
first six months of 2013. HDHPs typically have
lower premiums, but higher deductibles than
traditional health plans. In 2012, the minimum
deductible for an HDHP was $1,200 for
individual coverage. Beginning in 2014, some
private coverage options offered through the
Health Insurance Marketplace have deductibles
as high as $6,350 for individuals. Consequently,
a further increase in the percentage of
individuals covered by HDHPs may occur. In the
early 2000’s, research showed that enrollees in
HDHPs tended to be younger, healthier, and
wealthier. However, with the expansion of
HDHPs, the population of those enrolled in
HDHPs may change. Additionally, data show
that only about one-third of those with HDHPs
have health savings accounts (HSAs) or health
reimbursement accounts (HRAs) that allow pretax dollars to be set aside for out-of-pocket
health care expenses. Characteristics of those
who have HDHPs without an HSA/HRA, HDHPs
with an HSA/HRA (also known as consumerdirected health plans, or CDHPs), and low
deductible health plans (LDHPs) were identified.
Patterns of health care use and financial burden
of health care among those who have HDHPs,
CDHPs, and LDHPs were examined.
Study Design: This analysis used 2012 data
from the cross-sectional National Health
Interview Survey (NHIS). Key measures include
demographic characteristics, health status,
indicators of financial burden of medical care,
and various health care utilization measures.
Chi-square tests were used to examine
differences between adults with an HDHP
without an HSA/HRA, CDHP, and LDHP.
Population Studied: U.S. civilian noninstitutionalized privately insured adults aged 1864.
Principal Findings: Results show that in 2012,
6.7% of adults with HDHPs without an HSA/HRA
were in fair or poor health, compared to 5.1% of
adults with CDHPs and 5.5% of adults with
LDHPs. Adults with HDHPs without an
HSA/HRA were more than twice as likely as
those with LDHPs to delay medical care due to
cost (10.7% and 5.0%, respectively). Those with
HDHPs without an HSA/HRA were almost twice
as likely to be in a family having problems
paying medical bills (20.2%) compared with
those enrolled in LDHPs (11.6%). Adults with
HDHPs without an HSA/HRA were also more
likely than those with LDHPs to skip medication,
take less medication, and delay filling a
prescription to save money on prescription drug
costs (6.8%, 7.4%, and 9.1% compared with
4.0%, 4.6%, and 5.7%, respectively). Comparing
those with CDHPs and those with LDHPs, there
were no differences in the percentages of adults
who were in a family having problems paying
medical bills or skipping medication, taking less
medication, or delaying filling a prescription to
save money on prescription drug costs.
Conclusions: Adults with HDHPs without
HSA/HRAs were more likely to delay care and
experience financial burdens of medical care
than those with LDHPs.
Implications for Policy, Delivery, or Practice:
The NHIS provides a data source to examine
utilization and financial burden of health care
among adults with HDHPs with and without an
HSA/HRA.
Funding Source(s): CDC
Poster Session and Number: A, #188
Religion and Risky Health Behaviors among
U.S. Adolescents and Adults
Sanjeev Kumar, Yale School of Public Health;
Jason Fletcher, University of Wisconsin
(Madison)
Presenter: Sanjeev Kumar, Postdoctoral
Fellow, Yale School of Public Health
sanjeev.kumar@yale.edu
Research Objective: Risky health behaviors,
like smoking, binge drinking, and illicit drug use
during adolescence and early adulthood lead to
adverse health outcomes in older age. Our
research objective is to explore if some of the
existing institutions like religion could be
leveraged to help adolescents and young adults
make healthier life-style choices. To investigate
that this study measures the effects of a broad
set of measures of religiosity—religious
attendance, prayer frequency, and self-reported
importance of religion—on risky health
behaviors defined as the propensity to indulge in
addictive substance use at different stages of
the life course.
Study Design: We used the restricted version
of the National Longitudinal Study of Adolescent
Health dataset, a longitudinal study of a
nationally representative sample of 7th-12th
grade students (N=20,745) surveyed through
their 30s (N=15,701). We used data from Wave
1 (1994-1995), Wave 3(2001-02), and Wave 4
(2007-08). Our measures of the risky behaviors
included usage of both licit and illicit
substance—Cigarette, Binge Drinking,
Marijuana, Cocaine, Methamphetamine,
Ecstasy, Inhalants, LSD, Heroin, PCP, and
Other illegal drugs. We analyzed the selected
data using the ordinary linear regression (OLS)
with state-, family-, and twin-fixed effects
separately. In addition, we controlled for sex,
race, income, assets, education, birth-weight,
PVT score, parental religiosity, education and
working status of mother; we also use some
neighborhood characteristics: proportion of
individual of same race, age-groups, and religion
living in the respondents’ area.
Population Studied: Around 12,000 individuals
and 3,000 siblings were found in all three waves.
Only 368 respondents reported being an
identical twin in our sample.
Principal Findings: Religiosity was found to be
associated with anywhere between 4% to 7%
reduction in smoking, 3% to 4% reduction in
binge drinking, 2% to 3.3% reduction in
marijuana smoking, and 1% to 2% reduction in
cocaine, methamphetamine, and other illicit
drugs. Around their college going age,
respondents showed a higher likelihood to
indulge in binge drinking. By the time
respondents reached around age 30—an agebracket with a deeper attachment with labor
market—religiosity once again showed positive
effects. Specifically, we found a strong and
persistent association of the variable,
Importance of religion, with smoking.
Conclusions: Irrespective of their religious
affiliations, religious adolescents were found to
be less likely to indulge in risky health behaviors.
Also, indulgence in risky health behaviors was
found to be a medium through which young
adults seemed to assert their autonomy.
Interestingly, it was the intrinsic dimension of
religiosity—prayers and beliefs—that showed
stronger association with the avoidance of risky
health behaviors.
Implications for Policy, Delivery, or Practice:
Given the low price elasticity of addictive
substance, the risk of the emergence of
underground illegal market owing to regulation,
high deadweight loss involved in price based
regulation of alcohol, religion could potentially
complement the existing prevention efforts to
rein in use and abuse of cigarettes, alcohol, and
illicit drugs.
Funding Source(s): AHRQ
Poster Session and Number: A, #189
Trend in Bilateral Salpingo-oophorectomy
among Taiwanese Women Receiving Benign
Hysterectomy and Its Associated Factors
Jerry Cheng-yen Lai, National Yang-Ming
University; Yiing-Jenq Chou, Institute of Public
Health & Department of Public Health, School of
Medicine, National Yang-Ming University; KungLiahng Wang, Department of Nursing, Mackay
Junior College of Medicine, Nursing, and
Management; Department of Obstetrics and
Gynecology, Mackay Memorial Hospital and
Mackay Medical College; Department of
Obstetrics and Gynecology, Taipei Medical
University, Taipei, Ta; Hsiao-Yun Hu, Institute of
Public Health & Department of Public Health,
School of Medicine, National Yang-Ming
University; Department of Education and
Research, Taipei City Hospital; I-Ting Chen,
Institute of Public Health & Department of Public
Health, School of Medicine, National Yang-Ming
University; Nicole Huang, Institute of Hospital
and Health Care Administration, School of
Medicine, National Yang-Ming University;
Department of Education and Research, Taipei
City Hospital
Presenter: Jerry Cheng-yen Lai, Student,
National Yang-Ming University
chengyen@hotmail.com
Research Objective: The 2008 guideline of
American Congress of Obstetricians and
Gynecologists (ACOG) recommends elective
BSO for post-menopausal women, and
advocates ovarian conservation for premenopausal women. Past researchers have
shown that insurance status is an important nonclinical predictor, which influences women's
decision on elective bilateral salpingooophorectomy (BSO) at benign hysterectomy.
The National Health Insurance (NHI) program,
however, is an universal, single-payer insurance
program that provides comprehensive coverage
for all civilian residents in Taiwan. It is important
to investigate women's decisions on concomitant
BSO procedure in systems where financial
barriers to elective procedure may be small. Our
objective was to examine the recent trends in
the performance of elective BSO under the NHI
program in Taiwan from 2000 to 2010, and to
identify patient and provider-related
characteristics associated with BSO at benign
hysterectomy.
Study Design: We conducted a pooled crosssectional study using the population-based
dataset from the Taiwan NHI program. Incident
cases of benign hysterectomy were identified
from the inpatient admission claims. The effects
of both patient and provider-related
characteristics on BSO rate were investigated
using generalized estimating equation for
multilevel analyses (first level: patient; second
level: provider) to adjust for women decision
induced through shared physicians.
Population Studied: All selected inpatients
were between 20 and 79 years of age;
undergone incident hysterectomy for benign
diseases; treated by physicians with board
certification of obstetrics and gynecology. We
excluded cases with incomplete information;
removal of remaining ovaries, unilateral
(salpingo-) oophorectomy, partial oophorectomy
prior or within the same inpatient admission
claims; or BSO in prior inpatient admission
claims.
Principal Findings: Approximately 15 percent
of 181,531 adult women received elective BSO
at hysterectomy during the study period. The
overall BSO rate declined steadily from 22
percent in 2000 to 9.9 percent in 2010,
particularly in women aged 40-49 (decreased by
80 percent). Women aged 55 years or older
were significantly more likely to receive elective
BSO at hysterectomy. Presence of co-morbid
illness; prior catastrophic illness; abdominal or
laparoscopic surgery; and regional or medical
center inpatient were associated with a greater
likelihood of BSO at hysterectomy for women of
all ages. Nearly one-third of these patients who
had BSO at hysterectomy were diagnosed of
CIS of uterine cervix (31.9 percent), whereas
women with pre-operative diagnosis of uterine
prolapse had a 94 percent relative reduction in
their BSO rate.
Conclusions: The apparent decreasing trend of
elective BSO performance suggests that both
Taiwanese women and their providers are in
favor of ovarian preservation. Age, hysterectomy
types, and disease diagnoses influenced
women's decision on elective BSO at
hysterectomy in Taiwan, a country with national
health insurance. The current findings provide
valuable information for policy makers regarding
the role of physician and hospital characteristics
in women's decision under a universal, singlepayer insurance program.
Implications for Policy, Delivery, or Practice:
This study highlights the importance of how
patient and provider-level characteristics may
influence decision on prophylactic oophorectomy
for women undergoing benign hysterectomy.
Being the only effective intervention for reducing
ovarian cancer, clinical practice and patient
counseling of elective BSO based on ACOG
guideline requires a strong reconsideration, and
can have significant health implications.
Funding Source(s): No Funding
Poster Session and Number: A, #190
Factors Associated with Physician Switching
Doohee Lee, Marshall University; Charles
Begley, University of Texas School of Public
Health
Presenter: Doohee Lee, Associate Professor,
Marshall University
leed@marshall.edu
Research Objective: Maintaining
physician/patient relationships is an important
objective of health care reform. Understanding
the factors that lead to physician switching is
needed to achieve this objective but limited
information is available in the literature. The
present study explores various health services
factors associated with physician switchers
compared to non-switchers at the national level.
The following research questions are addressed:
What is the current prevalence rate of physician
switching at the national level? What are the
characteristics of those patients willing to switch
physicians? Are switching behaviors associated
with resource use and usual sources of care?
What is the relative importance of health service
factors versus patient and physician
characteristics?
Study Design: This is a cross-sectional study
describing and comparing factors associated
with physician switching and non-switching. We
performed multivariate regression to estimate
determinants of physician switching in relation to
covariates (resource use, usual source of care,
length of stay in hospitals) while controlling for
personal factors. The data analysis was fully
adjusted, using the weight variable given in the
data and STATA ‘svy’ commands, in order to
represent a national sample and correct the
complex survey design.
Population Studied: We analyzed “the 2010
Health Tracking Household Survey public use
data” (n=16,671 patients) collected by the
Center for Studying Health System Change
(HSC) and sponsored by the Robert Wood
Johnson Foundation (RWJF).
Principal Findings: The physician switching
prevalence rate was 9.2%. Physician switching
was associated with difficult access to care after
regular hours, unmet medical needs, delay of
care, longer length of stay, and less use of
electronic medical communications with
providers. Results from a multiple regression
analysis reveal that unmet medical needs
(B=1.04, p=0.033), difficult access to care after
regular hours (B=-.44, p=0.002), and less loyalty
to providers (B=-.57, p<0.001) remained
significant in relation to patients’ switching
behaviors while adjusting for age, gender, and
race.
Conclusions: Our empirical findings show that
patients’ physician switching behaviors were
linked to poor health outcomes including usual
source of care and resource use at the national
level.
Implications for Policy, Delivery, or Practice:
Our national study shows a significant number of
patients considering provider switching. The
concept of switching provider behaviors is not
well researched but is linked to negative medical
outcomes such as unmet medical needs and
delay of care that may be against our cost
containment efforts. Health care providers and
policymakers hence may want to consider
addressing these factors to improve continuity of
care. Also, examining the overall ability of
sustaining patients who may switch or consider
switching providers may help prevent unsatisfied
patients from physician switching.
Funding Source(s): No Funding
Poster Session and Number: A, #191
Does Adding Medicare Part D Gap Coverage
for Generic Medications Increase Use of
Generic Drugs among Medicare Advantage
Enrollees?
Chang Liu, Duke-NUS Graduate Medical School
Singapore; Vincent Mor, Brown University;
Lewis Kazis, Boston University; Alan Zaslavsky,
Harvard University; John Ayanian, University of
Michigan; Amal Trivedi, Brown University
Presenter: Chang Liu, Assistant Professor,
Duke-NUS Graduate Medical School Singapore
chang.liu@duke-nus.edu.sg
Research Objective: The Affordable Care Act’s
strategy to close the coverage gap immediately
discounts brand medications by 50% while
providing a more modest discount of 7%
annually for generic medications. This approach
contrasts with the strategy employed by some
Part D plans prior to 2011 that provided gap
coverage for generic medications exclusively.
The objective of our study is to examine the
effect of adding gap coverage exclusively for
generic drugs on beneficiaries’ switching
behavior from brand to generic drugs in three
drug categories: angiotensin converting enzyme
(ACE) inhibitors and angiotensin receptor
blockers (ARBs), statins, and selective serotonin
reuptake inhibitors (SSRIs) and
serotonin/norepinephrine reuptake inhibitors
(SNRIs).
Study Design: Quasi-experimental pre-post
study with concurrent controls.
Population Studied: 621 beneficiaries who
reached coverage gap in 8 Medicare Advantage
plans that added Part D gap coverage for
generic medications and 1,129 patients in 18
control MA plans that retained the coverage gap.
Principal Findings: Among beneficiaries who
exclusively filled branded prescriptions of the
three-class drugs before reaching the gap, the
share of patients who filled generic drug
prescriptions increased by 2 percent in case
plans from 2006 to 2007, but decreased by 10
percent in control plans over the same period
(with a difference-in-difference odds ratio
estimate of 2.64 (95%CI: 1.08, 6.45)), while the
differences between case and control plans
were significant and more profound for statins.
Moreover, the mean ratio of generic drugs to
total prescription in case and control plans
started to diverge in the pre-gap period in 2007,
and beneficiaries in case plans on average
delayed entered coverage gap in 2007 as
compared with beneficiaries in control plans.
Conclusions: Adding gap coverage of generic
medications increased the likelihood of switching
from brand to generic drugs and delayed entry
to the coverage gap.
Implications for Policy, Delivery, or Practice:
The increase in the use of less expensive
generic alternatives could potentially reduce
both Medicare program and patient out-ofpocket medication spending without
compromising the access and quality of care.
Funding Source(s): NIH
Poster Session and Number: A, #192
Using Novel Health Messages to Influence
Food Choice in a Hospital: Repetition May
Matter More than the Specific Message
Mary Carol Mazza, Palo Alto Medical
Foundation/Stanford Clinical Excellence
Research Center; Robert Siegel, Cincinnati
Children's Hospital; Linda Dynan, Cincinnati
Children's Hospital; Anita Tucker, Harvard
Business School; Kathleen McGinn, Harvard
Business School
Presenter: Mary Carol Mazza, Post-Doctoral
Fellow, Palo Alto Medical Foundation/Stanford
Clinical Excellence Research Center
marycarolmazza@post.harvard.edu
Research Objective: To educate consumers
about healthier food choices and ultimately
combat obesity, nutrition labels and calorie
counts have been placed on packaged food
items and menus in a wide variety of venues.
However, the majority of labeling efforts have
focused on discrete calorie counts, an
information form which may be less applicable to
consumers than other means of expressing the
same information. This field study examines two
forms of nutritional health messages to assess
their impact on food and beverage choices in a
hospital cafeteria.
Study Design: Targeting chip and beverage
products, the health messages stated either the
amount of exercise (walking) one would need to
engage in to burn off the calories in an item or
the percentage of recommended daily calories
an item contained. Given that the average
calories in both products (an individual bag of
chips and a 20-ounce bottle of regular soda)
were equivalent, the messages were the same
across the two product categories (“almost an
hour of walking” and “12.5% of your
recommended daily calories”). The pair of
messages, one for chips and one for beverages,
was presented for 3 weeks followed by a 3-week
“no message” period and then the second pair of
messages. The order of the messages was
counterbalanced across the two product
categories such that the exercise message was
shown first for chips and the percentage of
recommended daily calories message appeared
first for beverages.
Population Studied: Cafeteria data in the form
of sales receipts were obtained as part of a
larger 21-month field study at an employee
cafeteria inside a large children’s hospital in the
Midwestern United States.
Principal Findings: For both beverages and
chips, the second health message resulted in
significantly (p < .05) greater healthful
purchases; the first health message yielded no
change. Given that the two types of health
messages were counterbalanced across the two
categories, this suggests that it may not be the
exact message- exercise or percentage of
recommended daily calories- so much as the
repetition of novel health messages that
improved the healthfulness of product sales.
Conclusions: The exact health information
conveyed in dietary health messages may be
less important than repeating the health
information in new, readily-understood and
applicable ways.
Implications for Policy, Delivery, or Practice:
Recognizing that caloric information can be
conveyed through means other than direct
calorie count may allow public health officials
and policymakers more opportunities to
influence consumer food choice and improve
obesity rates.
Funding Source(s): No Funding
Poster Session and Number: A, #193
Better to Avoid Tempting than Offering an
Alternative: Encouraging Healthful Choices
in a Cafeteria through Choice Architecture
Mary Carol Mazza, Palo Alto Medical
Foundation/Stanford Clinical Excellence
Research Center; Anita Tucker, Harvard
Business School; Kathleen McGinn, Harvard
Business School
Presenter: Mary Carol Mazza, Post-Doctoral
Fellow, Palo Alto Medical Foundation/Stanford
Clinical Excellence Research Center
marycarolmazza@post.harvard.edu
Research Objective: Consumers are thought to
construct heuristics and decision making
strategies that are highly context dependent,
suggesting that the environment partially
determines the mental shortcuts and strategies
used. This research seeks to modify consumer
food choice by altering environmental factors
such as the arrangement of food by testing two
seemingly opposing theories.
Research in social psychology suggests that
when two choices are evaluated jointly rather
than separately, preferences may reverse.
When evaluating items individually, there is no
basis of comparison and short-term desires and
impulses may have great strength given there is
little to force a cost-benefit analysis. In contrast,
when two choices are evaluated jointly rather
than separately, the joint evaluation of the two
prompts a more reasoned, rational analysis as
there is a choice that must be made between the
two items. This would suggest that offering a
less healthful item alongside a healthful item
would help the consumer make the reasoned,
healthful choice.
An alternative approach would suggest grouping
food into smaller subsections by healthfulness,
thus narrowing the options to all of one type,
healthful or less healthful, ultimately allowing a
consumer theoretically to focus on foods in the
healthful section and avoid temptation of less
healthful items.
Study Design: We focused our research on the
beverage and chip sections of a hospital
cafeteria. For the joint evaluation condition, we
posted a sign that said “Instead of this, try this”
with reference to a picture of Coca-Cola and
Diet Coke for the beverages and Doritos and
pita chips for the chips.
For the grouping condition, we organized both
beverage and chips into three distinct physical
groupings based on healthfulness, placing the
most healthful items in a green-labeled section,
the next most healthful in a yellow-labeled
section, and the least healthful in a red-labeled
section.
Population Studied: Cafeteria data in the form
of sales receipts were obtained as part of a 21month field study at an employee cafeteria
inside a large children’s hospital in the
Midwestern United States.
Principal Findings: Neither intervention
impacted chip sales. In contrast, the percentage
of healthful beverage sales decreased under the
joint evaluation condition (p < .05) and increased
in the grouping condition (p < .001).
Conclusions: Various food and beverage
product categories may be impacted differently
by environmental interventions. Additional
research is needed to further our understanding
of the psychology of food choice and uncover
the ways in which consumer behavior may
diverge from that predicted by current theories.
Implications for Policy, Delivery, or Practice:
The ease of introducing such environmental
changes coupled with their effectiveness in
impacting choice suggests that making changes
to the food environment may be one of the more
fruitful avenues to changing consumer dietary
choices. However, such changes should be
used with caution, as additional research is
required to examine boundary conditions on
current theories, especially those that seem to
support opposite environmental changes.
Funding Source(s): No Funding
Poster Session and Number: A, #194
Emergency Care Sought by Transplant
Recipients: A Review
Kathryn Schmidt, Northwestern University
Feinberg School of Medicine; Christopher
Richards, Northwestern University Feinberg
School of Medicine; Lisa McElroy, Northwestern
University Feinberg School of Medicine; Jane
Holl, Northwestern University Feinberg School of
Medicine; Daniela Ladner, Northwestern
University Feinberg School of Medicine; James
Adams, Northwestern University Feinberg
School of Medicine
Presenter: Lisa Mcelroy, Postdoctoral Research
Fellow, Northwestern Unviersity
lisa.mcelroy@northwestern.edu
Research Objective: Transplant recipients
present unique challenges when presenting for
emergency evaluation. Little is known about the
care of this population in the Emergency
Department (ED). The aim of this study was to
summarize the findings describing the utilization
of emergency care by transplant recipients.
Study Design: A search of PubMed was
performed using the Mesh terms “Emergencies",
"Emergency Medical Services", "Emergency
Service, Hospital", "Evidence-Based Emergency
Medicine", "Emergency Treatment", "Emergency
Nursing", "Emergency Medicine" combined with
“Liver transplantation” or “Kidney
transplantation”. ?Articles that described the
characteristics and care of adult transplant
recipients in the ED setting were identified by
three independent reviewers. Included
publications underwent thematic analysis.
Population Studied: Adult solid organ
transplant recipients
Principal Findings: Six publications met
inclusion criteria. All identified reports were
single-center pilot studies. Major themes
presented in the literature include: 1) transplant
recipients commonly present to the ED with a
wide variety of medical problems; 2) the majority
of transplant recipients present within the first
postoperative 60 days; 3) transplant recipients
have serious illnesses with nonspecific
presentations; 4) transplant recipients seeking
emergency care often experience extensive
diagnostic testing and long lengths of stay; and
5) transplant recipients experience high
admission rates.
Conclusions: Transplant recipients often
require emergency care, but for a wide range of
conditions. The extent of care received in the
emergency department is great, yet few studies
have examined the system of care delivery to
complex surgical patients in the ED.
Implications for Policy, Delivery, or Practice:
Further research is needed to assess the ED
process of transplant patients and the
communication practices between emergency
physicians and the transplant team, and to
enable the development of an algorithm that will
allow for patient-centered, quality care of
transplant recipients.
Funding Source(s): AHRQ
Poster Session and Number: A, #195
ACA-Mandated Elimination of Cost-sharing
for Preventive Screening has had Limited
Impact
Shivan Mehta, Perelman School of Medicine,
University of Pennsylvania; Daniel Polsky,
Perelman School of Medicine, University of
Pennsylvania; Jingsan Zhu, Perelman School of
Medicine, University of Pennsylvania; James
Lewis, Perelman School of Medicine, University
of Pennsylvania; Jonathan Kolstad, Wharton
School, University of Pennsylvania; George
Loewenstein, Social and Decision Sciences,
Carnegie Mellon University; Kevin Volpp,
Perelman School of Medicine, University of
Pennsylvania
Presenter: Shivan Mehta, Instructor, Perelman
School of Medicine, University of Pennsylvania
shivan.mehta@uphs.upenn.edu
Research Objective: The Patient Protection
and Affordable Care Act (ACA) of 2010
contained provisions to eliminate cost-sharing
for evidence-based preventive care. In this
study, we examined the impact of the elimination
of cost-sharing for colonoscopy and
mammography among small business plan
beneficiaries of Humana, a large national private
health insurance plan, using grandfathered
plans as a comparison group.
Study Design: We conducted an interrupted
time series analysis to examine whether the
change in cost-sharing policy was associated
with a change in screening utilization for the
small business population. Small business plans
were categorized as non-grandfathered
(intervention) or grandfathered (control), as
denoted by the insurer, based on whether they
were required by the policy to have no costsharing for preventive procedures. We
compared each intervention and control group
with itself, before and after the policy change.
The primary outcome variables were rates of
colonoscopy and mammography per person
month among the eligible population. We also
controlled for additional individual variables in
the adjusted analysis including age, gender,
education, race, and median income.
Population Studied: We analyzed de-identified
claims data from Humana for all patients aged
50-64 who were enrolled in small business plans
at any time between October 2008 and May
2012. Eligible patients for colorectal cancer
screening included men and women aged 5064, and breast cancer screening included
women between 50-64, as per the USPSTF
recommendations. We only included
beneficiaries enrolled in plans that were in
existence on March 2010, when grandfathered
status was determined, and who were
continuously enrolled for 2 years prior to the
change in the policy and for a minimum of 6
months afterwards.
Principal Findings: The increase in the
proportion of colonoscopies which required no
cost-sharing was small, as only 60% of
colonoscopies after the policy change were
recorded as preventive and those not coded as
preventive were not required to have zero cost
sharing. Few mammograms required costsharing before the policy change. There were no
consistent increases in utilization for
colonoscopy or mammography after the policy.
Unadjusted and adjusted regression analyses
for the small business cohort showed no
significant relative change in the small business
intervention plans relative to the control plans for
colonoscopy (coefficient .0005, p-value .12;
coefficient .0006, p-value .09) and
mammography utilization (coefficient .0019, pvalue .09; coefficient .0019, p-value .11). There
was also no significant relative change among
only those colonoscopies coded as preventive
(coefficient .0004, p-value .10; coefficient .0005,
p-value .09).
Conclusions: The results suggest that the
policy is not having its intended effects, as costsharing rates for colonoscopy and
mammography did not change substantially and
utilization of colonoscopy and mammography
changed little following the implementation of
this new policy approach.
Implications for Policy, Delivery, or Practice:
Additional policies should be considered to
increase screening rates such as waiving of
cost-sharing for preventive, diagnostic, and
therapeutic colonoscopies in eligible screening
populations, make communication of the
financial incentives more salient to patients, or
lower the price of a colonoscopy below zero to
offset the time and hassle costs of the
procedure.
Funding Source(s): Other Humana
Poster Session and Number: A, #196
Antibiotics As Perceived Service:
Stewardship through the Disparity Lens
Jake Morgan, Boston University
Presenter: Jake Morgan, Researcher, Boston
University
jakem@bu.edu
Research Objective: Inappropriate prescribing
of antibiotics is a major public health concern. It
remains widespread despite substantial
research and intervention efforts to improve
practice. Much of the current literature has
focused on spreading knowledge of clinical
guidelines in order to promote stewardship,
largely ignoring the role that racial, age, gender,
or socioeconomic disparities might have in the
antibiotic prescribing process. In particular, the
medical consumer has influence on whether or
not a medical visit yields a prescription, and
framing the issue as one of consumer choice
can highlight the case where patients who are
more highly educated, richer, or influential may
demand an antibiotic leading to the "worse"
health outcome of an inappropriate prescription.
This study adopts a behavioral economics
framework to understand the relation between
classic ideas of health disparity and the function
of antibiotics as a health "service."
Study Design: We conduct an in-depth
retrospective analysis of outpatient visits for
respiratory tract infections in an urban safety net
hospital. Using abstracted electronic medical
record data, we categorize the prescribing of an
antibiotic as appropriate or inappropriate and
comparing this to overall rates of prescribing.
We analyze these outcomes using bivariate
methods including chi-square analysis and build
a multivariable model to explain the
interrelatedness of the factors predicting the
outcome.
Population Studied: Our single site
retrospective analysis uses abstracted electronic
medical records from a large urban safety net
hospital, which provides an excellent opportunity
to exam the question of disparity as the patient
population is highly heterogeneous with regard
to race, socioeconomic status, and other
potential indicators of disparity.
Principal Findings: We find that certain nonclinical factors, including race, gender, and
insurance status, were significantly associated
with overall antibiotic prescribing rates. This
association, however, was not indicative of more
appropriate health prescribing. While uninsured
patients were less likely to be underprescribed
compared to commercially insured individuals
because they were less likely to receive an
antibiotic, for example, they were also more
likely to be underprescribed (p<0.001), indicating
that disparity, rather than concerns for clinical
appropriateness, may be driving outcome
heterogeneity.
Conclusions: The focus solely on clinical
outcomes such as antibiotic appropriateness
may mask underlying health disparities that are
important to address. Using the one view of
antibiotics as a clinical service rather than a
specific treatment reveals that health disparity
remains a significant barrier in health delivery.
Implications for Policy, Delivery, or Practice:
Future research and policy analysis can address
the barrier of disparity, and the urgent problem
of inappropriate antibiotic use and emerging
resistance, by including disparity factors when
considering clinical outcome success and
framing outcome studies as a consumer choice.
Funding Source(s): No Funding
Poster Session and Number: A, #197
Examining Potential Receptivity among LowIncome Women to a Voluntary Restaurant
Recognition Program in Los Angeles County
Seeking to Encourage Patron Selection of
Smaller Portion Sizes at Restaurants
Brenda Robles, Los Angeles County
Department of Public Health; Elaine Lai, Los
Angeles County Department of Public Health;
Amelia DeFosset, Los Angeles County
Department of Public Health; Lauren Gase, Los
Angeles County Department of Public Health;
Lauren Dunning, Los Angeles County
Department of Public Health; Tony Kuo, Los
Angeles County Department of Public Health
Presenter: Brenda Robles, Research Analyst,
Los Angeles County Department of Public
Health
brrobles@ph.lacounty.gov
Research Objective: To promote healthy eating
among Los Angeles County residents at risk of
overweight and obesity, the Los Angeles County
Department of Public Health launched the
Voluntary Restaurant Recognition Program
(VRRP) in 2013 to promote healthier meal
choices among restaurant patrons across Los
Angeles County. The salient feature of the
program is the requirements that restaurants
offer smaller portion size options at a lower
price, chilled water free of charge, and expand
children’s meal option to include fruits and
vegetables, healthy beverages, and non-fried
foods. Low-income women with young children,
in particular, are an important target population
for obesity prevention efforts as they often play a
critical role in influencing the eating behaviors,
habits, and attitudes of their children. This study
explores key factors that should be taken into
consideration to encourage low-income women
to make healthier restaurant selections for
themselves and their children at these
participating VRRP restaurants.
Study Design: A qualitative two-group
comparison design was used to examine unique
characteristics of the study population. Semistructured qualitative interviews were conducted
with the following groups: 1) 23 women with
children 0-5 part of the Early Head Start
program; and 2) 30 Los Angeles County adult
riders of Metro buses and railways. Semistructured interviews focused on food
consumption behaviors, restaurant food
purchasing behaviors, taste preferences, eating
behaviors, and receptivity to components of the
VRRP.
Population Studied: Low-income women with
children 0-5.
Principal Findings: Content analyses of
qualitative data identified five broad themes
across both groups that influence restaurant
food selection choices (in order of frequency): 1)
food quality; 2) price; 3) taste preference; 4)
restaurant convenience; and 5) food quantity
offered at restaurants. Compared to adult riders
of Metro buses and railways, low-income women
also reported children and pre-existing health
conditions as a driving force influencing family
food decisions.
Conclusions: Major factors that influence
restaurant food selection decisions appear to be
comparable among low-income women and
socio-economically diverse group of Los
Angeles County adults.
Implications for Policy, Delivery, or Practice:
Low-income women with young children are an
important target population for obesity
prevention efforts as many bear the dual
responsibility of caregiver and nutritional
gatekeeper. While the VRRP has the potential to
benefit this population, the program should be
tailored and marketed to address key concerns
around food quality, price, taste preference,
convenience, and quantity. Additionally, public
education efforts should occur in conjunction
with the program to encourage low-income
women to be champions of healthy eating for
their children
Funding Source(s): No Funding
Poster Session and Number: A, #198
Affective Factors Associated with
Medication-Taking Following Myocardial
Infarction
Lisa Rosenbaum, University of Pennsylvania;
Kevin Volpp, University of Pennsylvania, CHIBE;
David Asch, University of Pennsylvania, CHIBE,
Director RWJ Clinical Scholars; Loren Robinson,
University of Pennsylvania, RWJ Clinical
Scholar; Marcus Bachhuber, University of
Pennsylvania, RWJ Clinical Scholar
Presenter: Lisa Rosenbaum, Robert Wood
Johnson Clinical Scholar, University of
Pennsylvania
lisarose@mail.med.upenn.edu
Research Objective: Medication nonadherence impacts every disease but is
particularly concerning among patients who
have had a myocardial infarction (MI), as chronic
medication therapy achieves a clear reduction in
cardiovascular related morbidity and mortality.
Nevertheless, approximately one half to two
thirds of patients discontinue some or all of
these medications 1-2 years following MI.
Reasons for cardiovascular medication nonadherence are not well understood. Existing
interventions tend to target more obvious
barriers to adherence, such as cost, but these
interventions have only modest success,
perhaps because they fail to account for some of
the poorly understood emotional barriers
specifically related to medication-taking for
cardiovascular disease.
The purpose of this study was to better
understand the experience of patients who have
had an MI, and how this experience informs their
perspectives and feelings regarding medication
use.
Study Design: In person semi-structured
interviews of 20 patients hospitalized with
myocardial infarction were conducted, with
telephone follow up interviews one month and
six months later. Patients were asked about their
life following myocardial infarction, about
barriers to medication taking, ideas about how to
stay “heart-healthy,” and why they think others
may struggle with medications. All interviews
were audio-recorded and transcribed and have
been reviewed for salient themes by two
independent physician-researchers who did not
participate in the interview process.
Population Studied: Patients who have had
myocardial infarction in the Philadelphia region,
and who live independently.
Principal Findings: Several insights have
emerged from the analysis. Above all, many
patients harbor a visceral aversion to taking
medications for cardiovascular disease. This
aversion is fueled by a preference to manage
risk factors more “naturally,” such as via diet and
exercise, or with supplements such as herbs.
Indeed, when asked about the most important
thing they need to do to stay healthy, most
respond “diet and exercise,” as opposed to
medication-taking. Additionally, even those who
are not experiencing side effects note that the
possibility of side effects, in the long term, is a
concern. Some patients see medications as
something to be taken only for a defined period,
and believe they can and should be “weaned
off” medications once they have recovered.
Many articulated that cardiovascular disease is a
result of having lived an unhealthy life. Some
speculated that others’ unwillingness to take
medications may be because they see needing
to take medications for CV disease as a sign of
weakness.
Conclusions: These results point to the
emotional antagonism many harbor toward
medications for cardiovascular disease, in large
part driven by a preference for achieving
cardiovascular health through lifestyle changes
which are both “natural” and lack side effects.
Closely related is a sense that weaning off
medications is a sign of success, whereas taking
medications may be seen as a sign of
weakness.
Implications for Policy, Delivery, or Practice:
Current adherence interventions tend to focus
on practical barriers such as cost, regimen
complexity, memory lapses, and poor
understanding. Such interventions may be
improved by creative approaches that respond
to emotional barriers that are particularly
relevant to cardiovascular disease and its
treatment.
Funding Source(s): CMS
Poster Session and Number: A, #199
Are Chronically-Ill Adults who Advise
Friends and Family about Health Issues More
Likely to be Engaged in Their Own Health
Care than Those Who Do Not Assume This
Social Role?
Yunfeng Shi, Pennsylvania State University;
Jessica Mittler, Pennsylvania State University;
Strumpf Erin, McGill University
Presenter: Yunfeng Shi, Research Associate,
Pennsylvania State University
yus16@psu.edu
Research Objective: Overall, friends and family
are the second most cited source of health
information for individuals behind health care
professionals. Roughly two-fifths of U.S. adults
seek health information to support someone else
in a given year. This study examines whether
being a source of advice about health issues for
friends and family is associated with being more
engaged in one’s own health and health care.
Study Design: We use data from a randomdigit-dial survey of chronically ill adults between
June 2007 -2008. Respondents (n= 6792) are
from 14 Aligning Forces for Quality sites and a
national comparison sample. Being an adviser is
determined by the survey item “Friends or family
members ask me for my advice on health care
issues.” For analysis, the four response
categories (strongly disagree, disagree, agree
and strongly agree) are dummy variables. The
outcome measures are dichotomous variables
that assess respondents’ self-management of
their chronic illness(es), participation in healthcare decisions, and use of public reports about
providers’ performance. Relationships between
behaviors and being an advisor are estimated
using logistic regression models that control for
age, gender, chronic illness, education,
insurance status, marital status, individual
activation level and geography. Disagreeing with
being an adviser was the referent group.
Population Studied: Our study sample included
6729 individuals who responded to all the survey
items used in this study. Subsamples of each
AF4Q site and the national comparison group
are of similar sizes. Overall, 60% of respondents
reported being advisers.
Principal Findings: Advisers are 23% (agree
an adviser) to 30% (strongly agree an adviser)
more likely to be aware of quality information of
health care providers than those who do not give
advice to friends and family (disagree an
advisor). Similarly, advisers are more persistent
in asking questions of their providers compared
to non-advisers. Advisers are also 14% (agree)
to 17% (strongly agree) more likely to take
preventive actions for their chronic illnesses
compared to non-advisers (disagree). Being an
adviser is negatively related to exercising on a
regular basis.
Conclusions: Being an adviser to friends and
family on health issues is most strongly and
consistently associated with information-seeking
behaviors.
This is logical since (1) practicing informationseeking behaviors is likely a core criteria others
use assessing an individual’s credibility as an
adviser, and (2) perceiving oneself as adviser
may encourage the adviser’s informationseeking behaviors.
Implications for Policy, Delivery, or Practice:
These results suggest that interventions
targeting advisers provides a new, and
potentially more efficient way, to capitalize on
the information-seeking tendencies of this
population. First, a focus on improving advisers’
identification, comprehension, and application of
correct and relevant information should improve
the adviser’s own health and that of his/her
social partners. Second, current interventions
typically appeal to individual motivation (i.e. help
yourself, be healthier). Identifying and
capitalizing on the adviser role (i.e. help your
friends and family) holds promise for gaining
more widespread engagement.
Funding Source(s): RWJF
Poster Session and Number: A, #200
Associations between Pursuit of HealthRelated Information, Behavioral Changes,
and Health Service Utilization
Hyo Jung Tak, University of North Texas Health
Science Center; Gregory Ruhnke, University of
Chicago
Presenter: Hyo Jung Tak, Assistant Professor,
University of North Texas Health Science Center
hyojung.tak@unthsc.edu
Research Objective: Patients have access to
increasingly diverse sources of information that
may influence health-related behaviors and
health service utilization. Information derived
from such sources could serve as a less
expensive substitute for consultation with health
professionals, thereby lowering visit frequency.
Alternatively, it could complement and
potentially induce conventional visits if greater
awareness creates attitudes and concerns that
increase professional visit frequency. However,
research on the effect of health-related
information gathering on utilization has drawn
disparate conclusions. We examined the
association of information pursuit and related
behavioral changes with health service
utilization.
Study Design: We used the nationallyrepresentative, individual-level 2010 Health
Tracking Household Survey to analyze the
relationship between accessing alternative
information sources and utilization, as measured
by number of physician visits, having a
procedure, hospitalization, emergency room
(ER) visits, and visitation to non-physician
providers. We extracted (a) alternative source(s)
of health care information accessed (internet,
friends/relatives, newspapers/books/magazines,
TV/radio, and other) and (b) behavioral changes
in response to information obtained
(understanding treatment, approaches to health
maintenance and lifestyle, coping with a chronic
condition, and decisions to visit a doctor, ask a
doctor questions, and seek a second opinion).
Controlling for demographic characteristics,
socioeconomic and health status, we estimated
the association of information pursuit, behavioral
changes, and utilization overall, and stratified by
health status, race, and education level, using
negative binomial and logit models, each
separately specifying one utilization measure as
the dependent variable.
Population Studied:
Principal Findings: Among all respondents
(n=13,366, mean age 47.2, 52.1 percent
women, 66.8 percent white), 50.2 percent
obtained information from an alternative source
(32.0 percent and 28.9 percent from the internet
and friends/relatives, respectively). Of this
subset, 84.7 percent reported that such
information influenced their behavior –
understanding of treatment and approach to
health maintenance were influenced among 60.7
percent and 56.4 percent, respectively.
Obtaining information from both the internet and
friends/relatives was most strongly predictive of
physician visit frequency (incidence rate
ratio[IRR]=1.35;P=0.001), having any procedure
(odds ratio[OR]=1.80;P=0.001), and visitation to
non-physician providers (OR=1.69;P=0.001).
However, obtaining information only from
friends/relatives was most strongly predictive of
ER visitation (OR=1.55;P=0.001). Traditional
media (TV/radio) had no independent effect on
any utilization measure. Regarding behavioral
changes, the decisions to visit a physician and
seek a second opinion were most strongly
predictive of physician visit frequency
(IRR=1.24;P=0.001) and having a procedure
(OR=1.39;P=0.001), respectively. ER visitation
was reduced by understanding of treatment
(OR=0.75;P=0.049) and modification in
approach to health maintenance
(OR=0.64;P=0.002). The identified associations
were accentuated among respondents who were
healthier, White, and more educated.
Conclusions: Information obtained from the
internet and friends/relatives, as well as
decisions to ask questions and seek a second
opinion were predictive of greater physician visit
frequency and having procedures. Modifications
in treatment understanding and approach to
health maintenance reduced the likelihood of ER
visitation.
Implications for Policy, Delivery, or Practice:
Certain types of information gathering increase
health service utilization, most likely by raising
previously unrecognized health concerns,
especially among those with better health and
socioeconomic status. However, to the extent
that information improves treatment
understanding and approaches to health
management, improved health literacy may
decrease discretionary utilization.
Funding Source(s): No Funding
Poster Session and Number: A, #201
The Effect of Multiple Employer-Sponsored
Health Insurance Plans on Utilization among
Two-Worker Families
Hyo Jung Tak, University of North Texas Health
Science Center; Gregory Ruhnke, University of
Chicago
Presenter: Hyo Jung Tak, Assistant Professor,
University of North Texas Health Science Center
hyojung.tak@unthsc.edu
Research Objective: An estimated 7.6 percent
of the non-elderly population is in a family
having multiple employer-sponsored health
insurance plans (ESHIPs) concurrently. A family
might have multiple ESHIPs when a husband
and wife are each eligible for separate
employer-sponsored plans. Little empiric
research has estimated the impact of multiple
ESHIPs on utilization. We hypothesized that
families would enroll in multiple plans only in the
presence of significant premium subsidies
and/or the accrual of additional net benefits with
dual coverage, which could result in moral
hazard. We investigated whether non-elderly
individuals in a family with multiple ESHIPs
report higher levels of health service utilization.
Study Design: We used the 2006-2010 Medical
Expenditure Panel Survey Household
Component, restricting the sample to families in
which both spouses are offered health insurance
by their employers (15,349 individuals in 4,979
families). The primary independent variable was
a family-level binary indicator reflecting multiple
ESHIPs. Utilization was measured by the annual
number of physician and non-physician visits in
the calendar year prior to the interview. We
estimated the relationship between multiple
ESHIPs and the utilization measures with a
negative binomial model, and a two-stage
residual-inclusion model in which the
characteristics of each spouses’ employers were
used as instrumental variables (IVs). We
performed these estimates for the total sample
population and stratified by the presence of
children in the family, as dependents might
modify coverage choice incentives.
Demographic and socio-economic
characteristics, health status, and location of
residence were also controlled.
Population Studied:
Principal Findings: On average, respondents
visited physicians 2.78 times and non-physicians
1.60 times in the calendar year. Among those
eligible for multiple ESHIPs, 38.7 percent of
families were enrolled in two ESHIPs. Without
controlling for potential endogeneity, among
respondents with children, those covered by
multiple ESHIPs visited physicians 0.26 times
(average marginal effect (AME)=0.26, pvalue=0.02) more per year, compared to
individuals enrolled in a single ESHIP. In the
non-linear IV estimation, the constructed AME
was 0.45 (p-value=0.34) among respondents
with children. The estimation results were not
statistically significant for the total sample
population, and among families without children.
Notably, the non-linear IV model did not pass
the endogeneity test (p-value =0.06), implying
that enrollment in multiple ESHIPs is more likely
due to large subsidies rather than selection.
Conclusions: Enrollment in multiple plans is
associated with a greater frequency of physician
visits among families with children. In contrast to
previous literature showing endogeneity
between enrollment in a single ESHIP and
utilization, we find no evidence of such
endogeneity in the decision to enroll in a second
ESHIP.
Implications for Policy, Delivery, or Practice:
Although policymakers generally focus on the
consequences of being uninsured, our study
suggests that multiple ESHIPs within families
may lead to inefficient welfare losses via moral
hazard. Given frequent claims that certain
groups utilize excessive and perhaps
unproductive volumes of medical care, our
results have important implications.
Furthermore, given our findings that the primary
driver of dual coverage decisions is generous
premium subsidies from employers, policies
might address the inefficiencies created by the
tax incentives to provide such subsidies.
Funding Source(s): No Funding
Poster Session and Number: A, #202
Not Just “Informed” Decision-Making:
Emphasizing Support and Empathy in
Shared Decision-Making
Manasi Tirodkar, National Committee for Quality
Assurance; Holly Spalt, National Committee for
Quality Assurance; Nadia Alsado, Columbia
University Mailman School of Public Health;
Margaret O'Kane, National Committee for
Quality Assurance; Sarah Hudson Scholle,
National Committee for Quality Assurance
Presenter: Manasi Tirodkar, Research Scientist,
National Committee for Quality Assurance
tirodkar@ncqa.org
Research Objective: While evidence-based
decision aids and processes for shared
decision-making (SDM) exist, they are rarely
used in clinical practice. The purpose of this
study was to describe SDM initiatives used in
cancer treatment settings beyond support from
research or demonstration projects from both
the perspective of the practices and their
patients.
Study Design: We conducted case studies of
six organizations with sustained SDM initiatives
for breast or prostate cancer. We conducted
brief phone interviews with physicians,
administrators and clinical staff to elicit which
care team members were involved, their roles,
SDM workflow, and what information was
documented in the record. We also conducted
focus groups with patients who had been
diagnosed with breast or prostate cancer at
three of the organizations to explore how they
experienced the decision-making process.
Interview and focus groups were transcribed and
analyzed using Dedoose, a qualitative software
program.
Population Studied: Three organizations were
academic centers, two were medical groups,
and one was a private community-based
oncology practice. Focus groups were
conducted with 24 breast and 16 prostate
cancer patients with a diversity of age, race and
treatment choices.
Principal Findings: All organizations that had
sustained SDM processes had four elements in
common: identification of patients to receive
information and SDM support, distribution of
information and help to prepare for consultation,
having a conversation with the provider
regarding the treatment decision, and follow-up
and support from clinical support staff including
documentation of patient values and
preferences in the medical record. Organizations
were motivated to establish SDM processes in
order to provide more patient-centered care and
improve patient experiences. Implementation of
SDM was facilitated by culture change and
physician support of the SDM model. Key
informants cited lack of sustainable funding as a
key barrier to widespread implementation of
SDM. The most common data elements that key
informants cited as being documented in the
clinical record were whether the patient received
and reviewed the information, values and
preferences that might influence the patients’
decision, and the final decision on course of
treatment.
Focus group participants talked about the need
for emotional support and logistical support as
well as information relevant to their decision.
They expressed a need for empathy, stress
management and support to develop confidence
in their decision. Preparing for their consultation
(e.g. developing questions for the physician)
was cited as being very useful. Some patients
found having multiple options presented to them
through materials or specialist consultations
useful. However, patients also sought
information from many other sources such as
the internet, family/friends, and second opinions
and expressed that they are not sure what
information to trust; some patients desired a
recommendation from the physician.
Conclusions: Sustained shared decisionmaking processes for breast and prostate
cancer go beyond simply distributing information
to patients. A conversation with the provider is
crucial for successful SDM. Patients require
support to prepare for consultations and
empathy from providers to feel confident in their
decision.
Implications for Policy, Delivery, or Practice:
Efforts to support standard processes for
implementing and documenting SDM in the
medical record are needed along with
measurement of patient experiences of SDM.
Funding Source(s): RWJF
Poster Session and Number: A, #203
Estimating the Effect of Primary Care and
Specialist Copayment Levels on Primary
Care Utilization
Laurel Trantham, BCBSNC; Daryl Wansink,
BCBSNC
Presenter: Laurel Trantham, Informatics
Scientist, BCBSNC
laurel.trantham@bcbsnc.com
Research Objective: To determine the
relationship between primary care and specialist
copayment amounts and the likelihood of
seeking care from primary care providers.
Study Design: Retrospective study. The first
model predicted the likelihood of seeking care
from a primary care provider (PCP) as a function
of primary care copayment. The second model
had the same dependent variable of interest, but
used specialist copayment as the primary
independent variable. All models were estimated
using logistic regression and controlled for
gender, age, risk, comorbid conditions,
physician access, plan type, and severity of
condition (for specialist models only).
Additionally, we estimated the models
separately by gender and the presence of
chronic conditions (primary care models only).
Population Studied: More than 1,000,000
covered lives in a state-wide health insurance
program in North Carolina. The study population
was limited to adults age 18-64 with continuous
coverage through a PPO plan throughout 2012.
For the specialist utilization analysis, we limited
the sample to members who sought care for one
of a set of selected conditions.
Principal Findings: The primary care
copayment value did not have the expected
negative relationship with probability of primary
care utilization. Plan members with 30 dollar
copayments were more likely to seek care from
a PCP than members with 25 dollar
copayments. The effect was consistent across
men and women as well as those with and
without chronic disease. The effect of specialist
copayment on primary care utilization varied
with the preference sensitivity of the condition.
For non-preference sensitive conditions such as
fractures and diverticulitis, members were
equally likely to seek care from a PCP rather
than a specialist regardless of specialist
copayment level. For preference sensitive
conditions like back pain and gastroesophageal
reflux disease, members with high specialist
copayments were more likely than members
with low specialist copayments to seek care
from a PCP rather than a specialist.
Conclusions: Small increases or decreases in
copayment levels may not result in changes in
care utilization, particularly for primary care and
conditions that are not preference sensitive.
Nearly two-thirds of the members in the primary
care analysis had copayments between 20 and
30 dollars, and the lack of variation in
copayments may make it difficult to detect an
effect. In the specialist analysis, we found most
conditions to be preference sensitive, however,
the effect is most evident at higher copayment
levels. That is, primary care utilization was not
strongly related to specialist copayment at
copayments less than 70 dollars. Orthopedic
and dermatologic conditions were the exception
and showed a strong relationship between
specialist copayment and primary care utilization
at all copayment levels.
Implications for Policy, Delivery, or Practice:
As insurers are increasingly attempting to steer
members towards preferred providers through
differential copayments, it is essential to
understand that small changes in copayment
level may not have the desired effect. Insurers
and employer groups looking to implement
value-based benefits should investigate
additional levers to ensure their members
receive care from high-quality, efficient
providers.
Funding Source(s): Other BCBSNC
Poster Session and Number: A, #204
Characteristics and Utilization Among
Consumers Who Choose to Enroll in a
Consumer-Directed Health Plan
Margaret Warton, Kaiser Permanente Divison of
Research; Jie Huang, Kaiser Permanente
Divison of Research; Ilana Graetz, Kaiser
Permanente Divison of Research; Richard
Grant, Kaiser Permanente Divison of Research;
Mary Reed, Kaiser Permanente Divison of
Research
Presenter: Margaret Warton, Data Consultant,
Kaiser Permanente Divison of Research
margaret.m.warton@kp.org
Research Objective: Consumer directed health
plans with high deductibles are increasingly
offered to health insurance enrollees in both
employer-sponsored and new health insurance
exchange plans. Little is known about which
types of consumers choose a plan with a
deductible when also offered the choice of
enrolling in a traditional non-deductible benefit
plan. We describe patient characteristics and
healthcare utilization among consumers offered
a choice between deductible consumer directed
and non-deductible health plans.
Study Design: Within an integrated healthcare
delivery system (IDS), we conducted a
descriptive study of patients insured under
employer-sponsored health plans without a
deductible during 2010 who were offered a
choice between deductible and non-deductible
plans in 2011. Multivariate logistic regression
estimates of adjusted odds of choosing a
deductible plan used patient covariates from
2010. We also compared utilization between the
deductible and non-deductible groups in 2012
using t-tests.
Population Studied: We studied patients age
18 and older with employer-sponsored nondeductible plans in 2010 who were under age
65, had continuous coverage in the IDS, and
had no evidence of Medicare or COBRA
coverage through the end of 2012. Among all
37,217 patients offered a choice between
deductible and non-deductible plans, 54% were
female, 51% were of non-white race/ethnicity,
22% lived in a low SES neighborhood, and 18%
worked for a small business. Mean age was
42.4 years (SD 11.7).
Principal Findings: Among consumers with a
choice of plans, 20.1% (7,482) chose a
deductible-based plan in 2012. In multivariate
analyses, those choosing a deductible over a
non-deductible plan were more likely to be
younger (Odds Ratio 0.78, 95% CI 0.72-0.84 4664 vs. 18-30 years), Asian (OR: 1.15, CI: 1.071.25 vs. non-Hispanic white), live in a low SES
neighborhood (OR: 1.08 CI: 1.01-1.15), and
work for a small business (OR: 1.99, CI: 1.872.11). Patients choosing a non-deductible plan
were less likely to be black (OR: 0.63, CI: 0.550.72 vs. non-Hispanic white), have one or more
chronic conditions (OR: 0.80, CI: 0.74-0.86 vs.
no chronic conditions), and have frequent office
visits (OR: 0.81, CI: 0.74-0.88 for >5 office visits
vs. none) and ED visits (OR: 0.84, CI: 0.77-0.94
for at least 1 ED visit vs. none) in 2010, the year
before their plan choice. In the year following
their plan choice, those with deductible plans
continued to be have significantly lower rates of
office visits (mean 2.7 visits vs. 5.3 visits in
2012) and ED utilization (mean 0.15 vs. 0.21
visits in 2012) than those with non-deductible
plans.
Conclusions: Consumers with employersponsored coverage who chose deductible
plans differed significantly from those who chose
non-deductible coverage. Members who chose
non-deductible plans had fewer chronic
conditions and lower utilization.
Implications for Policy, Delivery, or Practice:
Differences in choice of plan are likely driven in
part by cost and utilization considerations. Such
self-selection may be a source of bias in
analyses of health plan characteristics and
healthcare outcomes.
Funding Source(s): Other Kaiser Permanente
Community Benefits Grant
Poster Session and Number: A, #205
Evaluating the Effect of the Enhanced
Benefits Rewards Program Participation on
Florida Medicaid Enrollees’ Acute Care
Utilization
Shuo Yang, Department of Health Services
Research, Management and Policy, University
of Florida; Jeffrey Harman, Department of
Health Services Research, Management and
Policy, University of Florida; Allyson Hall,
Department of Health Services Research,
Management and Policy, University of Florida;
R. Paul Duncan, Department of Health Services
Research, Management and Policy, University
of Florida
Presenter: Shuo Yang, Health Economist,
Department of Health Services Research,
Management and Policy, University of Florida
shuoyang2000@gmail.com
Research Objective: The Enhanced Benefits
Reward (EBR) program is a major component in
Florida Medicaid Reform that intends to uses
financial incentives to promote healthy behaviors
among Medicaid beneficiaries, such as
participation in preventive care, smoking
cessation, or disease management programs. It
is expected, by encouraging people to engage in
healthier behaviors, the program will be able to
improve enrollees’ health and lower Medicaid
enrollees’ health care expenditures and
utilization, especially acute care utilization.
However, to date few studies have been
conducted to systematically examine the effect
of financial incentive programs in public sector.
This study examines the difference in Florida
Medicaid enrollee’s acute care utilization
associated with different levels of EBR program
participation.
Study Design: The study period is from fiscal
year 2006-2007 to 2009-2010. Outcome
measures, including beneficiaries’ annual total
inpatient days, number of emergency
department (ED) visits, and avoidable
hospitalizations, were calculated using Medicaid
claims data. The amount of reward credits
earned by Medicaid beneficiaries participating in
the program was extracted from Florida’s
Enhanced Benefits Information System data and
used to determine beneficiaries’ EBR program
participation level. Florida Medicaid Reform
enrollees who participated in the EBR program
with different participation levels were compared
to non-participating Medicaid Reform enrollees.
Logistic models were used to examine the
association between EBR program participation
level and the odds of having any ED visits,
inpatient days, or avoidable hospitalizations.
And negative binomial models were adopted to
analyze the relationship between EBR
participation level and outcome measures for
individuals with positive utilization. All
parameters were estimated using GEE method.
Sensitivity analyses were conducted to assess
whether how participation level was defined and
enrollment length limitations had any impact on
the results of the study.
Population Studied: The study population
included Florida Medicaid Reform mandatory
participants in urban reform counties (Broward
and Duval) who enrolled in a Provider Services
Network plan.
Principal Findings: Participating in EBR
program is associated with lower likelihood of
having any inpatient days, ED visits, and
avoidable hospitalizations, especially
participation levels from two years ago. For the
enrollees who had positive utilization, EBR
program participation is associated with lower
number of annual total inpatient days and ED
visits, but the effect on avoidable
hospitalizations is not clear. The findings from
sensitivity analyses were consistent with the
main study findings.
Conclusions: Results from this study suggest
that participating in EBR program is associated
with lower Medicaid acute care utilization. As
expected, the effect of the program is not
instantaneous and its influence accrues over
time.
Implications for Policy, Delivery, or Practice:
Findings from this study suggest that other
states may want to consider implementing
similar programs. To further improve the
program, modification of the current incentive
structure may be warranted.
Funding Source(s): No Funding
Poster Session and Number: A, #207
Key Factors in Women’s Decisions when
Choosing a Pediatrician: Considering
Diverse Patient Perspectives
Yara Youssef, Baystate Medical Center/Center
for Quality of Care Research; Penelope Pekow,
Baystate Medical Center/Center for Quality of
Care Research; Jasmin Roberts, Baystate
Medical Center/Center for Quality of Care
Research; Peter Lindenauer, Baystate Medical
Center/Center for Quality of Care Research;
Katharine White, Baystate Medical Center;
Sarah Goff, Baystate Medical Center/Center for
Quality of Care Research; Tetine Sentell,
University of Hawaii; Jill Miyamura, Hawaii
Health Information Corporation
Presenter: Yara Youssef, Research Assistant,
Baystate Medical Center/Center for Quality of
Care Research
yyoussefpc@gmail.com
Research Objective: Choosing a pediatric
practice is a decision made by millions of
families each year. Little is known about what
factors matter to women when choosing a
pediatrician or how this varies among ethnic
groups. We interviewed ethnically diverse
women about key factors in the selection of a
pediatric practice.
Study Design: As part of a post-delivery survey,
women in Honolulu, HI (n=116) and Springfield,
MA (n=113) were asked to rate how various
factors affected their choice of pediatrician using
a four-point Likert scale or selecting “did not
consider.” Chi-squared tests were used to
assess differences in how women in the two
study locations rated each factor. Kruskal-Wallis
rank tests were used to compare the importance
of the factors by ethnicity.
Population Studied: English-speaking women
ages 18-50 between 20-34 weeks of gestation in
Springfield, MA and Honolulu, HI.
Principal Findings: HI and MA populations
differed at baseline on median age (29.0 HI vs.
25.0 MA years, p<0.05), education level (62.0%
HI vs 2.6% MA received a college degree or
higher, p<0.05), yearly income (44.0% % HI vs.
9.7% MA earned $40,000-$99,999 annually,
p<0.05). Finally, the HI population was 32.8%
White, 28.5 Asian, and 27.6% Pacific Islander
while the MA population was 19.5% White,
59.3% Hispanic, and 15.9% Black (p<0.05).
Factors in which the highest percentage of all
the women (n=229) selected “mattered a lot”
included insurance acceptance (91.7%), office
staff accessibility by phone (78.2%), and how
close the office is to home (47.2%). Factors in
which the lowest percentage of women selected
“mattered a lot” included information found on
the pediatrician’s website (11.3%),
recommendation from a friend (22.7%), and
office accessibility by public transportation
(24.5%). 42.4% did not consider online care
quality data while it “mattered a lot” to 25.8% of
women who did consider it. 41.5% did not
consider online patient experience data while it
“mattered a lot” to 18.8% of women who
considered it.
Factors that varied significantly in importance
among MA and HI included: how easily it is to
get to the office using public transportation
(47.8% vs. 24.1%, p<0.05), recommendation
from a family member about a pediatrician
(51.3% vs. 29.6%, p<0.05), already knowing the
pediatrician (76.9% vs. 26.7%, p<0.05) and
already knowing the office staff (79.5% vs.
25.9%, p<0.05).
Three factors varied significantly in importance
across racial groups (p<0.05). “Office
accessibility by public transportation” mattered
more to Blacks compared to other ethnic groups.
“Already knowing the staff” and “online care
quality data” mattered more to Blacks,
Hispanics, and Pacific Islanders.
Conclusions: Among this diverse population of
women, insurance acceptance and office staff
accessibility by phone mattered most to women.
Online pediatric quality information was not
considered by the majority of women in this
study when choosing a pediatrician. The
relative importance of some factors considered
varied based upon study location and ethnicity.
Implications for Policy, Delivery, or Practice:
Factors other than care quality may matter most
to women when choosing a pediatrician.
Ethnicity was associated with the importance of
several of the factors and should be considered
when designing interventions to increase use of
quality data when choosing a pediatrician.
Funding Source(s): RWJF
Poster Session and Number: A, #209
Admission rates from the Emergency
Department as a Measure of Community
Access to Care: Identifying Community
Resources Sensitive Conditions (CRSC)
Sheryl Davies, Stanford University School of
Medicine; Benjamin Goldstein, Stanford
University; Kathryn McDonald, Stanford
University School of Medicine
Presenter: Sheryl Davies, Research Associate,
Stanford University School of Medicine
smdavies@stanford.edu
Research Objective: Admissions from the
emergency department (ED) have been shown
to vary substantially across providers and have
been implicated as “the most expensive decision
made in the ED.” Decisions to admit to the
hospital are influenced by the patient’s clinical
condition, resources available to the patient in
the community and practice patterns. We aimed
to identify diagnoses for which admission rates
from the ED were associated with fewer
community based resources for a significant
acute care need. These community sensitive
resource conditions (CRSC) could provide a
signal by which to assess community-level
access to care.
Study Design: We first identified AHRQ Clinical
Classification System (CCS) categories with
total admission rates between 5 and 75 percent
for all age groups. Using a Delphi review, a
panel of 9 ED clinicians rated the CCS
categories as to their appropriateness as a
CRSC and refined the included diagnosis codes,
resulting in a list of clinically-identified CRSC.
Using the Agency for Healthcare Research and
Quality (AHRQ) Healthcare Cost and Utilization
Project (HCUP) State Inpatient and State
Emergency Department Databases, 2008-2010,
we calculated admission rates for these
clinically-identified CRSC based on the patient’s
county of residence. We estimated the variance
explained by county and hospital effects using a
hierarchical logit model. As a second method of
identifying CRSC we empirically evaluated the
relationship between admission rate for each of
the CCS categories and resource factors
(county level poverty, lack of insurance, primary
care provider density, mental health provider
density, self-reported social support (source,
County Health Rankings, 2012)) using
standardized and weighted linear regression,
selecting CCS categories for which county
resource factors explained the most variance in
admission rates.
Population Studied: The HCUP databases
include 80,205,600 all-payer visits, in 28 states
and 1778 counties.
Principal Findings: The clinically-identified
CRSC included: asthma, chronic obstructive
pulmonary disease, diabetes, hypertension,
convulsions, deep vein thrombosis, pneumonia,
gallstones without infection, cellulitis of face
neck or trunk and pressure ulcer. The mean
county-level admission rate for CRSC was 3.0
percent (SD=3.2) Admission rates were highly
persistent from year to year (autoregressive
correlation = 0.93). In the hierarchical model,
county-to-county variance was 0.30 and
hospital-to-hospital variance, 0.22. Lower
community-based resources explained the most
variation in admission rate (R2 > 0. 25) for the
following empirically-identified CRSC: liver
disease, anemia, coronary atherosclerosis,
dehydration, nonhypertensive heart failure,
chest pain, thromboembolism, abdominal hernia
and coagulation disorders. Admission for
pneumonia, gallstones and cellulitis were
positively related to community resources, and
were removed from the CRSC list.
Conclusions: Admission rates from the ED vary
systematically nationwide, although these
admission rates are more related to proxy
measures of access to care for a subset of
diagnoses. Admission rates for the CRSC are
explained by both patient’s county of residence
and hospital-level effects, with county-level
effects being more explanatory.
Implications for Policy, Delivery, or Practice:
We identified 15 diagnostic groups for which
admission rates are either clinically or
empirically related to the availability of
community based resources to support acutely
ill individuals. This concept could be used to
measure access to care.
Funding Source(s): AHRQ
Poster Session and Number: A, #170
Opioid use among OEF/OIF Veterans with
Traumatic Brain Injury
Teresa Hudson, University of Arkansas for
Medical Sciences; Rebecca Pope, STAT
Corporation; Silas Williams, VA HSR&D Center
for Mental Health Outcomes and Research;
Bradley Martin, University of Arkansas for
Medical Sciences, College of Pharmacy; Mark
Sullivan, University of Washington; John
Fortney, University of Arkansas for Medical
Sciences, College of Medicine; Mark Edlund,
Research Triangle Institute
Presenter: Teresa Hudson, Assistant Professor,
University of Arkansas for Medical Sciences
hudsonteresaj@uams.edu
Research Objective: Injuries of the head and
neck, particularly due to explosions are more
common among veterans who served in Iraq
(Operation Iraqi Freedom – OIF) and
Afghanistan (Operation Enduring Freedom –
OEF) compared with veterans in previous
conflicts. An estimated 60% of Veterans
wounded by explosions suffer traumatic brain
injury (TBI) which is often associated with both
acute and chronic pain syndromes for which
opioids are commonly prescribed. OEF/OIF
Veterans have risk factors associated with
chronic opioid use and opioid misuse/abuse
including high rates of mental health and
substance use disorder. Post-traumatic Stress
Disorder (PTSD) has been associated with
negative outcomes among Veterans using
opioids and is common in Veterans with TBI.
This study describes opioid prescribing among
OEF/OIF Veterans with TBI who received care
within the Veterans Health Administration (VHA)
and, controlling for PTSD, identifies factors
associated with negative outcomes.
Study Design: This is a retrospective,
epidemiologic study of opioid use among
OEF/OIF Veterans using national VHA data
repositories.We calculated the proportion who
used any opioids and the proportion of use
opioids chronically (defined as 91 or more days
within the fiscal year) and modeled factors
associated with chronic use and with adverse
outcomes. Adverse outcomes were defined as:
opioid-related accidents, opioid overdose, selfinflicted injuries, violence-related injuries.
Population Studied: We identified all OEF/OIF
Veterans with a TBI diagnosis in FY’11. To be
included in the sample we required Veterans to
be listed in the OEF/OIF roster, to have two
records of any use of VA services with the fiscal
year and to have at least one ICD-9 CM code
consistent with TBI. We excluded Veterans who
received only inpatient opioid medications, who
had a cancer diagnosis or who were living in
hospice, skilled nursing facility, or domiciliary.
Principal Findings: A total of 99,648 OEF/OIF
Veterans with TBI received care in VA in FY’11.
The mean age was 33 years, 5.3% were female.
Approximately 74.1% were Caucasian, 16.2%
were other races; race was missing in 9.7% of
the sample. Approximately a third of the sample
(33.6) used any outpatient opioid medications
with only 13.6% use opioids chronically. The
morphine equivalent dose among chronic users
was 42.7 mg daily compared to 22.1 mg among
non-chronic users. Additional analyses are
underway to identify risk factors for chronic
opioid use and for negative outcomes among
opioid users with TBI.
Conclusions: The rate of chronic opioid use
was relatively low in OEF/OIF Veterans with TBI.
In previous analyses of Veterans using VHA
care we found that approximately 57.7% of all
opioid users received opioids chronically
regardless of diagnosis.The mean morphine
equivalent dose was also lower compared with
cohort of all opioid users mean=26.2mg (SD
30.6).
Implications for Policy, Delivery, or Practice:
The rate of chronic opioid use, and the mean
dose associated wtih chronic use among
OEF/OIF Veterans wtih TBi may be lower than
with Veterans overall. This may imply that opioid
prescribing in this popoulation is less
problematic than suggested in previous
publications.
Funding Source(s): NIH
Poster Session and Number: A, #173
Coverage and Access: Medicaid and
Exchanges
The Effect of Pre-PPACA Medicaid Eligibility
Expansion in New York State on Access to
Specialty Surgical Care
Oluseyi Aliu, University of Michigan; Katherine
Auger, Cincinnati Children’s Hospital Medical
Center; Gordon Sun, Partnership for Health
Analytic Research LLC; James Burke, The
University of Michigan Health System; Colin
Cooke, The University of Michigan Health
System; Kevin Chung, The University of
Michigan Health System, Ann Arbor, MI; Rodney
Hayward, The University of Michigan Health
System
Presenter: Oluseyi Aliu, Research Fellow,
University of Michigan
oluseyi@umich.edu
Research Objective: Critics argue that
expanding health insurance coverage through
Medicaid may not result in improved access to
care. The ACA provides reimbursement
incentives aimed at improving access to primary
care services for new Medicaid beneficiaries;
however, there are no such incentives for
specialty services. Using the natural experiment
of Medicaid expansion in New York State in
October 2001, we examined whether Medicaid
expansion increased access to common
musculoskeletal procedures for Medicaid
beneficiaries.
Study Design: We obtained estimates of the
quarterly probability of Medicaid beneficiaries
undergoing select musculoskeletal procedures
between January 1998 and December 2006. We
used these estimates in an interrupted time
series (ITS) model with variance weighted least
squares regression (VWLS) to examine the
association between Medicaid expansion in
October 2001 (main predictor variable) and the
percentage of Medicaid beneficiaries amongst
all patients who received the select
musculoskeletal procedures (outcome variable)
in each quarter of the study period. The
variables of interest were those central to ITS
analysis: a dichotomous variable for the pre- vs.
post-intervention; Medicaid expansion (January
1998 to September 2001 vs. October 2001 to
December 2006), time as a continuous variable
and an interaction term between the two. Using
the coefficients for these variables from the
VWLS models, we calculated the following: (1)
the abrupt change in percent Medicaid after
Medicaid expansion and (2) whether the
temporal slope changed post-intervention.
Population Studied: From the State Inpatient
Database for New York State, we identified 1964 year old patients who received; lower
extremity large joint replacement, spine
procedures and upper/lower extremity
fracture/dislocation repair from January 1998December 2006.
Principal Findings: Of the 254,650 patients
who underwent any one of the selected
musculoskeletal procedures, 23,442 (9.2%)
were Medicaid beneficiaries. Following Medicaid
expansion, there was a small but statistically
significant abrupt increase in the probability of a
musculoskeletal surgical patient being a
Medicaid beneficiary (0.5% [95% CI 0.1, 0.9]).
More notably, there was a reversal of the time
trend, from a slight decline in the proportion of
Medicaid beneficiaries prior to the expansion (0.08% [95% CI -0.12, -0.04] per quarter) to a
steady increase (+0.12% [95% CI 0.10, 0.14] per
quarter) after expansion. By roughly 5 years
after the Medicaid expansion, the proportion of
the musculoskeletal surgical patients who were
Medicaid beneficiaries was 4.7% higher [95% CI
3.7, 5.3] than expected based on the preexpansion time trend.
Conclusions: Medicaid expansion in New York
State appeared to significantly improve access
to common musculoskeletal procedures, for
Medicaid beneficiaries.
Implications for Policy, Delivery, or Practice:
As states expand Medicaid, policy analysts
should pay close attention to which providers
provide services to newly covered Medicaid
patients, with the potential that improved access
may largely result from greater payment equity
for those already pre-disposed to caring for the
poor.
Funding Source(s): No Funding
Poster Session and Number: B, #726
Enrollment and Utilization Following CHIP
Expansion:
David Becker, UAB School of Public Health;
Justin Blackburn, UAB School of Public Health;
Michael Morrisey, UAB School of Public Health;
Bisakha Sen, UAB School of Public Health;
Meredith Kilgore, UAB School of Public Health;
Cathy Caldwell, Alabama Department of Public
Health; Christopher Sellers, Alabama
Department of Public Health; Nir Menachemi,
UAB School of Public Health
Presenter: David Becker, Associate Professor,
UAB School of Public Health
dbecker@uab.edu
Research Objective: In October 2009 Alabama
expanded eligibility in its Children’s Health
Insurance Program, known as ALL Kids, from
200% to 300% of the federal poverty level (FPL).
We examine the expenditures, health services
utilization and re-enrollment behavior of ALL
Kids expansion enrollees (200-300% of FPL)
relative to traditional enrollees (100-200% FPL)
in the program.
Study Design: We use claims data to construct
person-month level expenditure and utilization
measures. We use a two part-modeling strategy
to examine differences in expenditures by ALL
Kids eligibility category after controlling for
enrollee characteristics. We use the enrollment
data to examine the growth in ALL Kids
expansion enrollment and differences in reenrollment behavior by ALL Kids eligibility
category. We use probit models to examine
patient characteristic adjusted differences in reenrollment behavior by eligibility category.
Population Studied: The study is based upon a
sample of 2,054,074 person-months of ALL Kids
coverage between October 2009 and December
2011 and 245,088 periods of enrollment initiated
after October 2009 and ending prior to
December 2011.
Principal Findings: From the expansion of ALL
Kids eligibility in October 2009, the expansion
population increased steadily to approximately
20% of total enrollment as of October 2012.
Expansion enrollees exhibit modestly higher
monthly expenditures than traditional ALL Kids
enrollees, with higher outpatient expenditures
($78.37 vs. $66.86 for low-fee and $72.56 for
fee-group enrollees) but lower emergency
department expenditures ($11.58 vs. $17.04 for
low-fee and $12.93 for fee-group enrollees).
The expansion enrollees exhibit marginally lower
utilization of ED services for low-severity
conditions and higher utilization of physician
outpatient visits. However, overall their
expenditures and utilization are fairly similar to
those of the fee-group enrollees. In our
analysis of re-enrollment behavior we find that
expansion enrollees are around 5-10 percentage
points more likely to re-enroll in ALL Kids than
traditional enrollees.
Conclusions: The expansion of ALL Kids
coverage has added approximately 16,000 new
enrollees to the program. They exhibit more
persistent enrollment in the program and
patterns of health expenditures and utilization
that are roughly similar to those of traditional fee
group enrollees who are subject to the same
levels of cost-sharing.
Implications for Policy, Delivery, or Practice:
Although states are prohibited from changing
CHIP eligibility until 2019, the costs associated
with the expansion population will be important
to future policy decisions. Our work also
provides a useful framework for states who are
considering the cost implications of Medicaid
expansion under the Affordable Care Act.
Funding Source(s): Other Alabama
Department of Public Health
Poster Session and Number: B, #727
Health-Related Expenditures and Financial
Burdens among Veterans pre -ACA
Didem Bernard, Agency for Healthcare
Research and Quality; Thomas Selden, AHRQ;
Susan Yeh, Johns Hopkins School of Public
Health
Presenter: Didem Bernard, Senior Economist,
Agency for Healthcare Research and Quality
didem.bernard@ahrq.hhs.gov
Research Objective: There are currently 24
million living veterans in the US, over 6 million
receive care from the Department of Veteran
Affairs (VA) every year, which costs
approximately $52 billion annually. Because of
limited resources, VHA uses a priority system to
establish which veterans can actually receive
care. Preference is given to those who have a
service-connected disability or low income, and
lower-priority veterans may have to pay
copayments for services.
The effect of the Affordable Care Act on the use
of veteran health services is unclear. On the
one hand, uninsured veterans who are currently
not using the VHA, may choose to enroll. On the
other hand, some veterans who are currently
using the VHA may switch to Medicaid or obtain
coverage through the exchanges if they quality
for subsidies. We examine health care
expenditure levels and health-care related
financial burdens among veterans by reliance on
VHA services. Understanding reliance on VHA
among veterans with other insurance coverage
is helpful to policymakers as the composition of
veterans who access the VHA and the utilization
levels may change with the ACA.
Study Design: We examine expenditures and
burdens for (1) veterans who utilize both VA and
non-VA health care, (2) veterans who utilizes VA
health care exclusively, and (3) veterans who
utilizes non-VA health care exclusively. High
burden is defined as spending on health care
and health insurance premiums greater than
20% of family income.
Population Studied: We use pooled data from
the Medical Expenditure Panel Survey (MEPS)
for 2006-2011. The sample includes persons
aged 18 to 64 who were honorably discharged
from military services.
Principal Findings: Preliminary results based
on 2006-2010 data show that mean
expenditures are: $10,776 among veterans who
use a combination of VA and non-VA health
care; $2,527 among veterans who only use VA
care, and $2,965 among veterans who only use
non-VA health care. Among veterans who use
VA and non-VA health care, the VA, private
insurance, Medicare, Medicaid and Tricare
account for 23%, 53%, 7%, 4% and 4%,
respectively, of expenditures with 10% paid outof-pocket. For veterans who use only VA health
care, VA accounts for 86% of expenditures.
Family income is significantly lower among
veterans with only VA use ($36,878) compared
to veterans with VA and non-VA use ($68,312)
and veterans with only non-VA use ($75,927).
Mean out-of-pocket expenditures are $1,503
among veterans with only VA use, $3,581
among veterans who use both VA and non-VA
health care, and $3,511 among veterans who
only use non-VA health care.
We find that 10.7 % among veterans who
utilizes both VA and non-VA health care,
compared to 5.3% among veterans who use
only non-VA health care had high burdens in
pooled data from 2006-2010. The difference in
the prevalence of high burdens between
veterans with both VA and non-VA use (10.7%)
and veterans with only VA use (8.2%) is not
statistically significant.
Conclusions: Health care expenditures and
burdens vary significantly among veterans by
reliance on VHA health care, suggesting that the
change in coverage options with the ACA can
lead to significant changes in the utilization of
VHA services.
Implications for Policy, Delivery, or Practice:
The preliminary results are for all veterans. The
proposal will extend these results by examining
expenditures and burdens by ACA-relevant
subgroups, such as those who will be eligible for
Medicaid, and those who will be eligible for
subsidies for coverage through the Exchanges.
Funding Source(s): AHRQ
Poster Session and Number: B, #728
Public Health Care Spending among Those
Eligible for Medicaid and Premium Tax
Credits in the Insurance Marketplaces under
the ACA
Didem Bernard, Agency for Healthcare
Research and Quality; Thomas Selden, AHRQ;
Yuriy Pylypchuk, Social & Sceintific Systens,
Inc.
Presenter: Didem Bernard, Senior Economist,
Agency for Healthcare Research and Quality
didem.bernard@ahrq.hhs.gov
Research Objective: U.S. health care spending
was $2.7 trillion or 17.9% of GDP in 2012.
Combining public outlays with implicit public
spending through tax expenditures, the public
share of total health spending was 63.0% in
2012. This study combines estimates from the
National Health Expenditure Accounts and data
from the Medical Expenditure Panel Survey
(MEPS) to examine the variation in the
percentage of health care paid for by the public
sector by insurance status and among
subgroups who will gain access to coverage
through adult Medicaid expansions and premium
tax credits in the new insurance marketplaces
under the Affordable Care Act (ACA).
Study Design: We use household and
employer survey data, national expenditure
benchmarks, and microsimulation modeling. We
align 2010 MEPS by type of service and source
of payment with 2010 NHEA benchmarks to
account for underreporting in surveys. Next, we
allocate amounts in NHEA that are outside the
scope of MEPS such as administrative costs,
DSH payments, research and investment.
Finally, we estimate a comprehensive array of
tax expenditures. Eligibility for Medicaid and
premium tax credits is simulated.
Population Studied: Our sample includes
persons aged 18-64.
Principal Findings: Based on analysis using
2010 data, under the ACA, 12.3 million who are
not currently enrolled in public coverage will be
eligible for Medicaid under the 138% threshold
and live in states that will expand Medicaid.
Public spending accounts for 41.1% of total
health care spending ($4,277) among those who
will be eligible for Medicaid. Another 9.7 million
are currently not enrolled in Medicaid, and would
have been eligible for Medicaid but live in nonexpansion states. Public spending already
accounts for 37.2% of total health care spending
($5,340) among this group. Under the ACA, 19.6
million will be eligible for premium tax credits in
the exchanges. Public spending accounts for
39.3 % of total health care spending ($3,794)
among this group. Currently, 96.4 million have
health insurance offers (own employer or
spouse) and are not eligible for exchange
coverage. Even among this group, public
spending accounts for 43.1% of total spending.
Another 10.0 million have incomes higher than
the subsidy threshold in the exchanges (400%
federal poverty line) but do not have offers. This
group will be able to participate in the
marketplace but without subsidies. Public
spending currently accounts for 33.3% of total
health care spending ($5,689) among this group.
Conclusions: Under the ACA, 31.9 million will
be eligible for Medicaid or premium tax credits in
the marketplaces. ACA implementation will
dramatically enhance insurance availability and
will increase public spending on health care
among these populations.
Implications for Policy, Delivery, or Practice:
This study provides a baseline to evaluate the
impact of the ACA on the level and incidence of
outlays (Medicaid/CHIP and Medicare),
exchange subsidies, and tax expenditures.
Funding Source(s): AHRQ
Poster Session and Number: B, #729
Does Disability Affect Whether Access
Problems Predict Health Insurance
Switching?
Elizabeth Blodgett, University of North Carolina
Presenter: Elizabeth Blodgett, Doctoral Student,
University of North Carolina
e.geneva.blodgett@gmail.com
Research Objective: To determine whether
healthcare access problems predict switching
within the private health insurance market for
people with disabilities.
Study Design: I conducted quantitative analysis
of 5 consecutive rounds (one panel) of the
Medical Expenditure Panel Survey (MEPS). I
performed logistic regression models to test
whether an increase in the reported rate of
access problems early in the reference period
was associated with an increased likelihood of
switching private health insurance plans (either
to a different source, or to a different plan from
the same source) later in the reference period.
Additionally, I tested whether there was a
different marginal effect of access problems on
switching for disabled adults as compared to
nondisabled adults. Each model controlled for
demographic and socioeconomic factors that
could facilitate or impede insurance coverage
switching.
Population Studied: My sample included 7,951
privately-insured working-age noninstitutionalized adults surveyed in the MEPS.
Participants were classified as disabled or
nondisabled based on their work limitation
status.
Principal Findings: People with disabilities
were significantly more likely to report problems
in accessing medical care, dental care, or
prescription medications. However, only nondisabled individuals showed a significantly
increased likelihood of switching health
insurance plans associated with increasing rates
of access problems. Higher rates of access
problems did not predict increased insurance
switching among individuals with disabilities. All
models controlled for age, sex, race, ethnicity,
self-rated health, years of education received,
marital status, income, employment status, and
whether the individual changed jobs during the
reference period.
Conclusions: The post-ACA private health
insurance market assumes that individuals can
and will change insurance when needed in order
to find the best fit for their health and finances.
However, switching insurance can be difficult for
Americans with disabilities. This population has
more health problems and a greater need for
continuity of care, both of which we know to be
associated with less willingness to change
insurance. Additionally, evidence from other
countries indicates that people with disabilities
tend to see more barriers to insurance switching,
even when these barriers are not objectively
present. The current study extends this literature
to Americans with disabilities, finding that
healthcare access problems do not prompt
these individuals to change their healthcare
coverage. Given these findings, there is strong
reason to be concerned about how people with
disabilities will adapt to the major changes
underway in our health insurance system.
Implications for Policy, Delivery, or Practice:
In other countries’ health insurance markets
under managed competition, consumers –
including those with disabilities - generally
address access problems by switching their
insurance. If Americans with disabilities are less
likely to switch their coverage in response to
access problems, they may not benefit from the
better coverage and lower cost promised by the
Affordable Care Act. Disability advocates and
policymakers should specifically target people
with disabilities in the private insurance market
to ensure that the greater consumer mobility and
better, more affordable care promised by the
ACA are equally available to all Americans.
Funding Source(s): No Funding
Poster Session and Number: B, #730
Impact of the Affordable Care Act on
Patients’ Out-of-Pocket Burden
Christine Buttorff, Johns Hopkins Bloomberg
School of Public Health; Kevin Riggs, Johns
Hopkins Medical Instituions; Caleb Alexander,
Johns Hopkins School of Public Health
Presenter: Christine Buttorff, Phd Candidate,
Johns Hopkins Bloomberg School of Public
Health
cbuttorf@jhsph.edu
Research Objective: Out-of-pocket (OOP)
spending caps under the Patient Protection and
Affordable Care Act (PPACA) are designed to
limit high health care costs for individuals and
their families. Individuals and families who
obtain private insurance outside of health
insurance exchanges will have caps equal to
that of the Health Savings Account qualified
plans ($6350 for individuals and $12,700 for
families for 2014). For coverage purchased in
the exchanges, OOP caps are discounted from
these maximums on an income-basis so that
patients with lower incomes have lower OOP
caps. The goal of this study was to estimate the
proportion of Americans who have OOP costs
greater than the OOP spending caps.
Study Design: The primary outcome was the
proportion of individuals whose spending
exceeded the uniform OOP cap for insurance
purchased outside of health insurance
exchanges. To examine the effect of OOP caps
for individuals with varying incomes, we
categorized individuals’ incomes relative to the
federal poverty limit (FPL). We assessed the
proportion of privately insured that would be
over the 2014 limits, treating the whole sample
as if it were in and out of the exchange. This
estimate will provide boundaries on the impact
of the OOP max since it is impossible to tell
which of the currently privately insured may
move to the exchanges in 2014. Including the
currently uninsured would bias upward the
proportion since with insurance, their OOP
spending will be different. Costs were inflated to
2014 dollars. Lastly, we examined how selfreported cost-related access problems (unable
to obtain medical care or prescription drugs for
reasons “could not afford” or “insurance would
not cover”) varied by OOP costs.
Population Studied: We used the 2011 Medical
Expenditure Panel Survey, a large nationally
representative survey to examine OOP costs for
medical care and prescription drug utilization.
We limited our analyses to adults age 18-64 with
private health insurance.
Principal Findings: There were 12301 privately
insured adults in the survey, representing a
weighted population of over 136 million
individuals. Median household income was
$77,755. The proportion of individuals with OOP
costs greater than the uniform cap was 0.93%,
and the proportion exceeding the cap decreased
with increasing family income (2.10% of those
<100% of FPL, and 0.84% of those >400% of
FPL). Applying the exchange’s income-based
caps, the proportion with OOP costs greater
than the cap was 2.29%. Specifically, 3.75
times as many individuals <400% FPL would
benefit from the income-based caps, compared
with the uniform caps. Self-reported cost-related
access problems were low (1.99% overall),
although problems decreased as income
increased (5.21% for <100% FPL and 1.02% for
>400% FPL).
Conclusions: Individuals with lower income
experience greater self-reported cost-related
access problems, and they are more likely to
have high OOP costs.
Implications for Policy, Delivery, or Practice:
The uniform OOP spending caps under PPACA
will impact a modest proportion of individuals,
with those in the lowest income categories most
likely to benefit. However, extending the
income-based spending caps to include
insurance purchased outside of the exchanges
would have a much larger impact. Additionally,
policy should address the gap in subsidy for
those under 100% of the FPL living in states that
decided not to expand Medicaid, allowing them
to be eligible for the premium and cost-sharing
subsidies, as well as out-of-pocket maximums.
Funding Source(s): No Funding
Poster Session and Number: B, #731
The Seasonality of Medicaid Enrollment:
Implications for Medicaid/Marketplace
Planning and Outreach
Cheryl Camillo, NORC at the University of
Chicago; Cheryl Camillo, Samuel Stromberg,
NORC at the University of Chicago; Vikki
Wachino, NORC at the University of Chicago
Presenter: Cheryl Camillo, Principal Research
Scientist, NORC at the University of Chicago
camillo-cheryl@norc.org
Research Objective: To identify and explain
seasonality in Medicaid applications and
enrollment for the purposes of interpreting
Medicaid and Marketplace enrollment dynamics,
including churning; refining enrollment
projections; and informing outreach strategies.
Study Design: Using validated state Medicaid
Statistical Information System (MSIS) data
obtained from the Centers for Medicare &
Medicaid Services (CMS), we computed monthto-month changes in enrollment for each study
population over the last decade and analyzed
them in the context of national economic and
policy changes. We identified trends and
compared and contrasted them across
populations. We drew on state Medicaid
eligibility experience and collected and reviewed
qualitative data to hypothesize explanations.
We tested some using regression analysis.
Population Studied: We studied four main
populations: non-disabled adults (who constitute
most of the "newly eligible); non-disabled
children; the disabled; and the aged. We also
examined subsets of these populations.
Principal Findings: There is strong evidence
that Medicaid application and enrollment varies
seasonably for certain populations, particularly
non-disabled adults and children, typically
peaking in October. This pattern is consistent
across states and time periods.
Conclusions: Seasonal effects could be
explained by institutional demands (for example,
school vaccination requirements),
employment/economic cycles,and the U.S.
cultural calendar.
Implications for Policy, Delivery, or Practice:
Seasonal variation must be taken into account
when interpreting Medicaid/Marketplace
enrollment data. State outreach and staffing
strategies should be developed with seasonal
effects in mind. In addition, policymakers should
be mindful of these effects when designing
policies to address churning between Medicaid,
the Marketplace, and uninsurance.
Funding Source(s): Other
Poster Session and Number: B, #732
Potential Adult Medicaid Beneficiaries under
the Patient Protection and Affordable Care
Act Compared with Current Adult Medicaid
Beneficiaries
Tammy Chang, University of Michigan; Matthew
Davis, University of Michigan
Presenter: Tammy Chang, Assistant Professor,
University of Michigan
tachang@med.umich.edu
Research Objective: Under healthcare reform,
states will have the opportunity to expand
Medicaid to millions of uninsured U.S. adults.
Information regarding this population is vital to
physicians as they prepare for more patients
with coverage. Our objective is to describe
demographic and health characteristics of
potentially eligible Medicaid beneficiaries.
Study Design: Cross-sectional study of the
National Health and Nutrition Examination
Survey (2007-2010), to identify and compare
adult U.S. citizens potentially eligible for
Medicaid under provisions of the Affordable
Care Act (ACA) to current Medicaid
beneficiaries. We compared demographic
characteristics (age, gender, race/ethnicity,
education) and health measures (self-reported
health status; measured body mass index [BMI],
hemoglobin A1C, systolic and diastolic blood
pressure, depression screen [Patient Health
Questionnaire, PHQ-9], tobacco smoking and
alcohol use).
Population Studied: Nationally representative,
cross-sectional sample of US civilian, noninstitutionalized adults age 19-64.
Principal Findings: Potentially eligible
individuals are expected to be more likely male
(49.2% potentially eligible vs 33.3% current
beneficiaries; p<0.001), more likely white and
less likely black (58.8% white, 20.0% black vs
49.9% white, 25.2% black; p<0.02), and have no
significant difference in educational attainment.
Overall, potentially eligible adults are expected
to have better health status (34.8% “excellent” or
“very good,” 40.4% “good”) than current
beneficiaries (33.5% “excellent” or “very good,”
31.6% “good”; p<0.001). The proportions obese
(34.5% vs. 42.9%; p<0.001)and with depression
(15.5% vs. 22.3%; p=0.002) among potentially
eligible individuals are significantly lower than for
current beneficiaries, while there are no
differences in the expected prevalence of
diabetes or hypertension. Current tobacco
smoking (49.2% vs. 38.0%; p=0.002), and
moderate and heavier alcohol use (21.6% vs.
16.0% and 16.5% vs 9.8%; p<0.001,
respectively) are more common among the
potentially eligible population than among
current beneficiaries.
Conclusions: Under the ACA, physicians can
anticipate a potentially eligible Medicaid
population with equal if not better current health
status and lower prevalence of obesity and
depression than current Medicaid beneficiaries.
Implications for Policy, Delivery, or Practice:
Therefore, federal Medicaid expenditures for
newly covered beneficiaries may not be as high
as anticipated in the short term. However, given
the higher prevalence of tobacco smoking and
alcohol use, broad enrollment and engagement
of this potentially eligible population is needed to
address their higher prevalence of modifiable
risk factors for future chronic disease.
Funding Source(s): No Funding
Poster Session and Number: B, #733
American’s Understanding of Health
Insurance Terminology and the Affordable
Care Act: A Preliminary Study with Opt-in
Online Panels
Robin Cohen, National Center for Health
Statistics; Sarah Joestl, National Center for
Health Statistics; Kathleen O'connor, National
Center for Health Statistics
Presenter: Robin Cohen, Statistician, National
Center for Health Statistics
rcohen@cdc.gov
Research Objective: To obtain some early
information on the public’s knowledge of the
provisions of the Affordable Care Act (ACA) and
health insurance terminology using a
commercial opt-in online panel to inform
questionnaire and instrument content for
national health surveys.
Study Design: In November 2013, the National
Center for Health Statistics (NCHS) launched
the Health Insurance Terminology Survey
(HITS). HITS is a web-based survey of
approximately 1,000 respondents. The sample
was purchased through a commercial survey
research company. Individual knowledge of
health insurance and ACA terminology was
assessed through a series of true/false, multiple
choice, and “select the best answer” questions.
Population Studied: A commercial opt-in online
panel of adults
Principal Findings: Interim results found that
among HITS respondents aged 18-64, 92%
correctly identified the term ‘copayment’, 85%
correctly identified the term ‘deductible’, and
80% correctly identified the term ‘premium’. In
regard to a question about the definition of a
health insurance subsidy where respondents
could indicate more than one response, just
under three-quarters (73%) of respondents
indicated that a health insurance subsidy is ‘a
benefit for people below a certain income’ and
32% indicated ‘a benefit given by some
employers’. Just over three-quarters (76%) of
respondents heard of the Health Insurance
Marketplace, and 91% of these individuals knew
that ‘this is where people and employers can
buy health insurance policies’. When presented
with the following statement, ‘The ACA creates a
new government-run insurance plan to be
offered along with private plans’, 35% of
respondents indicated this was a true statement,
38% indicated it was false, and 27% were not
sure.
Conclusions: The commercial opt-in online
panel provided NCHS with a timely sample
frame for obtaining early information and
understanding of health insurance terminology
and ACA during the early stages of
implementation. The findings identified some
areas for questionnaire improvement. These
results will be augmented by more formal means
of question and questionnaire development,
such as key informant interviews, cognitive
interviews, focus groups, and field tests.
Implications for Policy, Delivery, or Practice:
The use of internet panels can be useful for
understandings what people know and can
contribute to the development of future survey
questions about health insurance.
Funding Source(s): CDC
Poster Session and Number: B, #734
Health Care Utilization Among Children
Enrolled in Medicaid and CHIP Via Express
Lane Eligibility
Margaret Colby, Mathematica Policy Research,
Inc.; Brenda Natzke, Mathematica Policy
Research, Inc.
Presenter: Margaret Colby, Senior Health
Researcher, Mathematica Policy Research, Inc.
mcolby@mathematica-mpr.com
Research Objective: Express Lane Eligibility
(ELE) enables state Medicaid and/or Children’s
Health Insurance Programs (CHIP) to use
another agency’s eligibility findings to qualify
children for public health insurance coverage,
thus facilitating enrollment of eligible uninsured
children. Lack of prior enrollment may reflect
limited awareness about the enrollment process
or lower demand for health care. Because ELE
sometimes requires few affirmative enrollment
steps from families, a related issue is whether
enrollees understand their coverage. We
examined utilization by ELE enrollees to assess
whether: (1) they understand how to access
care and (2) their demand for care differs from
children enrolling via standard pathways.
Study Design: We compared first-year
utilization among non-disabled children who
enrolled through ELE and standard pathways.
Two-step estimation was used, first examining
the likelihood of utilization and then service
volume and cost among users. We examined
several types of services: inpatient,
outpatient/physician, and emergency room
visits; prescription drugs, vision care, dental
care; and behavioral health. Outcomes included
any service use, the number of uses or visits,
cost of care, and exclusive use of that service
type without other claims. We also examined
length of time until first service receipt.
Regression-adjustment was used to account for
demographic characteristics and enrollment
month.
Population Studied: We included non-disabled
children ages 0-18 enrolled in Medicaid or CHIP
through ELE or standard pathways in Alabama,
Iowa, Louisiana, and New Jersey during 2009 2012. Sample members were continuously
enrolled for six months and had no prior public
coverage or had a gap in public coverage of at
least two months. ELE sample sizes ranged
from 1,800 to 61,000 across states.
Principal Findings: Most ELE enrollees (65 to
94 percent across states) accessed services in
their first year of enrollment, and many (46 to 63
percent) accessed services within the first two
months. ELE and other enrollees tended to use
a variety of services, and users had multiple
visits. However, ELE enrollees were somewhat
less likely to use each service type studied, and
users often accessed fewer services than other
enrolled children in their state, by small but
significant margins. For example, ELE enrollees
who used services averaged 10 to 44 percent
fewer outpatient/physician visits. Regressionadjusted fee-for-service costs for the first year
were 14 to 52 percent lower for ELE enrollees
using services, compared to children enrolling
via standard pathways.
Conclusions: Utilization patterns suggest ELE
enrollees are aware of their coverage and able
to access services. Results are consistent with
the theory that eligible children who do not enroll
directly may have a lower demand for health
care than their enrolled peers.
Implications for Policy, Delivery, or Practice:
Findings provide support for the expansion of
ELE as an enrollment simplification tool and do
not support concerns that ELE enrollees are
unaware of their coverage. Most ELE enrollees
access a variety of Medicaid/CHIP services,
suggesting that public health insurance is
providing substantial value to ELE families.
However, states considering ELE—especially
those that negotiate contracts with capitated
managed care organizations to deliver
services—may expect that eligible but uninsured
children will be less expensive to cover than
existing beneficiaries.
Funding Source(s): Other ASPE
Poster Session and Number: B, #735
CHIPRA Express Lane Eligibility (ELE)
Evaluation: Lessons Learned from Case
Studies of Eight States
Brigette Courtot, Urban Institute; Sheila Hoag,
Mathematica Policy Research; Ian Hill, Urban
Institute; Jennifer Edwards, Health Management
Associates; Margo Wilkinson, Urban Institute
Presenter: Brigette Courtot, Research
Associate, Urban Institute
bcourtot@urban.org
Research Objective: CHIPRA permitted ELE,
which can streamline children’s access to
coverage by letting states rely on eligibility
findings of other public agencies to determine
whether a child qualifies for Medicaid or CHIP.
We examined implementation of ELE in eight
states, as part of a broader, mixed-methods
evaluation of ELE.
Study Design: We conducted in-depth site
visits to 8 states—Alabama, Iowa, Louisiana,
Maryland, Massachusetts, New Jersey, Oregon,
and South Carolina—that implemented ELE
between 2009 and 2012. Approximately 20
interviews were conducted in each state with
Medicaid and CHIP officials, partner agency
staff, eligibility workers and others, as well as
focus groups with parents of children enrolled
via ELE.
Population Studied: CHIP and Medicaid
program administrators, other state agency
officials, state legislators, child and family
advocates, and consumers of state Medicaid
and CHIP programs.
Principal Findings: States have implemented
three types of ELE: (1) automated processes
that enable states to use data linkages with
partner agencies (such as those administering
SNAP) to automatically enroll and/or renew
children in coverage; (2) mailings-based ELE
processes that function as a form of outreach by
identifying children likely to be eligible and
sending a simplified Medicaid/CHIP application;
and, (3) other ELE processes such as electronic
referrals between children’s coverage programs.
Almost all ELE processes aim to reduce
application processing times; most also simplify
the application experience for enrollees and
improve outreach. The states that experienced
the greatest enrollment or renewal gains and
administrative savings implemented automatic
ELE processes.
Experiences of the study states underscore the
significance of allowing states flexibility in
implementing ELE. For instance, ELE can be
phased in, allowing states to test it and resolve
operational issues, or to expand the policy as
resources and support permit. State
circumstances are also a key consideration
when determining whether ELE is a viable
process for initial enrollment, renewal, or both.
Selecting an appropriate partner agency is
important, but the process states implemented
for using partner data has a greater effect on the
number of children enrolled or renewed through
ELE. ELE saves time for staff processing
applications and/or renewals; as a result, ELE
has helped states deal with hiring freezes and
staff layoffs. Because ELE is more efficient than
standard application or renewal processes, it
expedites coverage for beneficiaries.
Conclusions: ELE can be an effective tool for
facilitating the enrollment of children into
Medicaid and CHIP, and for helping them retain
coverage. Case study findings reflect an
inherent value of ELE; the policy is adaptable
and states that have used it did so in ways that
suited their circumstances.
Implications for Policy, Delivery, or Practice:
The ELE evaluation provides an important
opportunity to document ELE implementation
and understand the implications of adopting the
policy. There is no single way to implement ELE,
and the way ELE is implemented can profoundly
affect its potential benefits. Evaluation findings
suggest four ELE best practices to maximize
coverage: (1) adopt automated ELE processes;
(2) use ELE for renewal; (2) choose Express
Lane partners with centralized, linkable data;
and, (4) consider ELE processes that remove
administrative barriers for families.
Funding Source(s): Other DHHS Assistant
Secretary for Planning and Evaluation (ASPE)
Poster Session and Number: B, #736
How Do Medicaid Premiums Affect Take-up
and Retention? Evidence From Wisconsin
Childless Adults and Parents
Laura Dague, Texas A&M University; Thomas
DeLeire, Georgetown University; Lindsey
Leininger, University of Illinois-Chicago;
Marguerite Burns, University of WisconsinMadison
Presenter: Laura Dague, Assistant Professor,
Texas A&M University
dague@tamu.edu
Research Objective: Some state Medicaid
programs, when using waivers to expand
coverage to low-income but not poor enrollees,
charge premiums as a way of offsetting the cost
of care and of making the program more like
private insurance. Knowledge about how
variation in premiums affects take-up and
retention is currently limited, especially among
newly covered populations of adults without
dependent children. We study how variations in
premiums affect the enrollment, retention, and
access to health care for all adults.
Study Design: In June 2012, Wisconsin
introduced premiums for all new and current
adults in BadgerCare Plus (the state’s Medicaid
program) with family income from 133%-150%
of the Federal Poverty Level, increased premium
amounts for those with incomes above 150%,
and expanded the consequences of
nonpayment of premiums by increasing the
restrictive re-enrollment period from six to 12
months. We use difference-in-differences and
regression discontinuity empirical analyses to
study the effects of these changes.
Population Studied: Universe of BadgerCare
Plus-enrolled parents, caretakers, and adults
without dependent children in Wisconsin, from
2010-2013.
Principal Findings: Preliminary results indicate
that the newly introduced premiums resulted in
significant disenrollment, while changing the
amount of the premium had a smaller effect. The
number of new enrollees also declined.
Conclusions: Introducing and changing
premiums has important negative effects on the
enrollment and retention of higher-income
Medicaid enrollees.
Implications for Policy, Delivery, or Practice:
Premiums will affect take-up and continuity of
coverage for relatively low-income enrollees
and, beyond Medicaid this has implications for
subsidized exchange-based insurance. Lowincome enrollees are likely to forego premium
payments, as they did in Wisconsin’s Medicaid
program even when facing significant lock-out
periods. Premiums, rather than serving as a
significant revenue offset of Medicaid costs, may
yield cost savings indirectly by deterring
participation in the program.
Funding Source(s): Other State of Wisconsin
Department of Health Services
Poster Session and Number: B, #737
When It Can’t Wait Until Morning: Low Acuity
ED Use in an Insured Population
Andrea DeVries, WellPoint; Winnie Chi,
HealthCore, Inc; Lori Uscher-Pines, RAND
Corporation; Ateev Mehrotra, Harvard Medical
School
Presenter: Andrea DeVries, Director, WellPoint
adevries@healthcore.com
Research Objective: A recent study on
insurance expansion indicates that increased
coverage may actually increase Emergency
Department (ED) use - an alarming possibility
given the impact of the Affordable Care Act and
already overwhelmed EDs. The reported
increase in ED use is surprising given that
excessive use of the ED (for example, visiting an
ED for a low-acuity condition such as bronchitis
or strep throat) is typically attributed to a lack of
access to primary care and/or lack of insurance.
In this study we compared characteristics and
preferences of people visiting an ED or primary
care physician (PCP) for low-acuity conditions
with the goal of informing strategies to mitigate
inappropriate ED use in a newly insured
population.
Study Design: A 45-question survey was
administered to commercially insured members
(18-64 yrs) of Blue Cross Blue Shield plans who
either visited an ED or PCP for a low-acuity
condition between Nov 2010 and April 2011.
Differences in demographic characteristics and
responses were compared using nonparametric
analysis of variance (ANOVA) methods for
continuous variables and ?2 tests for categorical
variables.
Population Studied: Respondents lived in
metropolitan areas from 11 states in the
Western, Midwestern, and Southeast United
States. The ED cohort included 451
respondents and the PCP cohort included 397
respondents. Both cohorts had similar insurance
benefits.
Principal Findings: In both groups, the majority
of patients reported having a usual source of
care (94.5% ED vs. 95.5% PCP, p>.05) and that
their doctors’ office was a usual source of care
(85% ED vs. 95% PCP, p<.05). Before seeking
treatment, 44.9% of respondents in the ED
cohort called their physician office before going
to the ED. Among those that did not call their
physician office before going to the ED, the
majority said it was because the office was
closed. The respondents who used ED were
more likely to have used EDs frequently in
childhood (14.0% ED vs. 8.9% PCP, p<.05) and
were more likely to report that ‘people around
them’ were likely to use the ED for medical care
(26.9% ED vs. 16.2% PCP. p<.01). The ED
cohort was significantly more likely to seek care
within 24 hours of experiencing symptoms
(43.2% ED vs. 11.8% PCP, p<.01) and
significantly more likely to report that they
needed to be treated within 2 hours for their
most recent acute visit (64.2% ED vs. 40.5%
PCP, p<.01).
Conclusions: We identify several key
mechanisms that might drive people to go to the
ED instead of a PCP for a low-acuity reason
including: the PCP office sent them to the ED, a
lack of after-hours options, social norm that
going to the ED is acceptable, and differences in
perceived urgency of the medical problem.
Implications for Policy, Delivery, or Practice:
While insurance coverage clearly offers
increased access to care for previously
uninsured individuals, there remains significant
inappropriate ED use even within insured
populations. Strategies to reduce low-acuity ED
may include increased guidance and information
regarding what is an emergent problem, other
alternatives for care outside the ED, and offering
expanded office hours and /or triage within a
physician practice.
Funding Source(s): Other WellPoint, Inc.
Poster Session and Number: B, #738
Sampling and Analysis Issues related to the
CMS Consumer Experience Surveys for
Health Insurance Exchanges and Qualified
Health Plans
Christian Evensen, American Institutes for
Research; Mike Cohen, American Institutes for
Research; Steven Garfinkel, American Institutes
for Research
Presenter: Christian Evensen, Senior Research
Scientist, American Institutes for Research
cevensen@air.org
Research Objective: We developed two
surveys based on the Consumer Assessment of
Healthcare Providers and Systems (CAHPS) to
assess consumers’ on-going experiences with
the new Health Insurance Exchanges
(Marketplaces) and Qualified Health Plans
(QHPs). Both surveys will be field tested in 2014
and the sampling design for each is challenging
given complexities in defining and identifying
applicants, enrollees, and QHPs. The research
objective is to develop and implement a
sampling design that produces data required for
evaluating measurement properties of both
surveys.
Study Design: The sampling design has been
developed, but is subject to change given the
nature of the Marketplaces. Sampling frames will
be constructed using CMS databases. For the
Marketplace survey, potential respondents will
be randomly sampled from each of the 51
Exchanges. For the QHP Enrollee survey,
potential respondents will be randomly sampled
from 30 QHPs across the nation.
Population Studied: The study population for
the Marketplace survey includes adults who
have at a minimum provided their contact
information, regardless of how far they got in the
application and enrollment process. The study
population for the QHP Enrollee survey includes
adults enrolled in a QHP for 5 months or longer
with no more than one 30-day break in
enrollment.
Principal Findings: Challenges associated with
sample frame construction and sampling for the
field tests of both surveys have implications for
analysis. With respect to sampling frame
construction, there are 16 separate sources of
data – one from each of the 15 state-based
marketplaces (SBMs) that have their own Web
site (includes D.C.), and a single frame from the
remaining 36 federally facilitated marketplace
(FFM) states using HealthCare.gov. SBMs
provide data to CMS for enrollees or effectuated
enrollees only, thus limiting the ability of the
Marketplace survey field test to evaluate
consumer experiences from those who
accessed a Marketplace but never enrolled in a
QHP.
While we use the term QHP as a semantic
convenience, the definition of the sampling and
reporting unit presents challenges. Early data
indicate that, just in the 36 FFM States, there
are over 4,400 unique QHP products offered by
approximately 200 issuers. If a QHP is defined
by combining all of an issuer’s offerings in a
given State within metal levels within product
types, there are approximately 965 such units in
the 36 FFM States.
Conclusions: For the Marketplace survey field
test, the main challenge for sampling is
capturing the full range of consumer experience
and minimizing bias related to Marketplace type.
For the QHP survey, the main challenge is to
define a sampling unit that will allow future
reporting at a level that is meaningful to
consumers and other stakeholders while
avoiding undue burden on issuers in future
implementations by requiring them to conduct
duplicative data collections from enrollees in
products that are virtually identical.
Implications for Policy, Delivery, or Practice:
Precise and practical definitions of applicants,
enrollees, and QHPs are needed to minimize
bias in survey results. Findings from the field
test procedures and results from the survey data
analysis will inform necessary changes for
national implementation of both surveys and
public reporting of QHP data.
Funding Source(s): CMS
Poster Session and Number: B, #739
Estimating the Effect of Medicaid Expansion
on Veterans Health Administration
Enrollment and Utilization
Austin Frakt, Deptartment of Veterans Affairs &
Boston University; Amresh Hanchate,
Department of Veterans Affairs & Boston
University; Steven Pizer, Northeastern
University
Presenter: Austin Frakt, Associate Professor,
Deptartment of Veterans Affairs & Boston
University
frakt@bu.edu
Research Objective: Low income Veterans are
an important subpopulation affected by the
Affordable Care Act (ACA). Though many have
access to Veterans Health Administration (VA)
care, the ACA’s Medicaid expansion potentially
offers another option, which could affect VA
enrollment and utilization. Our research
objective was to estimate the historical
relationship between state Medicaid expansions
and VA enrollment and utilization. Based on this
estimate, we aim to predict the likely impact of
the ACA’s Medicaid expansion.
Study Design: Our study is retrospective and
observational. First, using historical Medicaid
eligibility rules from the Urban Institute’s TRIM
model and a standardized population from the
Medical Expenditure Panel Survey, we
constructed a policy-driven measure of Medicaid
eligibility—the proportion of a standardized
population eligible for each state’s program in
each year—that is not affected by demographic
or economic variation. Next, we used this
measure as the key independent variable in
region-year level Poisson regressions of VA
enrollment and utilization, controlling for
economic and demographic factors, as well as
year and state fixed effects.
Population Studied: Non-elderly, U.S.
Veterans in years 2002-2008.
Principal Findings: We found that historical
increases in eligibility for Medicaid were
statistically significantly associated with lower
VA enrollment and lower outpatient utilization,
but there was no statistically significant effect on
inpatient utilization. Our preliminary estimates
are that a 10 percentage point increase in the
proportion of the population eligible for Medicaid
is associated with about a 1.1 and 1.4
percentage point decrease in VA enrollment and
outpatient utilization, respectively.
Conclusions: The ACA’s Medicaid expansion
alone has the potential to reduce demand for VA
enrollment and outpatient care. However, there
are three countervailing factors. First, VA
enrollment satisfies the ACA’s individual
mandate, which might encourage greater
enrollment. Second, general awareness about
the coverage requirement and availability of
resources pertaining to coverage could increase
interest in the VA as an option for Veterans.
Navigators and enrollment offices/programs
could screen for Veteran status to help identify
those who are eligible for VA services. Third,
half of U.S. states are not expanding their
Medicaid programs.
Implications for Policy, Delivery, or Practice:
There is a potential for the ACA’s Medicaid
expansion to reduce demand for VA care, which
could improve access to VA care for other
Veterans. However, countervailing forces could
reverse this effect. Policymakers should closely
monitor demand for VA care as the ACA is
implemented.
Funding Source(s): VA
Poster Session and Number: B, #740
Waiting, Here Or There: The Relationship
Between Primary Care Access and
Emergency Department Wait Times
Ari Friedman, University of Pennsylvania; Daniel
Polsky, Leonard Davis Institute, University of
Pennsylvania; Brendan Saloner, Leonard Davis
Institute, University of Pennsylvania; Karin V.
Rhodes, Leonard Davis Institute, University of
Pennsylvania
Presenter: Ari Friedman, MD/PhD Student,
University of Pennsylvania
abfriedman@gmail.com
Research Objective: Wait times for
uncomplicated care in emergency departments
(EDs) have risen dramatically for more than a
decade, a reflection of increased demand for ED
services. This ED crowding has been
repeatedly linked to adverse outcomes,
including mortality. A lack of access to primary
care has been thought to be one of the primary
drivers of ED crowding, but little data has been
available to study the relationship between the
two until now. Because patients without a
longitudinal relationship with a primary care
provider (PCP) may be the most likely to use an
ED in lieu of primary care, we then examine the
association of these ED wait times with directlyassessed new appointment availability at nearby
primary care clinics in two states, one postexpansion and the other pre-reform.
Study Design: Primary care appointment
availability was assessed through a simulated
patient/audit methodology. Appointment
availability was determined and modeled
separately for simulated patients with private
insurance or Medicaid. We used a spatial
bivariate generalized additive model to define
the availability of primary care in areas
surrounding EDs.
We obtained a novel dataset of hospital-level
wait times by use of the Supervised Learning
Outputting Waittimes (SLOW) algorithm, which
we developed and validated. This technique
allowed imputation of each visit's wait times in
the Healthcare Cost and Utilization Project's
State Emergency Department Database (HCUP
SEDD), which provides hospital identifiers and
has sufficient patient visits to generate stable
estimates.
The association between ED wait times and
primary care appointment availability was
assessed using robust linear regression models
adjusted for county income and uninsurance
rate and state fixed effects.
Population Studied: The study utilized data
from a census of 125 EDs (approximately 4
million visits) in 2011, and 4,196 audited primary
care clinics in Massachusetts and New Jersey in
2012/13.
Principal Findings: A 10 percentage-point (one
quartile) increase in primary care appointment
availability for privately-insured patients was
significantly associated with 0.5 - 1.2 minute
lower emergency department wait times,
compared to a historical national annual
increase in ED wait times of 0.75 minutes. The
association for Medicaid appointment availability
was weaker and inconsistent.
Conclusions: Primary care access and ED
crowding are strongly associated for privatelyinsured patients who may have more choice in
their location of care. This crowding-access
association was weaker in Massachusetts-which may reflect the effect of prior insurance
expansion--and non-existent for Medicaid.
Implications for Policy, Delivery, or Practice:
Insurance expansion in Massachusetts
decreased visits to emergency departments
(Miller 2011), whereas for Oregon Medicaid it
increased them (Taubman et al. 2014). This
differential impact may reflect differences in
primary care appointment availability, both
between states and between the general
population and a pre-PPACA Medicaid
population. For insurance expansion to improve
ED crowding, PCP capacity must be available
near crowded EDs. Primary care capacity
should be monitored specifically for those who
will be newly seeking primary care who are near
crowded EDs. Given the heterogeneous effects
observed in this study, policymakers should
exercise caution in extrapolating from the effect
of a single state's Medicaid expansion on ED
crowding to a national, all-income insurance
expansion.
Funding Source(s): RWJF
Poster Session and Number: B, #741
Impact of State Health Insurance Rate
Review on Health Insurance Premiums and
Coverage, 1998-2012
Brent Fulton, University of California, Berkeley;
Pinar Karaca-Mandic, University of Minnesota;
Richard Scheffler, University of California,
Berkeley
Presenter: Brent Fulton, Assistant Adjunct
Professor Of Health Economics And Policy,
University of California, Berkeley
fultonb@berkeley.edu
Research Objective: The research objective is
to estimate the impact of states' health
insurance prior-approval authority and their
medical loss ratio (MLR) requirements on health
insurance premiums and coverage, in the
individual and small group markets from 19982012.
Study Design: The impact of states' health
insurance rate review and MLR regulations is
estimated using difference-in-differences
models. We put together a rich dataset that
provides a full characterization of rate review
authority and MLRs from 1998-2013. This data
was collected from the states' grant applications
to the Center for Consumer Information &
Insurance Oversight to bolster their rate review,
Consumers Union, state legislative librarians,
and a survey we administered to the state
regulators, which also asked about their rate
review activity. The impact of these regulations
on premiums is estimated from changes in
regulatory authority and activity during the study
period. The impact of premium changes on
coverage is estimated using price elasticity of
demand for insurance estimates. Models control
for insurer demographics such as HMO status,
ownership type, business tenure, group
affiliation and overall size and presence in other
markets and other insurance segments
(individual market, small group, large group);
state regulations (e.g. rate-band restrictions,
high risk pools); and state health insurance and
provider market concentration characteristics.
Population Studied: The population includes
enrollees in the individual and small group
markets from 1998-2012. Individual market
health insurance premiums are from the 20022012 Medical Expenditure Panel Surveys-Household Component (MEPS-HC) and five
waves of the Community Tracking Study-Household Survey (CTS-HS) during 1998-2010.
Small group market premiums are from the
MEPS-IC (Insurance Component), available
from MEPS-IC State Summary Tables provided
by AHRQ. Data on insurance premiums are also
from the Supplemental Health Care Exhibit of
the National Association of Insurance
Commissioners for 2010-2012.
Principal Findings: Our preliminary findings
reveal that change in premiums, adjusted for
claims spending, decreased during 2010-2012
across all states, but the decreases were larger
in states with baseline prior approval, compared
with states that had baseline file-and-use or no
rate review regulations. Moreover, among states
with either the baseline prior approval or file and
use regulation, the declines in adjusted
premiums were larger in states with less
stringent or no MLR thresholds, potentially
because of their transition to the higher federal
MLR that began in 2011. The full 1998-2012
period is still under investigation.
Conclusions: State-level health insurance prior
approval authority and more stringent MLR
requirements moderated health insurance
premium increases, in the individual and small
group markets from 2010-2012. This led to an
increase of health insurance coverage.
Implications for Policy, Delivery, or Practice:
The impact of rate review and MLR regulation
on premiums, and ultimately, coverage, holds
significant policy interest. At least 21 states have
recently introduced rate review regulation bills,
such as prior-approval authority (e.g.,
California). The ACA requires the U.S.
Department of Health and Human Services to
work with state insurance departments to
conduct an annual review of rate increases of
10% or more. However, the ACA itself does not
grant states or the federal government new
authority to disapprove rate increases
determined to be unreasonable.
Funding Source(s): RWJF
Poster Session and Number: B, #742
Insuring Kids After Banning Pre-Existing
Condition Exclusions in the ACA
Gilbert Gonzales, University of Minnesota
School of Public Health
Presenter: Gilbert Gonzales, Doctoral Student,
University of Minnesota School of Public Health
gonza440@umn.edu
Research Objective: As of September 23,
2010, the Affordable Care Act (ACA) mandated
that group health insurance plans (i.e. employersponsored insurance) and new plans purchased
on the individual were no longer able to exclude
children 19 years or younger from accessing
coverage because of a pre-existing condition.
While several studies have documented the
ACA’s early impact on health insurance
coverage for young adults, this paper evaluates
the effect of banning pre-existing condition
exclusions for children.
Study Design: This study relied on data from
the 2007-2012 National Health Interview Survey
and uses a difference-in-differences analysis to
estimate the effects of banning pre-existing
condition exclusions for children under the ACA.
The primary outcome of interest was children’s
insurance status (any insurance, private
insurance, dependent coverage). We then
estimated the ACA’s impact on four measures of
access to care: usual source of care (besides a
hospital), delayed care due to cost, forgone care
due to cost, and having a checkup in the past 12
months. All models controlled for each child’s
race, age, sex, health status, citizenship, survey
language, family income, parents’ educational
attainment, parents’ work status, family structure
(single or two-parent), region and time fixed
effects. Each outcome was also estimated
separately for toddlers (2-3), young children (411) and adolescents (12-17).
Population Studied: Children with chronic
conditions (n=15,661) were identified if they
were previously diagnosed or limited by one of
several chronic conditions including attentiondeficit/hyperactivity disorder, mental retardation,
Down syndrome, asthma, cerebral palsy, sickle
cell anemia, muscular dystrophy, autism,
congenital or other heart disease, and diabetes.
The comparison group not affected by the ACA’s
policy change consisted of children without any
of these chronic conditions (n=50,835).
Principal Findings: Dependent private
insurance increased 4% (p<.001) for young
children (4-11 years) with a chronic condition
following the ACA’s ban on pre-existing
condition exclusions. Young children with
chronic conditions were also 2% (p<.05) less
likely to delay care due to cost. Similar results
were not found for toddlers and adolescents.
Conclusions: Banning pre-existing condition
restrictions in the Affordable Care Act (ACA)
helped some young children (4-11 years) gain
health insurance who would have been
previously denied coverage. This study adds to
the growing body of evidence in support of the
ACA’s impact on health insurance coverage and
better access to health care.
Implications for Policy, Delivery, or Practice:
Providing health insurance to children is an
important health policy goal, as coverage
improves childhood health and prevents early
mortality. Children’s health insurance also
increases the likelihood that children receive
well-visits and preventive medicine, which is
particularly important for children with chronic
conditions or special health care needs. Yet,
early findings indicate that not all children with
chronic conditions have benefited from the
ACA’s ban on pre-existing condition restrictions.
Enrollment campaigns should remind potential
enrollees that individuals previously denied
health insurance for a pre-existing condition may
now be eligible, especially now that insurance
companies can no longer deny health plans to
children and adults entering individual or group
insurance markets.
Funding Source(s): No Funding
Poster Session and Number: B, #743
Taking Stock of SHOP: Lessons from
California and Colorado
Leif Wellington Haase, New America Foundation
Presenter: Leif Wellington Haase, Senior
Fellow, New America Foundation
haase@newamerica.net
Research Objective: As states and the federal
government scramble to recover from the rocky
launch of the Affordable Care Act's health
exchanges, most policymakers and journalists
have focused on how many Americans are
enrolling in individual exchanges and whether
they will have access to timely and affordable
care.
The rollout of Small Business Health Options
Program (SHOP) exchanges has drawn far less
attention. Out of the spotlight, 16 states are
forging ahead with the launch of SHOP
exchanges in 2014. These marketplaces aim to
provide affordable high-quality health insurance
for small business owners and their employees.
Study Design: Using structured interviews with
key policymakers, small business owners,
insurers and brokers, this case study focuses on
the opportunities and challenges experienced
during the implementation of SHOP exchanges
in California and Colorado. These states
enacted SHOP exchanges that feature options
designed to make insurance coverage more
appealing to small business. These include
employee choice, premium aggregation, and
streamlined administration of benefits. The
author is conducting these interviews in
collaboration with Small Business Majority, a
national advocacy group for employers.
Knowing why small business owners chose or
chose not to participate in SHOP is critical to
policymakers who are seeking the best ways to
encourage enrollment in the marketplaces. We
asked what small employers think of the quality
of information provided by the exchanges, how
easy or difficult it was for them to navigate
websites and to compare the premiums,
benefits, and networks of plans. We asked
whether the availability of tax credits played a
role in the firms' decision-making. And to the
extent possible we tried to identify the
characteristics of small firms that purchase
through SHOP in the respective states and how
they compare to small employers purchasing in
the non-exchange market. Was price a key
factor? If prices are comparable with plans
outside the exchange do SHOP's distinguishing
features tip the balance in its favor?
Finally, since both in California and in Colorado
insurance brokers and general agents have the
exclusive responsibility for marketing SHOP
plans (though direct enrollment is possible), we
examined the role that brokers play, whether this
role is changing, and how brokers' commitment
to SHOP compares to their interest in promoting
non-exchange small group plans and those
available through private exchanges.
Population Studied: Both California and
Colorado planned extensively for the launch of
SHOP exchanges and consulted extensively in
advance with stakeholders and small business
owners. Both states have adequate insurer
participation as well as networks and premiums
that are competitive with existing small group
coverage. Early take-up and interest in the
SHOP exchanges was comparatively high in
both states (over 500 businesses had applied for
coverage in CA by the end of 2013 and 3683
employer accounts were created in CO)and both
states had solid enrollment in individual
exchange-based coverage.
While these state SHOP marketplaces have
much in common, differences between the two
states can also inform researchers to draw early
conclusions about how SHOP is working in
practice. For instance, employer choice in
California is more limited than in Colorado,
where employers are able to select plans from
multiple tiers rather than a single tier of
coverage chosen by the employer.
Principal Findings: In neither state were small
business owners aware of SHOP in any but the
most general way nor were they well-informed,
on the whole, of its distinctive features. For
those who enrolled in SHOP or considered
enrolling, most small business owners in both
California and Colorado became aware of SHOP
through the advice of their brokers rather than
through the website directly. Once aware, they
found the websites easy to navigate. Few
businesses actually took, or are planning to
take, the tax credit but a substantial number
were drawn to SHOP because of the possibility
of credits and then enrolled in SHOP plans
because of other advantages. Employer/
employee choice and streamlined administration
were the principal attractions. Business owners
in Colorado reported that their employees had
selected plans on different tiers and were
satisfied with their choices. California business
owners reported that their employees wanted
greater choice but didn't believe that expanding
choice would make a difference to their decision
to purchase a SHOP plan in the first place. In
California, but not apparently in Colorado, more
small businesses with higher numbers of
workers chose to enroll directly through the
exchange and bypassed brokers.
Conclusions: More than half of all uninsured
Americans are small business owners,
employees, or their dependents. Proponents of
SHOP exchanges argued that choice, premium
aggregation, tax credits, and other reductions in
administrative burdens for employers would
result in wider insurance coverage for this group
relative to that available on existing small-group
markets. By studying the SHOP exchanges in
two states where the rollout has been relatively
untroubled, we conclude that the distinctive
features of SHOP have prompted some small
businesses to purchase exchange coverage, but
that hurdles principally related to communication
impede take-up by larger number of firms. The
challenges of effective communication may help
explain why the very smallest businesses with
low-wage employees, those for whom the
benefits of SHOP are most advantageous on
paper, appear to have been the least likely to
participate.
Implications for Policy, Delivery, or Practice:
With the full implementation of SHOP delayed in
federally run exchanges, those federal
exchanges and state exchanges which are
getting underway in 2015 will be seeking
models. Many small businesses that took early
renewal options this year will be considering
exchange plans for the first time. For these
groups and for policymakers that are designing
exchange policies, the early experience of fullfeatured versions of SHOP, such as those
operating in California and Colorado, will be
vital. The findings of this study have both
regional and national significance.
Funding Source(s): CWF
Poster Session and Number: B, #744
Community Clinic Readiness for Health Care
Reform
Max Hadler, UCLA Center for Health Policy
Research; Brittany Dixon, UCLA David Geffen
School of Medicine; Nadereh Pourat, UCLA
Center for Health Policy Research
Presenter: Max Hadler, Research Associate,
UCLA Center for Health Policy Research
mhadler@ucla.edu
Research Objective: Community clinics,
particularly federally qualified health centers
(FQHCs), are the cornerstone of safety net
providers. ACA is anticipated to change their
provider mix because the newly insured may
leave community clinics for private providers.
This study assesses the readiness of community
clinics for health care reform on the eve of ACA
implementation.
Study Design: We surveyed Los Angeles (LA)
County community clinics (12/2013-1/2014) on
NCQA patient-centered medical home (PCMH)
recognition, attestation for meaningful use, and
participation in quality improvement
collaboratives. We also asked (1) if they had
applied for PCMH or were planning to do so, (2)
types of electronic patient records or systems in
use, and (3) types and number of quality
improvement initiatives implemented. We
supplemented this data with publicly reported
2011 utilization data to identify public managed
care participation and size of their primary care
workforce. We scored community clinics for
readiness for health reform using these data
from one (low) to five (high).
Population Studied: All community clinics in LA
County offering comprehensive primary care
were included in the study. 204 (out of 271
licensed clinics representing 71 clinic
organizations) had open licenses, were
operating, and offered comprehensive primary
care to the general population. 49% of clinics
responded to the survey.
Principal Findings: Among the clinics studied,
20% received the highest ACA readiness score
of 4-5, 31% received a score of 3, and 45% had
a score of 1-2. Examining the subcomponents of
the score showed that 21% had already been
recognized as PCMH, 34% had applications
pending, and the rest had no plans (21%) or
planned to apply at some future date (24%).
Also, 29% had providers that had attested for
meaningful use, 39% had electronic health
records but no providers had attested, and the
rest had some electronic systems but had no
plans or were going to apply for meaningful use.
91% had participated or conducted quality
improvement initiatives and 17% of clinics had
participated in multiple and overarching such
initiatives. 28% of clinics had 25-45% of patients
with public managed care coverage, another
58% had 1-25% of patients with public managed
care coverage, and the rest did not have any
such patients. About 56% of the clinics had
primary care to patient population ratios below
the median value of 1,373 patients per full time
equivalent primary care provider.
Conclusions: Los Angeles community clinics
have made significant progress in improving
care processes, delivery and infrastructure
required to become providers of choice after the
implementation of ACA. This progress included
gaining PCMH recognition from NCQA and
providers with meaningful use attestation,
participation in major quality improvement
collaboratives, and contracts with Medicaid
managed care organizations.
Implications for Policy, Delivery, or Practice:
Readiness for ACA requires significant effort by
community clinics. PCMH recognition,
meaningful use attestation, and participation in
quality improvement initiatives require a
significant influx of resources. LA County clinics
leveraged federal and foundation grants that
provided consulting and infrastructure support to
prepare for ACA. Community clinics elsewhere
in the U.S. will benefit from such support to
remain viable and survive in the post-ACA
climate.
Funding Source(s): Other California
Community Foundation
Poster Session and Number: B, #745
Decrease in Emergency Department
Utilization for Young Adults under the
Healthcare Reform
Tina Hernandez-Boussard, Stanford School of
Medicine; Carson Burns, Stanford School of
Medicine; N. Ewen Wang, Stanford School of
Medicine; Laurence Baker, Stanford School of
Medicine; Benjamin Goldstein, Stanford School
of Medicine
Presenter: Tina Hernandez-Boussard, Assistant
Professor, Stanford School of Medicine
boussard@stanford.edu
Research Objective: A provision of the
Affordable Care Act (ACA) extended healthcare
coverage eligibility to young adults age 19 to 25
years of age under their parents’ health
insurance plans. Gaining insurance may
change use of the emergency department (ED)
and other healthcare services for affected young
adults. Our research objective was to determine
the relationship between the ACA provision and
ED utilization in the affected population.
Study Design: We conducted an observational
study using all-capture, state administrative
databases. We compared changes in ED
utilization in 19-25 year olds before and after the
ACA provision with changes for 26-31 year olds,
who were not covered by the provision. In our
main analysis we study population-level rates of
ED visits. In a second analysis we used
regression models to estimate the probability of
having a visit, controlling for patient sex,
race/ethnicity, age and state.
Population Studied: We evaluated 10,158,254
million ED visits in California, Florida, and New
York from September 1, 2009 to December 31,
2011 for adults 19 to 31 years of age.
Principal Findings: Following the ACA
provision, 19-25 year olds had a decrease of 2.7
ED visits per 1,000 population compared to 2631 year olds, a relative change of -2.1% (95%
Confidence Intervals [CI], -2.5 to -1.7). The
probability that a 19-25 year old would use the
ED at all also slightly decreased relative to 2631 year olds by -0.36% (CI, -0.72 to -0.01). The
largest decreases were found in females (-3.0%;
CI, -3.5 to -2.4) and Blacks (-3.4%; CI, -4.3 to 2.5). This decrease in ED utilization implies a
total reduction of more than 60,000 visits across
the three states.
Conclusions: Expansion of insurance eligibility
under the ACA provision was associated with
decreased ED utilization in young adults.
Implications for Policy, Delivery, or Practice:
As insurance coverage is expanded under the
health care reform, changes in health care
utilization may be expected.
Funding Source(s): AHRQ
Poster Session and Number: B, #746
US Veterans’ Enrollment in Medicaid:
Implications for the Expanded Medicaid
Eligibility Under the ACA
Denise Hynes, VA Information Resource Center,
Edward Hines, Jr. VA Hospital; Kristin de Groot,
VA Information Resource Center, Edward Hines,
Jr. VA Hospital; Melissa Joyce, VA Information
Resource Center, Edward Hines, Jr. VA
Hospital; Thomas Weichle, VA Information
Resource Center, Edward Hines, Jr. VA
Hospital; Linda Kok, VA Information Resource
Center, Edward Hines, Jr. VA Hospital; Donghui
Kan, VA Information Resource Center, Edward
Hines, Jr. VA Hospital
Presenter: Denise Hynes, Scientist/professor,
VA Information Resource Center and University
of Illinois at Chicago
dhynes@uic.edu
Research Objective: To describe and examine
US veterans’ enrollment in Medicaid and
establish a baseline for predictors of enrollment
prior to the implementation of the Affordable
Care Act.
Study Design: In a retrospective study design,
we focused initially on those veterans enrolled in
the US Department of Veterans Affairs Health
Administration (VHA) as of 2008. We linked
VHA enrollment files to the Medicaid Person
Summary files. We examined demographics,
including gender, age, race, and state of
residence; VHA priority level; reasons for
Medicaid eligibility; and concurrent Medicare
enrollment and as predictors of Medicaid
enrollment.
Population Studied: Initial analysis included all
Veterans enrolled in VHA during September
2008. Detailed analysis included only Veterans
who were dually enrolled in VHA and Medicaid
during the month.
Principal Findings: In September 2008, there
were 7.4 million Veterans enrolled in VHA. Of
these, 5.3% were also enrolled in Medicaid.
Seventy-two (72%) of the VHA-Medicaid dually
enrolled Veterans were also enrolled in
Medicare. The states with the highest percent of
Veterans enrolled in Medicaid were Maine
(14.6%), District of Columbia (10.6%), Vermont
(10.6%), and Massachusetts (10.3%). The
states with the lowest percent of Veterans
enrolled in Medicaid were Montana (3.0%),
Virginia (3.0%), and Utah (3.1%). Approximately
50% of VHA-Medicaid dual enrollees were
eligible for Medicaid due to old age; 36% were
eligible to due disability; 14% were eligible for
other reasons. Almost all (99%) Veterans who
were eligible for Medicaid due to old age and
60.3% of Veterans eligible for Medicaid due to
disability were also enrolled in Medicare.
Patterns of Medicaid enrollment varied greatly
by age and gender. Female enrollment in
Medicaid peaked in the youngest age group (1824; 11.2%) and declined with age to only 5.3%
for women ages 45-64, then increased again
after age 65 to 10.4%. In contrast, males had
very low levels of Medicaid enrollment (less than
3%) until after age 45. Between ages 45 and
64, 5.0% of Veterans were VHA-Medicaid dually
enrolled and it increased slightly after age 65 to
6.2%.
Conclusions: Approximately 5% of Veterans
enrolled in the Veterans’ Health Administration
(VHA) are also enrolled in Medicaid. Women,
the disabled, and the elderly were more likely to
be enrolled in Medicaid. This research also
provides a framework for considering factors
that may affect Medicaid enrollment for the
larger veteran population and deserves attention
as the ACA is rolled out.
Implications for Policy, Delivery, or Practice:
Prior research has found that over a million nonelderly veterans and their families lacked health
insurance coverage. While VHA coverage is
available to many veterans, priority and access
are based on service-connected disabilities,
income level, and other factors. As our research
shows, even veterans enrolled in VHA may also
seek Medicaid enrollment. Those dually enrolled
in VHA and Medicaid often have situations or
conditions that require specialized care. The
impact of the expanded Medicaid eligibility under
the Affordable Care Act on veterans’ choice of
health insurance coverage deserves further
attention.
Funding Source(s): VA
Poster Session and Number: B, #747
Variation in the Tobacco Surcharge and Plan
Affordability for Tobacco-Users under the
Affordable Care Act
Cameron Kaplan, University of Tennessee
Health Science Center; Ilana Graetz, University
of Tennessee Health Science Center; Teresa
Waters, University of Tennessee Health Science
Center
Presenter: Cameron Kaplan, Assistant
Professor, University of Tennessee Health
Science Center
ckaplan@uthsc.edu
Research Objective: New federal guidelines for
individual health plans sold on the health
insurance exchanges established under the
Affordable Care Act (ACA) allow insurers to
charge tobacco users up to 50% more for
insurance premiums. States may set more
restrictive limits, and insurers are free to set
tobacco surcharges at any level up to those
limits and have differential surcharges by age.
This current policy environment makes it likely
that a range of tobacco surcharges will be
implemented across the U.S. Differences in
tobacco surcharges could lead to sorting of
tobacco users and non-users into different
plans, raising concerns for market instability.
This study examined the variation in tobacco
surcharges and plan affordability for tobaccousers across states.
Study Design: Using data we collected from a
variety of internet-based resources, we
examined insurance premiums from individual
health insurance exchanges to describe
variation in tobacco surcharges by state and
across insurance plans within states. We then
calculated the premiums that would be paid after
federal subsidies for both tobacco users and
non-tobacco users in order to examine who
lacked access to affordable coverage.
Affordable coverage is defined by the ACA as
having at least one plan available with premiums
less than 8% of income, and those without
affordable coverage are exempted from the
mandate.
Population Studied: We studied examined
insurance premiums for individuals in the health
insurance exchanges in 36 states. These
premiums primarily impact those who lacked
insurance prior to the implementation of
healthcare reform.
Principal Findings: The median plan in the
health insurance exchanges charged only 10%
higher premiums to tobacco users, and nine in
ten plans had a lower surcharge than was
allowed at the state level. Even with the lower
than allowed surcharges, tobacco users lacked
affordable coverage in more states than did nontobacco users. For example, a 45-year old
smoker with an income of $35,000 lacked
affordable coverage in one-third of the states in
our sample, while a non-smoker of the same
age and income lacked affordable coverage in
only two of the thirty-six states in our sample.
Conclusions: Our results suggest that the
variation in tobacco surcharges may result in
sorting of tobacco users and non-users into
different plans, and could be influential in
tobacco users’ decisions to opt out of coverage
altogether.
Implications for Policy, Delivery, or Practice:
Each state may set it’s own limit for the tobacco
surcharge, or decide to disallow tobacco
surcharges altogether. Our study suggests that
most insurers do not charge tobacco users the
maximum surcharge allowed in the state, but
even small surcharges were enough to create
differences in affordability of plans between
tobacco users and non-users. This research will
be important for state and federal policy makers
in their decision of what tobacco surcharge to
allow in the future.
Funding Source(s): No Funding
Poster Session and Number: B, #748
The Scope and Distribution of Diagnostic
and Therapeutic Imaging Services at Critical
Access Hospitals in the U.S.
Amir Khaliq, University of Oklahoma Health
Sciences Center; Eugene Nsiah, Harvey L.
Neiman Health Policy Institute; Nadia Bilal,
Harvey L. Neiman Health Policy Institute; Danny
Hughes, Harvey L. Neiman Health Policy
Institute; Richard Duszak, Emory University
School of Medicine
Presenter: Amir Khaliq, Associate Professor,
University of Oklahoma Health Sciences Center
amir-khaliq@ouhsc.edu
Research Objective: The purpose of this study
was to understand the availability, geographic
distribution and scope of imaging services at
Critical Access Hospitals (CAHs) throughout the
United States.
Study Design: We merged the American
Hospital Association Annual Survey data for
6,317 hospitals for the year 2011 with the U.S.
2010 census data. Imaging services survey data
included mammography, ultrasound, computed
tomography (CT), magnetic resonance imaging
(MRI), single photon emission computed
tomography (SPECT) and combined positron
emission tomography (PET) CT. Availability and
characteristics of imaging services at the 1,060
CAHs in 45 states for which sufficient data were
available were studied. The results of descriptive
statistical analysis are reported.
Population Studied: 1060 Critical Access
Hospitals (CAHs) in 45 states
Principal Findings: The most widely available
of all imaging services, mammography,
ultrasound and some form of Computerized
Tomography (CT)are available in all CAHs in
only 13%, 33%, and 56% of the 45 states in
which designated Critical Access Hospitals exist.
In none of the 45 states 64+ slice CT, MRI,
SPECT, and PET/CT services were available at
any of the CAHs.
Conclusions: An overall scarcity of access to
imaging services exists at CAHs throughout the
United States.
Implications for Policy, Delivery, or Practice:
With 19.3% of the population (more than 60
million people)residing in rural areas and being
almost entirely dependent on CAHs for health
services, the policy implications for imaging
access could be profound.
Funding Source(s): Other Harvey Neiman
Health Policy Institute
Poster Session and Number: B, #749
Evaluating the Breast and Cervical Cancer
Early Detection Program (BCCP) in the
context of Medicaid
Siran Koroukian, Case Western Reserve
University; Paul Bakaki, Case Western Reserve
University; Mark Schluchter, Case Western
Reserve University; Cynthia Owusu, Case
Western Reserve University, University
Hospitals of Cleveland; Gregory Cooper, Case
Western Reserve University, University
Hospitals of Cleveland; Susan Flocke, Case
Western Reserve University
Presenter: Siran Koroukian, Associate
Professor, Case Western Reserve University
skoroukian@case.edu
Research Objective: The BCCP is a screening
program for uninsured, low-income women.
Once diagnosed with cancer, BCCP women are
eligible to receive treatment through Medicaid.
Non-BCCP, low-income women diagnosed with
cancer might enroll in Medicaid upon being
diagnosed with cancer. These women, whom
we refer to as Safety Net Medicaid Beneficiaries
(SNMBs), experience worse cancer outcomes
than women insured through Medicaid prior to
cancer diagnosis, as shown previously. The
objective of this study is to compare cancerrelated outcomes between BCCP women and
SNMBs. We hypothesize that BCCP women
would be more likely than SNMBs to experience
favorable cancer-related outcomes.
Study Design: We used linked data from the
2002-2008 Ohio Cancer Incidence Surveillance
System, Medicaid, the BCCP database, and
Ohio death certificates (through 2010). We
examined the following outcomes: a) regional- or
distant-stage (advanced-stage) cancer at
diagnosis; b) treatment delays after being
diagnosed with cancer; c) receipt of standard
treatment in patients with local-or regional-stage
cancer; and d) overall and cancer-specific
survival. In addition to descriptive analysis, we
conducted multivariable logistic regression and
time to event analysis to examine the
association between BCCP and the outcomes of
interest, controlling for age, race, marital status,
comorbidities, socioeconomic status (SES) at
the census block group level, residence in
Medically Underserved Areas (MUAs), and
county of residence. Survival models also
adjusted for cancer stage.
Population Studied: Women 40-64 years of
age and diagnosed with incident breast cancer
during the years 2002-2008 and enrolled in
Medicaid as BCCP or SNMBs upon being
diagnosed with cancer, or within a 3-month
window before or after cancer diagnosis.
Principal Findings: We identified 433 women
and 686 in the BCCP and SNMB groups,
respectively. The two groups were similar in
their distribution by SES and residence in MUAs.
The proportion of BCCP and SNMB women
diagnosed with advanced-stage cancer was
48.7% and 63.4% respectively (p < 0.001).
Standard treatment was received by 62.5% of
BCCP women and 59.0% of SNMBs (p=0.29).
Adjusting for confounders, BCCP women were
47% less likely than SNMBs to be diagnosed
with advanced-stage cancer (adjusted odds ratio
(AOR): 0.53 (95% confidence Interval 0.42,
0.69)). In addition, while time to treatment
initiation was shorter among BCCP women
(Adjusted Hazard Ratio (AHR): 1.27 (1.10,
1.46)), BCCP women were equally likely as
SNMBs to receive standard treatment (AOR:
1.20 (0.91, 1.59)). With respect to survival, we
observed both overall and disease specific
survival advantage in BCCP women (AHR: 0.50
(0.37, 0.68) and 0.61 (0.43, 0.85), respectively).
Additional analysis examining outcomes in firsttime and repeat-BCCP users indicated
accentuated effects in the latter group.
Conclusions: Compared to SNMBs, BCCP
women experienced breast cancer stage,
shorter time to treatment, and survival benefits.
However, we observed no differences between
the two groups in receipt of standard treatment.
Implications for Policy, Delivery, or Practice:
Despite the similarities between the two groups
with regard to their SES, BCCP women
experienced more favorable cancer-related
outcomes. In addition to the patient navigator
program, this may be attributed to differences in
knowledge and attitude with regard to screening
benefits, for which we could not account in this
study.
Funding Source(s): Other American Cancer
Society
Poster Session and Number: B, #750
Utilization of Extended Hours in Safety Net
Clinics in Kansas City through the Kansas
City Safety Net Expansion Project
Danielle Liffmann, NORC at the University of
Chicago; Natacha Clavell, NORC at the
University of Chicago; Cameron Johnstone,
NORC at the University of Chicago; Jane
Mosley, Health Care Foundation of Greater
Kansas City; Graciela Couchonnal, Health Care
Foundation of Greater Kansas City; Adil
Moiduddin, NORC at the University of Chicago
Presenter: Danielle Liffmann, Senior Research
Analyst, NORC at the University of Chicago
liffmann-danielle@norc.org
Research Objective: In the greater Kansas City
area, 1/5 of the population is uninsured,
underinsured, or covered by Medicaid.
Employment makes it difficult to see health care
providers during traditional clinic hours. From
October 2009 through September 2013, the
Health Care Foundation of Kansas City
distributed grants to five safety net clinics in
metropolitan Kansas City to add additional hours
on top of their existing schedules during nights
and weekends. We examine the volume and
types of visits typically utilized during the
extended hours to determine who benefited from
the program.
Study Design: We collected self-reported data
from the clinics on the demographic, clinical, and
geographic characteristics of the patients
utilizing the extended hours. In addition, we
collected the most frequent diagnoses during
visits. In years three and four, we also collected
the top CPT office visit procedure codes, as a
proxy for complexity to determine whether
patients were new to the panel. Data were
aggregated across clinics and over time to
characterize patients utilizing extended hours
across the clinics. ZIP codes were mapped to
demonstrate geographic distribution of patients
utilizing after hours and overlap between clinics.
Population Studied: Utilizers of safety net clinic
extended hours from October 2009 through
September 2012.
Principal Findings: 20,757 patients used the
clinics during extended hours over three years.
The majority (80%) of patients treated were
uninsured. In years two through four (data were
not collected in year one), most patients treated
ranged between ages 20 and 55 (76%) and had
incomes below the federal poverty line (67%).
The majority of patients using extended hours
clinics lived in the urban core of Kansas City and
multiple clinics saw patients from the same ZIP
codes. Although the diagnoses treated varied
year to year, essential hypertension, diabetes
mellitus and special investigations and
examinations (including routine gynecological
visits) were in the top 10 diagnoses treated. In
years three and four (data were not collected in
years one and two), visits were a mix of new and
established patients. In year 3, at least 29% of
visit codes indicated that the patient required 25
minutes or more with a physician, which
indicates a moderately high level of complexity.
In year 4, 95% of visit codes indicated at this
level of complexity, although this increase may
be due to two clinics no longer reporting on this
measure.
Conclusions: Extended hours were used
primarily by the uninsured and low-income
population. The moderate severity of visits
suggests that individuals were appropriately
using the clinic in lieu of alternatives such as the
emergency room. In addition, the high overlap of
patients suggests that patients might seek care
at multiple clinics, rather than seeking a regular
source of care, and warrants further research on
whether the same patients were visits multiple
clinics.
Implications for Policy, Delivery, or Practice:
Safety net clinics should consider providing care
during non-traditional hours. Further research is
warranted to determine if additional access to
safety net clinics can act as a lower cost
alternative to emergency room care.
Funding Source(s): Other Health Care
Foundation of Greater Kansas City
Poster Session and Number: B, #752
Impact of the Affordable Care Act for
Members with “Canceled” Individual Health
Plans in Kaiser Permanente of the MidAtlantic States
Jersey Chen, Kaiser Permanente, Mid-Atlantic
Permanente Medical Group, Mid-Atlantic
Permanente Research Institute; Jared Lane
Maeda, Kaiser Permanente, Mid-Atlantic
Permanente Medical Group, Mid-Atlantic
Permanente Research Institute
Presenter: Jared Lane Maeda, Research
Scientist, Kaiser Permanente
jared.l.maeda@kp.org
Research Objective: The cancellation of
individual health plans because of noncompliance with provisions of the Affordable
Care Act (ACA) is a contentious issue. Although
health plans already in existence at the time the
ACA was enacted are grandfathered, plans
offered after ACA enactment must meet new
requirements for specific essential health
benefits. Because of controversy surrounding
this requirement there has been a proposed 1year extension for ACA non-compliant plans, but
health plans need to be compliant after 2014.
Despite the possible cancellation of noncompliant health plans, little is known regarding
the proportion of plans that are not compliant,
reasons for non-compliance, impact on patient
costs, and how costs vary according to patient
characteristics. In this study, we examined the
characteristics of Kaiser Permanente of the MidAtlantic States (KPMAS) members from the
individual market whose health plans were
cancelled and modified due to the ACA. We also
examined members’ health care utilization and
change in health plan costs.
Study Design: We identified members from the
KPMAS individual market and accessed their
electronic medical records to derive members’
socio-demographics, premiums, deductibles,
and utilization. We then examined the change in
members’ premiums and deductibles from 2013
to the most equivalent ACA-compliant health
plan in 2014.
Population Studied: All KPMAS members from
the individual market in 2013, classified into a
grandfathered or non-grandfathered, ACA noncompliant health plan.
Principal Findings: A total of 7,247 members
from the individual market were enrolled in an
ACA non-compliant plan, which represented
about 1.5% of overall KPMAS membership in
2013. An additional 5,351 members were
enrolled in a grandfathered plan. Members
enrolled in a non-compliant plan were older
compared to those enrolled in a grandfathered
plan (39.2 years vs. 34.2 years, p<.001) and
were less likely to be female (46.8% vs. 51.1%,
p<.001). About 5.5% of members enrolled in a
non-compliant plan had 3 or more comorbidities
compared to 4.1% of members enrolled in a
grandfathered plan (p=.005). In addition,
members enrolled in a non-compliant plan had
9% fewer outpatient visits (14.6 vs. 16.0 per 100
members) and 10% more hospitalizations (481.5
vs. 457.6 per 100 members) than members in a
grandfathered plan. Additional analyses
currently underway will be presented on the
change in health plan cost. The major reason for
non-compliance with the ACA included the lack
of pediatric dental and pediatric vision benefits
and the elimination of some high deductible
plans, with some changes to specific drug
coverage.
Conclusions: Preliminary analyses reveal that
only a small proportion of KPMAS members
were affected by the ACA plan cancellations.
Reasons for ACA non-compliance were few as
KPMAS individual insurance plans already
provided comprehensive benefits. Members
enrolled in a non-compliant plan were older and
had more comorbidities than members in a
grandfathered plan, which might suggest
adverse selection.
Implications for Policy, Delivery, or Practice:
Our results may help to inform policymakers
regarding how the ACA impacted the changes in
health plan benefits and affordability and which
segment of the population was affected the
most.
Funding Source(s): Other Kaiser Permanente
Community Benefits Program
Poster Session and Number: B, #753
Adapting the CAHPS® Child Medicaid Survey
for use in the Health Insurance Marketplaces:
Methodological challenges and opportunities
Coretta Mallery, American Institutes for
Research; Brandy Farrar, American Institutes for
Research; Daniel Harwell, American Institutes
for Research; HarmoniJoie Noel, American
Institutes for Research; Steven Garfinkel,
American Institutes for Research
Presenter: Coretta Mallery, Senior Research
Scientist, American Institutes for Research
cmallery@air.org
Research Objective: Overall uninsurance rates
among children are low and the highest rates of
uninsurance are households that do not qualify
for Medicaid or CHIP. The Health Insurance
Marketplaces (Marketplaces) have the potential
to close this gap. The purpose of this study is to
identify the key issues consumers will consider
when deciding whether to enroll their children in
health insurance through the Marketplaces. The
goal of the overarching project is to design a
survey, based on the Consumer Assessment of
Healthcare Providers and Systems (CAHPS®),
which will assess parent/ guardian experiences
with their child’s QHPs. Designing a survey
specifically for the child population is challenging
given the unique coverage situations and
specialized health issues that children face.
Study Design: Our team conducted nine oneon-one interviews, a literature review, and
conducted a panel meeting of pediatric
stakeholders in August-October 2013 to (1)
identify key issues for parents/ guardians
seeking health plans for their children; (2) to
detect potential gaps in our understanding of this
health plan market; and (3) to better understand
parents/ guardians’ perspectives about the
Marketplaces.
Population Studied: The interviews were
conducted prior to open enrollment for
Marketplaces. Thus, we recruited consumers
with characteristics similar to parents/ guardians
who would potentially be eligible to enroll their
children in health insurance through the
Marketplace. This included parents/ guardians of
children who were on Medicaid, enrolled in a
child-only health plan, or uninsured.
Principal Findings: • Children face a number
of complex coverage situations due to factors
including changes to their parents/ guardians’
eligibility for health insurance through their
employer, changes in their parents/ guardians’
income which affects their eligibility for Medicaid
or CHIP, or changes in their parents/ guardians’
employers health insurance offerings. Churning
will thus affect the sample for further testing and
implementation.
• Parents/ guardians with uninsured children
wanted coverage and were uncomfortable
without it, but could not afford it.
• Generally, our respondents had very little
experience shopping for and seeking information
on plans.
• Respondents generally were not aware of the
Health Insurance Marketplaces and those who
were aware only had minimal amounts of
information about them. Despite that, after
hearing a brief description most respondents
indicated that they would be interested in using
the Marketplaces to at least research health
plans that may be available.
• Concerns included whether the coverage
offered would be affordable, ease of enrollment,
and privacy of information.
• Respondents said they would need information
on costs and provider networks in order to select
a plan and they wanted a combination of online,
phone, and in-person assistance available.
Conclusions: Based on the formative research,
the decision was made to proceed with the
CAHPS Child Medicaid Survey as the basis for
developing the Child QHP Survey with the
addition of domains unique to the Marketplace
population.
Implications for Policy, Delivery, or Practice:
The Marketplaces will allow access to health
insurance for millions of Americans including
children, many for the first time. Conducting this
research allows us to better understand the
needs of parents/ guardians shopping for health
insurance for their children and the issues that
are most important to them when searching for a
health plan.
Funding Source(s): CMS
Poster Session and Number: B, #754
Can Technology Utilization Increase
Enrollment and Retention in Public
Insurance Programs?
James Marton, Georgia State University; Angela
Snyder, Georgia Health Policy Center; Susan
McLaren, Georgia Health Policy Center
Presenter: James Marton, Associate Professor,
Georgia State University
marton@gsu.edu
Research Objective: CHIPRA 2009 and the
ACA have provided state Medicaid programs
with significant federal dollars to help support
eligibility system upgrades and improvements
that streamline and simplify the eligibility
process for families applying for public health
care coverage (Medicaid and CHIP). The
federal government has given states limited
access to query federal agency data like the
Social Security Administration (SSA) and the
Department of Labor (DOL) to assist with data
verification in the application process. Our
objective is to evaluate the impact of the use of
technology to increase enrollment and retention
in public insurance programs by using federal
data matches to streamline the documentation
requirements for citizenship, identity and income
in one southern state.
Study Design: We used state administrative
public insurance data to retrospectively evaluate
the recent procedure change to match and
permanently store the results of applicant’s
citizenship and identity data with the SSA, using
a “pre vs. post” design. To prospectively
investigate income verification using DOL
income data, we manually matched income
reported to Medicaid with DOL income records
over a six quarter study period (July 2010 –
December 2011) for all enrollees with a valid
social security number, aged 18 and above,
enrolled in family Medicaid for at least one
month during that time frame.
Population Studied: The population studied
consists of recent Medicaid recipients in one
southern state.
Principal Findings: Implementation of matching
and permanently storing the results of
applicant’s citizenship and identity data with
SSA resulted in mixed effects. Application
processing days decreased, while the number of
cases denied due to “failure to verify” slightly
increased.
With respect to prospectively analyzing income
matching, we found that only 33 percent of the
family Medicaid cases matched to DOL. Of the
third that did match 92 percent have a higher
income reported in DOL with the remaining 8
percent having a higher income reported within
Medicaid. Of the 92 percent with a higher DOL
income, half had zero reported Medicaid
income. Their average quarterly wages in DOL
were about $2,000 per quarter. More than half
of the remaining cases with income reported by
both sources differed by more than 100 percent.
Conclusions: This suggests that the applicant’s
citizenship and identity verification are not the
primary cause for eligibility to be denied under
failure to verify. The more likely reason may be
due to the state’s inability to verify income
through electronic matches with DOL. However,
using DOL data to verify income may not be
very effective for enrollees in family Medicaid,
who have incomes below the ACA Medicaid
expansion population.
Implications for Policy, Delivery, or Practice:
States that choose not to expand their Medicaid
programs in response to the ACA may not see
as much benefit from electronic income
verification as states that do expand.
Funding Source(s): CMS
Poster Session and Number: B, #755
Insurance Status of Hispanic Adults in
Massachusetts More than Seven Years PostHealth Reform
Karen Schneider, JSI Research and Training
Institute, Inc.; James Maxwell, JSI Research and
Training Institute, Inc.; Dharma Cortés,
Northeastern University; Rodolfo Vega, JSI
Research and Training Institute, Inc.; Christine
Barron, JSI Research and Training Institute,
Inc.; Catherine West, Center for Health
Information and Analysis
Presenter: James Maxwell, Director of Health
Policy and Management Research, JSI
Research and Training Institute, Inc.
jmaxwell@jsi.com
Research Objective: Research shows that
Hispanics are significantly more likely than other
populations to remain uninsured and to
experience barriers to care in Massachusetts
post-health reform. Understanding coverage and
access among the Hispanic population is
important, as Hispanics are the largest and
fastest growing ethnic minority group in
Massachusetts. The aim of this research is to
describe the remaining uninsured and the
experience of currently insured Hispanics in
Massachusetts.
Study Design: Cross-sectional survey with a
convenience sample of Hispanics recruited from
different community-based venues across
Massachusetts between October 2013 and
December 2014.
Population Studied: Surveys completed by 411
non-elderly adult (18 to 64) Hispanic residents of
Massachusetts.
Principal Findings: Overall, one-third of
respondents were male, 39 percent were 18 to
34 years of age, and 84 percent completed the
survey in Spanish. Nine percent reported never
having insurance as an adult in the U.S. and an
additional 10 percent were currently uninsured.
Among the currently uninsured, a slightly higher
percent were 18 to 34 years (43 percent) and
male (55 percent) than the overall sample; 47
percent never had insurance as an adult. Of
those currently insured, the majority reported
having coverage for more than 2 years (81
percent), while 10 percent had it for 12 months
or less and 8 percent had it for 13 to 24 months.
Nearly half reported Medicaid coverage, 13
percent reported insurance through an employer
or directly purchased from a carrier, and 6
percent reported Commonwealth Care (which
ended in January 2014 when the Affordable
Care Act took effect). Seven percent responded
with names of commercial plans that may be
Commonwealth Care plans or acquired through
other means. Other respondents reported
temporary and/or less comprehensive coverage:
6% reported using the Health Safety Net, which
in itself is not insurance, but covers the cost of
medically necessary services at community
health centers and hospitals; 1% reported
MassHealth Limited, emergency medical
coverage for non-citizens and undocumented;
and 1% reported international health plans from
their home countries.
Conclusions: Hispanics continue to face issues
with gaining and maintaining health coverage
more than seven years post-health reform.
Nearly one in five Hispanics surveyed never had
insurance or were currently uninsured, which is
higher than estimates from other surveys. This is
due to the venues from which participants were
recruited (the goal was to capture the remaining
uninsured). While a high percent of insured
maintained their coverage for more than two
years, one in five had a change in plan in the
past two years. In addition, the types and
comprehensiveness of coverage for the insured
greatly varied. At least one in ten had plans that
provided less comprehensive coverage.
Implications for Policy, Delivery, or Practice:
In 2012, 17% of the US population was Hispanic
and 29% were uninsured. Therefore, there is a
high need for assistance and information in
Spanish. By collecting previously unavailable
data on coverage and access issues among
uninsured and underinsured Hispanics, this
research has direct implications for improving
outreach and enrollment strategies so that
Hispanics are enrolled or reenrolled in health
insurance in a seamless fashion as the ACA
continues to roll out.
Funding Source(s): Other Center for Health
Information and Analysis
Poster Session and Number: B, #756
What Explains Variation in Premiums under
Marketplace Plans in the ACA?
Timothy McBride, Washington University; Keith
Mueller, University of Iowa; Abigail Barker,
Washington University; Leah Kemper,
Washington University
Presenter: Timothy McBride, Professor,
Washington University in St. Louis
tmcbride@wustl.edu
Research Objective: The objective of this work
is to explore, describe and seek to explain the
variation in premiums for private marketplace
plans offered under the Affordable Care Act
(ACA)
Study Design: Using a database linking data on
the marketplace plans to information on their
premiums, to information on the areas in which
they are located this project looks first at
descriptive data and then multivariate analysis to
explore factors that might explain the variation in
premiums and plan characteristics across
marketplace plans.
Population Studied: The research uses planlevel data (since data on individuals in the plans
has not been released as of this time). However
the data is linked to aggregated data on the
characteristics of individuals in the rating areas - including their basic socioeconomic
characteristics, health spending, health use, and
health status.
Principal Findings: As has been noted in the
popular press, and in other publications, there is
sometimes significant variation in premiums
across geographic areas, within and between
states. Much speculation has attributed this to
various factors. However, this analysis controls
for cost of living differences and other cost
variations, and much of the obvious variation is
eliminated. After this control, we find no
systematic explanation for any remaining
variation that can be explained with the available
data.
Conclusions: Variation in premiums may be
explained by differences in costs of living and
costs across regions of the country, and not
other factors, such as urban/rural status, and
health status.
Implications for Policy, Delivery, or Practice:
The findings suggest that at least in the initial
offering of plans in the marketplaces there is not
significant bias in premiums based on
characteristics that would concern policymakers
(e.g., health status, poverty). While this may not
sustain itself over time, it suggests that that
some of the policies put in place during ACA
implementation may have worked to protect
against geographic variation.
Funding Source(s): HRSA
Poster Session and Number: B, #757
Understanding the Demand for Health Care
Services among the Newly Insured: Lessons
Learned from Expanding Health Coverage to
Low Income Workers in Arkansas
Michael Motley, Arkansas Center for Health
Improvement; Heather Rouse, Arkansas Center
for Health Improvement; Rhonda Hill, Arkansas
Center for Health Improvement; Joseph
Thompson, Arkansas Center for Health
Improvement
Presenter: Michael Motley, Prevention
Specialist, Arkansas Center for Health
Improvement
mwmotley@uams.edu
Research Objective: Health care demand by
those newly insured under the Affordable Care
Act (ACA) is uncertain, particularly for Medicaid
expansions. For example, some estimates from
other states that have previously expanded
coverage indicate pent up demand for services
during the first few months of enrollment,
whereas others do not. A 1115 Medicaid waiver
in Arkansas established an employer based
limited benefit program in 2005 for previously
uninsured, low income residents, called
ARHealthNetworks (ARHN). The purpose of this
study was to examine service utilization and cost
during enrollees first year to inform projections
for ACA expansions.
Study Design: ARHN medical and pharmacy
claims from 2007 through 2011 were examined.
Enrollment patterns and annual use and costs
were compared among income groups (low,
under 138 percent of the federal poverty level
(FPL); middle, 139 to 200 percent FPL; high,
above 200 percent FPL). Total cost was defined
as amount paid by the plan. Members with no
claims were assigned zero cost.
Population Studied: The study population
consisted of 14,291 individuals, 19 to 64 years
old, who were the primary contract holders and
were enrolled for at least 12 contiguous months
during the study period.
Principal Findings: The study population was
representative of the overall ARHN population
(average age of 41 years and 59 percent
female). The average length of enrollment was
23.8 months. Most (53 percent) were low
income, with less than 8 percent in the high
income group.
Income and age were both significantly related
to total annual costs (p less than .0001). The
average annual cost per enrollee for the low
income group (2,312 dollars) was significantly
greater (p less than .01) than for the middle
income group (2,002 dollars), whereas average
annual cost of the high income group (2,383
dollars) was not significantly different.
In their first year of enrollment, 80 percent of the
population used at least one benefit. Low
income enrollees were significantly less likely to
use at least one benefit compared with high
income enrollees. On average among all income
groups, the benefit use was lowest in the first
month (n=4,060, 28 percent) and increased
gradually across the first year of coverage.
Conclusions: The average annual total cost for
low income working individuals with new
coverage was not significantly different than
their high income counterparts but their health
care use was significantly less. There was no
evidence of pent up demand in the first few
months of enrollment. Overall, the number of
individuals using any benefit increased gradually
throughout the first year of enrollment.
Implications for Policy, Delivery, or Practice:
Continued investigation of cost and use for low
income, newly insured individuals will support
resource planning for potential Medicaid
expansions. Efforts to expand coverage for low
income populations should anticipate non
coverage related barriers to care (e.g.,
transportation, opportunity costs of missed work)
and maintenance efforts to retain coverage once
established.
Funding Source(s): Other Arkansas
Department of Medicaid
Poster Session and Number: B, #758
Comparing Cost and Utilization Among
Medicaid Enrollees Receiving Primary Care
at Federally-Funded Health Centers Relative
to Other Settings
Robert Nocon, University of Chicago; , ; Ravi
Sharma, Health Resources and Services
Administration; Quyen Ngo-Metzger, Agency for
Healthcare Research and Quality; Dana
Mukamel, University of California Irvine; Leiyu
Shi, Johns Hopkins Bloomberg School of Public
Health; Laura White, University of California
Irvine; Marshall Chin, University of Chicago;
Elbert Huang, University of Chicago
Presenter: Robert Nocon, Senior Health
Services Researcher, University of Chicago
rnocon@uchicago.edu
Research Objective: The Affordable Care Act
(ACA) calls for significant expansions in
Medicaid and increased funding for federallyfunded health centers (HCs), which will likely
lead to increased use of HCs. Given the
national focus on restraining rapidly growing
costs of care, it is critical to examine the
association between care in HCs and utilization
and cost. Existing studies of this topic are
limited by analyses that cover a small number of
states or use data that may not reflect current
practice patterns. This study compares
utilization and costs between HC and non-HC
Medicaid enrollees using data from 13 diverse
states and the most recent data available prior
to the onset of ACA insurance coverage
expansions.
Study Design: A cross sectional comparison
using data from the 2009 Medicaid Analytic
eXtract files. We categorized patients as either
HC or non-HC based on whether more than half
of their primary care visits occurred in a HC. We
compared HC and non-HC patients along 11
outcomes: primary care visits, primary care cost,
non-primary care outpatient visits, non-primary
care outpatient cost, prescription drug cost,
emergency department visits, emergency
department cost, inpatient admissions, days of
inpatient stay, inpatient cost, and total cost of
care. Our analyses adjusted for patient
demographics, Medicaid coverage
characteristics, disease burden (using the
Chronic Illness & Disability Payment System),
and state. We used generalized linear models
with log link assuming a gamma distribution for
cost outcomes and negative binomial distribution
for utilization. We used generalized estimating
equations with compound symmetry working
correlation to account for the clustering of
patients within primary care service areas. We
conducted sensitivity analyses that tested
different thresholds for assigning patients to HC
versus non-HC groups.
Population Studied: Two million adult fee-forservice Medicaid enrollees who had primary
care utilization in 2009 in 13 states. We
excluded long-term care recipients, dual
Medicaid/Medicare eligibles, individuals who
died in 2009, and those with anomalous or
missing values for analysis variables.
Principal Findings: In multivariate analysis for
all states combined, HC patients had more
primary care visits (5.24 vs 4.73, p<0.0001 for all
comparisons shown) and higher average total
primary care costs (913 vs 776). HC patients
had fewer non-primary care outpatient visits
(3.99 vs 5.07), emergency department visits
(3.63 vs 3.83), and inpatient admissions (0.78 vs
0.82). Costs for all non-primary care utilization
was also lower for HC patients relative to nonHC, resulting in lower average total costs of care
per beneficiary for HC users (3,204 vs 3,618).
Overall findings were robust to multiple methods
of assessing HC use.
Conclusions: Among Medicaid enrollees in our
study population, those who received the
majority of their primary care in health centers
use more primary care visits at a higher cost, but
that is offset by lower use and costs across all
other services, resulting in lower total costs of
care.
Implications for Policy, Delivery, or Practice:
While this study should be viewed alongside
literature that addresses relative HC quality of
care, our findings suggest that investments in
comprehensive primary care through HCs may
be associated with savings in other areas and
lower overall cost.
Funding Source(s): HRSA
Poster Session and Number: B, #759
Consumer Understanding of the Health
Insurance Marketplaces: Insight into How
Marketplaces Can Successfully Attract and
Retain Consumers
Harmonijoie Noel, American Institutes for
Research; Daniel Harwell, American Institutes
for Research; Steven Garfinkel, American
Institutes for Research; Coretta Mallery,
American Institutes for Research; Graciela
Castillo, American Institutes for Research; Cong
Ye, American Institutes for Research
Presenter: Harmonijoie Noel, Survey
Methodologist, American Institutes for Research
hnoel@air.org
Research Objective: As of October 2013, each
state has a Health Insurance Marketplace
(Marketplace) where consumers can buy health
insurance from Qualified Health Plans (QHPs).
While there is a pool of potential enrollees
created by the legislative mandate, it is still a
challenge to best reach consumers for
enrollment. The biggest incentive to enrolling
through a Marketplace is the premium tax
subsidy; however, all consumers are not aware
of this benefit. This research was done as part
of a larger effort to design two surveys, based
on the Consumer Assessment of Healthcare
Providers and Systems (CAHPS) Principals,
which will assess consumers’ experiences with
the Marketplaces and QHPs. This research will
present results from our cognitive testing with
consumers, which may shed light on some of
the key issues Marketplaces will face as they
seek to attract and retain consumers for
enrollment.
Study Design: Cognitive testing was conducted
in two rounds with 52 individuals across three
languages: English, Spanish, and Chinese. All
three languages testing were done in
Massachusetts in the first round and in
Minnesota, Maryland and New York in the
second round to account for potential regional
variations in the populations and State Based
Marketplaces. In both rounds, the interview
notes were analyzed within each language first
and then a comparison was conducted between
the findings for each language to determine
similarities or differences in the findings.
Population Studied: The first round of cognitive
interviews was conducted with individuals who
had enrolled in a health plan through the
Massachusetts Health Connector before the
Marketplaces opened for enrollment October 1,
2013. The second round of testing was
conducted in three states with state-based
Marketplaces: Minnesota, Maryland and New
York with people who had at least compared
health plans and gave their income information
to the Marketplace. Respondents ranged in age
from 19-63 with a mean age of 42 across both
rounds.
Principal Findings: Respondents were not
always familiar with the specific name of the
State Based Marketplace or the general word
“Marketplace” or “Health Insurance
Marketplace,” but often were more familiar with
“Obamacare.”
Respondents’ experiences and their
interpretation of the questions varied by whether
or not they got help from a person either over
the phone or in person compared to using the
website on their own.
A number of respondents interacted with the
Marketplace using more than one mode, such
as using the website and calling the customer
service Help Line.
Respondents had a difficult time distinguishing
and defining the application and enrollment
processes.
Respondents often did not understand health
insurance terms (e.g., claims).
Respondents would sometimes think about
experiences outside of the Marketplace when
responding to survey items.
Conclusions: The financial assistance offered
by the Marketplaces provides a clear incentive
to consumers to enroll in insurance through the
Marketplace. However, lack of awareness and
inconsistencies in language may inhibit
enrollment.
Implications for Policy, Delivery, or Practice:
The Marketplaces will allow access to health
insurance for millions of Americans, many for
the first time. These findings will help illuminate
some of the difficulties with enrollment and help
Marketplaces target areas for improvement.
Funding Source(s): CMS
Poster Session and Number: B, #760
Development of the Health Insurance
Literacy Measure (HILM): Conceptualizing
and Measuring Consumer Ability to Choose
and Use Private Health Insurance
Kathryn Paez, American Institutes for Research;
Coretta Mallery, American Institutes for
Research; Harmoni Joie Noel, American
Institutes for Research; Chris Pugliese,
American Institutes for Research; Jennifer
Lucado, American Institutes for Research;
Eloesa McSorley, American Institutes for
Research; Deepa Ganachari, American
Institutes for Research
Presenter: Kathryn Paez, Principal Researcher,
American Institutes for Research
kpaez@air.org
Research Objective: The Affordable Care Act
(ACA) mandated that as of October 2013,
consumers would have the opportunity to
purchase health insurance through Health
Insurance Marketplaces. Successful
implementation of this legislation hinges upon
consumers’ understanding of health insurance
and their options for coverage. The research aim
is to develop a validated measure to assess the
health insurance literacy (HIL) of consumers
purchasing insurance in the private market. This
research reports the results of the field test and
the psychometric properties of the measure.
Study Design: We conducted factor analysis in
order to determine how health insurance literacy
items may be summarized into a smaller set of
domains or composite measures. We used the
Rasch model to calibrate each item in terms of
its difficulty level and estimate each item’s fit
statistics. To assess the validity of the health
insurance literacy measure, we performed
bivariate correlations and ANOVA analyses
between the HIL items or subscales and a scale
assessing health insurance literacy knowledge
and skills. We developed a knowledge and skills
scale consisting of multiple choice questions
with correct and incorrect response choices. The
correct responses were summed to create
scores.
Population Studied: Participants enrolled in
GfK’s online panel were used for the field test.
The GfK panel is a national probability sample of
the U.S. adult population. The target population
was adults ages 22 through 64 who reside in the
United States and who had private insurance or
Medicaid. Anyone enrolled in Medicare was
excluded from the study. Among the 937
respondents who consented to take the survey,
828 cases met the qualification criteria.
Approximately 27.4% of the sample was
nonwhite, 69.4% had a high school degree or
less and 35.2% had a household income of less
than $50,000.
Principal Findings: The HIL measure
demonstrated four reliable subscales: 1)
Choosing Health Insurance: General
Understanding, 2) Choosing Health Insurance:
Comparing Plans, 3) Using Health Insurance:
Knowing How, 4) Using Health Insurance: Being
Proactive. We estimated reliability for each
subscale using the Cronbach’s Alpha score. All
reliability coefficients were above .9 indicating
high internal consistency within the subscales.
All items met appropriate Rasch infit and outfit
statistic thresholds. Additionally, all subscales
were positively and significantly related to the
health insurance knowledge and skills scale
providing validity evidence for the measure.
Conclusions: These findings suggest that
greater endorsement of the attitudes and
behaviors measured by the HIL subscales are
more likely to be related to actual health
insurance knowledge and skills. This preliminary
field test shows the HIL measure to be a reliable
and valid tool for assessing HIL.
Implications for Policy, Delivery, or Practice:
A body of evidence developed from a validated
measure of health insurance literacy will move
the discussion from the nonspecific-- “most
consumers know little” and “health insurance is
just too complicated”-- to targeted action where
information and outreach can be systematically
tailored to the audience.
Funding Source(s): Other Missouri Foundation
for Health
Poster Session and Number: B, #761
What do Small Businesses Think of the
Affordable Care Act (ACA) of 2010?
Donna Perlmutter, IMPAQ International LLC;
Jacob Benus, IMPAQ International LLC; Manan
Roy, IMPAQ International LLC; Scott Davis,
IMPAQ International LLC; Neha Nanda, IMPAQ
International LLC; Futoshi Yumoto, IMPAQ
International LLC
Presenter: Donna Perlmutter, Research
Associate, IMPAQ International LLC
dperlmutter@impaqint.com
Research Objective: The Affordable Care Act
of 2010 (ACA) includes sweeping reforms
designed to fundamentally alter how the United
States health insurance market operates. One
significant component of the ACA is the
requirement that firms with 50 or more full-time
employees either provide such employees with a
health insurance package or else pay an annual
penalty of $2,000 per worker. The popular press
has offered wide-ranging opinions by well-known
economists as to how employers will react to
this new requirement when it goes into effect in
2015. Some have suggested that the ACA will
have a negative impact on hiring – particularly
employers with 40 to 50 employees.
To assess the validity of this speculation, a
sample of small businesses were surveyed to
determine their: (1) knowledge of the ACA, (2)
understanding about the provisions of the ACA,
(3) plans for hiring full-time employees, and (4)
plans for adjusting workers’ hours. This study
provides timely evidence on if small businesses
plan to adjust their work force and health
coverage in response to ACA’s provisions.
Study Design: IMPAQ International fielded a
national survey of small business employers
who are most likely to be affected by the ACA.
We examined whether employers’ hiring and
health insurance coverage decisions are
affected by the ACA. A mail survey was sent to
10,000 small businesses between May and June
2013. The survey was conducted prior to the
postponement of the ACA implementation for
small businesses.
Population Studied: The survey collected data
from three samples: (1) National sample of
employers with 40-60 employees (2) State
sample of all employers in Wisconsin (3) County
sample of employers in Howard County,
Maryland.
Principal Findings: Preliminary results from the
survey show that a vast majority of employers
are aware of the ACA (95 percent); most know
about the ACA employer requirements and the
associated penalty in case of noncompliance (96
percent); and most know about the penalty
amount (95 percent). The survey results also
show that larger employers are more likely to
provide health insurance benefits. While few
employers plan to reduce employment in the
near future, those who do plan to reduce
employment, 70 percent said that ACA is
entirely or partly the reason for their decision.
Conclusions: Small businesses are: (1) aware
of the ACA and its upcoming implementation
and (2) are effected by the planned
implementation of the ACA. Inasmuch as small
businesses employ the majority of the US
workforce, is important to continue to monitor
the effect of ACA implementation on the small
business community.
Implications for Policy, Delivery, or Practice:
The small business provisions of the ACA have
been postponed for one year. As a result, the
implications of the ACA small business
provisions will continue to influence the small
businesses hiring decisions. Policy makers must
have a better understanding of the impact of
these provisions on US employment.
Funding Source(s): No Funding
Poster Session and Number: B, #762
Projecting the Use of Inpatient and
Emergency Department Services After the
Affordable Care Act Medicaid Expansion
Gary Pickens, Truven Health Analytics; Ginger
Carls, Truven Health Analytics; Christine Eibner,
RAND Corporation; Joanna Jiang, Agency for
Healthcare Research and Quality; Zeynal
Karaca, Social & Scientific Systems, Inc.;
Audrey Weiss, Truven Health Analytics; Herbert
Wong, Agency for Healthcare Research and
Quality
Presenter: Gary Pickens, Vice President,
Truven Health Analytics
gary.pickens@truvenhealth.com
Research Objective: Medicaid expansion under
the Patient Protection and Affordable Care Act
(ACA) will add new enrollees to Medicaid
programs in states that elect to expand eligibility.
While there is substantial amount of research
has been conducted on the expected size and
composition of the newly covered Medicaid
population under different assumptions, less is
known about the use of healthcare—particularly
hospital care—that is likely to occur in this newly
insured population. The objective of this study is
to provide projections of inpatient hospital and
emergency department (ED) use after ACA
Medicaid expansion.
Study Design: This was a retrospective
observational study measuring inpatient
discharges, and ED visits. Regression models
estimated utilization measures from predictor
variables. Our models incorporate population
characteristics and state context factors that
may be influenced by policy. Hospital utilization
metrics were total discharges, preventable
admissions, and emergency department visits.
Discharge and ED visit rates were estimated
using the state- and year-specific Medicaid
enrollment estimates. ED visit records were
summarized in a similar fashion for 2007–2010.
State Medicaid expansion stances were
determined. Data for Medicaid patients were
aggregated by state, year, and type of service.
The utilization metric for the ED encounters was
visit volume, and visit rates were computed
using the same Medicaid enrollment estimates
employed for inpatient discharge volumes.
Population Studied: Hospital Inpatient and ED
records were extracted from Healthcare Cost
and Utilization Project (HCUP) State Inpatient
Databases (SID) for the years 2007–2011and
State Emergency Department Databases
(SEDD) for the years 2007–2010. The
enrollment estimates were based on the Centers
for Medicare & Medicaid Services (CMS)
Medicaid statistics and information from the
American Community Surveys for 2007–2011.
Medicaid enrollee characteristics, program
factors, and state variables were collected.
Inpatient discharge records were aggregated by
the state of the patient’s residence, year, and
major service line including Medicine, Surgery,
Maternity & Newborn, Injuries and Mental
Health. Data were restricted to adults aged 19–
64 years who reported a primary expected
payment source of Medicaid, because this age
group is likely to contribute the vast majority of
new Medicaid enrollees.
Principal Findings: Our models project that
change in population composition alone results
in a 22% increase in inpatient discharges and a
30% increase in ED visits, while use rates fall
6% and 0%, respectively. With the additional
capacity, reimbursement, and innovation policy
effects in place, inpatient discharges increase by
7% and ED visits by only 1%, while use rates fall
18% and 22%. Among the policy effects,
increases in primary care physicians’
acceptance rates of Medicaid patients appears
to have a greater impact on hospital use than
increases in capacity (physician and bed supply)
or delivery system innovation (Medicaid
managed care penetration).
Conclusions: Medicaid expansion will increase
inpatient and ED volumes, but utilization rates
will be below current levels. States can limit
increases through provider capacity, Medicaid
managed care, and increasing physician
acceptance of Medicaid patients.
Implications for Policy, Delivery, or Practice:
The findings of this study will inform
policymakers with details of how Medicaid
expansion will affect the volume and case mix of
Medicaid inpatient across different states.
Funding Source(s): AHRQ
Poster Session and Number: B, #763
Does Recognition as Patient-Centered
Medical Home and Behavioral Health
Integration Improve Care Delivery in the
Safety Net? Evidence from Early Expansion
of Medicaid in California
Nadereh Pourat, UCLA Center for Health Policy
Research; Xiao Chen, UCLA Center for Health
Policy Research; Max Hadler, UCLA Center for
Health Policy Research; Brittany Dixon, UCLA
Center for Health Policy Research
Presenter: Nadereh Pourat, UCLA Center for
Health Policy Research
pourat@ucla.edu
Research Objective: The Affordable Care Act
(ACA) has accelerated the adoption of the
patient-centered medical home (PCMH) and
promoted support for care coordination and
integration. These concepts are anticipated to
promote the triple aims of ACA by improving
quality of care, patients’ health, and reduce
costs. This study examines the impact of PCMH
recognition and behavioral health integration on
service use of enrollees in the Low Income
Health Program (LIHP). LIHP was established
by a Medicaid 1115 waiver as a bridge to reform
and was implemented from January 2011 to
December 2013 and was implemented by 53
California counties.
Study Design: We examined enrollment and
claims data from the first seven quarters of
LIHP. From publicly reported data, we identified
the community clinics that were recognized as
PCMH and whether they employed behavioral
health(BH) providers. We used random effects
Poisson models to examine change over time in
service use-- number of evaluation and
management (E&M) visits, emergency room
(ER) visits, and hospitalizations-- by PCMH
status, controlling for age, gender, chronic
condition and duration of enrollment. We next
examined change over time in service use by
PCMH status and by employing BH staff for
enrollees with BH diagnosis. We calculated the
predictive margins for ease of interpretation.
Population Studied: We included LIHP
enrollees (total N=48,666 in 37 clinics; had BH
diagnosis N=16,847) from 41 participating
counties. LIHP eligibility was restricted to
individuals ages 19-64 who met residency
requirements, had incomes below 133% of
federal poverty level, and were not eligible for
other public programs. These criteria were the
same as eligibility for Medicaid under the ACA.
Principal Findings: PCMH Enrollees had
higher adjusted rates of E&M visits (1.3 visit per
person) in the first quarter, but reduced this rate
to 0.9 visits per person in the last quarter. The
adjusted rate of E&M visit also changed from
0.78 to 0.75 for non-PCMH enrollees. The rate
of ER visits dropped accordingly from 0.19 to
0.15 only for PCMH enrollees. Similarly, the
rates of hospitalizations dropped from 0.09 per
person to 0.04 for PCMH enrollees only.
Enrollees with BH diagnosis in PCMH clinics
with BH employees experienced the highest
drop in adjusted E&M rates (1.59 visit/person) in
the first quarter, to 1.24 visits/person in the last
quarter. Those with PCMH but no BH
employees also experienced a drop by it was
less steep and those without PCMH but with BH
employees had the least steep drop in rates of
E&M. No change was observed for those
enrollees in clinics without PCMH or BH
employees. No change in ER rates or
hospitalizations was observed for enrollees with
BH diagnosis.
Conclusions: PCMH recognition significantly
reduced the rates of primary care and ER visits
and hospitalization during early expansion of
Medicaid. BH Integration further reduced
primary care visits, but did not have an
independent impact from PCMH on ER visits
and hospitalization.
Implications for Policy, Delivery, or Practice:
The findings support the perception that PCMH
recognition and BH integration could lead to cost
savings. Medicaid programs can garner costs
savings by PCMH recognition and BH
integration.
Funding Source(s): Other Blue Shield of
California Foundation
Poster Session and Number: B, #764
Successful Patient Centered Medical Homepilot experiences from Arizona, in
partnership with a Medicaid plan
Priya Radhakrishnan, St. Joseph's Hospital and
Medical Center; Roshni Kundranda, St Josephs
Hospital & Medical Center; Binh Doung, St
Josephs Hospital & Medical Center; Jenni
Schroeder, St Josephs Hospital & Medical
Center; Michael Hedden, Mercy Care Plan;
Christi Lundeen, Mercy Care Plan; Bob
Reichert, mercy Care plan; Veena Dhillon, St
Josephs Hospital & Medical center
Presenter: Priya Radhakrishnan, Robert Craig
Academic Chair, St. Joseph's Hospital and
Medical Center
pradhakri@dignityhealth.org
Research Objective: To study the impact of
care coordination and NCQA certification on
ulitization and costs in a high cost medicaid
population in inner city Phoenix
Study Design: This is a study of Utilization, cost
analysis on claims based data.
we embarked on a pay for performace pilot
based on risk sharing with Mercy care plan - an
Arizona state Medicaid plan.
Interventions
1. Active case management of the top 10%
utilzers
2. Multidisciplinary team discussion
3. patient open access ambulatory clinic
schedule
4. extended hours
5. NCQA level III certification
Population Studied: Medicaid population (
Mercy care plan) assigned to the Internal
Medicine health center (IMHC), an academic
practice in Phoenix , Arizona.
Principal Findings: Active care coordination in
patients assigned to the IMHC by an
interdisciplinary team led to a signifcant
reduction in utilization and a corresponding
reduction in costs.
The IMHC pilot program with Mercy care plan (
MCP) began in 2011. During the first year (
calender year 2012 Jan 1st -Dec 31st) the
annual membership was 1027 patients.
The results are as follows
1. The per member per month cost reduced to
$636.96 compared to a baseline cost of
$833.88 ( reduction of 24%).
2. The patients assigned to the IMHC practice in
the first year of the performance pilot were 1027
( an increase from baseline of 480.
3. There was a 32% reduction in emergency
room visits (baseline 1645/K to 1117/k)
4. There was a corresponding reduction of 29%
in-patient admissions 459/k to 327/K.
5. The pharmacy costs also reduced from a
baseline of $447.76 to $225.
Conclusions: The PCMH pilot with a high risk
population (inner city medicaid population) has
shown that intensive care coordination and
improving access can reduce costs. We were
able to shift care to the primary care offices by
reaching out to patients who predominantly used
the emergency room for care. In addition within
the subset of patients with complex medical
conditions, we were able to demonstrate
reduciton in hospitalizations by managing mild
disease flares in the ambulatory setting.
Implications for Policy, Delivery, or Practice:
This pilot demonstrates the ability for practices
to provide high value care by ensuring
personalized medicial plans for the highest
utilizers. For the success of the affordable care
act, insurers and the local, state and federal
goverments must examine practice
dmeographics and link performace to pay. It is
possible for some practices to provide care for
complex medical conditions by increasing the
range of services provided in the mabulatory
setting. This study provides the basis for
bundling and risk sharing in the out-patient
setting for primary care.
Funding Source(s): Other Mercy Care Plan (
Arizona Medicaid plan)
Poster Session and Number: B, #765
Understanding the Emerging Medicaid
Populations, Their Outreach and Healthcare
Needs: An Ohio Population-Based Survey
Lorin Ranbom, The Ohio Colleges of Medicine
Government Resource Center; Mina Chang,
Ohio Department of Medicaid; Eric Seiber, The
Ohio State University; Tim Sahr, The Ohio
Colleges of Medicine Government Resource
Center; William Hayes, The Ohio State
University
Presenter: Lorin Ranbom, Chief of Health
Services Research and Program Development,
The Ohio Colleges of Medicine Government
Resource Center
mina.chang@medicaid.ohio.gov
Research Objective: To develop comparative
health demographic and service profiles among
adults 19 to 64 years of age of the emerging
Medicaid populations as authorized by the
Affordable Care Act (ACA).
Study Design: The authors used data from the
2012 Ohio Medicaid Assessment Survey
(OMAS) to examine variations in health care
access, health services utilization, health status
(general, dental, vision, and mental health), and
special health care needs for Ohioans reporting
annual family income =138% Federal Poverty
Level (FPL) by the insurance categories
Medicaid enrolled and uninsured. All analyses
incorporated survey weights and complex
design characteristics. Analyses estimated
variation across demographic characteristics
such as race/ethnicity, age, geographic
residence, educational attainment, family
composition and working status. Analyses were
developed into a health profile for Ohio’s
Medicaid enrolled compared to Ohio’s newly
eligible populations under Medicaid extended
coverage.
Population Studied: The population for this
study consisted of residential Ohioans 19-64
years of age with reported 2011 annual family
income (income reported for the year prior to
being surveyed) =138% FPL ($25,571 for a
family of three). Data were collected using the
2012 OMAS, a dual-frame complex designed
telephone survey of Ohio’s non-institutionalized
residential adults – African-Americans were
oversampled and Hispanics and AsianAmericans were surname sampled. This study
sample was 4,730 adults and Medicaid
enrollment and uninsured status were selfreported, as were health care system
participation and reported health statuses.
Principal Findings: We estimated that for
Ohioans with income =138% FPL, 32.4%
(737,071) were uninsured; 26.5% (604,094)
reported having insurance other than Medicaid;
and 35.3% (803,549) reported Medicaid
enrollment – 5.8% (131,966) reported
government insurance other than Medicaid.
Overall, the currently enrolled Medicaid
population reported better access to and routine
use of health care services, a higher rate of
special health care needs, a lower prevalence of
poor-to-fair self-rated general, dental, and vision
health statuses and moderately higher level of
mental health impairment than the uninsured
within the same income category. Analyses of a
subgroup of uninsured women of child bearing
age =138% FPL reported significantly less
health care use than the Medicaid enrolled
mothers.
Conclusions: The new eligibles under Medicaid
extended coverage who were uninsured had
moderately higher estimated rates of selfreported risk behaviors, poorer general, dental
and vision health status, chronic health
conditions, and moderately lower level of mental
health impairment. A subgroup of uninsured
nulliparous women ages 19 to 44 years =138%
FPL had lower rates of health care access and
health care use, but higher special health care
needs than women with children – similar
findings were estimated for Medicaid enrolled
nulliparous women within =138% FPL.
Implications for Policy, Delivery, or Practice:
The profiles of the aforementioned emerging
Medicaid populations present key opportunities
for healthcare delivery: 1) for the subgroup of
uninsured with higher risk profiles, early
outreach and engagement in appropriate care
and disease management becomes critical; 2)
for the subgroup of uninsured nulliparous
women with a lower prevalence of special
needs, the extension of Medicaid coverage to
these newly eligible will help bridge service
gaps.
Funding Source(s): Ohio Department of
Medicaid
Poster Session and Number: B, #766
Labor Market Effects of the EmployerSponsored Health Insurance Tax Subsidy:
Evidence from Canada
Preethi Rao, The Wharton School, University of
Pennsylvania; Nora Becker, The Wharton
School, University of Pennsylvania; Elena
Prager, The Wharton School, University of
Pennsylvania
Presenter: Preethi Rao, Student, The Wharton
School, University of Pennsylvania
preethir@wharton.upenn.edu
Research Objective: We study the impact of a
change in the tax subsidy to employersponsored health insurance on labor market
outcomes. Though the Canadian government
provides health insurance to its citizens, a
majority of the population chooses to enroll in
supplemental coverage, primarily through
employer-sponsored health insurance. Like in
the United States, insurance obtained as a
benefit through an employer is not taxed as
income, and can therefore cause distortions to
the demand for insurance and more broadly, to
labor market outcomes.
Study Design: In 1993, the provincial
government of Quebec discontinued this tax
subsidy, providing us with a natural experiment
to study these distortionary effects. We extend a
widely used theoretical model in the literature to
predict that, following an increase in the effective
price of employer-sponsored health insurance,
hours worked should fall, wages should rise, or
both. Previous empirical work in this area has
used difference-in-differences analysis to isolate
the labor market impact of a rise in health
insurance costs, but has not sufficiently
addressed the potential endogeneity of health
care costs to labor market outcomes. Therefore,
we use pooled cross-sectional data from two
waves of the Canadian General Social Survey to
conduct a two-stage least squares analysis in
which we instrument for the potentially
endogenous cost of health insurance with the
exogenous policy change.
Population Studied: We study a population of
working-age Canadians (aged 25 to 64). We
exclude those from the province of Ontario,
which enacted a similar reform during our study
period.
Principal Findings: We find that following an
increase in the effective cost of health
insurance, hours worked fall significantly; for a
ten percentage point increase in the effective
price of insurance, hours worked fall by 1.1
hours per week. The effect is less pronounced,
but still significant, when we limit our sample to
individuals who work full-time (at least 40 hours
per week). We also find that a one percentage
point increase in the cost of insurance is
associated with an insignificant increase in
wages of 41-44%. However, the wage variable
is noisy and this result should be interpreted with
caution. These findings are consistent with our
theoretical predictions.
Conclusions: We find that following an increase
in the effective price of employer-sponsored
health insurance, hours worked fall and wages
rise.
Implications for Policy, Delivery, or Practice:
With an unprecedented number of individuals
set to purchase individual insurance in the U.S.
under health care reform, the distortions caused
by tax-subsidized health insurance may become
less pronounced in the coming years. The
removal of the tax subsidy in Quebec may
represent the largest similar change to date. It
may therefore provide an understanding of the
labor market effects of decreasing tax subsidies
for employer-sponsored health insurance and
the aggregate impacts of the ACA. To our
knowledge, this is the first research to examine
this question directly.
Funding Source(s): No Funding
Poster Session and Number: B, #767
Young Adults’ Experiences with Health Care
Affordability in the United States and Nine
Other Countries
Petra Rasmussen, The Commonwealth Fund;
Petra Rasmussen, The Commonwealth Fund;
Tracy Garber, The Commonwealth Fund;
Michelle Doty, The Commonwealth Fund; Sara
Collins, The Commonwealth Fund
Presenter: Petra Rasmussen, Senior Research
Associate, The Commonwealth Fund
pwr@cmwf.org
Research Objective: Young adults in the
United States have historically had high rates of
uninsurance. However, the Affordable Care
Act’s (ACA) health insurance coverage
provisions – the opportunity to remain on a
parent’s policy until age 26, subsidized private
health plans available through state
marketplaces, and expanded eligibility for
Medicaid – offer young adults the ability to
maintain health coverage at key life transition
points, including graduation from high school
and college, when millions have lost insurance
in the past. As the ACA’s coverage provisions
lead to a greater number of insured young
adults, the experiences of young adults in other
industrialized countries can be used as a
benchmark to measure improvements in health
care affordability.
Study Design: This study analyzes data from
the 2013 Commonwealth Fund International
Health Policy Survey, conducted in Australia,
Canada, France, Germany, Netherlands, New
Zealand, Sweden, Switzerland, the U.K., and the
U.S. among nationally representative sample of
adults ages 18 and older. Social Science
Research Solutions and country contractors
conducted interviews by telephone between
March 4 and June 5, 2013. The final samples
were weighted to reflect the distribution of the
adult population for each country
Population Studied: Adults ages 19-34 in ten
countries; total sample size is 3,857 with country
sample sizes ranging from 219 to 850.
Principal Findings: Overall, young adults in the
U.S. are by far the most likely to forego needed
care due to costs (46%), to have serious
difficulty paying medical bills (25%), and to have
high out-of-pocket medical expenses of $1,000
or more (29%). However, once insured, young
adults in the U.S. begin to look more like their
peers in other industrialized countries, while
uninsured U.S. young adults have significantly
more affordability problems. More than three of
five (65%) American young adults who were
uninsured during the year reported experiencing
an access problem due to cost, compared to
30% of continuously insured U.S. young adults
and between 27% (Netherlands) and 6% (United
Kingdom) of young adults in other countries.
Two of five (39%) uninsured U.S. young adults
reported having a serious problem paying for
medical bills in the past year compared to 13%
of U.S. young adults who were insured all year,
a rate that is better than young adults in France
(20%) and comparable to Switzerland (12%),
New Zealand (11%), and the Netherlands
(10%). Adjusted percentages, controlling for
income and health status, will be presented in
final product.
Conclusions: While U.S. young adults who
have been uninsured during the year stand out
as reporting problems affording health care at
high rates when compared to young adults in
other countries, U.S. young adults who have
been insured all year report experiences similar
to those in peer countries.
Implications for Policy, Delivery, or Practice:
As more young adults in the U.S. gain coverage
under the ACA, it is likely that the group as a
whole will report fewer affordability issues in
receiving health care. By using young adults in
other industrialized countries as a benchmark,
we will be able to measure how much of an
impact the ACA’s coverage provisions have on
young adults’ experiences with health care
affordability.
Funding Source(s): CWF
Poster Session and Number: B, #768
Emergency Department Transfers and
Transfer Relationships in U.S. Hospitals: A
Data-Driven Approach
Dana Sax, The Permanente Medical Group;
Ryan Mutter, The Substance Abuse and Mental
Health Services Administration; Robert
Houchens, Truven Analytics; Marguerite Barrett,
ML Barrett, Inc.; Jesse Pines, Department of
Emergency Medicine and Health Policy, George
Washington University Hospital
Presenter: Dana Sax, Emergency Department
Physician, The Permanente Medical Group
danakindermann@gmail.com
Research Objective: We describe and
characterize transfers out of hospital-based
emergency departments (ED) in the United
States; specifically, differences between sending
and receiving hospitals, and average distances
between hospitals. We also develop a new
measure called a “diversity index” to assess the
stability of ED transfer relationships between
hospitals. The overall goal is to develop a novel
measure for ED regionalization.
Study Design: We linked ED records at
transferring hospitals to ED and inpatient
records at receiving hospitals using the 2010
Healthcare Cost and Utilization Project (HCUP)
State Emergency Department Databases and
State Inpatient Databases, the American
Hospital Association Annual Survey, and the
Trauma Information Exchange Program.
Records were included where a sending and
receiving record were available; these data were
tabulated to describe ED transfers and their
hospital-to-hospital distances. To further
characterize the ED transfer relationship, we
created a novel measure, called the “diversity
index”, which estimated the effective number of
“transfer partners” from sending EDs. We used
logistic regression to analyze factors associated
with higher diversity indices.
Population Studied: We studied encounters
from hospital-based EDs in nine US states in
2010. Using the Clinical Classification Software
(CCS) to categorize conditions, we studied the
50 disease categories with the highest transfer
rates.
Principal Findings: A total of 97,021 ED
transfer encounters were included in the
analysis in the 50 high transfer rate disease
categories; among these, transfer rates ranged
from 1% to 13%. Circulatory conditions made up
about half of all transfers. Receiving hospitals
were more likely non-profit, teaching, trauma,
urban and had more beds, greater specialty
coverage and more advanced diagnostic and
therapeutic resources. The median transfer
distance was 23 miles; 25% traveled at least 4050 miles. Sending hospitals had a median of 3.5
effective transfer partners. In general, a higher
proportion of publicly insured patients was
associated with a with higher diversity index
(10% and 12% increases in the Medicare and
Medicaid share of ED encounters, respectively,
were associated with 10% and 14% increases in
the effective number of transfer partners). For
nervous system and trauma-related conditions,
both a greater share of Medicaid and uninsured
patients was associated with more transfer
partners.
Conclusions: Patterns of ED transfers in the
United States tend to follow logical patterns with
regard to available hospital resources. The
average ED transfer travels more than 20 miles,
with some travelling great distances for services.
Many EDs tend to transfer patients to multiple
hospitals, and the stability of this relationship –
assessed by the diversity index – differs by
condition. Less stable transfer relationships (i.e.
hospitals with greater numbers of transfer
partners) were more common in EDs with higher
proportions of publicly insured and uninsured
patients.
Implications for Policy, Delivery, or Practice:
At the hospital-level, a higher Medicaid share
may require additional coordination by ED
physicians (i.e. a greater search for an accepting
hospital), possibly because of lower payments
for care. This effect is important because as the
Affordable Care Act (ACA) is implemented and
millions gain Medicaid insurance, this could
potentially stress ED regionalization efforts.
Funding Source(s): AHRQ
Poster Session and Number: B, #769
Migrating to Medicaid? The Potential
Spillover Effects of Expanding Medicaid
Under Health Reform
Aaron Schwartz, Harvard University; Benjamin
Sommers, Harvard School of Public Health
Presenter: Aaron Schwartz, Graduate Student,
Harvard University
aschwart@fas.harvard.edu
Research Objective: Starting in 2014, many
low-income residents of states that forgo the
Medicaid expansion of the Affordable Care Act
(ACA) will be eligible for that program if they
move to another state. Some of these people
may migrate to receive coverage, thereby
increasing costs for states that have expanded
the program. This is known as the “welfare
magnet” hypothesis, a claim that geographic
variation in social programs induces the
migration of welfare recipients to places with
more generous benefits or eligibility. In order to
study the likelihood of such effects in the context
of the ACA, we examined whether recent public
insurance expansions were associated with
changes in the migration patterns of low-income
individuals.
Study Design: Using the 1998-2012 Current
Population Survey, we conducted difference-indifference analyses of low-income in-migration
and out-migration in states that expanded public
insurance (Arizona, Maine, Massachusetts and
New York) and matched control states (New
Mexico, Nevada, Pennsylvania, New
Hampshire, Connecticut and Rhode Island). Our
study period spanned five years before and five
years after each state’s insurance expansion. In
addition, event study regressions were
performed in order to test for immediate, lagged
or anticipatory migration effects. Because the
CPS solicits respondents’ main reason for
migrating, we also examined migration for health
reasons as a secondary outcome. For all
regression specifications, we included personlevel sociodemographic covariates and lagged
state-level economic covariates, and we tested
for non-parallel pre-expansion migration trends
between expansion and control states.
Population Studied: Our primary study sample
consisted of non-elderly adults with incomes
less than 200% of the federal poverty level who
resided in states undergoing public insurance
expansions and matched control states (n =
62,737 for in-migration sample, n = 61,991 for
out-migration sample). In robustness checks,
alternate subgroups were examined including
adults who were younger (aged 19-30),
childless, with poor or fair self-reported health,
or with lower incomes.
Principal Findings: Difference-in-difference
analyses indicated no statistically significant
changes in in-migration or out-migration
associated with the insurance expansions. This
result was consistent for our primary sample
(annual in-migration effect = -.47 percentage
points, 95% CI -1.46 to 0.54; annual in-migration
effect = 0.55 percentage points, 95% CI -0.30%
to 1.40%) and all examined subgroups. Event
study analysis of our primary sample did not
indicate the presence of immediate, lagged or
anticipatory migration effects. There was also no
association between the expansions and inmigration for health reasons or out-migration for
health reasons. Our preferred estimate for the
annual net migration effect associated with the
expansions (-1.01 percentage points, 95% CI 2.14 to 0.12) is precise enough to rule out net
migration effects larger than 1,600 people per
year in an expansion state.
Conclusions: Recent state public insurance
expansions were not associated with substantial
in-migration of low-income individuals.
Implications for Policy, Delivery, or Practice:
Although there are important distinctions
between the ACA and the policies we examined,
our findings suggest that migration will not be a
common way for people to obtain Medicaid
coverage under the current expansion, and that
interstate migration is not likely to be a
significant source of costs for states choosing to
expand their programs.
Funding Source(s): NIH
Poster Session and Number: B, #770
Effect of the Affordable Care Act Dependent
Coverage Expansion on Young Adults:
Evidence from Trauma
John Scott, Brigham & Women's Hospital;
Thomas Tsai, Brigham and Women's Hospital;
Aaron Schwartz, Harvard Medical School; Zirui
Song, Harvard Medical School
Presenter: John Scott, Surgical Resident,
Brigham & Women's Hospital
jwscott@partners.org
Research Objective: The Affordable Care Act
expanded coverage to young adults
(dependents) by allowing individuals to remain
on their parents’ health plans until the age of 26.
The effect of this dependent coverage
expansion on health care utilization and
outcomes remains largely unknown. We
evaluated the effect of this policy in the context
of trauma care, which accounts for a significant
proportion of health care utilization among
young adults in the U.S.
Study Design: We used the National Trauma
Data Bank (NTDB), a unique dataset containing
the universe of all trauma patients who
presented to U.S. trauma centers. For this
analysis, we focused on comparing 2008-2009
against 2011-2012 data. We focused on patients
presenting to trauma centers between the ages
of 21 and 25 before and after the ACA
dependent coverage expansion. Variables of
interest included private insurance status, injury
severity, length of hospital stay, and outcomes
such as mortality and length of stay. In
sensitivity analyses, we included data prior to
2008 and conducted subgroup analyses.
Population Studied: The universe of all trauma
patients cared for by over 700 U.S. Hospitals in
the NTDB.
Principal Findings: We found that the average
proportion of 21-25 year old patients presenting
with trauma who had private insurance
increased from 22.2% during 2008-2009 to
27.6% during 2011-2012. The proportion of
these patients presenting with low-severity
injuries increased from 52.8% to 56.3% during
this same period. Length of stay and 30 day
mortality fell from 4.9 days to 4.5 days and from
2.2% to 2.0%, respectively.
Conclusions: After the Affordable Care Act, the
proportion of young adults on private insurance
increased with a coinciding decrease in severity
of injury upon presentation to U.S. trauma
centers. These data suggest that the dependent
coverage expansion lead to meaningful changes
in health care access and may have increased
trauma center utilization in this population.
Implications for Policy, Delivery, or Practice:
Expansion of dependent health insurance
coverage increases access to care for young
adults. This may induce changes in the severity
of patients presenting for care. In the trauma
setting for young adults, the increase of patients
of lesser severity may increase volume and
exert capacity pressure on U.S. hospitals.
Funding Source(s): NIH
Poster Session and Number: B, #771
The Impact of Restrictive Antidepressant
Formulary Policies on Outcomes for Patients
with Major Depressive Disorder
Seth Seabury, University of Southern California;
Darius Lakdawalla, University of Southern
California; Deborah Walter, Takeda
Pharmaceuticals America, Inc.; John Hayes,
National Network of Depression Centers;
Thomas Gustafson, Arnold & Porter, LLP; Anshu
Shrestha, Precision Health Economics; Dana
Goldman, University of Southern California
Presenter: Seth Seabury, Associate Professor,
University of Southern California
seabury@usc.edu
Research Objective: Many state Medicaid
programs in the US have implemented
restrictive formulary policies designed primarily
to reduce spending on prescription drugs. For
patients with major depressive disorder (MDD),
formulary restrictions could severely limit access
to antidepressant therapies and cause potential
disruptions in treatment. The impact of these
restrictions on outcomes for MDD patients is
unknown.
Study Design: A retrospective analysis of state
Medicaid data on hospitalizations, emergency
room (ER) visits and health care expenditures.
Outcomes included the frequency of MDDrelated hospitalizations and ER visits per patient
and adoption of therapy. Multivariate regression
was utilized to identify associative effects of
state policies conditional on patient
characteristics, including age, gender and
health.
Population Studied: We linked data on patient
outcomes from 24 state Medicaid programs to
information on formulary restrictions, including
prior authorization and step therapy, from 2001
to 2008.
Principal Findings: Antidepressants subject to
step therapy and prior authorization were 56%
less likely to be adopted by MDD patients. Prior
authorization was associated with a 3.6%
increase/year in the number of Medicaid patients
with an MDD-related hospitalization and a 2.7%
increase/year in MDD-related ER visits. Adding
step therapy was associated with an additional
12.5% increase/year in MDD-related
hospitalizations and an 8.2% increase/year in
MDD-related ER visits by 8.2%. All differences
were statistically significant at the 5% level.
Conclusions: Formulary restrictions on
antidepressant therapy in state Medicaid
programs are associated with significantly worse
economic outcomes for patients with MDD.
Implications for Policy, Delivery, or Practice:
Funding Source(s): Other Takeda
Pharmaceuticals America, Inc
Poster Session and Number: B, #772
Use of HIT to Increase Primary Care Access
in Medicaid Patients: A Mixed Methods
Study
Cynthia Sieck, Ohio State University; Jennifer
Hefner, Ohio State University Department of
Family Medicine; Jennifer Lehman, Ohio State
University Department of Family Medicine; Ann
McAlearney, Ohio State University Department
of Family Medicine; Chris Taylor, Ohio State
University Department of Family Medicine;
Randell Wexler, Ohio State University
Department of Family Medicine
Presenter: Cynthia Sieck, Ohio State University
cynthiasieck@gmail.com
Research Objective: The Emergency
Department (ED) is often inappropriately used
for non-urgent or routine health service. This
occurs more frequently for patients covered by
Medicaid than for those with private insurance.
Encouraging appropriate ED use is critical to
control costs and achieve optimal health
outcomes. Our study developed, implemented,
and evaluated an ED-PCP (Primary Care
Provider) Connector Program aimed to: 1)
improve access to primary care for Medicaid
patients who routinely use the ED; and 2)
improve care coordination across healthcare
settings.
Study Design: We used health information
technology (HIT) to facilitate access by
scheduling follow-up PCP appointments during
ED visits. We also provided PCPs with access to
patients’ medical records through an integrated
electronic health record (EHR) system. In a
randomized controlled trial (RCT), we tested
whether the ED-PCP connector program
impacted healthcare utilization. Quantitative data
included measures of ED utilization and patient
satisfaction surveys. Qualitative data are being
collected through interviews with participating
providers, staff, and patients.
Population Studied: 137 patients participated
in the ED-PCP connector RCT; 69 received the
intervention. We will conduct a total of 30 followup interviews with subjects from both arms, and
with 25 administrative and 25 clinical key
informants from both the ED and primary care
settings.
Principal Findings: Of patients in the
intervention group, 22% attended their
scheduled primary care appointments. 60% of
the control group returned to the ED within 12
months for a non-urgent concern, versus 65% in
the intervention group. Early qualitative findings
suggest two categories of reasons for this
persistent use of the ED: (1) a preference for the
ED; and (2) a knowledge gap about PCP
services. Patients preferred the ED for a variety
of practical and social reasons, including viewing
the ED as a one-stop shop, location of the ED in
the neighborhood or near public transportation,
and seeing the ED as a social hub or as a safe,
warm place to go and where they might receive
a free meal. Further, there was a general lack of
awareness about the wellness and health
maintenance goals and availability of urgent
care services within primary care.
Conclusions: The Oregon Medicaid program
received national attention for the recently
published finding that Medicaid expansion
increased non-urgent ED visits, but lacked an
understanding of why enrollees used the ED
more frequently. Despite removing the traditional
barrier of PCP scheduling, we did not decrease
ED visits. This finding underscores the
importance of rarely discussed variables, such
as the practical social support functions the ED
provides to underserved populations.
Implications for Policy, Delivery, or Practice:
To solve this policy problem we must focus on
barriers to primary care beyond traditional health
services access variables, considering both
patients’ preferences for the ED, and their lack
of awareness of the goals of health maintenance
and the services available in a PCP office.
Funding Source(s): AHRQ
Poster Session and Number: B, #773
Health Care Access and Affordability among
Adults Potentially Eligible for Expanded
Medicaid Coverage
Donna Spencer, University of Minnesota School
of Public Health; Heather Dahlen, SHADAC;
Sharon Long, Urban Institute; Kathleen Thiede
Call, SHADAC
Presenter: Donna Spencer, Senior Research
Associate, University of Minnesota School of
Public Health
dspencer@umn.edu
Research Objective: In the states that are
choosing to expand Medicaid coverage under
the Affordable Care Act (ACA), a substantial
number of low-income, uninsured adults may
obtain new insurance coverage, along with the
associated gains in access to health care and
financial protections from high health care costs.
Medicaid coverage, however, is not without its
limits, as restrictions on benefits and barriers to
obtaining care are common. The purpose of this
presentation is to examine the current scope of
health care access and affordability among
Medicaid enrollees and estimate the access and
affordability needs among new enrollees under
the ACA expansion.
Study Design: Pooled 2011 and 2012 National
Health Interview Survey (NHIS) data were used
in the analysis. Data were extracted from the
Integrated Health Interview Series (IHIS), a
harmonized version of the survey. Analyses
focused on twelve measures of health care
access and affordability including long-standing
measures within the survey as well as new
measures added to the survey beginning in
2011 (e.g., individuals’ difficulty in finding a
doctor, difficulty in paying medical bills, reasons
for emergency department visit). The recycled
prediction method was used to estimate the
probability of each access and affordability
outcome under the hypothetical scenario that
low-income, uninsured adults gain Medicaid
coverage. Models controlled for key
demographic variables, socioeconomic factors,
and health and disability status. All analyses
accounted for the survey’s complex sample
design.
Population Studied: This study focuses on two
non-elderly adult coverage populations (aged 19
to 64 years) in the United States: individuals
with Medicaid coverage and low-income
individuals without health insurance coverage
who may qualify for expanded Medicaid
coverage.
Principal Findings: Low-income, uninsured
adults fare worse than adults covered by
Medicaid in terms of both access to and
affordability of health care. For example,
preliminary results show that a third of Medicaid
enrollees lacked a general doctor visit in the
prior year, whereas over 60% of low-income,
uninsured adults had not visited a doctor during
the same time frame. Additionally, almost half of
low-income, uninsured adults reported medical
debt; the proportion for adults with Medicaid
coverage was approximately 30%. Our
multivariate model results suggest significant
improvements across a number of access and
affordability measures for low-income, uninsured
non-elderly adults who are potentially eligible for
expanded Medicaid coverage.
Conclusions: Despite possible improvements in
health care access and affordability, preliminary
results indicate that a noteworthy proportion of
the Medicaid expansion population may have
limitations in health care access and
affordability.
Implications for Policy, Delivery, or Practice:
Insurance coverage is a critical step to
improving access to health care and, eventually,
improved health and well-being. If the full
potential of the expansion of Medicaid coverage
is to be realized, states need to address the
gaps in health care access and affordability
under the program, and the possibility that some
of those gaps may be exacerbated as their
Medicaid caseloads expand.
Funding Source(s): RWJF
Poster Session and Number: B, #774
Evaluating Continuity of Care between
Medicaid and Exchanges
Laura Spicer, The Hilltop Institute; Hamid
Fakhraei, The Hilltop Institute; Charles Betley,
The Hilltop Institute
Presenter: Laura Spicer, Senior Policy Analyst,
The Hilltop Institute
lspicer@hilltop.umbc.edu
Research Objective: With the new coverage
options in the 2014 health insurance
marketplace, there is concern about continuity of
care for individuals who experience a transition
in eligibility across insurance affordability
programs. These transitions may be particularly
difficult for individuals with serious acute or
chronic medical and mental health conditions.
To address this concern in Maryland, the
Maryland Health Benefit Exchange Act of 2012
required the Maryland Health Benefit Exchange
(MHBE) to study and report findings and
recommendations on “the establishment of
requirements for continuity of care in the State’s
health insurance markets.” The purpose of this
study was to fulfill this legislative charge and
evaluate options for continuity of care provisions
to assist beneficiaries who transition between
coverage options, including Medicaid and the
Exchange.
Study Design: This study analyzed Maryland
Medicaid eligibility, claims, and encounter data
for fiscal year (FY) 2011 to: (1) identify those
individuals who experienced a transition in
Medicaid/Children’s Health Insurance Program
(CHIP) eligibility; (2) identify the health and
service needs of the populations transitioning in
and out of the program through diagnosis,
procedure, and other coding on the
Medicaid/CHIP claims and encounters; and (3)
estimate a percentage adjustment to health plan
premiums to account for the movement of
people with specified health care needs into and
out of health plans. The health and service
needs examined included: pregnancy;
hospitalization; chemotherapy, radiation, and
dialysis; organ transplant; individuals with
ongoing care needs for durable medical
equipment, home health, and prescriptions for
management of chronic diseases; mental health
and substance use; HIV/AIDS; and pediatric
orthodontia. The listed needs were selected
through literature review and stakeholder input.
Population Studied: Four populations were
studied: (1) individuals who were continuously
enrolled in Medicaid/CHIP; (2) individuals who
newly enrolled in Medicaid/CHIP; (3) individuals
who lost Medicaid/CHIP eligibility; and (4)
individuals who gained and then lost
Medicaid/CHIP eligibility. Groups 2 through 4
were considered to be the transition population.
Principal Findings: Approximately 33% of
Maryland Medicaid/CHIP enrollees experienced
an eligibility transition during FY 2011, with 20%
newly gaining coverage, 12% losing coverage,
and 1% gaining and then losing coverage. The
majority of the population losing coverage did
not have one of the measured health
conditions/services needs, but the population
gaining coverage was more likely to have the
measured conditions/services needs, particularly
prescriptions and hospitalizations. The
estimated health plan premium impact of
providing continuity of care (both from
Medicaid/CHIP to the Exchange and vice versa)
was minimal.
Conclusions: This study developed a
methodology for estimating the health and
service needs of the population expected to
churn between Medicaid and Exchanges, which
can be replicated to monitor continuity of care in
2014 and beyond.
Implications for Policy, Delivery, or Practice:
The findings and policy options developed from
this study informed the MHBE Board’s
recommendations report to the Maryland
General Assembly, which were incorporated into
the Maryland Health Progress Act of 2013.
Funding Source(s): Other State
Poster Session and Number: B, #775
Assessing the Role of Obstetrics and
Gynecology Providers in Women’s Lives
Gabrielle Stopper, Planned Parenthood
Federation of America; Rachel Fleischer,
Planned Parenthood Federation of America;
Emily Stewart, Planned Parenthood Federation
of America
Presenter: Gabrielle Stopper, Public Policy
Manager, Planned Parenthood Federation of
America
gabrielle.stopper@ppfa.org
Research Objective: The purpose of this study
was to understand how women of reproductive
age access health care. This research aimed to
assess the relationship women have with
different primary and preventive healthcare
providers.
Study Design: We employed a 23-question
survey using the online survey panel,
KnowledgePanel, which is representative of the
United States population. Respondents are
randomly recruited through probability-based
sampling. Households are provided with a
computer and internet, if needed.
Population Studied: We surveyed 1,036
women ages 18–44. The race and ethnicity
breakdown was 58% white non-Hispanic; 19%
Hispanic; 14% Black/African American; and 9%
other races or 2+ races. Seventy-six percent of
the population was insured and 21% was
uninsured; 3% did not report insurance status.
Principal Findings: Findings show that 58% of
women report seeing an OB/GYN provider on a
regular basis and 35% of women view their
OB/GYN provider as their main source of care.
For 41% of women, an OB/GYN is the first
provider they chose as an adult. By a 16-point
margin, women say they are more likely to be
open and honest with their OB/GYN provider
than with other providers. For women of color,
their relationship with their OB/GYN provider is
even more profound. Latinas are more likely
(47%) to say their OB/GYN provider is their main
source of care, compared to 35% of women
overall. In addition, 64% of African-American
women report visiting an OB/GYN provider
regularly, compared to 58% of women overall.
Women report that OB/GYN providers are two
times more likely than other healthcare providers
to talk to them about HIV and birth control.
Conclusions: A majority of women routinely
access care though an OB/GYN provider and for
many women, they are their main provider.
Women feel more comfortable being more open
and honest with their OB/GYN provider than
other healthcare providers, demonstrating the
unique nature of these relationships. This study
underscores the importance of developing
policies that increase access to OB/GYN
providers.
Implications for Policy, Delivery, or Practice:
OB/GYN providers play a central role in the
delivery of primary and preventive healthcare to
women. Unfortunately, too often, the
conversation around improving healthcare for
women ignores the reality of how women access
healthcare in the United States. For example,
current federal programs and policies
predominantly focus on increasing access to
family practice and internal medicine providers.
As nearly 12 million women of reproductive age
become newly insured through healthcare
Exchanges and Medicaid under the Affordable
Care Act (ACA), healthcare industry experts
have voiced concern that access will be limited
because there will not be enough providers in
new plan networks. Based on the experience of
Massachusetts, when it expanded access to
health insurance coverage, OB/GYNs can
expect one of the greatest increases in wait
times for scheduling appointments. As the ACA
is implemented, it is critical that policymakers
and healthcare stakeholders have a thoughtful
process designed to ensure women have
access to OB/GYN providers. For example,
OB/GYN providers must be adequately
represented in Exchange health plan networks
and emerging coordinated care models must
include women’s healthcare providers.
Funding Source(s): No Funding
Poster Session and Number: B, #776
Low income men, Medicaid, and the
Affordable Care Act: Policy Implications
Amy Taylor, Agency for Healthcare Research
and Quality; Joel Cohen, Agency for Healthcare
Research and Quality
Presenter: Amy Taylor, Health Economist,
Agency for Healthcare Research and Quality
amy.taylor@ahrq.hhs.gov
Research Objective: Medicaid is a program for
low income adults and children in the US.
Although 70 percent of Medicaid recipients are
women and children, the remaining 30 percent
are men. While much has been written on
women and child enrollees in the program, not
much is known about the men who are enrolled
in Medicaid. The purpose of this study is to
examine the characteristics of male enrollees,
as well as their health care utilization and
expenditures. This will serve as a baseline for
determining the impact of Medicaid expansions
under the Affordable Care Act on medical
access, use, and expenditures for this
population. The study will then use multivariate
analysis to estimate the potential effects of the
ACA Medicaid expansions on health care use
and expenditures for this policy relevant group.
Study Design: This analysis uses pooled data
from the Medical Expenditure Panel Survey for
years 2005-2010. The study examines the
characteristics of men and women who would
and would not be eligible for Medicaid under the
provisions of the ACA Medicaid expansion.
Eligibility will be estimated based on a MEPS
micro-simulation model. The paper will also
present tables showing utilization and
expenditures for health care services by
Medicaid recipients in 2005-2010 and predicted
utilization after the Affordable Care Act is
implemented
Population Studied: A nationally representative
sample of non-institutionalized men and women,
age 18 and over, in the MEPS Household
survey for the years 2005- 2010.
Principal Findings: Preliminary data show that
while 56% of men currently on Medicaid are
between ages 21 and 44, over 65% of men
eligible for Medicaid under the ACA will be in
that age group. Men on Medicaid are more
likely to be in fair or poor health than their nonMedicaid counterparts (35 vs. 9%). Medicaid
eligible men under the ACA are less likely to
have an office based doctor visit than those
currently on Medicaid (52% vs. 71%). They are
also less likely to be hospitalized (7% vs. 13%).
In comparison, non-Medicaid men are less likely
than either current or newly eligible Medicaid
recipients to have either an inpatient hospital
stay or an emergency room visit.
Conclusions: Men enrolled in the Medicaid
program are sicker and more likely to receive
disability income than men not enrolled in the
program. Men currently on Medicaid are more
likely to use medical services than men not on
Medicaid or those who will become eligible
under the ACA. The medical care utilization of
the newly eligible Medicaid population looks
more like the non-Medicaid population than
those currently on Medicaid.
Implications for Policy, Delivery, or Practice:
In order to study the effects of Medicaid
expansions under the ACA, it is necessary to
have baseline information on health care use
and expenditures prior to those expansions.
This analysis will provide that baseline and
present estimates of the potential impact of the
ACA Medicaid expansions on health care for this
policy relevant group.
Funding Source(s): AHRQ
Poster Session and Number: B, #777
Unemployment-Related Transitions into
Health Insurance Marketplaces
Namrata Uberoi, Penn State University; Pamela
Short, Penn State University
Presenter: Namrata Uberoi, Phd Candidate,
Penn State University
namrata@psu.edu
Research Objective: The Affordable Care Act
(ACA) expands current public and private
sources of health insurance coverage, while also
creating new insurance marketplaces. The
Congressional Budget Office estimates that
employer-sponsored insurance (ESI) will
continue to cover 60% of Americans under the
ACA. Accordingly, changes in employment and
employment loss will continue to be primary
causes of gaps and transitions in health
insurance, and many people may look to the
new marketplaces to fill gaps in insurance
triggered by unemployment. This study
examines the number and characteristics of
individuals likely to utilize the new health
insurance marketplaces for unemploymentrelated transitions.
Study Design: Using the 2004 and 2008 panels
of the Survey of Income and Program
Participation, we undertook a descriptive
analysis to assess the risk of unemploymentrelated transitions into the marketplaces.
Changes in employment and insurance status
were tabulated by income categories,
corresponding to eligibility for Medicaid and
subsidies under the ACA: less than/equal 138%
(low), greater than 138% to less than/equal
250% (lower middle), greater than 250% to less
than/equal 400% (upper middle), and greater
than 400% FPL (high). Furthermore, we
estimated annual average risks of employment
loss, ESI policyholder prevalence, and the
likelihood of utilizing the marketplaces in the
context of different macroeconomic conditions
corresponding to timeframes before (20042006), during (August 2008-July 2009), and after
the Great Recession (2010-2012).
Population Studied: Working adults, ages 1864.
Principal Findings: Over a quarter million ESI
policyholders are at risk of utilizing the health
insurance marketplaces or Medicaid for
unemployment-related insurance transitions;
54% of these workers will be eligible for
subsidies under health reform. An annual
average of 9.0 million adult workers have lost
employment since the Great Recession. Ten
percent were ESI policyholders. Of those ESI
policyholders, 4% were low-income and 17%
were high-income workers. The risk of becoming
uninsured after losing employment decreased as
income increased; 23% for low-income to 6% for
high-income workers. The likelihood of
transitioning to being insured (other than ESI)
increased as income increased; 63% for lowincome and 74% for high-income ESI
policyholders. During the Great Recession, the
overall risk of employment loss increased by
70% compared to before the recession, with the
largest percentage changes for middle and highincome workers.
Conclusions: The risk of employment loss
decreases with income, but the prevalence of
ESI increases with income. As a consequence,
the majority of individuals who lose employment
and ESI are middle-income workers. Increased
unemployment risk and prevalence of ESI
interact to create the highest risk of
unemployment-related ESI losses in the middleincome group, and will also interact with that
group’s eligibility for assistance under the ACA.
Accordingly, ACA subsidies are well targeted to
protect the majority of newly unemployed
individuals who lose ESI.
Implications for Policy, Delivery, or Practice:
One can expect that over a quarter million ESI
policyholders will utilize the new marketplaces or
Medicaid for insurance under the ACA due to
unemployment-related transitions. Over half of
those individuals are eligible for subsidies.
Furthermore, establishing what role the
marketplaces will have for unemploymentrelated transitions will be a significant factor in
whether individuals will experience stable and
seamless health insurance coverage.
Funding Source(s): No Funding
Poster Session and Number: B, #778
Indigent Care Programs—Do People Know
They’re Uninsured?
Gregory Watson, UCLA Center for Health Policy
Research; Dylan Roby, UCLA Center for Health
Policy Research
Presenter: Gregory Watson, Research Analyst,
UCLA Center for Health Policy Research
gwatson@ucla.edu
Research Objective: Medically indigent
populations may mistake indigent care programs
for health insurance. The California Low-Income
Health Program (LIHP) provided health
coverage from 2011 through the end of 2013 in
participating counties to low-income uninsured
residents ineligible for Medicaid, Medicare or
other welfare programs. This study seeks to
ascertain whether a significant number of
individuals enrolled in LIHP indicated they were
uninsured or had public insurance in the 201112 California Health Interview Survey (CHIS), a
random-dial telephone survey of over 50,000
Californians conducted every two years since
2001. Such a misreporting would result in an
underestimate of uninsured and an overestimate
of other public coverage in LIHP counties, and
would reveal a serious misunderstanding of their
coverage status on the part of the medically
indigent participating in indigent care programs.
Study Design: Survey respondents in LIHP
counties in 2011-12 were identified as eligible
for LIHP based on their self-reported income
and documentation status and the county’s
income eligibility criterion. A hierarchical logistic
regression was conducted to test for an
association between LIHP eligibility and
tendency to report other public coverage. A
variety of demographic variables, including age,
sex, race/ethnicity and income, were included as
covariates in the regression model to control for
potential confounding effects. Respondent
county was also included to account for
geographic correlation.
Population Studied: The study included
childless adult CHIS respondents between the
ages of 19 and 64 who met residency
requirements and had incomes at or below
133% of the federal poverty level who reported
they were uninsured or had other public health
insurance.
Principal Findings: A statistically significant
association between LIHP eligibility and
propensity to report other public as opposed to
uninsured health insurance status was detected.
Significant effects were also present for various
other covariates, underscoring the importance of
controlling for them in the model.
Conclusions: This study provides evidence that
medically indigent persons may mistake indigent
care programs for health insurance, and
indicates that population surveys may
underestimate the number of uninsured among
populations eligible for indigent care programs.
Implications for Policy, Delivery, or Practice:
Misunderstanding of coverage status among the
medically indigent calls for improved education
of program participants, lest this misconception
render them less likely to pursue insurance
through alternate sources that could ameliorate
their indigent status. Survey underestimates of
the uninsured may deflate the apparent
importance of indigent care programs by
suggesting an erroneously small uninsured
population. It may also call into question local
and national estimates of the uninsured
population, a quantity with considerable
importance in matters of health policy.
Funding Source(s): Other The California
Endowment
Poster Session and Number: B, #779
The Impact of Medicaid Managed Care and
Home and Community-Based Alternatives on
Quality of Care for Adults with Disabilities
Martin Wegman, University of Florida; Garth
Graham, University of Florida; I-Chan Huang,
University of Florida; Jill Herndon, University of
Florida; Keith Muller, University of Florida;
Kimberly Case, University of Florida; Jason Lee,
University of Florida; Elizabeth Shenkman,
University of Florida
Presenter: Martin Wegman, MD-PhD Student,
University of Florida
mwegman@ufl.edu
Research Objective: Confronted with the dual
challenges of providing effective long-term
services to Medicaid beneficiaries with
disabilities and growing program costs, states
are increasingly turning to managed-care
models (1915(b) waivers) and home and
community-based service alternatives (1915(c)
waivers) to institutional care. Yet it is largely
unknown how these waiver programs – serving
over 3.3 million nationwide – impact the quality
of care provided to disabled individuals, one of
the most vulnerable and costly populations. We
examined the effects of the Texas’ 1915(b) and
(c) Medicaid waiver program for disabled
enrollees (STAR+PLUS) on 5 National
Committee for Quality Assurance Health Care
Effectiveness Data and Information Set (HEDIS)
indicators related to appropriate care for chronic
conditions.
Study Design: During January-February 2007,
all disabled Medicaid members >21 years were
transitioned from fee for service (FFS) or
primary care case management (PCCM) to
STAR+PLUS in 28 Texas counties. This natural
experiment allowed for pre-post comparisons of
quality of care for enrollees in the transitioned
counties relative to enrollees in counties
remaining in the FFS or PCCM Medicaid
models. County was used as the independent
sampling unit. For the primary outcomes of
interest, person-level claims and encounter data
were used to calculate county-level rates for 5
HEDIS quality measures: care for diabetes,
COPD, asthma (two measures), and
cardiovascular disease. These were regressed
on STAR+PLUS program by time trend
indicators. General linear mixed models were
constructed with a covariance model carefully
chosen to account for both longitudinality and
administrative clustering. All inference (tests,
standard errors, confidence intervals) used the
Kenward-Roger approximation for the Wald
statistic, based on the model-based covariance
estimates. The regression equations were
adjusted for enrollee race, age, gender, health
status, census tract poverty, county-year median
income, institutionalization and interactions
between health status and year.
Population Studied: Adults (ages 21-64) with
disabilities enrolled for at least one month
between January 2006 and December 2011 in
Texas Medicaid, not also qualifying for Medicare
and eligible for the respective HEDIS outcome
measure.
Principal Findings: Although quality of care
was similar among programs at baseline,
STAR+PLUS led to dramatic improvements in
use of beta blockers after discharge for
myocardial infraction (49% vs 81% adherence
post 2007 transition; p < 0.01) and appropriate
use of systemic corticosteroids and
bronchodilators after a COPD event (39% vs
68% adherence post transition; p < 0.0001). No
statistically significant program effects were
identified for measures examining the quality of
care for asthma, diabetes or LDL-cholesterol
measurement for patients with cardiovascular
disease.
Conclusions: Given the paucity of current
literature, this study provides new insights
regarding the impact of 1915(b) and (c)
Medicaid waiver programs on quality of care
relative to traditional Medicaid. Future research
will explore additional quality indicators, analyze
expenditure impacts and dissect the key
improvement drivers.
Implications for Policy, Delivery, or Practice:
The findings inform which health care conditions
may be most positively impacted by 1915(b) and
(c) Medicaid waiver programs. In particular, the
dramatic and sustained improvements in care
for COPD and acute myocardial infarction
indicate that substantial cost savings, mortality
reduction and quality of life benefits may be
achieved.
Funding Source(s): NIH
Poster Session and Number: B, #780
The Effect of the Affordable Care Act’s
Expanded Dependent Coverage Policy on
Healthcare Spending and Utilization
Christopher Whaley, University of California,
Berkeley
Presenter: Christopher Whaley, Student,
University of California, Berkeley
cwhaley@berkeley.edu
Research Objective: To examine the effects of
the Affordable Care Act's (ACA) dependent
coverage policy, which requires employersponsored insurance plans to offer coverage to
dependents until age 26 and resembles many
existing state dependent coverage policies, on
health care spending and utilization.
Study Design: State identifiable Medical
Expenditure Panel Survey Data (MEPS) data
from 2002-2011 were used to estimate the effect
the ACA's dependent coverage policy by using a
difference-in-differences regression to compare
spending between those ages 19-25 to those
ages 26-30. To control for existing state policies,
a triple-differences regression used state policy
eligibility criteria to identify those who are newly
eligible for dependent coverage under the ACA’s
policy. Total spending by public payers, private
insurers, and individuals were examined
separately. Spending on emergency room,
inpatient, outpatient, physician visits, and
prescription drugs by each payer were also
examined separately by payer. OLS regressions
controlling for demographics (age, sex, and
race), medical conditions (BMI, and existing
chronic conditions), and state fixed effects were
used to estimate each model. The policy’s
effective date is September 2010, so this
analysis estimates the effect of the first full year
of the policy.
Population Studied: Young adults ages 19-30.
Principal Findings: For all young adults, the
ACA's policy leads to a 45% increase in private
spending on physician office visits and a 40%
increase in private spending on prescription
drugs. However, there are no changes in
utilization for any service category.
After accounting for existing state dependent
coverage laws, emergency room and
prescription drug spending decreased by
approximately 25% for all payers. Public
spending on physician office visits and
outpatient visits fell by 31% and 21%,
respectively. Emergency room visits decreased
by 11% while prescription drug fills fell by 70%.
Conclusions: In the first year of
implementation, the ACA's dependent coverage
policy led to substantial decreases in both
individual and public health spending across
several health service types. As expected,
young adults not eligible for expanded
dependent coverage see the largest changes
following the ACA’s policy.
Implications for Policy, Delivery, or Practice:
These results suggest that the ACA’s expanded
dependent coverage policy has substantial
effects on health care spending and utilization.
Funding Source(s): N/A
Poster Session and Number: B, #781
Impact of High-deductible Health Plans on
Diabetes Monitoring and Outcomes
James Wharam, Harvard Medical School &
Harvard Pilgrim Health Care; Fang Zhang,
Harvard Medical School & Harvard Pilgrim
Health Care; Emma Eggleston, Harvard Medical
School & Harvard Pilgrim Health Care; Steve
Soumerai, Harvard Medical School & Harvard
Pilgrim Health Care; Dennis Ross-Degnan,
Harvard Medical School & Harvard Pilgrim
Health Care
Presenter: James Wharam, Assistant
Professor, Harvard Medical School & Harvard
Pilgrim Health Care
jwharam@partners.org
Research Objective: The Affordable Care Act
is expected to dramatically increase highdeductible health plan enrollment. This
expansion could reduce use of secondary
preventive services and access to care among
chronically ill patients. No studies have
examined the impact of high-deductible health
plans on intermediate outcomes such as
hemoglobin A1C levels or on hospitalizations
among patients with diabetes.
Study Design: Retrospective interrupted time
series with comparison series.
Population Studied: We studied administrative
claims among 33,639 commercially insured
patients with diabetes. To minimize selection
bias, we included only members who could
choose one health plan in their follow-up year.
Members were continuously enrolled for 2 years
through a large national insurer between 2004
and 2012. Among 11,213 study group members,
we analyzed monthly hemoglobin A1C testing
and hospitalization rates for one year before and
after their employers mandated a switch from
traditional low-cost-sharing plans to high
deductible plans, compared with rates among
22,426 propensity score matched
contemporaneous controls. We also examinined
mean monthly hemoglobin A1C values among a
subset of members with laboratory value data.
We subtracted mean rates of the control group
from those of the HDHP group to obtain a
before-after differenced trend, then used an
autoregressive model containing the covariates
of time in months, an indicator for follow-up
versus baseline period, and time in months in
the post period.
Principal Findings: Hemoglobin A1C testing
rates dropped in the high deductible group
relative to the control group from the baseline to
the follow-up year (absolute estimated level
change of -7.96%, [95% C.I., -15.56% to 0.36%]). Hemoglobin A1C value levels did not
change to a detectable degree (trend change of
-0.013 per month, [-0.032 to 0.005]).
Hospitalizations declined by 2.30 per 1000 (3.56 to -1.04) in the month after the HDHP
transition then did not change to a detectible
degree.
Conclusions: Switching from traditional to highdeductible health plans immediately reduced
hemoglobin A1C testing and hospitalizations
among patients with diabetes but did not worsen
disease control as measured by hemoglobin
A1C values.
Implications for Policy, Delivery, or Practice:
Our results raise concerns about access to care
for chronically ill patients with high-deductible
health insurance. Future analyses should
examine longer follow-up and key subgroups
such as the poor and sick. Policymakers might
consider special provisions for chronically ill
patients with high-deductible insurance, such as
reducing cost sharing for high acuity care.
Funding Source(s): CDC
Poster Session and Number: B, #782
Chronically Ill Uninsured Residents of
Massachusetts: Clinical Outcomes 5 Years
after State Health Reform
James Wharam, Harvard Medical School &
Harvard Pilgrim Health Care; Tomasz
Stryjewski, Massachusetts General Hospital;
Fang Zhang, Harvard Medical School & Harvard
Pilgrim Health Care Institute; Dean Eliott,
Massachusetts Eye and Ear Infirmary; J. Frank
Wharam, Harvard Medical School & Harvard
Pilgrim Health Care Institute
Presenter: James Wharam, Assistant
Professor, Harvard Medical School & Harvard
Pilgrim Health Care
jwharam@partners.org
Research Objective: The impact of state health
insurance exchanges on clinical outcomes such
as chronic disease control is unknown. Our
objective was to determine whether
Massachusetts Health Reform improved health
outcomes in uninsured patients with
hyperlipidemia, diabetes, or hypertension.
Study Design: We examined 1,463 patients
with hyperlipidemia, diabetes, or hypertension
who were continuously uninsured in the 3 years
before the 2006 Massachusetts Health Reform
implementation. We assessed mean quarterly
total cholesterol, glycosylated hemoglobin, and
systolic blood pressure in the respective cohorts
for 5 follow-up years compared with 3,448
propensity-score matched controls who
remained insured for the full 8 year study period.
We used person-level interrupted time-series
analysis to estimate changes in outcomes
adjusting for sex, age, race, estimated
household income, and comorbidity. We also
analyzed the subgroups of uninsured patients
with poorly controlled disease at baseline, no
evidence of established primary care in the
baseline period, and those who received
insurance in the first follow-up year.
Population Studied: Patients in the Partners
HealthCare network, the largest delivery system
in Massachusetts and second largest provider of
care to the uninsured prior to the 2006 health
reform.
Principal Findings: In five years after
Massachusetts Health Reform, patients who
were uninsured at baseline did not experience
detectable trend changes in total cholesterol (0.39 mg/dl per quarter, 95% confidence interval,
[-1.11 to 0.33]), glycosylated hemoglobin (0.02% per quarter, [-0.06 to 0.03]), or systolic
blood pressure (-0.06 mmHg per quarter, [-0.29
to 0.18]). Analyses of uninsured patients with
poorly controlled disease, no evidence of
established primary care in the baseline period,
and those who received insurance in the first
follow-up year yielded similar findings.
Conclusions: In uninsured, chronically ill
patients, Massachusetts Health Reform was not
associated with a change in hyperlipidemia,
diabetes, or hypertension control after five years
of follow-up. The findings may not be
generalizable to other states because of the
generous uncompensated care pool, low rates
of uninsured, and high concentration of health
providers present in Massachusetts prior to
health reform.
Implications for Policy, Delivery, or Practice:
Interventions beyond insurance coverage may
be needed to improve clinical outcomes of
chronically ill uninsured persons in
Massachusetts.
Funding Source(s): No Funding
Poster Session and Number: B, #783
Establishing Benchmarks to Understand
Hospital Utilization Following Medicaid
Expansion under the Affordable Care Act
Herbert Wong, Agency for Healthcare Research
and Quality; Audrey Weiss, Truven Health
Analytics; Ginger Carls, Truven Health Analytics;
Joanna Jiang, Agency for Healthcare Research
and Quality; Zeynal Karaca, Social & Scientific
Systems, Inc.; Gary Pickens, Truven Health
Analytics
Presenter: Herbert Wong, Sr. Economist,
Agency for Healthcare Research and Quality
Herbert.Wong@ahrq.hhs.gov
Research Objective: In 2014, many states will
initiate Medicaid expansion under the Affordable
Care Act (ACA); some states will not. Medicaid
expansion is expected to increase hospital
utilization as previously uninsured adults
become covered under Medicaid. In this study,
we define benchmarks for the rates of hospital
inpatient and emergency department (ED) use
following Medicaid expansion.
Study Design: We first examined whether
hospital utilization rates differed based on states’
stance on Medicaid expansion. We standardized
the hospital use metrics by computing separate
index values for the Medicaid and uninsured
population metrics relative to the national mean
so that all measures had similar scale. We then
examined which state-level Medicaid program,
demographic, and health status characteristics
were related to the states’ expansion stance. For
each specific characteristic found to be strongly
related to states’ likelihood to expand, we then
separated states into terciles —based on values
for each characteristic — in order to estimate the
relationship of that characteristic to hospital
utilization. We estimated current Medicaid and
uninsured utilization rates for each tercile of
these characteristics.
Population Studied: We obtained hospital use
data from the Healthcare Cost and Utilization
Project (HCUP) 2010 State Inpatient Databases
(SID) and State Emergency Department
Databases (SEDD). We obtained state-level
data on Medicaid program characteristics and
population demographics and health status from
Centers for Medicare & Medicaid Services
(CMS) Medicaid statistics, American Community
Survey, and Behavioral Risk Factor Surveillance
Survey.
Principal Findings: We found that several state
health system infrastructure characteristics were
strongly related to both expansion likelihood and
hospital utilization. In particular, in states highly
likely to adopt Medicaid expansion in 2014, we
observed higher levels of Medicaid managed
care organization (MMCO) penetration, a lower
primary care physician (PCP) supply challenge,
a lower level of primary care case management
(PCCM), and a smaller expansion population
size relative to the current Medicaid population.
We also found that in those states that are
currently committed to Medicaid expansion had
substantially lower hospital inpatient and
emergency department (ED) use among
Medicaid-covered patients compared to states
that have not committed to expand. For
example, the states that are already committed
can expect a post-expansion ED visit rate
between 316 and 607 per 1,000 Medicaid
enrollees, while the states currently leaning
against expansion could expect a postexpansion ED visit rate between 541 and 991
per 1,000 Medicaid enrollees, given the
characteristics of the current Medicaid programs
and Medicaid and uninsured populations in
those states.
Conclusions: Our results revealed a lower
impact of Medicaid expansion on hospital
utilization among states that have elected to
expand than among states currently unlikely to
expand.
Implications for Policy, Delivery, or Practice:
We identified and described variations in State
Medicaid programs and current hospital care
use relative to States’ plans to opt-in or opt-out
of the Medicaid expansion. The findings of this
study suggests that states with a certain
infrastructure may be better able to
accommodate Medicaid expansion and will
experience less impact to their hospital system.
Funding Source(s): AHRQ
Poster Session and Number: B, #784
Satisfaction with and Affordability of Health
Insurance Following Massachusetts Health
Care Reform: Views of Safety Net Patients
Leah Zallman, Cambridge Health Alliance;
Rachel Nardin, Cambridge Health Alliance;
Assaad Sayah, Cambridge Health Alliance;
Danny McCormick, Cambridge Health Alliance
Presenter: Leah Zallman, Junior Scientist,
Cambridge Health Alliance
lzallman@challiance.org
Research Objective: Like the Affordable Care
Act (ACA), the 2006 Massachusetts (MA) health
reform law (fully implanted by 2008) expanded
Medicaid coverage and created a health
insurance exchange offering publicly subsidized
private health insurance plans called
Commonwealth Care (CWC). Low income
residents were eligible for one of three CWC
plans, with different cost sharing requirements,
based on income. Prior studies raise concerns
about affordability – in particular the calibration
of cost sharing with income - of insurance postreform but neither the affordability of nor
satisfaction with specific insurance types has
been directly compared previously.
Study Design: Between August and December
2013, we conducted 681 face-to-face structured
interviews to assess safety net patients’ views of
the affordability of care and satisfaction with
their insurance in order to understand whether
cost sharing was well calibrated to income. We
confirmed insurance type using a statewide
database that is updated daily. Medicaid and
CWC 1 plans were available for people with
incomes < 133 and 150% of the federal poverty
level (FPL), respectively and required no
premiums and modest copays for medications
only. CWC 2 and 3 plans were available for
people with incomes 150-300% FPL and
required monthly premiums and had significant
co-pays for most services; these were combined
for our analysis (CWC 2+3). We used chi-square
tests to compare outcomes by insurance type.
Population Studied: We interviewed patients
presenting to three emergency departments at a
large integrated safety net health care system in
the greater Boston area. We included patients
aged 18-64, who spoke English, Portuguese,
Spanish or Haitian Creole. We excluded
severely ill patients.
Principal Findings: Of the 681 participants
(81% response rate), 16 % were uninsured, 5%
had CWC 1, 9% had CWC 2+3, 42% had
Medicaid and 28% had private insurance.
Patients insured by CWC 2+3 (71%) and private
insurance (78%) were less likely to agree that
their insurance is affordable than those insured
by Medicaid (95%) and CWC 1 (97%) (p<0.001).
Patients insured by CWC 2+3 (79%) were less
likely to report overall satisfaction with their plan,
as compared to those insured by Medicaid
(95%), privately (90%) and CWC 1 (98%)
(p=0.003). Satisfaction with services covered
was also lower among those insured by CWC
2+3 (79%) and among the privately insured
(85%) as compared to those insured by
Medicaid (91%) and CWC 1 (95%), though this
was of borderline significance (p=0.057).
Uninsured patients (40%) and those insured by
CWC 1 (22%) and CWC 2+3 (19%) were more
likely to report having delayed any care due to
cost than were those insured by Medicaid (14%)
and the privately insured (16%) (p<0.001).
Uninsured (46%) and privately insured patients
(34%) were more likely to have unpaid medical
bills as compared to those insured by CWC1
(24%), Medicaid (23%) and CWC 2+3 (24%)
(p=0.0002).
Conclusions: Affordability and satisfaction with
insurance in post-reform MA differs significantly
by insurance type. Those insured under public
or subsidized plans with minimal cost-sharing
plans reported substantially higher rates of
satisfaction and affordability of care compared
with subsidized private insurance with higher
cost-sharing or unsubsidized private insurance.
Implications for Policy, Delivery, or Practice:
Careful calibration of cost-sharing to income is
likely to be important in maximizing afforadability
and satisfaction with insurance products offered
to low to middle income individuals through
health insurance exchanges under the ACA.
Funding Source(s): Other Center for Health
Information Analysis, Commonwealth of
Massachusetts
Poster Session and Number: B, #785
Health Insurance Plan Knowledge in
Massachusetts: A Comparison of
Consumers Obtaining Insurance from a
Health Exchange, Non-Exchange Source or
Medicaid
Leah Zallman, Cambridge Health Alliance;
Rachel Nardin, Cambridge Health Alliance;
Assaad Sayah, Cambridge Health Alliance;
Danny McCormick, Cambridge Health Alliance
Presenter: Leah Zallman, Junior Scientist,
Cambridge Health Alliance
lzallman@challiance.org
Research Objective: Like the Affordable Care
Act, the 2006 Massachusetts health care reform
law expanded Medicaid coverage and created a
state health insurance exchange offering
publicly subsidized private health insurance
plans (called Commonwealth Care (CWC)) to
low income residents. Informed decision-making
by those seeking insurance coverage requires
that the application process, as well as the cost
and coverage features of plans, be
comprehensible. Limited understanding of plan
features could adversely affect utilization of care
and could undermine the consumer-driven
competition between plans that is expected to
be a key driver of health plan affordability.
However, little is known about health plan
knowledge among those obtaining Medicaid or a
subsidized health plan via a health insurance
exchange.
Study Design: Between August and December
2013, we conducted 681 face-to-face structured
interviews with a convenience sample of 681
safety net patients. We confirmed respondents’
insurance type using a statewide database
updated daily. We examined subjects’
responses to questions about the application
process for their current insurance, about their
knowledge of their current plan’s costs and
benefits and about whether the extent of their
knowledge affected their utilization of care.
Comparisons across insurance types were
made using the chi-square test.
Population Studied: We interviewed patients
presenting to three emergency departments at a
large safety net health care system in the
Boston area. We included patients age 18-64
who had only one type of insurance and spoke
English, Portuguese, Spanish or Haitian Creole.
We excluded severely ill patients.
Principal Findings: Of the 681 participants,
16% were uninsured, 14% had CWC, 42% had
Medicaid and 28% had private insurance.
Overall, 21% of respondents reported not
understanding features of their coverage,
including costs, and 34% reported not being
confident in their knowledge; this did not vary by
insurance type. Among those who were not
confident, 33% delayed medical care due to this
uncertainty. Overall, 24% of respondents found
it difficult to apply for their current insurance; this
was significantly worse for those with publicly
subsidized than for those with private insurance
(Medicaid 25%, CWC 35%, private 14%, p=
0.03). Similarly, more respondents reported
difficulty completing the application process for
Medicaid (39%) and CWC (37%) than for private
insurance (16%, p = 0.003). Overall, 41% of
respondents said the information provided
regarding plan choices was hard to understand,
17% said they did not have all the information
they needed about costs when signing up for a
plan and 38% reported difficulty submitting the
paperwork necessary to keep their insurance
active; these measures did not vary by
insurance type.
Conclusions: A high proportion of safety-net
patients in Massachusetts report knowledge
deficits about their insurance plan features and
the application process, often resulting in
delayed medical care due to this uncertainty.
Those obtaining Medicaid and subsidized
insurance though the state exchange reported
the greatest difficulties.
Implications for Policy, Delivery, or Practice:
New federal and state insurance exchanges
created under the Affordable Care Act should
provide information about health plans and the
application process in a format that is readily
understandable to consumers. Relying on
informed consumer choice to maximize
competition in insurance exchanges may be
premature.
Funding Source(s): Other Center for Health
Information Analysis, Commonwealth of
Massachusetts
Poster Session and Number: B, #786
Disparities and Health Equity
Are There Disparities in Causes of Perceived
Unmet Need for Mental Health Care?
Sirry Alang, University of Minnesota
Presenter: Sirry Alang, University of Minnesota
alang002@umn.edu
Research Objective: Mental disorders are
among the leading causes of disability in the
United States. The Substance Abuse and
Mental Health Service Administration estimates
that over 10 million adults felt that even though
they needed treatment for mental health
problems, they received insufficient or no mental
health care - reporting unmet needs in 2011.
This study explores reasons for unmet need and
examines whether there are socio-demographic
disparities in reported causes of unmet need for
professional mental health services
Study Design: Cross-sectional were obtained
from the 2011 sample of the National Survey on
Drug Use and Health. Outcome variables were
five main causes of unmet need: cost, stigma,
minimization, low perceived treatment
effectiveness and structural barriers. Each
cause of unmet need was regressed separately
on socio-demographic, health and service use
characteristics of adults with unmet need for
mental health services.
Population Studied: The analytic sample
(N=2,564) consists of adults 18 or older who
reported unmet need for mental health services.
Principal Findings: Women had higher odds of
reporting cost barriers while men were
significantly more likely to report stigma barriers.
Odds of stigma-related unmet need were greater
among younger adults age 18-25, among Blacks
and Hispanics compared to Whites, and among
rural residents compared to their counterparts in
large metropolitan areas. Blacks were also
more likely than Whites to report structural
barriers to receiving mental health services.
Conclusions: There are disparities in reasons
for perceived unmet need for mental health
services. These findings are important for
monitoring and addressing leading causes of
untreated mental illness, expanding current
understanding of disparities in unmet need, and
informing policies to reduce the burden of
untreated or undertreated mental illness.
Implications for Policy, Delivery, or Practice:
This study has identified groups among which
specific barriers to treatment are common.
Overall, programs and policies to reduce mental
illness stigma should increase focus on men,
African Americans, Hispanics, young adults, and
in rural areas. Efforts should be made to
increase the diffusion of information about
mental health services especially among African
Americans. For the majority of the population,
situating mental health services in areas where
they can be accessed using public
transportation, and providing flexibility in
scheduling appointments will go a long way to
reduce unmet need.
Funding Source(s): AHRQ
Poster Session and Number: A, #212
The Impact of Social Inequality on Survival
for Children and Adolescents with Leukemia
Mary Austin, University of Texas MD Anderson
Cancer Center; Hoang Nguyen, University of
Texas MD Anderson Cancer Center; Jan Eberth,
University of South Carolina; Andras Heczey,
Baylor College of Medicine; Dennis Hughes,
University of Texas MD Anderson Cancer
Center; Anna Franklin, University of Texas MD
Anderson Cancer Center; Linda Elting,
University of Texas MD Anderson Cancer
Center
Presenter: Mary Austin, Assistant Professor,
University of Texas MD Anderson Cancer
Center
maustin@mdanderson.org
Research Objective: Racial and socioeconomic
disparities have been well documented for
adults with cancer. Low socioeconomic status
(SES) has been associated with more advanced
disease at presentation and increased mortality
in a number of adult malignancies. It is unclear
if similar disparities exist for children and
adolescents with cancer. Our primary objective
was to identify geographic, socioeconomic and
racial disparities in children with leukemia, the
most common childhood malignancy, and
examine their impact on survival.
Study Design: With IRB approval, we
conducted a retrospective cohort study using
state cancer registry data. Cox regression was
used to determine factors associated with 1year, 5-year and overall survival. Covariates
included sex, age at diagnosis, self-identified
race/ethnicity, type of leukemia, year of
diagnosis, driving distance from a patient’s
home to the nearest pediatric cancer treatment
center and SES quartile. SES was determined
using the Agency for Healthcare Research and
Quality (AHRQ) formula and 2007-2011 US
Census block group data. Statistical
significance was defined as p<0.05 and results
are reported as hazard ratios with 95%
confidence intervals (CI).
Population Studied: The study population
consisted of all incidence pediatric (age < 19
years) leukemia cases reported to the Texas
Cancer Registry between 1995 and 2009
(n=4,662).
Principal Findings: The majority of patients
were Hispanic (51%) followed by non-Hispanic
white (38%), non-Hispanic black (7%) and other
(4%). Most patients lived less than 50 miles
from the nearest pediatric cancer treatment
center (62%), yet 23% lived greater than 50
miles away. The driving distance to the nearest
pediatric cancer treatment center did not impact
survival. In all models, race was independently
associated with survival. After adjusting for all
covariates, Hispanic race was associated with
decreased 5-year (HR 1.3 [1.1, 1.6]) and overall
survival (HR 1.4 [1.2, 1.6]) and non-Hispanic
black race was associated with decreased
overall survival (HR 1.3 [1.0, 1.7]). There was a
significant association between race and SES
with an over-representation of minorities in the
lower SES quartiles. With race excluded from
the model, there is a significant association
between SES and 5-year and overall survival
with the lowest SES quartile associated with the
worst survival.
Conclusions: These data highlight statistically
and clinically significant racial and
socioeconomic disparities in survival of children
and adolescents with leukemia. The etiology of
these disparities remains unclear and is likely
multifactorial.
Implications for Policy, Delivery, or Practice:
Additional studies are needed to fully
understand the basis of these disparities and to
untangle the effects of race/ethnicity and SES
on survival outcomes. Nevertheless, our
findings suggest that clinical practice may need
to adjust to cultural differences between groups
and policies may be needed to address a lack of
resources for patients with lower SES. For
example, it may be beneficial to establish health
systems with more rigorous follow-up for low
SES patients thereby improving their access to
post-treatment care and ultimately improving
survival.
Funding Source(s): No Funding
Poster Session and Number: A, #213
Identifying Cardiovascular Barriers to Risk
Identification and Reduction in Women
Veterans
Bevanne Bean-Mayberry, VA Greater Los
Angeles HSR&D Center of Innovation; Melissa
Farmer, VA Greater Los Angeles HSR&D
Center of Innovation; Jessica Zuchowski, VA
Greater Los Angeles HSR&D Center of
Innovation; Julia Yosef, VA Greater Los Angeles
HSR&D Center of Innovation; Nataria Joseph,
VA Greater Los Angeles HSR&D Center of
Innovation; S. Callie Wight, VA GLA Women's
Health Program; Fatma Batuman, VA GLA
Women's Health Program; Alison Hamilton, VA
Greater Los Angeles HSR&D Center of
Innovation
Presenter: Bevanne Bean-Mayberry, Faculty
Physician, VA Greater Los Angeles HSR&D
Center of Innovation
bevanne.bean-mayberry@va.gov
Research Objective: Cardiovascular (CV)
disease is commonly presumed to be a man's
disease but is the leading cause of death in
women, who typically have limited
understanding of their CV morbidity and
mortality risks. Our objectives were to: 1)
identify barriers and facilitators to CV risk
identification and reduction among women
Veterans, and 2) identify key areas to target for
the development of educational tools addressing
women’s CV risk identification and
management.
Study Design: Focus groups with providers and
semi-structured interviews with patients at two
Veterans Affairs (VA) comprehensive women’s
clinics. Patients completed a one-page screener
for CV risk factors before participating in a
voluntary, brief interview with a nurse.
Interviews were performed until saturation of
topics was achieved. Interview transcripts were
summarized by key domains and then reviewed
and compared. Data were compiled by domain
into a spreadsheet, and then synthesized using
a consensus-building analytic process.
Population Studied: Providers included patientcentered medical home team members
(doctors/nurse practitioners, registered nurses,
health techs, clerical staff, dietitians, and health
coaches) at both clinics; patients included
women Veterans with appointments in either of
two VA women’s clinics.
Principal Findings: One hour-long focus group
with each care team was completed at each
clinic, and one combined session with health
coaches was conducted for a total of three focus
groups involving 21 providers/staff. Providers
identified the following institutional and providerlevel barriers to CV risk identification and
reduction among women Veterans: system
difficulties for promoting prevention activities,
difficulties disseminating CV information in the
clinic, communication challenges between
providers, and limited time to dialog with
patients. Providers also identified the following
patient-level barriers: patient education needs,
medication non-acceptance by women, patients
agreeing to behavior change with providers but
refusing other staff, and complexities due to
patient mental and physical health burdens.
Provider-identified needs for facilitating
improved CV risk identification and reduction
included strategies for patient
engagement/motivation, verified and organized
resources and tools on CV health, more health
coach integration, and technology-based
resources.
Nineteen patients (mean age 48, range 27-61;
32% Caucasian, 32% African American, 16%
Asian, 16% Hispanic) completed brief
interviews. In this sample, 79% reported
overweight/obese status, 42% high cholesterol,
37% smoking history, 26% hypertension, 16%
pregnancy-related risks, and 5% diabetes.
Patient-identified barriers to improving CV health
included poor motivation, difficulty prioritizing
self, and poor balance of health with competing
demands of physical and mental health.
Patients’ recommendations for facilitating
improved CV health included motivational
support from others, accountability partners of
any type (family, friends, providers, etc), more
knowledge about CV risks and complications in
women, and exercise programs that fit their lives
and competing demands.
Conclusions: Providers commonly expressed a
desire for more time with patients as well as
educational tools to utilize during the course of
regular visits. Patients requested better
personal and clinical support systems to discuss
CV risk-related issues, as well as more
accountability to enact lifestyle changes.
Implications for Policy, Delivery, or Practice:
Evidence-based strategies need to be
implemented at the local system level to
incorporate tools and processes that facilitate
provider-patient discussions, patient activation,
and accountability to promote CV disease risk
reduction for women.
Funding Source(s): VA
Poster Session and Number: A, #215
Racial and Rural Differences in Cervical
Cancer Prevention and Control Practices
Jessica Bellinger, University of South Carolina;
Swann Adams, University of South Carolina;
Alexa Gallagher, University of South Carolina;
Janice Probst, University of South Carolina
Presenter: Jessica Bellinger, Postdoctoral
Fellow, University of South Carolina
bellingj@mailbox.sc.edu
Research Objective: Access to preventive
services contributes to differences in cervical
cancer screening, treatment, and survival. We
examined access to advanced cervical cancer
prevention technologies, including liquid-based
Pap test cytology, HPV vaccination, and DNA
testing among rural versus urban women.
Study Design: We conducted a cross-sectional
study of 2006 - 2008 visit-level data from
National Ambulatory Medical Care Survey
(NAMCS) and National Hospital Ambulatory
Medical Care Survey (NHAMCS). Data were
linked to the 2009 Area Resource File (ARF)
based on both provider and patient/visit location.
Patient/visit and provider location were linked
using state and county FIPS codes.
Population Studied: To examine the likelihood
of receiving a liquid-based Pap test, the study
population was limited to visits by female white
and African American patients (9 – 70 years of
age) with record of a Pap test. To examine the
likelihood of receiving HPV DNA test, the study
population was limited to visits for preventive
screening or routine general exams rather than
visits for new problems or pre/post-surgery.
Records with missing information were
excluded.
To examine cervical cancer screening practices,
patients were categorized by Pap test cytology
(liquid-based, conventional, or unspecified) and
an HPV DNA test during their visit (yes/no).
Race/ethnicity was classified as white or
black/African American. Location was examined
based on patient’s county of residence and
physician practice site.
SAS-callable SUDAAN was used to account for
complex sampling required weighted analysis.
Descriptive statistics and bivariate comparisons
were computed using chi square tests.
Analyses incorporating restricted data
(environmental and geographic variables and
physician characteristics) were conducted at the
National Center for Health Statistics Research
Data Center.
Principal Findings: No significant differences
were observed for Pap test cytology by patient
residence (urban versus rural; p=0.21) or for
receipt of liquid Pap testing between white and
African-American women residing in urban or
rural counties (p=0.35).
A significantly higher proportion of women living
in rural counties (69.6%) received liquid based
Pap testing in hospital outpatient settings than
women in urban counties (39%; p=0.02). A
significantly higher proportion of women residing
in urban counties received HPV DNA testing
versus women residing in rural counties (10%
versus 3.3%, respectively). Report of HPV
vaccination was too low during the study period
to permit stable estimates for rural women, so
no rural-urban comparisons can be offered.
Differences in provider reimbursement were
noted with higher proportions of publicly insured
patients in rural practices than urban practices
(p<0.01).
Conclusions: The proportion of visits by rural
residents for preventive services was lower than
among their urban counterparts. Women
residing in rural counties did not differ from
urban in conventional or liquid-based Pap test
cytology; however, rural women were less likely
to receive HPV DNA testing. No racial
differences were detected. More research is
needed to determine if observed differences are
the result of provider or patient barriers and
acceptability.
Implications for Policy, Delivery, or Practice:
Increased coverage and expanded access to
cervical cancer preventive services may
increase uptake of innovative preventive
services, particularly liquid-based cytology and
HPV DNA testing.
Funding Source(s): HRSA
Poster Session and Number: A, #216
Differing Perceptions of Physician Supply
and Public Health Concerns on Long Island
[LI]
Lisa BenzScott, Stony Brook University; Norman
Edelman, Stony Brook University; Mahrukh
Riaz, Stony Brook University
Presenter: Lisa BenzScott, Director, Stony
Brook University
lisa.benzscott@stonybrook.edu
Research Objective: Analysis of public
perceptions of physician availability and health
concerns among residents in the two counties
on LI, NY. These counties are sociodemographically similar and equal in total
population but differ in physician supply and
publically supported healthcare.
Study Design: Questions concerning the
availability of primary care [PCP] and specialty
[SP] physicians as well as public health
problems were included in an annual household
telephone survey of a random sample of LI
adult residents. Respondents were asked
whether they thought there were “too few” “too
many” or “the right amount” of PCP and SP as
well as what was the greatest health problem.
County comparisons were by bivariate Chi
square analysis and socio-demographic
comparisons by logistic regression including as
major variables income, age, and gender; race
was not analyzed because of too few non-white
responders.
Population Studied: 812 adult residents of
Nassau and Suffolk counties. Female =49%;
income: <35k = 14%, 35k to 99k =44%,
=100k=42%. Age <65= 83%
Principal Findings: Residents of Suffolk
thought there were “too few” PCP and “too few”
SP to a significantly greater degree than those
of Nassau; 30.2 vs. 20.7% for PCP [p<.05] and
31.7 vs. 16.5% for SP [p<.001]. This parallels
the actual physician supply: 4.1/1000 [23% PCP]
in Suffolk vs. 7.3/1000 [21%PCP] in Nassau. In
multifactorial analysis, compared to higher
income residents, low income residents [<$35K]
were more likely to perceive too many PCP
[p<.01] but, in Suffolk, too few SP [p<.05], and
most likely to be concerned about environmental
health issues p[<.05]. Compared to men, women
were more likely to perceive too few physicians,
both PCP and SP [p<.05], and most likely to be
concerned about chronic illness such as cancer
[p<.01]. Younger people were more likely to
perceive obesity as a major health problem than
older people [p<.05]. Only 1.18% of the total
respondents listed heart disease as a major
concern.
Conclusions: The greater perception of lack of
physician availability mirrors the lesser supply in
Suffolk suggesting that physician supply in
Suffolk is insufficient to fully meet demand. The
surprising perception of too many PCP but too
few SP by low income respondents suggests
that the counties have largely succeeded in
meeting the needs of low income residents for
PCP but not for SP in Suffolk which, unlike
Nassau, provides only primary care for low
income residents. Gender differences likely
reflect women’s role as the primary family
healthcare manager.
Implications for Policy, Delivery, or Practice:
1] Public providers of healthcare to low income
residents of LI should address essential
specialty services as well as primary care
services. 2] In some suburban New York areas
physician supply may be inadequate to fully
meet demand despite being above the national
average; suggesting that even within a
metropolitan region uneven distribution of
physicians relative to population is a significant
problem. 3] Perceptions of major health
problems may not match reality and vary sociodemographically, suggesting the need for more
targeted health education initiatives.
Funding Source(s): Other Internal Funds
Poster Session and Number: A, #217
Should We Target Food Insecurity and CostRelated Medication Underuse in Diabetes
Population Management? Results from the
Measuring Economic iNsecurity in Diabetes
(MEND) Study
Seth Berkowitz, Massachusetts General
Hospital; James Meigs, Massachusetts General
Hospital; Steven Atlas, Massachusetts General
Hospital; Darren DeWalt, University of North
Carolina at Chapel Hill; Hilary Seligman,
University of California, San Francisco; Deborah
Wexler, Massachusetts General Hospital
Presenter: Seth Berkowitz, Fellow,
Massachusetts General Hospital
saberkowitz@partners.org
Research Objective: Population management
to address social determinants of poor diabetes
outcomes is an emerging strategy, but which
factors to target is unclear. Food insecurity, an
inability to consistently access nutritious foods,
and cost-related medication underuse (CRMU)
may be significant, yet modifiable, barriers to
diabetes management. To inform future
population management interventions, we
ascertained prevalence and tested hypotheses
that they are independently associated with poor
diabetes control.
Study Design: Cross-sectional assessment of
social circumstances and diabetes outcomes
Population Studied: We contacted a stratified
random sample of adult (age >20 years)
diabetes patients in 4 clinic sites (2 community
health centers, 1 academic internal medicine
practice, and a diabetes center) from June 15Sept 15th 2013. Participants completed
validated instruments assessing food insecurity
and CRMU in English or Spanish. We also
collected information on other social
circumstances, including educational attainment,
limited English proficiency (LEP), health literacy,
and nativity, along with age, gender,
race/ethnicity, insurance, diabetes duration,
Charlson comorbidity score, diabetes
medications, Hemoglobin A1c (HbA1c)
tests/year, healthcare visits and 30-day
readmissions. The primary outcome was poor
diabetes control (most recent HbA1c>9.0% or
LDL cholesterol >100 mg/dL). We estimated
prevalence using inverse probability weighting,
and used chi-squared and Wilcoxon testing for
unadjusted analyses. We performed
multivariable logistic regression analysis, using
generalized estimating equations to account for
clustering by clinic. We also used pseudo-R2
statistics to evaluate explained variation in
diabetes control.
Principal Findings: Overall, 412 patients were
included (response rate: 62%). Of these, 19%
reported food insecurity, and 28% CRMU.
Patients reporting food insecurity were more
likely to be younger (mean age 56 vs. 63 years,
p=.01), non-white (35% vs. 22%, p=.01), and
have Medicaid (22% vs. 11%, p=.03). In
unadjusted analyses, the prevalence of poor
diabetes control was higher in those with, vs.
without, food insecurity (53% vs. 28%, p=.01)
and CRMU (53% vs. 22%, p=.01). Those with,
vs. without, food insecurity had similar
comorbidity (median Charlson score 4 vs. 4, p
=.44) and HbA1c tests/year (an indicator of
engagement with care) (median 2.5 vs. 2.5, p =
.90), but more outpatient visits/year (median 8
vs. 6, p =.01). The pattern was similar for those
with, vs. without, CRMU. In a multivariable
regression analysis, adjusted for the covariates
listed above, food insecurity (adjusted OR1.62,
95%CI 1.18-2.23) and CRMU (aOR 2.59,
95%CI1.41-4.75) were associated with worse
diabetes control. Using Pseudo-R2 statistics, our
model explained 28% of total variation in
diabetes control; food insecurity and CRMU
accounted for 4% of total variation. By
comparison, education, health literacy, LEP, and
nativity (other indicators of social disadvantage)
together explained only 2%. Among those with
at least 1 inpatient admission, food insecurity
was associated with a higher 30-day
readmission rate (20% in food insecure vs. 7%
in food secure, p=.02).
Conclusions: Food insecurity and CRMU are
common in diabetes patients, and identify
patients with increased risk of poor diabetes
outcomes despite similar co-morbidities,
engagement with care processes, and higher
health service utilization.
Implications for Policy, Delivery, or Practice:
Food insecurity and CRMU may be important
diabetes intervention targets, especially for
improving metabolic control and reducing 30-day
readmissions.
Funding Source(s): HRSA
Poster Session and Number: A, #218
Which Area-Based Socioeconomic Status
Indicator Should be Used to Monitor
Disparities within Healthcare Systems?
Seth Berkowitz, Massachusetts General
Hospital; Carine Yelibi, Massachusetts General
Hospital; Daniel Singer, Massachusetts General
Hospital; Steven Atlas, Massachusetts General
Hospital
Presenter: Seth Berkowitz, Fellow In General
Medicine And Primary Care, Massachusetts
General Hospital
saberkowitz@partners.org
Research Objective: Many healthcare systems
wish to track socioeconomic (SES) disparities in
health outcomes, but do not have self-report
SES indicators. We sought to determine which
area-based SES indicator, and geographic level,
is best suited to monitor healthcare disparities
from a delivery system perspective.
Study Design: Cross-sectional, observational.
Population Studied: 142,659 adults seen in a
primary care network from January 1, 2009 to
December 31, 2011. Address data were used to
construct ‘geocoded’ area-based SES indicators
at 3 geographic levels: block group (BG), census
tract (CT) and ZIP code(ZIP). Indicators
constructed included median household income,
percentage living in poverty, percentage with
college diploma, percentage unemployed, and
two indices that combine multiple census
variables –the Neighborhood Deprivation Index
(NDI) and an index used by the Agency for
Healthcare Research and Quality (AHRQ). We
constructed health outcome indicators across 5
dimensions of care quality using electronic
health records: disease prevalence (% with
diabetes mellitus), disease management (% of
patients with diabetes or coronary heart disease
with LDL cholesterol < 100mg/dL), preventive
services (% of eligible patients up to date on
colorectal cancer screening), resource utilization
(% of patients with = 3 emergency department
visits in the last year) and patient communication
(% of patients reporting poor communication
with their doctor). Relative Indices of Inequality
(RIIs), which can be interpreted as prevalence
rations between the worst-off and best-off group,
were calculated using log-Poisson regression
with robust error variance to quantify disparities
detected by area-based SES indicators and
compared to RIIs from self-reported educational
attainment. We evaluated indicators based on
data missingness, disparity detection, and ease
of construction.
Principal Findings: ZIP indicators had less
missing data than BG or CT indicators (1% vs.
9% missing, p <.0001). Area-based SES
indicators were strongly associated with selfreport educational attainment (RII 50.5, 95% CI
46.1-55.5, p <.0001), suggesting that <high
school diploma educational attainment was 50
times more common in the least well-off ZIP
compared to the most well-off. ZIP, BG, and CT
indicators all detected expected SES gradients
in health outcomes similarly. For example, the
RIIs (95%CI) for the outcome of poor
communication were 2.87 (1.76 - 4.69) for
median household income by BG, 3.06 (1.81 5.20) by CT, 2.79 (1.72 - 4.50), 2.79 (1.72 4.50) for ZIP and 2.66 (1.92 - 3.69) by self-report
educational attainment. Single-item indicators
performed as well as multidimensional indices.
For example, for the outcome of frequent ED
utilization, median household income by ZIP had
an RII (95% CI) of 4.88 (3.84 - 6.19), compared
to 4.71 (3.67 - 6.05) for NDI Index by ZIP and
3.80 (2.96 - 4.88) for AHRQ Index by ZIP.
Conclusions: Area-based SES indicators and
self-reported educational attainment detected
health outcome differences similarly. Our
preferred indicator was ZIP level median
household income, because it detected
disparities well, had low missingness, and was
simple to construct.
Implications for Policy, Delivery, or Practice:
The use of ZIP-based SES indicators, which do
not require the more extensive geocoding
process needed for BG or CT indicators, may be
an easy to implement way to monitor SES
disparities within healthcare delivery systems.
Funding Source(s): HRSA
Poster Session and Number: A, #219
Out-of-Pocket Health Care Expenditure
Burdens Among Nonelderly Adults with
Heart Disease: 2006-2010
Didem Bernard, Agency for Healthcare
Research and Quality; Zhengyi Fang, Social &
Scientific Systems, Inc.; Susan Yeh, Johns
Hopkins School of Public Health
Presenter: Didem Bernard, Senior Economist,
Agency for Healthcare Research and Quality
didem.bernard@ahrq.hhs.gov
Research Objective: To examine the
prevalence of high out-of-pocket burdens and
financial barriers to care among patients with
treatment for heart disease.
Study Design: Main outcome measures are the
proportion of persons living in families with high
out-of-pocket burden associated with medical
spending relative to income, defining high health
care burden as spending on health care > 20%
of income and high total burden as spending on
health care and insurance premiums > 20% of
income.
Population Studied: Persons aged 18-64 with
treatment for heart disease from a nationally
representative sample of the US population from
the 2006-2010 Medical Expenditure Panel
Survey.
Principal Findings: The prevalence of high
total burdens was significantly greater for
persons with treatment for heart disease
(15.6%) compared to other chronically ill (9.6%)
and well patients (4.0%). Among patients with
heart disease treatment, those with private nongroup insurance were the most likely to have
high total burdens (48.9%), followed by the
uninsured (26.4%), those with public insurance
(19.8%), and those with private group insurance
(11.4%).
Among heart disease patients with high total
burdens, 18.5% said they were unable to get
care and 15.5% said they delayed care due to
financial reasons. Financial barriers were
highest among the uninsured and those with
public coverage: 44.8% among the uninsured
and 27.8% among those with public coverage
said they were unable to get care due to
financial reasons.
Conclusions: High burdens may deter patients
from getting needed care.
Implications for Policy, Delivery, or Practice:
High cost of care for persons with heart disease
are associated with reduced access to care.
Addressing financial barriers to care may
improve treatment adherence among patients
with heart disease.
Funding Source(s): AHRQ
Poster Session and Number: A, #220
Marital Status, Household Living
Arrangements, and Men’s Use of Preventive
Healthcare Services
Stephen Blumberg, National Center for Health
Statistics; Joseph Blumberg, Men's Health and
Wellness Center, Atlanta, GA; Anjel Vahratian,
National Center for Health Statistics
Presenter: Stephen Blumberg, Associate
Director For Science, National Center for Health
Statistics
sblumberg@cdc.gov
Research Objective: Previous research has
demonstrated that married men are more likely
than unmarried men to seek preventive
healthcare services (e.g., Stimpson & Wilson,
Prev Chronic Dis 2009;6(2):A55). These
researchers have suggested that women may
play a role in men’s health care by directly
encouraging men to seek preventive care and/or
by indirectly evoking in men a sense of
economic and social obligation to the family.
However, neither explanation requires a
marriage: If women play a health-promoting
role, one might expect that unmarried men living
with women (especially women identified as
partners) would be as likely as married men to
seek preventive healthcare services. This study
uses nationally representative data to examine
the association of marital status and use of
clinical preventive services among men and
women, and it advances previous work by
dividing the not-married group into subgroups
based on household living arrangements.
Study Design: With data from the 2012
National Health Interview Survey, we compared
selected measures of health care access and
use for three groups of men aged 18-64: married
men living with their spouses, not-married men
living in the same household as an adult woman
(who may have been a partner, mother, sister,
daughter, or other relative or non-relative), and
not-married men not living with a woman.
Sensitivity analyses were conducted by
restricting the sample to specific age and health
insurance status groups and by comparing
estimates for not-married men living with
partners to not-married men living with other
women. Differences observed among men were
compared to differences observed among
women.
Population Studied: U.S. civilian
noninstitutionalized population aged 18-64
Principal Findings: Compared to married men
living with their spouses, not-married men living
with women were less likely to have a usual
place for routine/preventive care, an office visit
to a doctor or other health care professional in
the past 12 months, and selected clinical
preventive services in the past 12 months, such
as blood pressure checks, cholesterol
screening, and diabetes screening. Among notmarried men not living with women, the
percentage with each of these selected
healthcare measures was generally between
(and significantly different from) the percentages
for the other two groups. These differences
were observed regardless of age or insurance
status, and no significant differences in use of
clinical preventive services were found when
estimates for not-married men living with
partners were compared with estimates for notmarried men living with other women. Estimates
for women revealed a similar pattern, but
differences between married women living with
their spouses and not-married women living with
an adult man were less than half the size of the
differences observed among men.
Conclusions: Marital status is associated with
men’s use of clinical preventive services, even
when comparing married men to unmarried men
living with women. The strength of this
predisposing factor is substantially stronger for
men than for women.
Implications for Policy, Delivery, or Practice:
Unmarried men living with women are a group
particularly at risk of not receiving clinical
preventive services recommended by the U.S.
Preventive Services Task Force.
Funding Source(s): No Funding
Poster Session and Number: A, #221
Comparative Mortality Rates for Veterans of
U.S. Military Service in Iraq and Afghanistan
Mary Bollinger, Audie Murphy VAMC; Mary Jo
Pugh, Audie L. Murphy VAMC; Laurel Copeland,
Olin Teague VAMC; Jacqueline Pugh, Audie L.
Murphy VAMC; Helen Parsons, University of
Texas Health Science Center, San Antonio TX;
Susanne Schmidt, University of Texas Health
Science Center, San Antonio TX
Presenter: Mary Bollinger, Investigator, Audie
Murphy VAMC
bollinger@uthscsa.edu
Research Objective: The aim of this study was
to assess the patterns of all-cause mortality in
Afghanistan and Iraq veterans to determine
whether mortality patterns were consistent with
the Healthy Soldier Effect (HSE)observed in
previous military cohorts.
Study Design: Cohort design using the
OEF/OIF/OND VHA Roster File and the Defense
Manpower Data Center (DMDC) reporting
system (DRS)
Population Studied: Veterans with an existing
relationship to VHA and active duty military
involved in the OEF/OIF/OND mission between
2002-2011.
Principal Findings: The overall VHA
Standardized Mortality Ratio (SMR) of 2.8 [95%
confidence interval (CI) 2.8-2.9] while the VHA
clinical group had an SMR of 3.2 (95% CI 3.03.3) and the non-clinical VHA group was not
significantly different from the U.S. For the DOD,
the calculated SMR was 1.5 (95% CI 1.44-1.5).
A clear gradient is evident within age groups
with the SMR highest in the youngest age
groups (17-24) and lowest in the oldest age
group (40+). Even among the non-clinical VHA
population which is not significantly different
from the U.S. population, mortality is significantly
higher than expected in the youngest age group.
The directly standardized relative risk (DSRR)
shows higher than expected mortality across
VHA clinical and DOD populations and in all
race, sex, and age categories. Indirectly
standardizing mortality on military specific
characteristics with the DOD rates, we see the
lack of a Healthy Soldier Effect may be
explained by rank, service component, and
branch of service.
Conclusions: Our results demonstrate that
among OEF/OIF/OND Veterans, VHA clinical
care is associated with significantly higher
mortality than both DOD service alone and being
enrolled in VHA without clinical care. The higher
number of deaths may be attributable to the
types of injuries this cohort of Veterans received
during their service, extended and repeated
deployments, and/or a lack of health insurance
coverage making non-VA care unlikely. In sum,
the higher or equivalent mortality among
Veterans and DoD military personnel during the
study period does not support the HSE.
Implications for Policy, Delivery, or Practice:
The Afghanistan and Iraq Veteran cohort has
experienced deterioration in the military mortality
advantage. The consistent and persistent
mortality disadvantage for DOD and VHA clinical
populations highlights a striking pattern that
needs to be investigated in more detail. It
appears that the mortality experience of this
cohort of Veterans is much more complex than
first thought and would benefit from taking a
modeling approach that adjusts for covariates
such as time in service, SES, health status at
discharge, combat or in-theater exposure, and
specific mortality cause. Further, examining
specific cause of death might yield additional
information as increased mortality from external
causes is consistent with patterns of postwar
mortality observed in veterans of previous wars.
Funding Source(s): VA
Poster Session and Number: A, #222
Improving Health in an Ethnically Diverse
Catchment Area: Findings from a Community
Health Needs Assessment in New York City
Anne Bozack, New York Academy of Medicine;
Kalpana Bhandarkar, Maimonides Medical
Center; Linda Weiss, The New York Academy of
Medicine, Center for Evaluation and Applied
Research
Presenter: Anne Bozack, Project Director, New
York Academy of Medicine
abozack@nyam.org
Research Objective: Research Objective:
Under the Affordable Care Act, non-profit
hospitals are required to conduct community
health needs assessments, which offer the
potential to inform and catalyze efforts to
address priority health needs within hospital
service areas. Brooklyn-based Maimonides
Medical Center has the challenge of serving one
of the most ethnically diverse areas in the
country, with significant numbers of Orthodox
Jewish, Chinese, Latino, Russian, Arab, and
South Asian residents. The communities served
have a high number of limited English speakers,
families living in poverty, families that follow
strict cultural traditions, and uninsured residents.
Maimonides is conducting a community health
needs assessment, inclusive of quantitative
health indicators, key informant interviews, and
resident and provider focus groups. Although
quantitative analyses suggest relatively good
health, qualitative methods were used to
understand health concerns and service needs
within subpopulations. This presentation focuses
on the findings from key informant interviews.
Study Design: Study Design: Semi-structured
interviews were conducted with health care
providers, social service providers, religious
leaders, educators, elected officials, Maimonides
staff members, and community leaders.
Interview topics included community health
concerns, barriers and facilitators to good
health, medical and other service needs, and
recommendations for services and activities that
may benefit residents’ health. Interview
transcripts were coded using hierarchical codes
and analyzed using standard qualitative
techniques, including pre-identified themes and
themes emerging from the data.
Population Studied: Population Studied:
Twenty-eight interviews were conducted with 31
key informants. Respondents worked health
care, n=17; social services, n=7; faith-based
institutions, n=3; government, n=3; and
education, n=1. Respondents identified as white
or Orthodox Jewish, n=18; Asian, n=3; Arab,
n=3; black, n=2; Latino, n=1; and Pakistani, n=1.
Principal Findings: Principal Findings:
Respondents noted several health concerns that
bridged ethnic and socioeconomic divides,
including diabetes, heart disease, and obesity.
However, they also described health and
behavioral concerns within specific populations.
Smoking has been commonly accepted in
Chinese and Arab communities, which was
associated with high smoking rates. Isolation
and depression were common among older
adults, particularly those who were recent
immigrants or had limited English proficiency.
Access to and use of health care and behavioral
services also differed between populations due
to knowledge, stigma, and the availability of
culturally-appropriate services. Respondents
reported that recent immigrants were less likely
to be aware of the importance of receiving and
availability of preventive services. Stigma
around seeking treatment for health issues,
particularly cancer, was common in the Chinese
and Orthodox Jewish communities; stigma
around mental illness was common across
subpopulations. Linguistic challenges also
posed barriers to care because community
members commonly preferred to receive
services in their native language. Across
groups, there were economic and cultural
barriers to healthful behaviors, including healthy
eating and physical activity.
Conclusions: Conclusions: Findings will help
inform the development of services and
programs, both within and outside of the clinical
setting, particularly those targeting specific
cultural or ethnic communities.
Implications for Policy, Delivery, or Practice:
Implications for Policy or Practice: The use of
qualitative data in a larger community health
needs assessment proved critical to highlighting
the concerns of subpopulations, which may be
obscured by aggregate data available from
surveys.
Funding Source(s): CMS
Poster Session and Number: A, #223
Does Selection Bias affect Children’s
Medicaid Mental Health Care? Enrollment
and Transitions of Children in Child Welfare
Derek Brown, Washington University; Ramesh
Raghvana, Washington University in St. Louis;
Benjamin Allaire, RTI International
Presenter: Derek Brown, Assistant Professor,
Washington University
dereksbrown@wustl.edu
Research Objective: To study child enrollment
patterns in managed care (MC) for Medicaid
mental health services to identify potential
selection bias.
Study Design: We studied enrollment and
transitions patterns among Medicaid children by
using an innovative linkage of Medicaid claims
(MAX) from 36 states and survey data on
abused or neglected children in child welfare.
Panel and cross-sectional logistic were used to
analyze enrollment in MC vs. fee-for-service
(FFS) plans. Survival models were also used to
the trend toward greater managed care over a 4year period. Controls included demographics,
foster care/out-of-home placement, health
status, rural, child behavior checklist (CBCL),
child maltreatment dummies, and state and
year.
Population Studied: Children in child welfare
who are enrolled in Medicaid.
Principal Findings: Cross-sectional models
show that Blacks and older children were more
likely to be enrolled in Medicaid behavioral
health; neglected children and those with higher
mental health needs (CBCL) were less likely.
White children, those with higher CBCL,
neglected or physically abused, and those
placed outside the home were more likely to
receive FFS; blacks were less likely. The FFS
population was more stable (fewer disenrolled);
65% never changed plans in the 4-year period
compared to 47% in behavioral MC. Fewer
factors predicted individual transitions, but older
children were more likely to join behavioral MC
plans and those with fair/poor health were much
less likely.
Conclusions: Some evidence of favorable
selection (or retention) in Medicaid managed
care plans and adverse selection (or retention)
in FFS was observed. Prior research has shown
a strong linkage between the predictors in this
analysis and increased Medicaid expenditures.
Future research will need to assess whether MC
enrollment is driven by caseworker behavior or
MC plan behavior and the implications for state
Medicaid finance.
Implications for Policy, Delivery, or Practice:
As both Medicaid enrollment and Medicaid
managed care expand, these trends should be
monitored for their impact on child health
outcomes, disparities, and financing.
Funding Source(s): NIH
Poster Session and Number: A, #224
Variation in Adult Obesity across the US: Do
Dentists Matter?
Maureen Canavan, Yale University; Jessica
Holzer, Yale University; Elizabeth Bradley, Yale
University
Presenter: Maureen Canavan, Associate
Research Scientist, Yale University
maureen.canavan@yale.edu
Research Objective: Dentistry literature has
argued that the prevalence of dentists may be
linked to lower obesity because of dentists’ role
in health education and encouraging healthy
eating habits. However, previous studies
identifying community level factors linked with
obesity have not evaluated this relationship.
Thus, we sought to examine the association
between the rate of dentists in a county and the
percent of adults who were obese, using countylevel data for health care resources, measures
of the built environment, and sociodemographic
and economic factors.
Study Design: We conducted cross-sectional
multivariable regression using data from the
2013 Robert Wood Johnson Foundation County
Health Rankings and Roadmap program. This
dataset integrated national representative
county-level data for any county that had its own
Federal Information Processing Standard
(FIPS). Our primary dependent variable was
percent of adults who were obese (body mass
index of 30 or greater) within a county. We
included several independent variables: health
care resources, specifically the rates of Dentists
and PCPs within the county, measures of the
built environment, and sociodemographic and
economic factors.
Population Studied: Data were available for
the 3,141 counties across the US. We excluded
299 counties due to missing data for our
dependent and independent variables for a final
analytic sample of 2,842 counties (inclusion rate
of 90.5%).
Principal Findings: In multivariable analysis,
adjusting for state-level fixed effects, having an
additional dentist or PCP per 10,000 people was
associated with a 1.0% and 0.7% decrease in
the percent of adults who were obese,
respectively (p-values <0.001). The association
between prevalence of dentists and obesity
rates was significantly modified by income levels
(interaction p-value = 0.042), in which the
magnitude of the effect of dentists was
magnified in more impoverished counties.
Among more impoverished counties (where
more than 25% of children lived in poverty),
each additional dentist per 10,000 people was
associated with a nearly 1.2% decrease in the
percent of adults who were obese (p-value =
0.009) (Table 4), whereas among less
impoverished counties (where 25% or less of the
children lived in poverty), each additional dentist
per 10,000 people was associated with only a
0.6% decrease in the percent of adults who
were obese (p-value = 0.049).
Conclusions: This study is the first to find an
association between prevalence of dentists and
percent of adults who are obese at the county
level and that this relationship is moderated by
income. Additionally, the strength of the
association suggests that future attention should
be focused to parse out the full extent of this
relationship.
Implications for Policy, Delivery, or Practice:
Findings may be useful in identifying the role of
dentists in combatting the obesity epidemic
within the U.S. Specifically, there is a strong call
for a longitudinal analysis to establish if a causal
relationship exists. If the findings bear a causal
relationship, then dentists should be included as
a key resource in the prevention and
management of obesity.
Funding Source(s): No Funding
Poster Session and Number: A, #225
Measurement Bias on the Consumer
Assessment of Healthcare Providers and
Systems (CAHPS) Survey and Its Influence
on Disparities in Patients’ Experiences.
Adam Carle, Cincinnati Children's Hospital and
Medical Center; Constance Mara, Cincinnati
Children's Hospital Medical Center
Presenter: Adam Carle, Assistant Professor,
Cincinnati Children's Hospital and Medical
Center
adam.carle@cchmc.org
Research Objective: The Consumer
Assessment of Healthcare Providers and
Systems (CAHPS) survey serves as a widely
used patient reported measure of patients’
experiences with their healthcare providers and
systems. Research using CAHPS has
documented differences (sometimes dramatic)
in the experiences of African-Americans,
Hispanics, and Asians as compared to Whites.
Based on CAHPS scores, minority groups tend
to report receiving poorer care (e.g., less timely
care, less adequate staff helpfulness, etc.).
However, the possibility exists that
measurement bias may influence our
understanding of these disparities.
Measurement bias refers to the possibility that
two individuals with equivalent patient
experiences but different racial or ethnic
backgrounds may nevertheless respond
differently to questions about their healthcare
experiences. Measurement bias can mask the
true size of disparities, as well as lead to less
reliable results for one racial ethnic group as
compared to another. To date, studies have
examined measurement bias on the CAHPS
across Whites and African-Americans and
Whites and Spanish speakers. However, to
date, no studies have evaluated the presence of
influence of measurement bias on the CAHPS
for Asian-Americans. And, no studies have
simultaneously included multiple racial and
ethnic groups. In this study, we address these
limitations of previous research and examine
whether measurement bias on the CAHPS
impedes valid measurement across White,
African-American, Asian-American, and
Hispanic patients.
Study Design: We used multiple group (MG)
multiple indicator multiple cause (MIMIC)
structural equation models (SEM) to examine
measurement bias. This approach combines
and capitalizes on the strengths of item
response theory (IRT) and confirmatory factor
analysis (CFA), and structural equation models
(SEM).
Population Studied: We examined
measurement bias across non-Hispanic White (n
= 30,357), non-Hispanic African-Americans (n =
5,825), non-Hispanic Asian-Americans (n =
252)and Hispanic (n = 3,079) adults. Patients
were all veterans who received care at a VA
hospital in 2012.
Principal Findings: MG-MIMIC demonstrated
statistically significant measurement bias across
groups. We observed statistically significant
differences in the discrimination and threshold
(location) parameters. We observed significant
measurement bias for each minority group
compared to Whites. Importantly,sensitivity
analyses indicated that measurement bias
meaningfully influence conclusions about
average experiences with care across nonHispanic White, non-Hispanic African-American,
non-Hispanic Asian, and Hispanic patients in our
sample.
Conclusions: CAHPS does not provide
similarly reliable and valid measurement across
Whites, Hispanics, African-Americans, and
Asian-Americans. Previously documented racial
and ethnic group differences in health care
experiences as measured by the CAHPS reflect,
in part, measurement bias. Disparities are likely
larger than previously reported.
Implications for Policy, Delivery, or Practice:
At a minimum, failing to account for
measurement bias in CAHPS survey items will
likely result in spurious conclusions about
disparities in health care experiences between
Asian-Americans, African-Americans, Hispanics,
and their White counterparts. At the least,
CAHPS users should use statistical techniques
that adjust for measurement bias and estimate
scores free of measurement bias.
Funding Source(s): Other VA
Poster Session and Number: A, #226
Effect of Acculturation on Variation in Having
a Usual Source of Care in Asian American
Versus Non-Hispanic White Adults
Eva Chang, Group Health Research Institute
Presenter: Eva Chang, Postdoctoral Fellow,
Group Health Research Institute
chang.eva@ghc.org
Research Objective: Having a usual source of
care (USC) is an important measure of access
because of associations with better health
outcomes and utilization of healthcare services.
Little empirical research has been conducted on
how key resources affect having a USC for
Asian Americans. The objective of this study
was (1) to determine whether Asian American
adults were less likely to have a USC compared
to non-Hispanic white adults (NHWs) and (2) to
examine how acculturation and key predisposing
and enabling resources differentially influence
having a USC in NHW and Asian adults.
Study Design: Data were from the 2005 and
2009 California Health Interview Survey. Using a
modified Anderson model of access to
healthcare, multivariate logistic regression
models were built to compare the odds of having
a USC between NHW and Asian adults and to
examine race-specific associations with key
predisposing characteristics and enabling
resources (educational attainment, employment
status, insurance status, household income) and
acculturation factors (English proficiency,
duration in US, ethnically concordant
neighborhood). Models were survey-weighted
and adjusted for age, gender, marital status,
health status, household size, and survey year.
Population Studied: The total sample used had
38,555 NHW and 7,566 Asian American adults
(18-64 years). The weighted sample
represented almost 27 million individuals from
California in 2005 and 2009.
Principal Findings: Although race-related
disparities between Asian Americans and NHWs
in having a USC were no longer significant after
acculturation factors were added to the model,
Asian American had 23% lower odds of having a
USC than NHWs. Within insurance categories,
only uninsured Asian adults had significantly
lower odds of having a USC compared to Asian
adults with employment-based insurance while
most insurance categories were significant
among NHWs. Uninsured NHWs and Asians
had 85-89% lower odds of having a USC
compared to those with employment-based
insurance (both p<0.01). Unlike NHWs, higher
levels of education and household income were
not associated with better access among
Asians. Asians with high school through college
degrees had 41-53% lower odds of having a
USC compared to those with graduate degrees
(all p<0.05). A significant regressive trend by
income was observed among NHWs, but no
significant differences by income were observed
among Asians. Self-employed Asians had over
two times greater odds of having a USC than
other employed Asians. Low English proficiency
and short duration in the US (less than 5 years)
were significantly associated with having a USC
in both groups.
Conclusions: Significant differences exist in
having a USC between Asian American and
NHW adults. Acculturation factors are key
drivers of disparities and should be included in
access to care models with AAs. Key
predisposing characteristics and enabling
resources are differentially significant for Asian
American compared to NHW adults.
Implications for Policy, Delivery, or Practice:
Policies and assumptions regarding health
access for Asian Americans based on studies
using NHWs should be made cautiously.
Differences in barriers to having a USC suggest
that unique policy interventions targeting Asian
American adults are necessary to address
disparities.
Funding Source(s): AHRQ
Poster Session and Number: A, #227
Variations in Factors Influencing Having a
Usual Source of Care among Asian American
Ethnic Adults
Eva Chang, Group Health Research Institute
Presenter: Eva Chang, Postdoctoral Fellow,
Group Health Research Institute
chang.eva@ghc.org
Research Objective: The research objectives
were to determine whether the odds of having a
usual source of care (USC) will vary among
Asian American ethnic subgroups (Chinese,
Filipinos, Japanese, Koreans, Vietnamese, and
South Asians), and to examine whether factors
influencing the odds of having a USC vary by
Asian ethnic subgroup.
Study Design: Data were from 2005 and 2009
California Health Interview Survey. Using a
modified Anderson model of access, logistic
regressions and pair-wise comparisons were run
to compare the odds of having a USC among
Asian American adults and to examine ethnicityspecific associations with acculturation (English
proficiency, duration in US, ethnically
concordant neighborhood) and key enabling and
predisposing resources (education, employment,
insurance, household income). Models were
survey-weighted and adjusted for age, gender,
marital status, health status, household size,
and survey year.
Population Studied: The total sample used had
7,566 Asian American adults (18-64 years),
including 1,918 Chinese, 882 Filipino, 467
Japanese, 1,138 Korean, 1,552 Vietnamese,
740 South Asian, and 869 Other Asian adults.
The weighted sample represented over 6 million
individuals from California in 2005 and 2009.
Principal Findings: Significant differences in
the magnitude of the disparity and factors
influencing having a USC were found across
Asian American ethnicities. Koreans had 3549% lower adjusted odds of having a USC
compared with all other subgroups except
Japanese; Japanese had 52% lower odds of
having a USC than Chinese (all p<0.05). Among
all ethnic subgroups, uninsured adults had 8594% lower odds of having a USC (all p<0.01).
All other trends and associations with having a
USC varied by ethnicity. Education, employment
status, duration in US and neighborhood
concordance were significant for Chinese,
household income for Filipinos, English
proficiency, and duration in US for Japanese,
and education for South Asians. Among
Koreans and Vietnamese, lack of insurance was
the only key factor significantly associated with
having a USC.
Conclusions: Asian ethnic subgroups are
significantly different from each other in having a
USC. While insurance is a key factor associated
with having a USC for all subgroups, patterns of
associations varied widely by subgroup. Poor
access among Korean and Japanese adults
need to be further investigated.
Implications for Policy, Delivery, or Practice:
Persistent variation and heterogeneous
associations suggest that targeted, ethnicityspecific policies and outreach are needed to
improve having a USC for Asian American
ethnic adults. Lack of significant associations
may suggest that previous experiences and
cultural norms from home countries cause
immigrants to underutilize USC. Complementary
strategies to educate immigrants on the value of
having a USC may be needed.
Funding Source(s): AHRQ
Poster Session and Number: A, #228
Trust and the Community Health Worker:
Exploring the Attributes of CHW-Patient
Trust
Ji Eun Chang, New York University
Presenter: Ji Eun Chang, Doctoral Student,
New York University
ji.chang@nyu.edu
Research Objective: Build a conceptual model
describing the attributes and factors that form
the basis of a trusting relationship between
patients and community health workers (CHWs)
through exploratory fieldwork
Study Design: Data was collected over 6 weeks
by regularly shadowing one CHW working in the
family practice department in a large South
Bronx hospital system. Observations took place
in the CHW's office, four health clinics, the
hospital maternity unit, a homeless shelter, two
patient homes, in the CHW's car, on the hospital
bus, and walking around the neighborhood.
Informal interviews were conducted with the
CHW, the director of family practice, other
CHWs, doctors, nurses, administrative staff, and
former and current patients who interacted with
the CHW.
Data analysis was conducted in concert with
data collection through the use of analytic
memos. Coding was conducted in three stages
– open coding of three observational sessions,
axial coding based on initial set of codes, and
selective coding around the category of trust.
Codes within the category of trust were
systematically compared and contrasted to
develop the theory.
Population Studied: CHWs working with
vulnerable populations in a primary care setting.
Principal Findings: Trust is an essential part of
the relationship between the CHW and both
current and former patients. Contrary to popular
belief, trust between the CHW and patients was
not based on cultural matching. Instead, trust
was based on cultural empathy – the capacity
and willingness identify with the feelings,
thoughts, and behavior of individuals from
different cultural backgrounds. In addition, trust
was based on three interrelated factors: (1) fulltime availability – the patient’s knowledge of the
willingness of the CHW to make herself
available to the patient at all times (2)
Competence – technical and interpersonal skills
(3) control over resources – patient’s knowledge
that the CHW has access to critical resources
that can influence outcomes such as direct
access to a physician.
Conclusions: The research objective was
achieved and a theoretical model was
developed to explain the attributes and factors
that form the basis of trust between CHWs and
patients.
Implications for Policy, Delivery, or Practice:
CHWs have been gaining popularity in recent
years as a means to reduce racial and ethnic
health disparities in the U.S. This paper sheds
light on the mechanisms that may allow CHWs
to build the oft-emphasized trusting relationship
with patients. Furthermore, the model provides
an alternative to the difficult process of matching
CHWs on specific aspects of patient culture.
Culture is a complex, multi-faceted concept that
goes well beyond the visible social markers of
race, ethnicity, disease, or socio-economic
status. By focusing on core competencies and
cultural empathy rather than relatively
permanent identities (i.e. race and ethnicity),
program administrators will be able to recruit
and train from a wider pool of potential
candidates without the need to identify a specific
cultural dimension required for matching. This
is particularly useful in the context of widespread
policy and health systems changes where CHW
programs are being scaled up to reach a larger
and more heterogeneous population. This
model can inform the design of current and
upcoming interventions to ensure greater
effectiveness in the future.
Funding Source(s): No Funding
Poster Session and Number: A, #229
State Variation in Disproportionate Share
Hospital Payments and its Implications for
Medicaid Expansion in the United States
Paula Chatterjee, Harvard School of Public
Health; Nandini Sarma, Harvard School of
Public Health; Ashish Jha, Harvard School of
Public Health
Presenter: Paula Chatterjee, Medical Student,
Harvard School of Public Health
paulachatterjee@gmail.com
Research Objective: There is a high degree of
variation in the allocation of disproportionate
share hospital (DSH) payments to hospitals that
provide uncompensated care. Understanding
this state-level variation will be critically
important, given the impending cuts to DSH
payments as a means to offset revenue from
Medicaid expansions. We examined the
distribution of DSH payments in the United
States, focusing on differences between DSH
hospitals in states expected to expand versus
not expand Medicaid.
Study Design: We identified all hospitals
receiving federally matched Medicaid DSH
payments in 2008. We then calculated the
proportion of hospitals within each state that
received DSH payments. Using data culled from
state legislative and executive sources, states
were divided according to their likelihood of
expanding Medicaid as specified in the
Affordable Care Act. We then compared
characteristics of hospitals receiving DSH
payments in both groups of states and how it
compares to the amount of uncompensated care
provided.
Population Studied: U.S. acute-care hospitals.
Principal Findings: We identified 4,503 acute
care hospitals in the U.S. of which 2,229
received Medicaid DSH payments in 2008. The
proportion of hospitals receiving DSH payments
within a state ranged from 0 to 100 percent, with
a mean of 53 percent and median of 52 percent.
The rates of receiving DSH payments was
comparable in the 25 states (and District of
Columbia) where Medicaid expansion is
occurring (52% of hospitals) compared to states
where there is no Medicaid expansion (51% of
hospitals, p-value for difference 0.54).
Approximately 59% of current DSH payments
are given to hospitals in expanding states.
Compared to DSH hospitals in states that do not
intend to expand Medicaid, DSH hospitals in
expanding states were more likely to be large
hospitals (16% of DSH hospitals in states
intending to expand vs. 12% of DSH hospitals in
states without intention to expand, p<0.001) and
of non-profit status (76% vs. 50%, p<0.001).
Conclusions: We found large variations in the
distribution of Medicaid DSH payments, both
within and between states. DSH hospitals in
states intending to expand Medicaid
fundamentally differ from DSH hospitals in
states that do not intend to expand. A majority of
the current DSH payments go to states that are
expanding Medicaid.
Implications for Policy, Delivery, or Practice:
While there is substantial concerns about cutting
DSH payments to hospitals that do not expand,
we find that a majority of current DSH payments
are occurring in states that plan to expand
Medicaid.
Funding Source(s): No Funding
Poster Session and Number: A, #230
Residential Segregation and 30-day Hospital
Readmission of Diabetic Medicare Home
Healthcare Beneficiaries
Hsueh-fen Chen, University of North Texas
Health Science Center; Sharon Homan,
University of North Texas Health Science Center
Presenter: Hsueh-fen Chen, Assistant
Professor, University of North Texas Health
Science Center
hsueh-fen.chen@unthsc.edu
Research Objective: To examine the
association between residential segregation and
30-day readmission of diabetic Medicare home
healthcare beneficiaries developing ambulatory
care-sensitive conditions (ACSCs) postdischarge.
Study Design: The study was a cross-sectional
study design. Several 2009 national data were
assembled, including Medicare Beneficiary
Summary File, Medicare Provider Analysis
Review file, Outcome Assessment Information
Set, Standard Analytical File, Dartmouth Primary
Care Service Area Project, American Hospital
Association annual survey, Provider of Services,
and Area Resources File. The dependent
variables are a dummy variable that represents
whether patients experienced 30-day
readmissions due to acute or chronic ACSCs.
There are three key independent variables:
black people, neighborhood composition defined
as black neighborhoods and integrated
neighborhoods, and the interactions of black
people and neighborhood composition. White
people and white neighborhoods were in the
reference groups. Black and white
neighborhoods were defined as neighborhoods
with 65% and above of blacks and whites,
respectively at the zip code level. The
integrated neighborhoods were defined as
neighborhoods that had at least 35% each of
whites and blacks population at the zip code
level, Multivariate logistic regression models
were used for attaining the research objectives.
Population Studied: The study sample was
Medicare beneficiaries with diabetes-related
conditions who received post-acute care from
home health agencies within 14 days of hospital
discharge. The study sample was further
restricted to those who were black or white and
lived in white, black or integrated
neighborhoods.
Principal Findings: There were 626 black
neighborhoods, 15,998 white neighborhoods,
and 1,086 integrated neighborhoods. 885 whites
and 5,797 blacks lived in black neighborhoods.
75,498 whites and 5,297 blacks lived in white
neighborhoods, and 3,497 blacks and 3,383
whites lived in integrated neighborhoods. Black
beneficiaries residing in black neighborhoods
were less likely to be readmitted for acute
ACSCs than white beneficiaries in black or white
neighborhoods. However, black beneficiaries in
black and integrated neighborhoods were more
likely to be readmitted for chronic ACSCs.
Conclusions: Living in a black or segregated
neighborhood may protect against readmission
for acute ACSCs among elderly black
beneficiaries; however, black-white race
disparities in hospital readmission rate for
chronic ACSCs persist, regardless of
neighborhood racial composition. Efforts to
address the disparity of care should consider
types of quality indicators as well as the
characteristics of patient and community.
Implications for Policy, Delivery, or Practice:
For clinical implication, the coordinated care
between hospitals and home health should take
patient risk factors and patient’s neighborhood
characteristics into account. Furthermore,
existing literature indicates that the prevalence
of ACSCs is higher in minorities than in whites.
Future studies for disparity of care for ACSCs
should consider acute and chronic ACSCs
separately. Finally, for policy implication, studies
that examine whether conducting transitional
care for minorities who lived in minority
neighborhoods is more costly than those for
whites who lived in white neighborhoods are
needed, which would provide useful information
for the policymakers to conduct fair payment for
safety-net hospitals with a large proportion of
minority patients who live in minority
neighborhoods under the ongoing health care
reform in the U.S.
Funding Source(s): Other UNTHSC
Poster Session and Number: A, #231
Does the 340B Program Reach Communities
in Need?
Chia-hung Chou, University of Chicago; Bobby
Clark, Walgreen Co.; John Hou, Walgreen Co.;
Elbert Huang, University of Chicago; Rena
Conti, University of Chicago
Presenter: Chia-hung Chou, Research
Assistant Professor, University of Chicago
chchou@medicine.bsd.uchicago.edu
Research Objective: Since 1993, the 340B
program has provided deep acquisition cost
discounts for outpatient prescription drugs for
qualified healthcare organizations (i.e., covered
entities) in the U.S. A 2009 regulatory change
expanded the program’s reach to include
dispensing through qualified contract
pharmacies. It is of significant national policy
debate whether and how current covered
entities serve the original mission of the program
to (1) serve the nation’s most vulnerable
populations and (2) provide comprehensive
services. This study is the first
contemporaneous, nationwide examination of
whether 340B qualified covered entities and
contract pharmacies serve poor, uninsured and
medically needy populations.
Study Design: In this cross-sectional study, we
used the publicly available 340B database
maintained by the Health Resources and
Service Administration (HRSA) to identify 340B
covered entities and contract pharmacies active
through December 2012. All 340B covered
entities and contract pharmacies were identified
by name, city, state, zip code, and year using
HRSA’s database. Demographic, economic, and
health insurance population characteristics were
obtained from the Census Bureau (2012
American Community Survey) for each primary
care service area (PCSA). A PCSA, defined by
Dartmouth Atlas of Health Care, is considered
the smallest discrete service area for primary
care. We follow the Census Bureau definition of
vulnerable populations as those
disproportionately comprised of the elderly,
minorities, the uninsured and individuals living in
poor households. We used chi-square tests to
compare these population-adjusted vulnerability
measures at the PCSA level served by covered
entities and contract pharmacies in 2012 to
those not served by the 340B program. All
analyses were performed in SAS 9.3.
Population Studied: 340B covered entities and
contract pharmacies at each zip code were
aggregated to the PCSA level (n=6,527) and
then matched to Census Bureau population data
(>99% match, all measures).
Principal Findings: When comparing PCSAs
with and without 340B covered entities, we
found 340B-covered PCSAs had significantly
higher percentages of vulnerable populations (all
measures, p<0.0001). Comparisons between
PCSAs with and without 340B contract
pharmacies showed similar results. Of 6,527
PCSAs, 66% (n=4,314) did not have covered
entities; 49% (n=3,205) did not have contract
pharmacies. Among no-340B-covered PCSAs,
many (ranging from 5.1% to 34.9% for all
measures) had concentrations of vulnerable
populations above the national average.
Conclusions: Our findings strongly suggest
currently 340B qualified covered entities and
contract pharmacies serve vulnerable
populations. Our results also suggest 340B
qualified contract pharmacies may have broader
geographical reach to serve vulnerable
populations than covered entities do. However,
the majority of primary service areas have
significant vulnerable populations without 340B
coverage by qualified healthcare organizations.
Implications for Policy, Delivery, or Practice:
This study provides the first national,
contemporaneous evidence to support the
contention that the current 340B program,
including recent expansions to include contract
pharmacies, likely serves its intended patient
population. Whether and how the program
ensures access to needed comprehensive
medical services is unknown. This information is
critical for policy makers to evaluate whether
and how the 340B program could be expanded
to serve vulnerable populations without current
340B coverage.
Funding Source(s): No Funding
Poster Session and Number: A, #232
National Outpatient Prescription Dispensing
Patterns through Contract Pharmacies
Serving the 340B Drug Discount Program in
2012
Bobby Clark, Walgreen Co.; John Hou,
Walgreen Co.; Chia-Hung Chou, University of
Chicago; Elbert Haung, University of Chicago;
Rena Conti, University of Chicago
Presenter: Bobby Clark, Senior Director,
Walgreen Co.
bobby.clark@walgreens.com
Research Objective: Since 1993, the 340B
program has provided deep acquisition cost
discounts for outpatient prescription drugs
dispensed by qualified healthcare organizations.
A 2009 Health Resources and Service
Administration (HRSA) regulatory change
expanded the program’s reach to include
contract pharmacies. It is of significant current
policy debate how qualified contract pharmacies
serve the program’s original mission to provide
access to comprehensive medical services
among medically needy populations. This study
is the first to describe national,
contemporaneous 340B qualified prescriptions
compared to all prescriptions dispensed in the
contract pharmacy setting.
Study Design: This is a retrospective repeated
cross-sectional study. The unit of analysis was
2012 prescriptions dispensed at Walgreens, a
large national retail chain and the market leader
in 340B qualified pharmacies. Drugs were
categorized by 340B qualification based on a
variable in Walgreen’s claims data required by
HRSA regulation. Drugs were categorized into
Anatomic Therapeutic Class and by primary
indication by the Medispan™ system and by
specialty drugs if they were considered to be
high value, high-touch and/or complex drugs.
Chi-square tests were used to test for statistical
differences between 340B prescriptions
compared to all prescriptions. Statistical
analyses were done in SAS 9.3.
Population Studied: We included all dispensed
prescriptions from patients of all ages, sexes
and insurance coverage in US states, the
District of Columbia and Puerto Rico.
Principal Findings: Approximately 500 million
prescriptions were identified for this analysis;
340B prescriptions amounted to less than onehalf percent of the total. By volume, the same six
therapeutic classes were ranked among the top
ten for 340B and all prescriptions: Antivirals,
antiasthmatics, antidabetics,
antihyperlipidemics, antihypertensives, and
antidepressants. The rankings by prescription
volume percentage differed between the two
groups; antivirals accounted for 10.6% of 340B
prescriptions compared to 0.9% of all
prescriptions (p<0.0001), antiasthmatics
accounted for 9.7% versus 3.2% (p<0.0001),
and antidiabetics accounted for 6.9% versus
3.6% (p<0.001). Specialty medications
accounted for 9% of 340B prescriptions and
0.4% of all prescriptions (p<0.0001). Generic
drugs accounted for 54% of 340B prescriptions
and 82% of all prescriptions (p<0.0001).
Conclusions: 340B prescriptions represent a
very small percentage of all prescriptions
dispensed by the market leader in 340B
qualified contract pharmacies. Drugs used to
treat chronic diseases, including hypertension,
cholesterol disorders, diabetes, and asthma
dominate all and 340B prescriptions. However,
antiviral drugs used to treat HIV/AIDs and
hepatitis B, antiasthmatic drugs, and antidiabetic
drugs are much more commonly dispensed
among 340B prescriptions. 340B prescriptions
also exhibit a much higher specialty drug
percentage and a much lower generic
percentage compared to all prescriptions.
Implications for Policy, Delivery, or Practice:
Our results strongly suggest contract
pharmacies dispense drugs that serve the
general chronic disease burden of the US
population. Among 340B prescriptions, contract
pharmacies also appear to disproportionately
serve one vulnerable population targeted by the
340B program, individuals suffering from
HIV/AIDS. Research is needed to better
understand the rationales underlying the low
generic percentage and high specialty
percentage of 340B prescriptions. This
information is critical for policy makers
considering how best to generate future program
savings.
Funding Source(s): No Funding
Poster Session and Number: A, #233
Medicaid Expansion and Disproportionate
Share Hospital Payments: Which Hospitals
are Financially Vulnerable?
Evan Cole, Georgia Health Policy Center; Daniel
Walker, Tulane University; Arthur Mora, Tulane
University; Mark Diana, Tulane University
Presenter: Evan Cole, Associate Project
Director, Georgia Health Policy Center
ecole@gsu.edu
Research Objective: The Affordable Care Act
reduces Medicaid Disproportionate Share
Hospital (DSH) payments that partially offset
uncompensated care costs incurred by hospitals
that treat uninsured and Medicaid populations by
$18.1 billion between FY 2016 and 2022.
However, the decision of several states not to
expand their Medicaid programs combined with
residual coverage gaps may leave as many as
30 million individuals uninsured, placing the
burden of uncompensated care costs on
hospitals. This analysis aims to identify those
hospitals that may be the most financially
vulnerable to Medicaid DSH payment
reductions.
Study Design: A descriptive analysis with
statistical testing of proportions of categorized
hospitals. The dataset used to measure the
financial condition of hospitals was the 2011
American Hospital Association (AHA) Annual
Survey and the 2011 Healthcare Cost Report
Information System. Data on Medicaid DSH
eligibility and payment amounts were obtained
from the 2008 Annual DSH report. Hospitals
were categorized by four variables that effect
their Medicaid DSH allotment and level of
uncompensated care provision: 1) Whether the
hospital resides in a high or low DSH state; 2)
whether the hospital resides in a Medicaid
expansion or non-expansion state; 3) how reliant
the hospital is on Medicaid DSH payments as a
source of revenue; and 4) whether the hospital
is considered to be a High Medicaid Volume or
High Low-Income Uncompensated Care
hospital. Systematic differences in the operating
margins of hospitals in each category were then
tested for as an indicator of financial condition.
Population Studied: Medicaid DSH eligible
hospitals in 45 states.
Principal Findings: Of the 4,407 acute care
hospitals in the AHA dataset, we identified 2,104
that were Medicaid DSH eligible. A significantly
smaller proportion of Medicaid DSH eligible
hospitals in states expanding their Medicaid
programs are in a strong (19.2% vs. 23.6%; p <
0.05) or weak financial condition (22.5% vs.
29.3%; p < 0.01) than those hospitals in nonexpansion states. 1,094 hospitals were
categorized into groups that will be particularly
affected by Medicaid DSH cuts, 357 of which are
in a weak financial condition, with operating
margins less than -8.00. Among hospitals in
these most vulnerable groups, the proportion
with weak financial conditions was similar
between expansion and non-expansion states.
Conclusions: The financial condition of
hospitals that receive Medicaid DSH payments
varies considerably. Accordantly, the impact of
Medicaid DSH reductions will vary as well and
will be greatly influenced by the decisions of
state policymakers on whether to expand their
Medicaid programs and how to distribute their
state’s Medicaid DSH funds under this new
policy.
Implications for Policy, Delivery, or Practice:
The chief implication to monitor is whether the
1,094 hospitals with the greatest sensitivity to
Medicaid DSH cuts will significantly reduce their
provision of uncompensated care or other
services that would adversely affect vulnerable
populations. State policymakers can either
choose to distribute the Medicaid DSH cuts
equally or in a targeted manner, with varying
implications. The financial distribution of
Medicaid DSH hospitals may systematically
change between expansion and non-expansion
states, and fewer hospitals that rely on DSH
payments may be able to achieve a strong
financial condition.
Funding Source(s): No Funding
Poster Session and Number: A, #234
Minnesota Health Systems’ Engagement with
Health Equity
Brooke Cunningham, Medica Research Institute;
Pamela Jo Johnson, Medica Research Institute;
Todd Rockwood, University of Minnesota
Presenter: Brooke Cunningham, Postdoctoral
Fellow, Medica Research Institute
brooke.cunningham@medica.com
Research Objective: Although Minnesota ranks
high nationally for overall health care quality, it
also has high levels of health disparities. This
study uses survey methods to examine the
perceptions of Minnesota health care leaders
regarding the engagement of health care
personnel and delivery systems with health
equity.
Study Design: One week after a health equity
forum held in St. Paul, Minnesota, invitations to
complete a web-based survey were sent via
email to those who had registered for the
conference. Respondents were asked to rate
senior leaders, mid-level managers, and frontline provider and staff’s awareness of health
inequities, engagement in addressing health
disparities, and sense of “safety” discussing
racial and ethnic disparities. They were also
asked to rate the factors that contribute to health
disparities and the effectiveness of their
organizations in addressing those factors.
Paired t-tests were used to compare how
respondents’ rated employee groups and their
perceptions of organizational engagement with
the different causes of health disparities.
Population Studied: Six health systems based
in Minneapolis, St. Paul, and Rochester,
Minnesota, invited senior leaders to attend a
health equity forum in the summer of 2013. As
part of the program evaluation, invitations to
complete a web-based survey were sent to
individuals who registered for the conference.
Principal Findings: The response rate was
41% (n=37). Most respondents were white
(89%), female (59%), had graduate training
(86%), and attended the health equity forum
(95%). Respondents perceived senior leaders to
be significantly more informed about health
equity, more engaged with addressing
disparities, and to feel safer discussing
racial/ethnic disparities than mid-level managers
and front-line providers and staff. Regarding
health system priorities, addressing disparities
ranked last, behind 1) reducing medical error; 2)
reducing costs/improving customer service; and
3) improving performance measures.
Respondents perceived social determinants to
be a significantly larger contributor to disparities
than any other factor (i.e., individual behavior,
health care systems, health insurance, and
access barriers). However, compared to these
other factors, they reported health systems to be
least effective at addressing social determinants
of health.
Conclusions: Findings suggest that leaders in
health care delivery systems perceive varied
levels of awareness and engagement around
health disparities in their organizations.
Respondents perceived front-line providers and
staff to have the lowest awareness of health
inequities, engagement with health disparities,
and safety discussing racial and ethnic health
care disparities. Social determinants of health
were considered the greatest contributor to
disparities but were least well addressed by
health care systems.
Implications for Policy, Delivery, or Practice:
Recent health policies seek to better promote
health equity and improve population health.
Given these goals, the high engagement of
health leaders with health equity is encouraging.
Leaders may need to develop additional
strategies to promote health equity with midlevel managers and front-line providers and
staff. Future research should explore the
perceptions of workers across all occupational
tiers; examine organizational factors associated
with health care personnel’s engagement with
health equity; and investigate how the ability of
health systems to address social determinants
of health could be improved.
Funding Source(s): No Funding
Poster Session and Number: A, #235
Association between Access to
Neurosurgical Care and Insurance Status on
Emergency Department Interfacility Transfer
for Severe Head Injury
M. Kit Delgado, University of Pennsylvania;
Daniel Holena, University of Pennsylvania;
Douglas Wiebe, University of Pennsylvania;
Brendan Carr, University of Pennsylvania
Presenter: M. Kit Delgado, Emergency Care
Research Scholar, University of Pennsylvania
mucio.kitdelgado@gmail.com
Research Objective: Treatment of severely
injured patients at trauma centers is associated
with improved survival. Uninsured patients with
trauma are more likely to be transferred to
trauma centers. We sought to determine
whether this association exists among a subset
of patients with isolated severe head injuries; the
importance of facility neurosurgical capabilities
was also explored.
Study Design: We performed a retrospective
analysis of the 2010 Nationwide Emergency
Department Sample. We used sample weighted
logistic regression to estimate differences in
transfer rates (vs. admission) by insurance
status, while adjusting for age, sex, mechanism
of injury, Injury Severity Score, weekend
admission, urbanicity, income, as well as
teaching status and US region. We considered a
non-trauma center to have neurosurgical
capabilities if it performed inpatient
neurosurgical procedures for traumatic brain
injury in patients admitted through the ED. We
examined the association between transfer rate
and insurance status, as well as the interaction
between insurance status and neurosurgical
capabilities.
Population Studied: We included all ED
encounters for major trauma (Injury Severity
Score >15) whose only severe injury
(Abbreviated Injury Scale>=3) was limited to the
head region in patients aged 18-64 years seen
in non-trauma centers. We excluded ED
discharges and ED deaths.
Principal Findings: There were 4,167
observations for analysis from 604 hospitals
representing a nationally weighted population of
17,672 non-trauma center ED encounters for
isolated severe head trauma (61% of all major
trauma encounters seen in non-trauma center
EDs); 47% presented to non-trauma center EDs
with neurosurgery capabilities. In non-trauma
centers with neurosurgical capabilities, 23% of
patients were transferred overall, and the
adjusted transfer rate was significantly higher for
uninsured patients compared to privately insured
patients (31% [95% CI: 23-39%] vs. 21% [95%
CI: 14-28%]). Transfer rates were
approximately 3-4 times higher overall in nontrauma centers without neurosurgical
capabilities, but the difference in transfer rates
between the uninsured and privately insured
was attenuated (86% [95% CI: 80-91%] vs. 82%
[95% CI: 78-87%]).
Conclusions: ED interfacility transfers for
patients presenting to non-trauma centers is
primarily driven by the availability of inpatient
neurosurgical care. However, in non-trauma
centers that have neurosurgical care, the
uninsured are transferred out at higher rates
than privately insured patients.
Implications for Policy, Delivery, or Practice:
The Emergency Medical Treatment and Labor
Act (EMTALA) requires EDs to stabilize all
patients regardless of ability to pay, and
hospitals must provide specialist care or arrange
transfer to a tertiary care center when specialist
care is unavailable. These data suggest
hospitals may be applying EMTALA selectively
to justify the transfer out of uninsured patients
when needed acute specialty care is available.
Paradoxically, being insured may be a barrier to
receiving optimized specialty care via transfer to
a designated trauma center. Given the nature of
severe head trauma, it is unlikely that patient
preferences account for these transfer
disparities. Efforts in monitoring and optimizing
trauma interfacility transfers and outcomes at
the population level are warranted.
Reimbursement policies aimed at counteracting
the financial disincentive to transfer out critically
injured patients with insurance may reduce
disparities in trauma regionalization.
Funding Source(s): NIH
Poster Session and Number: A, #236
Reducing Health Disparities: The Role of
Medical Education
Keren Dopelt, Ben Gurion University; Zehava
Yahav, Ben Gurion University; Jacob Urkin, Ben
Gurion University; Yaacov Bachner, Ben Gurion
University; Nadav Davidovitch, Ben Gurion
University
Presenter: Keren Dopelt, Lecturer, Ben Gurion
University
dopelt@bgu.ac.il
Research Objective: 1. To compare the
community orientation as reflected in the various
medical schools' curricula and the rates of
graduates working in the periphery.
2. To compare the social and community
orientation of graduates from the various
medical-schools.
Study Design: Online cross-sectional survey
conducted during May-June 2011.
The survey included questions dealing with
community orientation, social involvement,
specialty, dominant approach in medical
curriculum as perceived by graduates,
graduates' perception of the impact of medical
education on their social involvement etc.
The questionnaire was developed by the
researchers and underwent a pilot study with 7
experienced physicians (face validity).
Data was analyzed using uni-variate and multivariate analysis and Sobel Test.
Population Studied: 9,000 physicians who
graduated from 4 medical schools in Israel
(Hebrew University, Tel Aviv University,
Technion and Ben Gurion University.
Principal Findings: About 37% of the
Physicians reported that they were involved in
community programs. Physicians who are Israeli
born, working in the periphery, working in
primary health care, active in professional
positions conducting research, and those who
hold the perceptions regarding the role of the
physician to work towards the reduction of
health disparities are more likely to be socially
involved (C-statistic=0.72, p<0.001). Moreover,
the perceptions regarding the role of the
physician to work towards the reduction of
health disparities partially mediate the
relationship between medical school social
orientation and physicians' social involvement.
The relationship between the orientation during
studies and the social involvement of the
physician is reinforced when the physician
maintains the attitude that it is his/her role to
work towards the reduction of health disparities,
an attitude that by itself is influenced by a social
orientation of his/her studies.
Conclusions: This study shows the different
factors that can predict social and community
involvement among physicians in Israel. As
physicians' perception regarding their role in
reducing health disparities was found as an
important factor, it emphasizes the important
role of medical education in shaping physicians'
attitudes toward working in the periphery and
being more involved in community programs.
Implications for Policy, Delivery, or Practice:
The study findings are crucial when considering
changes in the medical schools' curriculum and
implementing a long-term national plan for
reducing health disparities. Furthermore,
Socially-oriented medical education has the
potential to induce a socialization process
reinforcing human values regarding doctor–
patient relationships and to produce positive
attitudes among future doctors about social
involvement.
Funding Source(s): N/A
Poster Session and Number: A, #237
Understanding the Effect of Insurance
Expansion on Utilization of Inpatient Surgery
Chandy Ellimoottil, University of Michigan; Sarah
Miller, University of Michigan; John Ayanian,
University of Michigan; David Miller, University
of Michigan
Presenter: Chandy Ellimoottil, Research Fellow,
University of Michigan
cellimoottil@gmail.com
Research Objective: Using Massachusetts
(MA) healthcare reform as a natural experiment,
we estimate the differential impact of insurance
expansion on the utilization of discretionary
versus non-discretionary inpatient surgery.
Study Design: We used the State Inpatient
Databases from MA and two control states to
identify non-elderly patients (19-64 years) who
underwent discretionary (DS) versus nondiscretionary surgery (NDS) during the years
2003-2010. We defined DS as elective,
preference-sensitive procedures (e.g., joint
replacement, back surgery), and NDS as
imperative and potentially life-saving procedures
(e.g., cancer surgery, hip fracture repair). Using
July 2007 as the transition point between pre
and post-reform periods, we performed a
difference-in-differences (DID) analysis to
estimate the effect of insurance expansion on
rates of DS vs NDS among the entire study
population, and for subgroups defined by
race/ethnicity, income and insurance status. We
then extrapolated our results from MA to the
entire US population.
Population Studied: All non-elderly patients
(19-64 years) who underwent surgery in
Massachusetts, New York and New Jersey.
Principal Findings: We identified a total of
836,311 surgeries during the study period. In
contrast to NDS, post-reform rates of DS
increased more in MA than in control states.
Based on our DID analysis, insurance expansion
was associated with a 9.3% increase in the use
of DS in MA (p=0.021). Conversely, the rate of
NDS decreased by 4.5% (p=0.009). We found
similar effects for DS in all subgroups, with the
greatest increase observed for African American
and Hispanic patients (19.9%, p<0.001). We
observed a 6.7% (p=0.004) and 10.6%
(p=0.017) increase in DS for patients from that
resided in counties with low income and high
numbers of newly insured, respectively. Based
on the findings in MA, we estimated that full
implementation of national insurance expansion
would yield an additional 465,934 discretionary
surgeries by 2017.
Conclusions: Insurance expansion in
Massachusetts was associated with increased
rates of discretionary surgery, and a concurrent
decrease in utilization of non-discretionary
surgery. The largest increase in discretionary
surgery rates were seen in African American
and Hispanic populations.
Implications for Policy, Delivery, or Practice:
If similar changes are seen nationally, the value
of insurance expansion for surgical care may
depend on the relative balance between
increased expenditures and potential health
benefits of greater access to elective inpatient
procedures.
Funding Source(s): AHRQ
Poster Session and Number: A, #238
Hepatitis C Infection among Hispanics in
California
Dennis Fisher, California State University Long
Beach; Catherine Cummins, California State
University Long Beach; Grace Reynolds,
California State University Long Beach; Erlyana
Erlyana, California State University Long Beach
Presenter: Erlyana Erlyana, Assistant
Professor, California State University, Long
Beach
erlyana.c@gmail.com
Research Objective: This paper aims to
investigate risk factors associated with Hepatitis
C infection among Hispanics and White-Not
Hispanics in California. Hispanics in California
are significantly more likely to be infected with
hepatitis C (HCV) than other ethnicities. Those
Hispanics who have been infected have had
their infection detected later, and they have
been less likely to be linked to treatment due to
reduced access to healthcare. A paucity of
clinical evidence exists regarding the
relationship between ethnicity, risk behavior
assessment, methadone use and HCV
prevalence.
Study Design: The basic research question
compared Hispanics with HCV infection to
Whites with HCV infection. Included was a
comparison in which HCV-positive Hispanics
were compared to HCV-negative Hispanics and
HCV-positive Whites were compared to HCVnegative Whites on bivariate tests of candidate
variables. Following Hosmer and Lemeshow,
those candidate variables that were significantly
different between the two groups (for each
ethnicity) were then considered for inclusion in
separate multivariate logistic regression models
to identify both risk and protective factors
separately for Hispanics and Whites. Both
bivariate and logistic regression analyses were
conducted using SAS 9.3.
Population Studied: The data were collected
from 573 individuals of Hispanic ethnicity, and
967 individuals of White-Not Hispanic ethnicity,
who were tested for antibodies to HCV from
August 31, 2000 through September 22, 2013 at
the Center for Behavioral Research and
Services (CBRS) which is an HIV/STD testing
site for Service Planning Area 8 (South Bay) of
Los Angeles County, CA. The Risk Behavior
Assessment that collects demographic
information and identifiable risk factors
associated with Hepatitis C infection was
administered by trained interviewers.
Principal Findings: Bivariate results show that
Hispanics who are HCV infected are more likely
to use illicit methadone more days, have been
on methadone detoxification or maintenance,
and incarcerated longer than other ethnicities
infected with HCV. White-Not Hispanics who
are HCV infected are more likely to be
homeless, give drugs to get sex and have
unprofessional tattoos more than their Hispanics
counterparts. In a comparison of multivariate
logistic regression models, Hispanics who were
HCV infected were more likely to use crack,
heroin, or speedball. Among White-Not
Hispanics, only those who use heroin or
speedball were more likely to be HCV-infected.
Being homeless and not having a paid job or
salary as their source of income were significant
predictors among White-Not Hispanics, but
those factors were not significantly associated
with HCV infection among Hispanics. Both
Hispanic and White-Not Hispanics were more
likely to have been in a prior methadone
detoxification or maintenance program with
Hispanics being twice as likely as their WhiteNot Hispanic counterparts.
Conclusions: Hispanics are the largest minority
in the USA and have high rates of HCV
infection. They are in urgent need of intervention
strategies.
Implications for Policy, Delivery, or Practice:
It is important to provide more information that
explains prevention, detection, and treatment of
HCV. Additionally, HCV testing and treatment
should be linked to methadone treatment as well
as for those who are incarcerated.
Funding Source(s): No Funding
Poster Session and Number: A, #239
Gaps in Health Needs and Services: A Zip
Code Analysis
Erlyana Erlyana, California State University,
Long Beach; Veronica Acosta-Deprez, California
State University Long Beach; Tony Sinay,
California State University Long Beach
Presenter: Erlyana Erlyana, Assistant
Professor, California State University, Long
Beach
erlyana.c@gmail.com
Research Objective: This paper aims to
determine major health issues, accessibility and
gaps in health services in the greater Long
Beach areas. The assessment provides
opportunity to assess area of perceived needs in
the community and to focus scarce resources to
the most vulnerable areas. Analysis was
undertaken to compare the most and less
vulnerable areas, designated based on their
community need index (CNI). CNI aggregates
five socioeconomic variables by zip codes.
Study Design: The data were collected through
a self-administered survey that were distributed
to a convenience sample at community forums,
events and health fairs within the city of Long
Beach from July through November, 2010. The
survey instrument was developed through an
iterative process and consisted of twenty-seven
questions covering topics such as; population
demographics, health concerns affecting adults,
teens and children and access to services and
providers. Zip code analysis was undertaken to
determine specific areas where there are gaps
in health services, as perceived by the
participants and community organizations and
hospitals.
Population Studied: Respondents were
resident of the greater Long Beach areas. Of the
481 respondents, 422 were included in the
analysis. About half (47%) was non-Hispanic
whites and Hispanic or Latino. Of those
identifying as Hispanic, African American or
Asian, nearly half live in areas deemed those
with the highest need.
Principal Findings: Upon stratification by zip
code, child abuse, gang activities, alcohol and
drug use were recognized as the top issues
living in areas of the highest need. Uninsured
population is significantly larger in most
vulnerable than less vulnerable areas. Among
adults, barriers in receiving medical care
including lack of insurance, language barriers,
and lack of knowledge on where to get care, as
well as transportation were significantly greater
in most vulnerable areas. On the other hands,
those barriers were found not significantly
different in most and less vulnerable areas
among teens and children. Among adults,
diabetes was reported to be the primary health
issue in both areas. Among children, obesity
was reported to be the primary health issue, and
significantly greater percentage was reported in
less vulnerable areas. Unmet needs are not
significantly different. These unmet needs
include mental health providers, dentists and
family doctors.
Conclusions: The needs of the population as
an aggregate are different than those perceived
in areas labeled as high need and by racial
stratifications, and age groups. Access to
insurance was reported to be the main barrier to
access care, particularly among Hispanic
population living in these high need areas.
Implications for Policy, Delivery, or Practice:
The findings helps provide community leaders
with long-term strategic planning initiatives
focused on the health status and needs of the
city. The findings also allow community partners
to identify gaps in services and to provide
opportunities for collaborative partnerships to
address the issues. These partnerships have the
potential to improve health status of the
community through program development,
access to services, and availability of services.
Funding Source(s): No Funding
Poster Session and Number: A, #240
Drivers to Hospital Inpatient Charges from
Socio-economic, Geographic and Healthcare
Workforce Perspectives
Raymond Fang, American Urological
Association; William Meeks, American
Urological Association; Kimberly Ross, Morgan
State University, School of Community Health
and Policy/American Urological Association; Tori
Pearson, American Urological Association;
Jaimie Toroney, American Urological
Association
Presenter: Raymond Fang, Director, American
Urological Association
raymond_fang99@yahoo.com
Research Objective: The U.S. Centers for
Medicare & Medicaid Services (CMS) released
the average hospital inpatient charges for
common services that show significant variation
across the country and within communities. The
objective of this study was to identify
neighborhood determinants of hospital inpatient
charges in order to find ways to make our health
care system more affordable and accountable.
Study Design: This study explored the average
hospital charges for the 100 most common
Medicare inpatient procedures paid under the
Medicare Severity Diagnosis Related Group
(MS-DRG) system for fiscal year 2011.
Providers determine the amount they will charge
for items and services provided to patients and
these charges are the amount billed. Hospital
neighborhood characteristics were examined
using data from the CMS, National Practitioner
Identifier (NPI) file and the U.S. Census.
Neighborhood data were linked with the hospital
charge data and analyzed by multivariate linear
regression models to determine the associations
between hospital charges and neighborhood
characteristics.
Population Studied: Study data covered 7
million discharges or about 60% of total
Medicare inpatient discharges.
Principal Findings: Hospital inpatient charges
for the same services varied from hospital to
hospital across the country. The ratios of
hospital charges from the highest rate to the
lowest rate ranged from 10.7 times for
permanent cardiac pacemakers implant with
major complicating condition to 47.1 times for
red blood cell disorders without major
complicating conditions. The gaps in hospital
charges ranged from $60,029 for fractures of hip
& pelvis without major complicating conditions to
$899,482 for respiratory system diagnosis with
ventilator support for 96+ hours. The major
neighborhood factors associated with higher
hospital inpatient charges include residing in
metropolitan areas or large towns; areas with
higher mean household income, physician
supply, unemployment rate, mean retirement
income and percent of families under poverty
line. In contrast, major neighborhood factors
associated with lower hospital inpatient charges
are more physician assistants and nurse
practitioners, families with food stamp benefits;
and residents with health insurance coverage.
Conclusions: Findings from this study indicate
significant variations in hospital inpatient
charges across the country. Elevated hospital
charges in areas with higher unemployment and
poverty rates can create a barrier to accessing
hospital care and further exacerbate
deteriorating health outcomes since vulnerable
populations need more acute care with chronic
conditions.
Implications for Policy, Delivery, or Practice:
Study findings suggest increasing the physician
assistant and nurse practitioner workforce and
expanding health insurance coverage will
positively result in improving population health,
reducing health disparities, lowering healthcare
costs and increasing healthcare system
affordability and accountability.
Funding Source(s): No Funding
Poster Session and Number: A, #241
Are there Racial Disparities in Quality of
Treatment for Locally Advanced Prostate
Cancer?
Kezhen Fei, Mount Sinai School of Medicine;
Rebeca Franco, Mount Sinai School of
Medicine; Rajwanth Veluswamy, Mount Sinai
School of Medicine; Nina Bickell, Mount Sinai
School of Medicine
Presenter: Kezhen Fei, Sr. Project Analyst,
Mount Sinai School of Medicine
kezhen.fei@mountsinai.org
Research Objective: African American men are
more likely to develop and die of prostate cancer
than white men. Past studies have shown lower
rates of invasive treatment among black men.
However these studies included early-stage
disease for which treatments may not be
beneficial. We undertook this study to assess
the quality of prostate cancer treatment among
men with locally advanced cancer who could
benefit from active therapies.
Study Design: A Steering Committee of experts
in prostate cancer care reviewed the evidence to
create quality measures of treatment of locally
advanced prostate cancer.
Population Studied: All black and a random
sample of white men with Gleason Scores of 7+,
diagnosed at an inner-city academic center
between 2007 and 2012, were identified from
pathology and charts abstracted for clinical,
pathologic and treatment data.
Principal Findings: Overall, 290 black and 255
white men were identified. The average age of
the entire cohort was 59 years old (SD=8.6) and
there was no statistical age difference between
races. Over 70% patients had commercial
insurance, and 18% had Medicare, there was no
racial difference on insurance coverage. Black
men were more likely to have comorbidities
(30% vs 9%; p<0.001) and higher prostatespecific antigen (PSA) levels prior to biopsy
(12.4 vs 6.9; p=0.0027) than white men, but
whites were more likely to present with stage 3
cancer (30% vs 22%; p=0.02) than blacks.
Using D’Amico risk criteria, 14% patients had
intermediate risk, 86% had high risk, and there
was no racial difference on D’Amico risk. Using
Schonberg e-prognosis mortality risk, black men
had higher 9 year mortality risk than whites
(26% vs 16%, p<.0001). Black men were less
likely than white men to undergo radical
prostatectomy (79% vs 91%, p<.0001), but more
likely to get external beam radiation therapy
(EBRT, 13% vs 5%; p=0.0009) and
brachytherapy (11% vs 4%; p=0.0037). Only
50% of men undergoing RT received androgen
deprivation therapy (ADT), with no racial
difference in ADT use. Overall, 6% of men with
locally advanced prostate cancer did not
undergo surgery or RT (definitive treatment),
black men were more likely to experience
underuse in unadjusted analyses (7% vs 4%;
p=0.0580). Multivariable logistic regression
showed that pathologic risk and mortality risk
predict underuse. Patients with intermediate
pathological risk were at greater odds in
experiencing underuse of definitive treatment
(OR=17; 95% CI: 6.8-45.1), and patients with
higher 9 year mortality risk were at greater odds
of underuse (OR=7; 95% CI: 1.48-30.79).
Conclusions: The rate of underuse of definitive
treatment in men with locally advanced prostate
cancer was low and was not driven by race.
However, underuse was significantly associated
with intermediate risk prostate cancer and higher
mortality risk.
Implications for Policy, Delivery, or Practice:
Men who could benefit from treatment of their
locally advanced prostate cancer are not getting
treated. Overall mortality risk does not
adequately explain underuse in this risk group.
Funding Source(s): Other Department of
Defense
Poster Session and Number: A, #242
Access To and Engagement in Substance
Use Disorder Specialty Treatment:
Comparing Justice-Involved and Other
Veterans
Andrea Finlay, VA Palo Alto Health Care
System; Jim McGuire, Veterans Justice
Programs, Department of Veterans Affairs; Joel
Rosenthal, Veterans Justice Programs,
Department of Veterans Affairs; Jessica BlueHowells, Veterans Justice Programs,
Department of Veterans Affairs; Tom Bowe,
Substance Use Disorder QUERI, VA Palo Alto
Health Care System; Alex Sox-Harris,
Substance Use Disorder QUERI, VA Palo Alto
Health Care System
Presenter: Andrea Finlay, Post-doctoral Fellow,
VA Palo Alto Health Care System
andrea.finlay@va.gov
Research Objective: More than half of all
Veterans involved in the criminal justice system
have an alcohol or drug use disorder, and many
have other mental health conditions, medical
conditions, and employment and housing
challenges. Veterans Health Administration
(VHA) has programs dedicated to connecting
justice-involved Veterans with VHA services to
treat substance use disorder (SUD) and mental
health symptoms and reduce their risk for
criminal justice recidivism and homelessness,
but the success of these programs in achieving
these goals is unknown. We examined the
effectiveness of VHA linkage programs on
access to and engagement in VHA SUD
treatment among justice-involved Veterans with
SUD conditions compared to non-justiceinvolved Veterans with SUD conditions.
Study Design: Using VHA health records, we
selected all Veterans who received a SUD
diagnosis in fiscal year 2012 (N = 501,593).
Measures included justice-involved status,
demographics (gender, age, race/ethnicity,
urban/rural), type of SUD condition (alcohol or
drug use disorder), presence of a co-occurring
mental health (PTSD, personality disorders,
depression, anxiety, schizophrenia, affective
psychosis, other psychosis), and a random
effect for VHA facility (N = 130). Outcome
measures included access to SUD treatment,
defined as receipt of any SUD services, and
engagement in SUD treatment, defined as the
number of SUD outpatient sessions, the number
of SUD inpatient/residential stays, or any use of
pharmacotherapy for alcohol or opioid
dependence, used within one year of initial
diagnosis. Multi-level model analyses were used
to test whether justice-involved status predicted
access to and engagement in SUD treatment.
Population Studied: All Veterans seen at a
VHA facility who received a SUD diagnosis in
fiscal year 2012, compared by justice-involved
(n = 28,608) or non-justice-involved (n =
472,985) status.
Principal Findings: Justice-involved Veterans
were more likely to access at least some SUD
treatment compared to non-justice-involved
Veterans (OR = 3.54, 95% CI: 3.44-3.64).
Engagement in addiction-related outpatient
treatment (b = 0.83, p < .001), defined as the
number of sessions, and inpatient/residential
stays (b = 0.63, p < .001), defined as the
number of days, was greater for justice-involved
Veterans. Among Veterans with alcohol
dependence, utilization of pharmacotherapy was
greater among justice-involved Veterans (OR =
1.39, 95% CI: 1.33-1.46). However, among
Veterans with opioid dependence, utilization of
pharmacotherapy was lower among justiceinvolved Veterans (OR = 0.82, 95% CI: 0.760.88).
Conclusions: Despite the complex needs of
this vulnerable population, VHA appears to be
effectively connecting justice-involved Veterans
with SUD treatment compared to other Veterans
with SUD conditions. Increasing use of opioid
pharmacotherapy may be especially important
for Veterans being released from prison as they
are at high risk for relapse and overdose.
Implications for Policy, Delivery, or Practice:
Determining the key aspects of the linkage
programs and expanding them to other Veterans
would help increase access to and engagement
in VHA SUD treatment services.
Funding Source(s): VA
Poster Session and Number: A, #243
Disturbing Portraits of Uninsured Minority
Children: Parental Awareness of and
Reasons for Children’s Uninsurance, and
Impact on Health, Access, Unmet Needs, and
Quality
Glenn Flores, University of Texas Southwestern
Medical Center and Children's Medical Center;
Candy Walker, UT Southwestern; Hua Lin, UT
Southwestern; Marco Fierro, UT Southwestern;
Monica Henry, UT Southwestern; Kenneth
Massey, UT Southwestern; Alberto Portillo, UT
Southwestern
Presenter: Glenn Flores, Professor And Chief,
University of Texas Southwestern Medical
Center and Children's Medical Center
glenn.flores@utsouthwestern.edu
Research Objective: Among US children,
Latinos (Ls) are at highest risk and AfricanAmericans (AAs) at second highest risk of any
racial/ethnic group of being uninsured, at 14.1%
and 9.3% lacking health insurance, respectively,
vs. 6.5% in whites (Ws). Indeed, the number of
uninsured L children (2.5 million) continues to
exceed the number of uninsured W children, and
L and AA children account for 53% of all
uninsured children, even though they comprise
only 38% of the total population of US children.
Not enough is known, however, about the
reasons for Medicaid- and CHIP-eligible minority
children being uninsured, parental awareness of
the child’s uninsured status, and the children’s
health, access to care, use of services, and
quality of care. The study aim, therefore, was to
examine parental awareness of and reasons for
uninsurance in Medicaid/CHIP-eligible minority
children, and children’s health, access, unmet
needs, and quality of care (QOC).
Study Design: As part of a randomized,
controlled trial of an insurance intervention for
uninsured children, the five communities in the
Dallas metropolitan area with the highest
proportion of uninsured and poor minority
children were targeted for study enrollment.
Recruitment occurred at 91 community sites,
including supermarkets, department stores,
public libraries, Goodwill stores, food banks,
health fairs, Boys and Girls clubs, churches,
schools, Laundromats, and housing projects.
Subjects were uninsured L and AA children
eligible for but not enrolled in Medicaid/CHIP
and residing in Dallas County, where 90% of
uninsured children are L or AA. Recruitment was
performed from June 2011 to January 2014 by
trained, bilingual L and AA staff. Characteristics
examined included sociodemographic features,
health status, use of health services, quality of
pediatric care (on a scale of 0-10, where 0 =
worst and 10 = best), quality of life, parental
satisfaction, and financial burden, using
validated instruments.
Population Studied: Uninsured minority
children eligible for but not enrolled in
Medicaid/CHIP.
Principal Findings: 48,107 potential caregivers
were screened for eligibility, yielding a final
sample size of 280 uninsured L and AA children
eligible for but not enrolled in Medicaid or CHIP.
Ls comprised 63% (N=176) and AAs 37%
(N=104) of the sample. Ninety-five percent of
caregivers were female, 38% were married and
living with their spouse, and 32% had limited
English proficiency. The mean annual family
income was $21,810 (range: $1,440-$64,000).
The mean age of the uninsured children was 7.4
years (range: 1-18 years); 50% were female,
and 95% were born in the US. Eighty-nine
percent of children had ever been insured
before, most often by Medicaid (74%), CHIP
(13%), and private insurance (13%). The mean
time without insurance was 14.2 months (range:
1-144 months). For those ever insured, the most
frequent reasons for loss of the child’s insurance
included insurance expired and never reapplied
(24%), parent told income was too high (13%),
missing paperwork (10%), and “don’t know why”
(10%). Among those never insured, reasons for
never having insurance included language
barriers, too much hassle, and “move to Texas,”
each at 20%. Only 48% of parents were aware
that their uninsured child was eligible for
Medicaid or CHIP. A total of 39% of children
were in suboptimal health (not excellent/very
good), about two-thirds had special healthcare
needs, 84% of parents reported worrying about
their child’s health more than other people, and
76% of parents reported worry or concern about
their child’s physical health. Only 37% of
children have a primary-care provider (PCP),
and 61% a usual source of preventive care.
Seventy-three percent have delayed or did not
receive needed healthcare, 53% have not
received all needed dental care, 9% have not
received all needed acute care, 13% have not
received all needed specialty care, and 12%
have not received all needed vision care.
Parental ratings of the quality of pediatric care
were low, with a mean of 5.0 overall, 3.7 for
primary care, and 1.8 for specialty care. Fortyfour percent of parents reported that they
needed additional income to cover the child’s
medical expenses, 34% that the child’s health
caused financial problems for the family, 23%
that they cut down their work hours to provide
healthcare for their child, and 10% ceased
working because of the child’s health. Compared
with AA parents, L parents were significantly
(P=.03) more likely to report worrying about their
child’s health more than others (87% vs. 73%),
and that the child has no PCP (70% vs. 52%) or
usual source of preventive care (48% vs. 23%),
and lacks 24-hour phone coverage for sick care
(96% vs. 71%).
Conclusions: Half of parents of uninsured
minority children are unaware their child is
Medicaid/CHIP-eligible. These children have
been uninsured for a mean of 14 months; 11%
have had no coverage in their lifetime. The most
common reason for insurance loss is not
reapplying after insurance expiration. These
uninsured children have poor health,
substantially impaired access, major unmet
needs, and poor QOC. Due to the child’s health,
>1/3 of parents suffer financial problems and 1
in 10 ceased work. L children are at significantly
higher risk of parental worry about the child’s
health and of lacking a PCP and 24-hour phone
coverage for sick care.
Implications for Policy, Delivery, or Practice:
Implications of these findings include: 1) parents
of uninsured minority children need better
education and awareness regarding
Medicaid/CHIP eligibility and the application
process; 2) improvements are needed in
Medicaid/CHIP outreach and enrollment, such
as greater use of known effective interventions,
including community health workers and
promotoras; 3) Medicaid and CHIP should be
maintained or expanded, and efforts to defund
CHIP—such as a recent House amendment to
the Continuing Resolution—should be avoided;
and 4) there is an urgent need to insure and
provide PCPs to uninsured L children, who are
at especially high risk for inefficient, costly use of
the ED for care, due to the vast majority lacking
access to 24-hour phone coverage for sick care.
Funding Source(s): NIH
Poster Session and Number: A, #244
The 2013 NCHS Urban-Rural Classification
Scheme: A Tool to Examine Health
Disparities
Sheila Franco, National Center for Health
Statistics and Centers for Disease Control and
Prevention; Deborah Ingram, National Center for
Health Statistics and Centers for Disease
Control and Prevention
Presenter: Sheila Franco, Health Statistician,
National Center for Health Statistics and Centers
for Disease Control and Prevention
sfranco@cdc.gov
Research Objective: To explore urban-rural
health disparities using the six-level 2013
National Center for Health Statistics Urban-Rural
Classification Scheme for Counties which was
recently updated to reflect the 2010 census.
Study Design: Researchers are often interested
in the level of urbanization because variations in
urbanization are associated with disparities in
health behaviors and health outcomes as well as
the availability of health care resources. A
variety of taxonomies to classify communities by
urban-rural level exist but may not be adequate
for distinguishing between inner cities and
suburbs which can differ substantially on healthrelated measures. NCHS developed a six-level
county-level urban-rural scheme which
distinguishes between inner cities and suburbs
and has recently updated this scheme using
2010 census data. 2010-2012 National Health
Interview Survey and mortality data from the
2008-2010 National Vital Statistics System were
used to assess the usefulness of this scheme
and to examine health disparities.
Population Studied: U.S. population, adults
aged 18 and over.
Principal Findings: Substantial differences in
health behaviors, status, and outcomes by
urbanization level were found. For most health
indicators, residents of the most rural counties
fared worse while suburban counties fared best.
For example, fair or poor health status was
lowest among adults aged 18–64 years residing
in large fringe metro (suburban) counties and
highest among those in the most rural counties,
noncore (rural) counties. The percentage of
uninsured adults aged 18-64 was lowest among
those residing in large fringe metro (suburban)
counties compared to all other urbanization
levels. Age-adjusted death rates from all causes
for those aged 25-64 years were lowest for large
fringe metro (suburban) counties, compared to
more rural counties, with the highest rates in the
most rural counties. Age-adjusted death rates
from motor vehicle accidents progressively
increase across the six urbanization levels, with
the age-adjusted motor vehicle accident death
rate lowest in large central metro (inner cities)
counties and about three times higher in
noncore (rural) counties, the most rural counties.
In contrast homicide rates were highest rates in
large central metro (inner cities) counties.
Conclusions: Substantial urban-rural
differences were observed for the health
measures examined. Of special note are the
findings that residents of large fringe metro
(suburban) counties often fare better on health
measures than residents of other urbanization
levels.
Implications for Policy, Delivery, or Practice:
This analysis demonstrates the usefulness and
importance of the NCHS taxonomy in identifying
differences in health measures across the six
urbanization levels.
Funding Source(s): CDC
Poster Session and Number: A, #245
Across State Lines: Implications of State
Decisions to Expand Medicaid on the
Poorest Residents
Tracy Garber, The Commonwealth Fund; Petra
Rasmussen, The Commonwealth Fund; Sara
Collins, The Commonwealth Fund; Michelle
Doty, The Commonwealth Fund
Presenter: Tracy Garber, Senior Policy
Associate, The Commonwealth Fund
tg@cmwf.org
Research Objective: In its 2012 decision, the
Supreme Court gave states the option to choose
whether or not to participate in the Affordable
Care Act’s expansion of Medicaid coverage to
all legal residents up to 138 percent of poverty.
To date, 26 states and the District of Columbia
have decided to expand. This means that
residents living under the poverty level in about
half of states may be finding themselves with no
new affordable health insurance options this
year. This analysis compares the pre-expansion
insurance coverage status of low income
residents in states expanding against those in
states not expanding Medicaid, and explores
how this disparity might be exacerbated by
states’ decisions to not expand the program.
Study Design: Data for this survey comes from
the Commonwealth Fund Health Insurance
Marketplace Survey, 2013, a nationally
representative telephone survey of adults ages
19-64 conducted by Social Science Research
Solutions from July to September 2013.
Population Studied: Adults ages 19 to 64
(N=6,132). Some analyses were restricted to
adults under 100 percent of poverty (N=1,113).
Principal Findings: In 2013, sixteen percent of
all adults ages 19-64 reported being uninsured
for the entire previous twelve months. Among
this group, 36 percent of those in non-expansion
states and 31 percent of those in expansion
states were under 100 percent of poverty.
Adults living in poverty in states that are not
expanding Medicaid had significantly greater
odds of being uninsured all year than their
counterparts in states that are expanding the
program, after adjusting for demographics (AOR
1.98, 95CI 1.35, 2.90). One third (33%) of
adults living in poverty in non-expansion states
were uninsured all year, compared to 22 percent
of those in expansion states.
Conclusions: Even before implementation of
the 2014 Medicaid expansion, there were
insurance coverage disparities among adults
living in poverty between states that have
decided to expand their programs and those that
have not decided to move forward this year.
Those in states not planning to participate were
more likely to be uninsured all year. While up to
31 percent of adults with incomes under poverty
in expansion states could gain access to
Medicaid coverage this year, 36 percent of
adults uninsured all year and living in poverty in
non-expansion states will have no new
affordable insurance options available.
Implications for Policy, Delivery, or Practice:
If the 24 states that have not yet committed to
expanding Medicaid under the Affordable Care
Act do not change course, it is likely that the
disparity in insurance coverage status of adults
living under the federal poverty level will grow
between residents of expansion states and
residents of non-expansion states. It is
imperative that states opt in to the Medicaid
expansion to allow their lowest income residents
to access health care.
Funding Source(s): N/A
Poster Session and Number: A, #246
Lifetime Sexual Assault, High-risk Behavior,
and Sexually Transmitted Infections Among
Women Veterans
Vinita Goyal, The Warren Alpert Medical School
of Brown University/Women and Infants
Hospital; Michelle A. Mengeling, Comprehensive
Access & Delivery Research and Evaluation
(CADRE), Iowa City VA Health Care System,
Department of Internal Medicine, University of
Iowa Carver College of Medicine, VA Office of
Rural Health (ORH); Brenda M. Booth, Center
for Mental Healthcare Outcomes and Research,
Central Arkansas Veterans Healthcare System
and Department of Psychiatry, University of
Arkansas for Medical Sciences; James C.
Torner, Department of Epidemiology, University
of Iowa College of Public Health, Departments of
Neurosurgery and Surgery, University of Iowa
Carver College of Medicine; Craig H. Syrop,
Department of Obstetrics and Gynecology,
University of Iowa Carver College of Medicine;
Anne G. Sadler, Comprehensive Access &
Delivery Research and Evaluation (CADRE),
Iowa City VA Health Care System, Department
of Psychiatry, University of Iowa Carver College
of Medicine
Presenter: Vinita Goyal, Assistant Professor,
The Warren Alpert Medical School of Brown
University/Women and Infants Hospital
vinitagoyalmd@gmail.com
Research Objective: Data on reproductive
health outcomes among the growing population
of women Veterans are lacking. Active duty
Servicewomen report elevated rates of high-risk
sexual behaviors including absent or
inconsistent condom use and having a new or
concurrent sexual partners, which may explain
the high prevalence of sexual transmitted
infections (STI) in this group. This study aimed
to examine high-risk sexual behaviors, STI, and
lifetime sexual assault (LSA) among women
Veterans.
Study Design: Data were from a retrospective
cohort study to determine lifetime prevalence of
STI from self-reports gathered via computerassisted telephone interview. Also collected
were data on sexual behavior and LSA before,
during, and after military service. Bivariate
analysis was used to evaluate the association
between LSA and STI history, reported as an
unadjusted odds ratio with 95% confidence
interval.
Population Studied: 1004 women Veterans
<51 years of age enrolled at two Midwestern VA
Medical Centers or outlying clinics between
2000-2008.
Principal Findings: Participants had mean age
of 38.3 years, were predominantly white (89%),
and the majority had college or technical training
(56%). Lifetime prevalence of gonorrhea was
5%, chlamydia was 15%, genital herpes was
8%, and <1% reported syphilis. Women
reported a mean of 5.1 years of unprotected
intercourse, 19% had unintended sex after
alcohol/drug use, and 31% had at least one nonmonogamous sexual partner after military
service. Women reporting an STI were
significantly more likely than those who did not
to receive treatment for drug or alcohol abuse
(26% vs. 12%, p<0.05), have depression (61%
vs. 50%, p<0.05), a younger age at first
intercourse (mean age 16.9 vs. 17.5 years,
p<0.05), a greater number of lifetime sexual
partners (mean 11.5 vs. 9.9, p<0.05), report no
condom use before military service (29% vs.
23%, p<0.05), report unintended sex after
alcohol/drug use before (37% vs. 29%, p<0.05),
during (51% vs. 30%, p<0.05), and after (25%
vs. 17%, p<0.05) military service, and nonmonogamous sexual partners before (48% vs.
36%, p<0.05), during (61% vs. 40%, p<0.05),
and after (40% vs. 28%, p<0.05) military service.
Women reporting LSA (62%) were significantly
more likely to report a history of STI compared
to those never assaulted (OR 1.89, 95% CI,
1.37-2.60). Women who reported LSA were
also more likely to report depression,
alcohol/drug abuse, and multiple high-risk
sexual behaviors before, during, and after
military service compared to their non-abused
counterparts.
Conclusions: High-risk sexual behaviors are
prevalent among women Veterans which may
explain the high rates of gonorrhea and
chlamydia infection in this population. Women
who have experienced LSA are a sub-group at
significant risk for STI, as well as depression
and alcohol/drug use that can contribute to the
cycle of high-risk behaviors.
Implications for Policy, Delivery, or Practice:
Reproductive health services for Servicewomen
and Veterans should emphasize assessment of
high-risk sexual behaviors and screening and
treatment for STI and associated mental health
conditions. These findings have important
implications for needed resources and
interventions to address high-risk behaviors and
adverse outcomes of untreated STI that can
impair Servicewomen’s lifelong health and
fertility.
Funding Source(s): VA National Institutes of
Health
Poster Session and Number: A, #247
Patient Web-Portal Use: Can Internet Access
Help Bridge the Divide?
Ilana Graetz, University of Tennessee Health
Science Center; Courtnee Hamity, Kaiser
Permanente Division of Research; Vicki Fung,
Harvard Medical School and Massachusetts
General Hospital; Nancy Gordon, Kaiser
Permanente Division of Research; Mary Reed,
Kaiser Permanente Division of Research
Presenter: Ilana Graetz, Assistant Professor,
University of Tennessee Health Science Center
igraetz@uthsc.edu
Research Objective: Patient web-portals have
the potential to improve access to care and
patient engagement. Meaningful use incentive
payments will require that patients have online
access to their health records and the ability to
exchange secure emails with providers. The
digital divide could limit access to web-based
portals among disadvantaged groups. We
examined the association between patient
characteristics and internet access, and how
they relate to use of a patient web-portal.
Study Design: A stratified random sample of
adult health plan members who were in the
plan’s chronic disease registries completed
surveys by mail, by internet-based survey, or by
telephone interview. During the study period, all
members could access the health plan’s webportal to send a secure email to a healthcare
provider via a web browser. Study participants
reported how often they used the internet, if they
used to their own computer or smartphone to
access the internet, and if they used the webportal to email a provider in the last year. All
analyses were weighted for sampling
proportions. We used multivariate logistic
regression to assess the association between
patient characteristics (sociodemographic and
health status) and internet access. To examine
characteristics associated with sending secure
email to healthcare providers, we examined two
model specifications: 1) adjusting for patient
characteristics only, and 2) adjusting for patient
characteristics, plus frequency of internet use
and devices used.
Population Studied: Among 1041 respondents
(87% response rate), 59% were white, 44%
were age 65+, 27% had household income
<$40,000, and 29% had a high school education
or less.
Principal Findings: Overall, 71% used the
internet regularly (daily, weekly or monthly), 72%
used their own computer and 21% used a
smartphone to access the internet, and 56%
used the web-portal to email a provider. In
multivariate analyses, respondents with lower
income and education were less likely to use
their own computer or smartphone for internet
access (p<0.05). Blacks and Hispanics were
less likely than whites to access the internet
using their own computers (p<0.001) and
respondents age 65+ (vs. 18-44 yrs) were less
likely to use smartphones to access the internet
(p<0.01). After adjusting only for
sociodemographic characteristics, those who
were male, age 55+ (vs. 18-44 years), Black or
Asian (vs. white), with lower income and
education were less likely to have emailed a
provider (p<0.05). After also adjusting for
internet use and devices, only differences by
education and gender remained (p<0.01).
Regular internet use and access via a personal
computer were associated with emailing a
provider (p<0.01); smartphone access was not,
but the portal mobile application was not
available during the study period.
Conclusions: Regular internet use and having
a personal computer could explain differences in
web-portal use to email providers by age, race,
and income. Education and gender-related
differences in use of email remained even after
controlling for internet access.
Implications for Policy, Delivery, or Practice:
As the availability and use of patient web-portals
increase, it is important to understand which
patients have limited access and what barriers
they face. Improving internet access and making
web-portals available across multiple platforms,
including mobile, may reduce disparities in use.
Funding Source(s): Other Kaiser Foundation
Research Institute
Poster Session and Number: A, #248
Geographic Barriers Impact on Access to
Percutaneous Cardiac Interventions Care
and Level I Trauma Care for Geographically
Isolated Urban Communities
Dennis Graham, Hunter-Bellevue School of
Nursing
Presenter: Dennis Graham, Assistant Professor
of Nursing, Hunter-Bellevue School of Nursing
grahamd@mskcc.org
Research Objective: To measure the affects of
geographic barriers on access to Percunteous
Cardiac Interventions Care and Level I Trauma
Care for geographically isolated urban
communities.
Study Design: Using Google Maps directions
application we collected the time and distance to
travel between New York City’s Rockaway
Peninsula five zip codes and the nearest PCI
and LITC. Time measurements were recorded at
non-rush hour (10 AM) and rush-hour (6 PM)
compared to the other zip codes in borough of
Queens NY.
Population Studied: Many areas in the US are
impacted by natural and urban geographic
barriers that limit access to Primary
Percutaneous interventions Centers (PCIC) or
Level I Trauma Care (LITC). Patients requiring
PCIC or LITC in these isolated communities
must be transported to existing PCIC and LITC
by ambulance. Many EMS departments do not
record or release the time it takes to travel to the
nearest hospital. There has been no research on
the time and distance to travel to gain access to
a PCIC or LITC for these isolated communities.
The Rockaway Peninsula is the most
geographically isolated community within NYC
and could serve as a model to study the affects
of geographic and urban barriers to time to PCIC
an LITC.
Principal Findings: The average time to a
PCIC/LITC for the Rockaway Peninsula in nonrush hour is 28.1 vs 12.4 minutes and during
rush hour 52.3 vs 20.3minutes for all of the
remaining Queens zip codes. locations.
Conclusions: There are many reasons for lack
of access to PCIC or LITC and the geographic
and urban barrier that isolated these
communities are frequently under appreciated.
Using this model health care planners could
anticipate the need for more access to PCIC and
LITC based on the geographic and urban
barriers.
Implications for Policy, Delivery, or Practice:
More research is needed to evaluate new
approaches to study using existing technology
such as automatic traffic light interrupters and
emergency traffic lanes, mobile emergency care
vehicles, that could reduce this time to
accessing PCIC or LITC.
Funding Source(s): No Funding
Poster Session and Number: A, #249
Augmenting Electronic Health Record Data
with Patient-Reported Outcome Measures to
Address Health Disparities in Clinical
Research
Sandra Griffith, Cleveland Clinic; Nicolas R.
Thompson, Cleveland Clinic; Jaivir S. Rathore,
Cleveland Clinic; Lara Jehi, Cleveland Clinic;
George E. Tesar, Cleveland Clinic; Irene
Katzan, Cleveland Clinic
Presenter: Sandra Griffith, Assistant Staff,
Cleveland Clinic
griffis5@ccf.org
Research Objective: Electronic health records
(EHR) present an opportunity to access large
stores of data for research, but mapping raw
EHR data to clinical phenotypes is complex.
Depending on the choice of mapping method,
disparities present in the healthcare delivery
system may be propagated to research data
sources. We propose adding patient-reported
outcome measures (PROMs) to the traditional
EHR data sources to improve phenotyping
performance and augment inclusion of
populations that may be underrepresented. We
use the clinical phenotyping of major depressive
disorder as a demonstration case.
Study Design: We compared four EHRphenotyping methods based on ICD-9 codes,
medication records, and the Patient-Health
Questionnaire 9 (PHQ-9) regarding the ability to
identify cases with depression and
characteristics of patients identified with
depression. We compared the demographic
composition of patients identified as depressed
by each of the four methods with respect to age,
sex, race, marital status, disability, healthrelated quality of life, and insurance. Several
factors based on ZIP code of residence,
including median income, percentage below
poverty, and rural status were also considered.
We based these categorizations on the priority
population standards endorsed by the Agency
for Healthcare Research and Quality. We also
assessed the diagnostic performance of each
method in a subset of 225 patients who had a
reference standard measurement for depression
available.
Population Studied: Our sample included
168,884 patients seen (2007 to 2013) at our
neurological institute where PROMs are
electronically collected during routine outpatient
care.
Principal Findings: ICD-9 codes identified the
fewest number of patients as depressed (4,658),
followed by PHQ-9 (46,565), and medication
data (50,505). The presence of at least one of
these criteria identified the largest number
(78,322). The PHQ-9 identified a higher
proportion of elderly (109.1% increase), disabled
(344.7% increase), married (13.2% increase),
Medicaid (28.2% increase), and rural patients
(98.9%), as compared to ICD-9 codes. The
population identified as depressed using the
PHQ-9 exhibited lower mean health-related
quality of life scores (0.52; SD: 0.23) than those
identified by ICD-9 codes alone (0.7; SD: 0.21).
When compared to a reference standard, ICD-9
codes were least sensitive (6.7% sensitivity),
whereas the method using at least one of the
criteria identified the highest number of truly
depressed patients (93.3% sensitivity); however,
specificity dropped from 97.7% to 58.1%.
Conclusions: Some EHR-based phenotyping
methods may disproportionately exclude patient
groups from research. Patient-reported data
holds potential to reduce some biases inherent
in EHR data and improve sensitivity while
maintaining an acceptable loss of specificity,
depending on the context.
Implications for Policy, Delivery, or Practice:
Where available, combing PROMs with EHR
data should be considered to ensure that priority
populations, as identified by AHRQ, are
adequately and accurately represented in EHRbased research studies.
Funding Source(s): No Funding
Poster Session and Number: A, #250
Women’s ACA Knowledge and Attitudes:
Implications for Health Care Access,
Utilization, and Equity
Kelli Hall, University of Michigan; A. Mark
Fendrick, University of Michigan; Vanessa
Dalton, University of Michigan
Presenter: Kelli Hall, Research Investigator,
University of Michigan
hkelli@umich.edu
Research Objective: Women’s health scholars,
clinicians and policy makers have placed our
highest hopes and expectations on universal
health care coverage, as a mechanism of
improved access that will result (directly) in
improved health service utilization, outcomes
and reduced disparities for women. Since the
Affordable Care Act (ACA) was passed and
more recently enacted, though, it is becoming
clear that achieving these end goals may be
even more complicated than even the most
astute scholars projected. Along these lines, we
sought to characterize women’s perspectives of
the ACA, specifically their ACA knowledge and
attitudes, and sociodemographic differentials
within them, among women in the United States.
Study Design: We conducted a populationbased, cross-sectional, internet survey of
women’s health care experiences and
preferences. The comprehensive survey which
we designed for this study included a series of
items assessing women’s knowledge and
attitudes of the ACA: 1) whether they had ever
heard of the Affordable Care Act, sometimes
referred to as “Obamacare,” the new health care
reform legislation passed by the U.S. Congress
in March 2010; 2) whether their health insurance
coverage has/will change for ten different types
of care as a result of the ACA; 3) how they
expect those changes will affect their use of
those health services; 4) what affect the ACA
will have on their ability to get their preferred
care (i.e. care they would most like to have); and
5) overall, whether they agree or disagree with
the passage of the ACA. We used weighted
proportions to describe ACA knowledge and
attitudes and chi-square and multivariable
logistic regression to examine associations
between sociodemographic characteristics and
ACA knowledge and attitudes.
Population Studied: U.S. women 18-55 years
were randomly sampled from an existing
representative probability panel (n=1,078).
Principal Findings: Most women had heard of
the ACA (81%), though 24% expected ACArelated insurance coverage changes. “Not
knowing” was a common response for specific
changes for preventive health (61%), women’s
health (62%), birth control (65%), breast exam
(66%) and mental health (76%) coverage. Few
women believed the ACA would improve their
ability to get their preferred care (14%; 22%
reported it would worsen their ability, 42% did
not know). One third of women disagreed with
the ACA overall; 23% did not know how they felt.
In the bivariate analysis, women’s ACA
knowledge and attitudes varied by nearly all
sociodemographic characteristics (all pvalues<0.05). For example, compared to their
counterparts, proportions of women who had
ever heard of the ACA were higher among older,
White, college-educated, higher-income,
politically-affiliated, married, employed, and
privately-insured women, women with a history
of recent health service use and lifetime
prescription contraception use. In multivariate
models, social determinants of ACA knowledge
and attitudes varied by ACA outcomes and
included age, race/ethnicity, education, income,
insurance status, political party, recent health
service use and prescription contraceptive use.
Conclusions: In our representative sample,
many women, especially socially-disadvantaged
women, lacked knowledge and had negative
attitudes about the ACA.
Implications for Policy, Delivery, or Practice:
Findings have implications for whether, when,
and how improved insurance access will
translate to improved service utilization,
outcomes and reduced disparities for women’s
health in the U.S.
Funding Source(s): NIH
Poster Session and Number: A, #252
Reports of Insurance-based Discrimination
and Implications for ACA Implementation
Xinxin Han, School of Public Health, University
of Minnesota--Twin Cities; Kathleen Call, School
of Public Health, University of Minnesota--Twin
Cities; Alisha Simon, Minnesota Department of
Health; Jessie Pintor, School of Public Health,
University of Minnesota--Twin Cities; Giovann
Espinoza, School of Public Health, University of
Minnesota--Twin Cities
Presenter: Xinxin Han, Graduate Student,
School of Public Health, University of
Minnesota--Twin Cities
hanxx725@umn.edu
Research Objective: While there are rich
studies examining racial/ethnic discrimination in
health care settings, little is known about
discrimination due to insurance type (e.g., public
versus private) or lack of insurance. This study
examines reports of insurance-based
discrimination by health care providers among
Minnesota non-elderly adults generally, and how
discrimination varies by types of insurance or
lack of insurance and other key
sociodemographic characteristics. We also
examine the impact of insurance-based
discrimination on access to services.
Study Design: Data are from the 2011
Minnesota Health Access (MNHA) survey.
MNHA is a statewide random-digit-dial dualframe telephone survey conducted by Minnesota
Department of Health (MDH) and University of
Minnesota School of Public Health since 2001.
Stratified random sampling was used to produce
reliable statewide, regional, and racial and
ethnic group estimates of insurance coverage
and access. Our key dependent variable is
report of unfair treatment by health care
providers because of type of insurance if insured
or no insurance if uninsured. We performed
bivariate analysis to describe sociodemographic
characteristics of those reporting insurancebased discrimination, and we modeled the
association between insurance type and unfair
treatment and the association between of unfair
treatment and access to services controlling for
race/ethnicity, age, education, income, marital
status, employment status, and health status.
Data were weighted to represent the state’s
population.
Population Studied: The 2011 MNHA consists
of data from a total of 11,355 Minnesotans. We
restrict the analysis to non-elderly adults (3943
age 18 to 64) who answered whether they were
treated unfairly by health care providers based
on their own experience by themselves rather
than by proxy reports.
Principal Findings: Overall, 10.3% of nonelderly Minnesota adults reported insurancebased discrimination from health care providers.
Reports of discrimination varied significantly by
different insurance status and type of insurance.
Specifically, 27.7% of those without insurance
reported being treated unfairly by health care
providers, followed by 24.5% of those who have
public insurance. Insurance-based
discrimination was significantly lower among the
privately insured: 3.8% for those with employersponsored insurance and 2.6% with selfpurchased insurance. Reports of insurancebased discrimination were also significantly
higher among adults from non-white racial/ethnic
communities, adults with lower socioeconomic
status (based on income and education) and fair
or poor health status. Multivariate analysis will
explore the relationships between insurancebased discrimination and insurance type after
controlling for demographic variation across
insurance type, and how insurance-based
discrimination impacts access to services.
Conclusions: Minnesota adults with public
insurance and who lack of insurance are more
likely to report receiving unfair treatment from
health care providers.. In addition, we find that
experiences of insurance-based discrimination
are significantly higher among disadvantaged
populations.
Implications for Policy, Delivery, or Practice:
The results present a baseline for understanding
how Minnesota adults experience the health
care system on the eve of full implementation of
the ACA. In addition to expanding coverage, the
ACA is also designed to reduce health
disparities. Yet insurance-based discrimination
may lead to under utilization of health care
services among those most in need, as a result,
it may undermine the health care access and
equity goals that are central to the ACA.
Funding Source(s): Other Minnesota
Department of Health
Poster Session and Number: A, #253
A First Look at PCMH Implementation for
Minority Veterans: Room for Improvement
Susan Hernandez, University of Washington;
Karin Nelson, VA Puget Sound Healthcare
System, Northwest HSR&D Center of
Excellence; Haili Sun, VA Puget Sound
Healthcare System, Northwest HSR&D Center
of Excellence; Chuan-Fen Liu, VA Puget Sound
Healthcare System, Northwest HSR&D Center
of Innovation; Edwin Wong, VA Puget Sound
Healthcare System, Northwest HSR&D Center
of Innovation; Christian Helfrich, 1.
VA
Puget Sound Healthcare System, Northwest
HSR&D Center of Excellence; Stephan Fihn,
VHA Office of Analytics and Business
Intelligence; Paul Hebert, VA Puget Sound
Healthcare System, Northwest HSR&D Center
of Innovation
Presenter: Susan Hernandez, Research
Associate, University of Washington
seh315@uw.edu
Research Objective: We test whether a patientcentered medical home (PCMH) model of care,
Patient Aligned Care Teams (PACT), was
implemented differently by facilities in relation to
the percent of minority Veterans comprising their
patients. The Veterans Health Administration
(VHA) rolled out the PACT initiative in 2010,
however there is limited research on the degree
to which PACT is reaching minority populations.
Study Design: Observational, facility level
analysis of VHA hospital-based and communitybased primary care clinics, using a crosssectional PACT Implementation Progress Index
(PI2) for 2012 constructed from VHA
administrative databases; a patient survey
modelled after CAHPS-PCMH; and a survey of
primary care providers and staff measuring team
functioning. For each facility we used an overall
PI2 score and a score for each PI2 domain:
access, continuity, care coordination,
comprehensiveness, self-management support,
patient-centered care (PCC) and
communication, shared decision making, and
team-based care. PI2 is a count of domain
scores in the top minus bottom quartiles, ranging
from 8 (best) to -8 (worst). PI2 scores 5 or
greater relative to scores -5 or lower are
correlated with higher patient satisfaction, lower
hospitalizations and emergency room utilization,
and lower provider burnout. Facilities
categorized as low (below 5-percent), medium
(5-15-percent), and high (above 15-percent)
based on the percent of minority patients
comprising their patient population. We used a
linear model, weighted least squares estimator
and robust standard errors to test for differences
in PACT.
Population Studied: Data was aggregated to
the facility level (n=832). The distribution of low,
medium and high minority facilities was 40percent, 30-percent, and 30-percent,
respectively; however, most minority Veterans
(78-percent) received care high minority
facilities.
Principal Findings: Preliminary findings show
medium (-.92, p=.002) and high (-1.73, p<.001)
minority facilities had lower PI2 scores than low
minority facilities. Medium and high minority
facilities had lower scores for access (-.112,
p=.009, -.265, p<.001); care coordination (-.205,
p<.001, -.307, p<.001); patient-entered care and
communication (-.222, p=.002, -.353, p<.001)
and shared decision making (-.168, p=.005, .187, p=.002) compared to low minority facilities.
Also, medium minority facilities had lower scores
for comprehensiveness (-.174, p=.008), selfmanagement (-.191, p=.004) and high minority
facilities had lower scores for continuity (-.143,
p=.0) compared to low minority facilities. Overall,
2.65-percent of minority Veterans versus 4.88percent of white Veterans received care at
facilities with a PI2 score 5 or greater (p<.001).
Conclusions: We found differences in PACT
implementation between medium and high
minority facilities compared to low minority
facilities. Medium minority facilities had the most
challenges with PACT implementation with six
out of eight domain scores lower than low
minority facilities; high minority facilities had five
out of eight scores lower than low minority
facilities. Medium and high minority overlapped
with lower scores in four domains. These
findings are important when considering 96percent of all minorities receive care at medium
and high percent minority facilities.
Implications for Policy, Delivery, or Practice:
This is the first study to characterize facility-level
PACT implementation in relation to the percent
of minority Veterans served. Further research
needs to investigate the relationship between
the racial composition of a facility and other
characteristics that may impede or improve
PACT implementation.
Funding Source(s): VA
Poster Session and Number: A, #254
Racial/Ethnic Disparities in the Receipt of
Prescriptions for Antidiabetic Medications by
Non-Institutionalized Survey Respondents
with Diabetes
Michael Hoffman, Tulane University School of
Public Health; Lizheng Shi, Tulane University
School of Public Health; Claudia Campbell,
Tulane University School of Public Health; Tina
Thethi, Tulane University Health Sciences
Center; Beth Nordstrom, Evidera
Presenter: Michael Hoffman, Graduate Student,
Tulane University School of Public Health
michael@hoffman@genzyme.com
Research Objective: The primary objective of
this study was to determine whether racial/ethnic
disparities exist in the receipt of antidiabetic
prescriptions in a non-institutionalized
population, and if so, how do individual
characteristics such as socioeconomic status
(SES) influence the differences.
Study Design: National survey data from the
2010 Medical Expenditure Panel Survey (MEPS)
were examined by applying a methodology
based on the Institute of Medicine (IOM)
definition of racial disparity that adjusts for
health status factors while allowing SES factors
to mediate differences between race/ethnicity
and the receipt of diabetes medication. Logistic
regression analyses were performed on
unadjusted data and on data transformed by a
rank-and-replace method to approximate the
IOM definition of disparity.
Population Studied: Survey respondents from
the 2010 Medical Expenditure Panel Survey
(MEPS) with self-reported diabetes.
Principal Findings: Among 1,844 survey
respondents with self-reported diabetes,
significant differences were found for
race/ethnicity, education, health insurance, and
the co-morbidities of heart disease and eye
problems/retinopathy. Race/ethnicity was a
significant predictor of the receipt of antidiabetic
medication prescription, with Hispanics being
more than 2 times as likely as non-Hispanic
whites to have received a prescription. This
difference was magnified in the IOM model that
controlled for health status. No significant
differences were observed between nonHispanic whites and non-Hispanic blacks or
other minorities. Having health insurance, higher
education, or eye problems/retinopathy were
also significant predictors of receiving an
antidiabetic prescription.
Conclusions: Using a methodology based on
the IOM definition of racial disparity that adjusts
for factors related to health status while allowing
factors related to SES to mediate racial/ethnic
differences, disparities were observed between
non-Hispanic whites and minorities, particularly
Hispanics, in the likelihood of receiving a
prescription for antidiabetic medication.
Implications for Policy, Delivery, or Practice:
Application of a rigorous definition of
racial/ethnic disparities, in addition to the
implementation of methodologies that allow for
mediation by SES factors by adjusting for health
status factors, are needed to identify and
address important gaps in the treatment of
diabetes.
Funding Source(s): No Funding
Poster Session and Number: A, #255
Physician Social Networks and Racial
Disparities in Access to High-Quality
Hospitals
John Hollingsworth, University of Michigan;
Russell Funk, Department of Sociology,
University of Michigan; Jason Owen-Smith,
Department of Sociology, University of Michigan;
Bruce E. Landon, Harvard Medical School; John
D., MD
Presenter: John Hollingsworth, Assistant
Professor, University of Michigan
kinks@med.umich.edu
Research Objective: Compared to white
patients, black patients, particularly those in
residentially segregated areas, are more likely to
receive surgical care at low-quality hospitals,
even when they live closer to high-quality ones.
We used social network analysis to explore one
potential explanation that physicians in
segregated areas are more isolated from each
other.
Study Design: Using Medicare data from
Michigan (2005 to 2011), we identified cohorts of
patients undergoing three common inpatient
procedures. After mapping the physician social
networks at the hospitals where these
procedures were performed, we characterized
them across a range of structural properties that
impact on coordination and information sharing.
Finally, we fitted multivariable models to
examine for associations between these
properties and the degree of residential
segregation in the hospital service area (HSA)
served by each network.
Population Studied: Beneficiaries age 66 and
older who underwent colectomy for colon
cancer, coronary artery bypass grafting, or hip
replacement in Michigan between January 1,
2005 and December 30, 2011.
Principal Findings: After accounting for
regional differences in health care capacity, the
social networks of physicians practicing in more
highly segregated areas varied in many
important respects from those of HSAs with low
segregation. Specifically, physicians serving
highly segregated HSAs had fewer repeated
interactions with each other than those in HSAs
with low segregation (P<0.001). When
physicians in highly segregated HSAs did
interact, they tended to congregate in smaller
groups (P<0.001). Moreover, they had fewer
interactions with physicians outside their local
community (P<0.05).
Conclusions: The structures of physician social
networks in HSAs with high and low segregation
differ in ways that likely relate to care
coordination and information sharing.
Implications for Policy, Delivery, or Practice:
Although their role in explaining differences in
access to high-quality hospitals has yet to be
determined, these findings suggest that planned
delivery system reforms that encourage
minorities to seek care within their established
networks may exacerbate surgical disparities.
Funding Source(s): AHRQ
Poster Session and Number: A, #256
Black-White Differences in Obstetric Quality
Indicators among New York City Hospitals
Elizabeth Howell, Icahn School of Medicine at
Mount Sinai; Jennifer Zeitlin, Epidemiological
Research Unit on Perinatal Health and Women’s
and Children’s Health at INSERM; Paul Hebert,
University of Washington School of Public
Health; Amy Balbierz, Icahn School of Medicine
at Mount Sinai; Natalia Egorova, Icahn School of
Medicine at Mount Sinai
Presenter: Elizabeth Howell, Associate
Professor, Icahn School of Medicine at Mount
Sinai
elizabeth.howell@mountsinai.org
Research Objective: Obstetrical quality
initiatives have focused on reducing
unnecessary interventions (overuse) which can
raise risks of morbidity for mothers and babies.
Rates of cesarean section, elective delivery, and
episiotomy have been proposed as measures of
obstetrical quality by the Joint Commission and
the National Quality Forum with this aim. Given
persistent higher perinatal and maternal
mortality and morbidity rates for blacks
compared with whites in New York City, we
aimed to assess how these indicators of quality
varied for black versus white women.
Study Design: This is a population-based study
examining three perinatal quality measures
(elective deliveries prior to 39 weeks, cesarean
sections in low-risk women, episiotomy rate).
Better hospital performance on these three
measures is indicated by lower rates.
Proportions were analyzed by Chi-square test
and multivariable analyses controlling for patient
age, comorbidities and hospital of delivery were
conducted using logistic regression analysis with
robust standard errors and conditional logistic
regressions stratified on hospital.
Population Studied: We used 2010 New York
City Vital Statistics linked with New York State
Discharge Abstract Data (SPARCs) to identify all
hospitalizations during which a delivery
occurred. We excluded 2 hospitals with annual
delivery volumes less than 5 births and included
40 hospitals in these analyses.
Principal Findings: Rates were higher for
whites than blacks for elective deliveries (21.9%
versus 15.5%; p=.003) and for episiotomies
(17.9% versus 8.6%; p<.0001). Although csection rates were similar for whites and blacks,
private hospitals had higher c-section rates than
public hospitals for both whites (21.5% versus
14.8%; p=.0004) and blacks (22.2% versus
14.7%; p=.0001). After adjusting for age and
comorbid conditions in logistic regressions the
black-white odds ratio for elective deliveries was
0.64 (95% CI: 0.56 - 0.72) Conditional logistic
regressions, which control for hospital level fixed
effects, yielded an odds ratio on black race that
was no longer significant (OR= 0.98; 95% CI:
0.83 - 1.16). After adjusting for age and
comorbid conditions, the black-white odds ratio
for episiotomy was 0.42 (95% CI: 0.39 - 0.45)
and remained significant in conditional logistic
regressions (OR=.60; 95% CI: 0.56 -0.65).
Conclusions: Two of three overuse measures
demonstrated higher rates among whites than
blacks in New York City hospitals. Hospital-level
factors explain the higher rates of elective
delivery among white women but do not account
for their higher rates of episiotomies. C-section
rates among low-risk women are higher at
private than public hospitals.
Implications for Policy, Delivery, or Practice:
Higher rates of overuse among whites than
blacks is consistent with literature on disparities
in other areas of medicine. Although black-white
differences in c-section rates did not exist,
efforts to reduce c-sections are needed in
private hospitals. While a focus on overuse may
be important for improving some parameters of
obstetrical quality, our data suggest it will likely
impact obstetrical care more for whites than
blacks. Our results raise the hypothesis that
obstetric quality measures related to underuse
may be better able to address persistent blackwhite disparities in perinatal outcomes.
Funding Source(s): NIH
Poster Session and Number: A, #257
Socioeconomic Status and Readmissions:
Evidence from an Urban Teaching Hospital
Jianhui Hu, Henry Ford Health System; Meredith
Gonsahn, Henry Ford Health System; David
Nerenz, Henry Ford Health System
Presenter: Jianhui Hu, Research Associate,
Henry Ford Health System
jhu1@hfhs.org
Research Objective: To understand how
elements of individual and neighborhood
characteristics and socioeconomic status
influenced probability of readmission under a
single, fixed organizational and staffing
structure, where variations in practice patterns
across hospitals had been eliminated.
Study Design: Retrospective cohort study.
Multivariate logistic regressions were used to
examine the associations between 30-day
readmissions and patient and neighborhood
characteristics.
Population Studied: All Medicare fee-forservice beneficiaries aged 65 years or older who
were discharged during the 2010 calendar year
from Henry Ford Hospital. The following
patients/admissions were excluded: (1)
admissions with an in-hospital death, (2)
patients discharged against medical advice, (3)
admissions for medical treatment of cancer, (4)
admissions for primary psychiatric disease, and
(5) admissions for rehabilitation care, fitting of
prostheses and adjustment device.
Principal Findings: After adjustment for
patients’ demographics and clinical conditions,
patients living in low-socioeconomic
neighborhoods (i.e., neighborhoods with high
poverty, low education, or low median
household income) were at greater odds of
being readmitted in our three individual-effect
models, and patients living in high-poverty
neighborhoods were twenty-four percent
(p=0.012) more likely to be readmitted than
others in our final combined model. Patients who
were currently married were less likely to be
readmitted (OR=0.80, p=.001).
Conclusions: Our findings added to the sparse
empirical evidence that socioeconomic
disparities in readmission risk still exist, even
after variations in practice patterns across
hospitals have been eliminated. The effects of
the socioeconomic status variables, measured
at neighborhood level, may reflect community
phenomena like access to primary and/or postdischarge care, availability of community
resources such as public transportation, grocery
stores and pharmacies, presence of social
support connections among neighbors, and
health-related features of the built environment
(e.g., age of housing stock).
Implications for Policy, Delivery, or Practice:
The recent CMS Hospital Readmissions
Reduction Program has focused attention on
ways to reduce readmissions and on factors
contributing to readmissions. One key debate
around the policy is the absence of adjustment
for patient socioeconomic status when
calculating the readmission rates, since it could
possibly mask disparities in quality of care. The
present study found that individual and
community-level socioeconomic factors were
related to probability of readmission, even in a
data set in which potentially relevant factors like
hospital medical staff structure, medical record
system, nurse staffing, and discharge planning
resources were controlled. Current risk
adjustment models used for measures like
hospital readmission include clinical variables
such as comorbidity and disease severity,
reflecting a view that hospitals and other
providers should not be held accountable for the
effects of those factors on quality measures like
readmission. The question of whether hospitals
should be held accountable for the effects of
factors like poverty, illiteracy, lack of English
proficiency, or lack of social support in the
patients and communities they serve has not yet
been resolved.
Funding Source(s): No Funding
Poster Session and Number: A, #258
Use Of Electronic Health Records and
Gender Differences in Healthcare Utilization
Jie Huang, Kaiser Permanente Northern
California; Richard Brand, Department of
Epidemiology and Biostatistics, University of
California at San Francisco; John Hsu, 6Mongan
Institute for Health Policy, Massachusetts
General Hospital, Harvard Medical School; Ilana
Graetz, Department of Preventive Medicine,
University of Tennessee Health Science Center;
Marc Jaffe, Department of Medicine and
Endocrinology, The Permanente Medical Group,
South San Francisco; Dustin Ballard,
Department of Emergency Medicine, The
Permanente Medical Group, San Rafael; Bruce
Fireman, Division of Research, Kaiser
Permanente Northern California; Mary Reed,
Division of Research, Kaiser Permanente
Northern California, Oakland
Presenter: Jie Huang, Statistical Demographer,
Kaiser Permanente Northern California
jie.huang@kp.org
Research Objective: Studie