Disparities

advertisement
Disparities
Call for Papers
Findings in Health Care Disparities on the Road Less
Traveled
Chair: Vanessa Sheppard, Georgetown University
Medical Center
Sunday, June 26 • 10:30 am – 12:00 pm
●Disparities in the Management of Early Breast Cancer
among Asian Americans and Pacific Islanders
Rebecca P. Gelber, M.D., James W. Davis, Ph.D., Todd B.
Seto, M.D., MPH
Presented By: Rebecca P. Gelber, M.D., General Internal
Medicine Fellow, Internal Medicine, Harvard Medical School,
Massachusetts Veterans Epidemiology Research and
Information Center; University of Hawaii John A. Burns School
of Medicine, 790 Boylston Street, 24F, Boston, MA 02199; Tel:
(808)382-7767; Email: rgelber@hsph.harvard.edu
Research Objective: Despite consensus guidelines for the
management of breast cancer, many patients do not receive
standard therapy. Although this discrepancy may be partly
related to patient ethnicity, little is known about breast cancer
care among Asian Americans and Pacific Islanders (AAPIs).
Objective: To evaluate the management of early stage breast
cancer among AAPI women.
Study Design: Retrospective analysis using data from the
Hawaii Tumor Registry, a Surveillance, Epidemiology, and End
Results (SEER) Program population-based registry, linked to
administrative data from Hawaii's largest health insurer.
Population Studied: We studied the first course of therapy for
2038 women (937 Japanese, 144 Chinese, 236 Filipino, 295
Hawaiian, 426 white; mean age, 59+/-13 years) diagnosed with
early stage breast cancer (stages I, IIA or B, IIIA) in Hawaii
from 1995-2001. We evaluated the use of standard-of-care
treatments as defined by NCI/NIH guidelines, including (1)
breast-conserving surgery (BCS); (2) radiation therapy
following BCS; and (3) chemotherapy for women with nodal
spread. We used logistic regression to examine the
association between ethnicity and quality of care.
Principal Findings: Overall, 1227 (60.2%) women had stage I
disease, 752 (36.9%) stage IIA or B, and 59(2.9%) stage IIIA.
Only 1623 (79.6%) of all women received BCS, and 64.9% of
these received radiation therapy. Of those with node-positive
disease (n=526), 82.7% received chemotherapy. Japanese and
Hawaiian women were significantly less likely than whites to
receive BCS after controlling for age, year of diagnosis, rural
residence, tumor size, grade, nodal status, estrogen-receptor
status, history of prior cancer, Charlson comorbidity index,
health plan typ (HMO, fee-for-service, Medicare), and income
(OR, 95%CI: 0.61, 0.42-0.88, for Japanese; 0.63, 0.40-0.98, for
Hawaiians; 0.76, 0.42-1.35, for Chinese; 0.68, 0.42-1.10, for
Filipinos). Filipino women were significantly less likely than
whites to receive radiation after BCS (OR, 0.60; 95%CI, 0.390.91), while other AAPIs were treated similarly to whites (OR,
95%CI: 0.84, 0.61-1.16, for Japanese; 1.06, 0.69-1.62, for
Hawaiians; 1.46, 0.83-2.56, for Chinese). AAPI women were as
likely as white women to receive chemotherapy for nodepositive disease (OR, 95%CI: 1.94, 0.84-4.48, for Japanese;
0.92, 0.36-2.33, for Hawaiians; 0.64, 0.19-2.12, for Chinese;
0.81, 0.32-2.10, for Filipinos).
Conclusions: We found significant disparities in the
management of early stage breast cancer among AAPI
women, with Japanese, Hawaiians, and Filipinos being
particularly less likely to receive standard-of-care treatment, as
compared to white women.
Implications for Policy, Delivery, or Practice: Further study
is needed to determine the reasons for the observed
differences in care and to examine the impact of these
disparities on health outcomes.
Primary Funding Source: AHRQ
●Medicare+Choice Enrollment and Plan Benefits Among
Black, Hispanic and White Medicare Enrolled VA-Using
Veterans in CY2000
Dolly A. John, MPH, Raji Sundaravaradan, Jessica Davila,
Ph.D., Iris I. Wei, DrPH, Margaret Byrne, Ph.D., Laura
Petersen, M.D., MPH, Debora A. Paterniti, Ph.D., Nora
Osemene, PharmD, Robert O. Morgan, Ph.D.
Presented By: Dolly A. John, MPH, Project Coordinator,
MEDVAMC (152), Houston Center for Quality of Care and
Utilization Studies, 2002 Holcombe Boulevard, Houston, TX
77030; Tel: (713) 794-8663; Fax: (713) 748-7359; Email:
dollya.john@med.va.gov
Research Objective: Prior studies examined enrollment of
Medicare enrolled VA-using veterans (MEVV) in
Medicare+Choice (M+C, now called Medicare Advantage)
plans but not enrollees’ plan benefits. We examined
enrollment in M+C plans of all MEVV in CY2000 and
enrollees’ monthly premiums, prescription drug coverage and
plan benefit packages. We compared these across
race/ethnicity groups.
Study Design: VA-using (any use between CY1996 and
CY2000) veterans were matched to CY2000 Medicare
Denominator, Group Health Plan and Benefits Information
Form (BIF) plan benefits data to identify veterans (age>=65)
enrolled in Medicare and in a M+C plan in CY2000. CMS’s
plan service area data identified 261 active M+C contracts
operating in 817 counties. Generosity of benefits was scored
on a combination of physician and hospital co-pays,
prescription drug and supplemental benefits (vision, dental,
hearing and podiatry) available for that plan. We predicted
enrollment in a M+C plan using univariate and multivariate
logistic regression models. We also controlled for income, VA
priority level and metropolitan status of the county.
Population Studied: 1,919,790 Black, Hispanic and White
MEVV in CY2000; 282,939 were M+C enrollees who represent
87% of all Black, Hispanic and White MEVV enrolled in any
type of Medicare managed care plan. We excluded from our
analyses those in other race/ethnic groups and those in
hospice.
Principal Findings: Compared to 60% of all White MEVV
living in a county with an active M+C plan, a greater
proportion of Black (72%) and a smaller proportion of
Hispanic (51%) enrollees lived in a county with an active M+C
plan in CY2000. Overall, among those with an active M+C
plan in their county, 25% of White (n=251,596), 21% of Black
(n=26,639) and 34% of Hispanic (n=4,704) MEVV were
enrolled in a M+C plan. Even after adjusting for the presence
of multiple plans, availability of benefits ($0 monthly
premium, prescription drug and supplemental benefits), age,
gender, income, Medicare state buy-in, VA priority status,
CY2000 VA use and the metro status of their county of
residence, Black enrollees were less likely (OR=.73, 95%CI: .71.74) and Hispanic enrollees were more likely (OR=1.62, 95%CI:
1.56-1.69) to be enrolled in a M+C plan than White enrollees.
Furthermore, more Black and Hispanic enrollees than White
enrollees were in plans with a $0 monthly premium (80% and
83% versus 70%, p<.0001), some drug coverage (96% and
94% versus 85%, p<.0001) and a high benefits package (82%
and 91% versus 78%, p<.0001) than White enrollees.
Conclusions: The majority of MEVV enrolled in a M+C plan in
CY2000 paid nothing, had some drug coverage and a high
benefits package. Compared to White M+C enrollees, Black
and Hispanic M+C enrollees were more likely to be in plans
without monthly premiums, with some prescription drug
coverage and a high benefits package.
Implications for Policy, Delivery, or Practice: Black and
Hispanic VA-using veteran M+C enrollees may have been
more vulnerable to M+C plan cost increases and benefit
reductions in recent years than White enrollees. How this
may have impacted their continuity and quality of care and
whether this results in disparities in health services use and
health outcomes needs further exploration.
Primary Funding Source: VA
●Racial Disparities In Primary Care and Utilization of
Health Services at the End of Life
Andrea Kronman, M.D., Jim Burgess, Ph.D., Arlene Ash, Ph.D.
Presented By: Andrea Kronman, M.D., Internal Medicine,
Boston University Medical Center, 91 E Concord Street MAT 2,
Boston, MA 02118; Tel: (617)414 6811; Fax: (617)414 6817;
Email: andrea.kronman@bmc.org
Research Objective: To determine the role of primary care in
the racial disparities of healthcare utilization at the end of life.
Study Design: This study is analyzes Medicare data to
examine racial differences in primary care visits and
hospitalizations for ACSC during the last 18 months of life.
Hospital admissions for two specific ACSC: diabetes (DM)
and congestive heart failure (CHF) were drawn from the final
6 months of life, while outpatient visit frequencies were from
the 12 months prior to that period. Utilization data was
obtained from the MedPAR, carrier and hospice files of
beneficiaries who died in the last 6 months of 2001, and
linked to NDI records to determine cause of death.
Demographics were obtained from the CMS 2000
Denominator file. Univariate and bivariate analysis was used
to assess differences between races.
Population Studied: The study population of Medicare
beneficiaries who died in the last 6 months of 2001 was
obtained from a random sample of 1 million beneficiaries
stratified by decedent status and race. Beneficiaries excluded
from the study were less than 66 years of age, not entitled to
both Medicare A and B, and not enrolled in a Medicare Fee for
Service plan continuously.
Principal Findings: Study sample (n=162,992) characteristics:
mean age 81 (range 66 - 98), female 56%, black 36%,
Hispanic 10%, Other 14%. Racial differences were found in
hospice use (24% white, 18 % black, 20 % Hispanic, p
<0.0001) and in-hospital deaths (38 % white, 46 % black, 52
% Hispanic, p<0.0001). Outpatient visits differed significantly
between races; particularly among those who had not had any
ambulatory care visits (16 % white, 26 % black, 24 % Hispanic
p<0.0001) or primary care visits (30 % white, 39 % black, 37
% Hispanic, p < 0.0001). DM was the cause of 4% of the
49,598 beneficiaries with DM who were hospitalized. There
were significant differences between races for those
hospitalized for DM (whites 2.7%, blacks 5.1 %, Hispanic 5.0
%, p<0.0001) and those hospitalized more than once (whites
7.7%, blacks 10.1 %, Hispanics 7.1 %, p<0.0001). CHF was
the cause of 18% of the 48,770 hospitalized beneficiaries with
a CHF diagnosis. There were racial differences for those
hospitalized for CHF (white 16%, black 18%, Hispanic 21%,
p<0.0001) and those with more than one admission (white 23
%, black 26%, Hispanic 26%, p <0.0001).
Conclusions: There are significant disparities between races
in frequency of primary care visits toward the end of life.
There are also significant racial disparities in the use of
hospice, and hospitalizations for diabetes mellitus and
congestive heart failure during the last six months of life.
Implications for Policy, Delivery, or Practice: This study
implies that receipt of primary care may contribute to
differences in care and higher costs of minorities at the end of
life. Increasing the provision of primary care to elderly minority
groups could improve management of chronic disease,
decrease costs, and improve care quality at the end of life. The
relationship between primary care and health care utilization
at the end of life warrants further study.
Primary Funding Source: No Funding
●Racial Disparities Among Medicaid Beneficiaries with
Mental Disorders
Mihail Samnaliev, Ph.D.
Presented By: Mihail Samnaliev, Ph.D., Project Associate,
Commonwealth Medicine Center for Health Policy and
Research, University of Massachusetts Medical School, Center
for Health Policy and Research, University of Massachusetts
Medical School, 222 Maple Avenue, Chang Building,
Shrewsbury, MA 01545; Tel: (508) 8561765; Email:
Mihail.Samnaliev@umassmed.edu
Research Objective: The ECA study and the National Health
Interview Survey (NHIS) have documented lower prevalence
of mental disorders among African Americans and Hispanics
compared to Caucasians. Yet, it is unclear if and to what
extent these disparities are the result of state healthcare
system factors as opposed to confounding ethnic differences
in health behavior and in the epidemiology of mental
disorders. In this study we explore interstate variation in rates
of diagnosis of mental disorders among racial and ethnic
groups. Second, we explore correlation patterns between
Medicaid eligibility, detection and service utilization rates. We
study all racial and ethnic groups, however, the primary focus
in this article is on the Hispanic population.
Study Design: A quasi-experimental design where states
represent the ‘experimental’ units. As a baseline for
comparison we use the 1999 NHIS which is a nationally
representative report of occurrence of mental disorders by
race. States were selected in a stratified sampling mode to
capture the variation in Medicaid programs regarding health
care coverage. Mental disorders were identified using ICD-9
codes and prescribed medications in 1999. Logit models were
used to estimate the probability of mental health diagnosis
and hospitalization.
Population Studied: Medicaid beneficiaries with mental
disorders in six states: Arkansas, Colorado, Georgia, Indiana,
New Jersey, and Washington in 1999.
Principal Findings: (1) Beneficiaries from Hispanic origin had
much lower rates of diagnosis with mental illness, compared
to Caucasians. This difference was as large as 12% and for
most states it was much larger than the approximately 2%
difference documented in the NHIS. There was a significant
interstate variation in rates of diagnosis of Hispanic
beneficiaries, ranging from 2.5% (in Georgia) to 11.9% (in
Arkansas). (2) Rates of diagnosis with mental illness among
Hispanics were positively correlated with the proportion of
Hispanics who were classified as ‘Disabled’ in the Medicaid
eligibility category. However, this finding should be viewed
with caution because of missing data on race in some of the
states. (3) Compared to Caucasians, the probability of
hospitalization was significantly greater among Hispanics and
there was a well expressed pattern where lower rates of
diagnosed mental illness among Hispanics was correlated
with greater probability of hospitalization for Hispanics vs.
Caucasians.
Conclusions: The greater disparities between Caucasians and
Hispanics as compared to findings from nationally
representative surveys, suggests that State Medicaid policies
can play an important role in alleviating or perpetuating
existing racial disparities. Significant interstate variation in
diagnosis rates further suggests that Medicaid programs in
some states are more effective than others in identifying the
Hispanic population with mental disorders. Further research
needs to explore the extent to which racial disparities are
driven by State disability standards. The correlation between
rates of diagnosis and probability of hospitalization raises a
question of whether underdiagnosis of MI among Hispanics
may have an adverse effect on health outcomes.
Implications for Policy, Delivery, or Practice: States with
relatively low detection rates of mental illness among the
Hispanic population may benefit from a more assertive
approach in identifying mental disorders among the Hispanic
population.
Primary Funding Source: RWJF
●Racial/Ethnic Disparity in Access to New Prescription
Drugs
Junling Wang, MS, C. Daniel Mullins, Ph.D., MS
Presented By: Junling Wang, MS, Doctoral Candidate and
Research Assistant, Pharmaceutical Health Services Research,
University of Maryland, 515 West Lombard Street, Room 272,
Baltimore, MD 21201; Tel: (410)706-1418; Fax: (410)706-5394;
Email: jwang005@umaryland.edu
Research Objective: To determine whether there are racial or
ethnic disparities in the utilization of new prescription drugs
and to examine the impact of the generosity of drug coverage
on these disparities.
Study Design: This is a non-experimental study.
Population Studied: The main data sources were Medical
Expenditure Panel Survey (1996-2001) and Mosby’s GenRx
Database. The former provided information on patient
characteristics and, for each prescribed medicines event
(when prescribed medicines were purchased or otherwise
obtained), drug name, NDC and payment sources; the later
provided information on drug approval dates. Drugs were
considered “new” if approved fewer than or equal to five years
ago at the time of survey data collection. A negative binomial
model was used to test the disparities across racial/ethnic
groups when controlling for other covariates; the dependent
variable was number of prescribed medicines event; the
independent variables were dummy variables for (1) black,
non-Hispanics, and (2) white Hispanics (white, non-Hispanics
was the reference group); generosity (percentages of drug cost
paid by health insurance); interaction terms between
generosity and the dummy variables for racial/ethnic groups;
age; gender; marital status; education; and dummy variables
for census regions, metropolitan statistical area, poverty
status, and self-perceived health status. The interaction terms
between generosity and racial/ethnic groups were included in
order to test whether effects of generosity differed across
racial/ethnic groups. Sensitivity analyses were performed
using variable definitions for the number of years post-launch
that a drug is considered new.
Principal Findings: Among 47,115 prescription users, 31,853
were white, non-Hispanics, 5,904 were black, non-Hispanics,
and 7,337 were white Hispanics. According to the least
stringent five-year criterion for defining new drugs, the mean
numbers of prescribed medicines events were 1.67, 1.36, and
1.07 among white, non-Hispanics, black, non-Hispanics, and
white Hispanics, respectively. The differences between white,
non-Hispanics and the two minority populations were
significant in the main model and most of the sensitivity
analyses. The negative binomial model showed that the
number of prescribed medicines events was 29-45% lower
(depending on the criterion for defining new drugs) among
black, non-Hispanics than among white, non-Hispanics
(P<0.01); the number of events among white Hispanics was
10-23% lower than among white, non-Hispanics but it was not
always significant. The interaction terms between
racial/ethnic groups and generosity were not significant,
suggesting that the effects of insurance generosity did not
differ across racial/ethnic groups. The gap between white,
non-Hispanics and black, non-Hispanics widened when the
criteria for new drugs became more stringent and the gap
between white, non-Hispanic and white Hispanics narrowed.
However, neither trend in the gaps was found significant in a
nested model.
Conclusions: There are racial disparities in the use of new
drugs but no ethnic disparities are confirmed. The generosity
of drug insurance coverage is positively correlated with the
use of new drugs and this correlation is fairly consistent
across racial/ethnic groups.
Implications for Policy, Delivery, or Practice: New
prescription drugs have potential benefits in improving life
expectancy and quality of life. Thus, measures should be
taken to address the racial/ethnic disparities in the use of new
drugs.
Primary Funding Source: No Funding
Call for Papers
Can Quality Improvement Programs Reduce
Health Disparities?
Chair: Kaytura Felix-Aaron, Bureu of Primary Health Care
Monday, June 27 • 9:00 am – 10:30 am
●Racial and Ethnic Differences in Use of High Volume
Hospitals and Surgeons
Andrew Epstein, MPP, Ph.D., Mark Schlesinger, Ph.D.,
Bradford Gray, Ph.D.
Presented By: Andrew Epstein, MPP, Ph.D., Assistant
Professor, Health Policy and Administration, Yale University,
60 College Street, New Haven, CT 06520; Tel: (203)785-6924;
Fax: (203)785-6287; Email: andrew.epstein@yale.edu
Research Objective: To quantify the direction and magnitude
of racial and ethnic differences in use of high volume hospitals
and surgeons among New York City-area residents
undergoing one of 10 procedures for which a volume-mortality
association has been identified for both hospitals and
surgeons.
Study Design: Using New York state hospital discharge data
from 1995-96 and 2000-01, we classified hospitals and
surgeons as high volume based on thresholds published in
volume-mortality studies. Linear probability models were
used to assess the overall independent association between
patient race/ethnicity and treatment by a high volume provider
for each time period, controlling for patient sex, age,
admission type/source, insurance status and number of
comorbidities. To decompose the influence of inter- and
intra-hospital referral patterns, patient residence ZIP Code
fixed effects were included in both hospital and surgeon use
models, distance to the closest hospital and closest high
volume hospital were included in the hospital use model, and
hospital fixed effects were included in the surgeon use
models.
Population Studied: Adults living in New York City, Nassau
County or Westchester County treated at hospitals in the same
area for: abdominal aortic aneurysm repair, carotid
endarterectomy, percutaneous coronary intervention, coronary
artery bypass graft surgery, total hip replacement, or for
surgery for pancreatic, breast, colorectal, gastric or lung
cancer.
Principal Findings: Black, Latino and Asian patients were
significantly less likely to use high volume hospitals in 8-9 of
the 10 procedures during 2000-2001. Although these
differences were reduced with the inclusion of patient
residence ZIP Code fixed effects, minorities remained less
likely to use high volume hospitals for 5-8 procedures. Similar
patterns were found for use of high volume surgeons: black
patients were significantly less likely to be treated by high
volume surgeons for all 10 procedures, Asian patients for 6,
and Latino patients for 7. After controlling for patient ZIP
Code and hospital fixed effects, the number of procedures
with significant racial differences (2, 3 and 3 respectively) and
the magnitudes of the differences became smaller. Ongoing
work explores the influence of specific hospital and
neighborhood characteristics on the use of high volume
providers. Analysis of the 1995-96 data yielded the same
pattern of findings.
Conclusions: For a range of surgical procedures for which
prior research suggests a volume-mortality association,
minority patients in the New York City area are less likely to
use high volume hospitals and surgeons. This relationship
can be attributed primarily to racial/ethnic differences in the
geographic distributions of patients and providers. For a
small number of procedures, however, minority patients are
less likely to be treated by high volume surgeons than are
white patients from the same ZIP Code admitted to the same
hospital.
Implications for Policy, Delivery, or Practice: Given
substantial differences in referral patterns by patient
race/ethnicity, strategies for centralization of surgical
procedures may have disproportionate effects for minorities.
An alternative strategy for improving survival for minority
patients would focus on quality improvement interventions
among providers primarily serving minorities.
Primary Funding Source: RWJF
●Disparities in Inpatient Quality of Care Measures by Race
and Ethnicity
Romana Hasnain-Wynia, Ph.D., David W. Baker, M.D., MPH,
Raj Behal, M.D., MPH, Joe Feinglass, Ph.D., David Nerenz,
Ph.D., Joel S. Weissman, Ph.D.
Presented By: Romana Hasnain-Wynia, Ph.D., Senior
Director, Research and Evaluation, Health Research and
Educational Trust, One North Franklin, 30th floor, Chicago, IL
60606; Tel: (312)422-2643; Fax: (312)422-4568; Email:
rhasnain@aha.org
Research Objective: This study addresses quality of care in
hospitals. Evidence indicates that quality measurement
initiatives, when linked to data on race and ethnicity, can
improve quality and reduce disparities. This study of 89
hospitals nationwide examines inpatient quality measures for
three conditions by race and ethnicity to see whether
disparities in care exist. This measure set is serving as the
basis for the national RWJF disparities initiative in cardiac
care.
Study Design: The study examines racial disparities in 18
inpatient quality of care measures for acute myocardial
infarction, heart failure, and pneumonia. Eighty-nine hospitals
are reporting the measures with patient race/ethnicity as part
of CMS’s overall Hospital Quality Initiative. We examined
mean rates and median times (in minutes for continuous
measures) for each of the measures for White, Black,
Hispanic, and Asian groups focusing on meaningful clinical
differences. For adjustment of the data, we examine
breakdowns by age, sex, admission source, admission type,
and payor type. We conduct multivariate analyses and
estimate separate models for each of the measures.
Population Studied: Eighty-nine hospitals reported measures
to the University Healthsytem Consortium from the third
quarter of 2002 to the first quarter of 2004 for a total of
118,279 admissions.
Principal Findings: In preliminary analyses we found
differences in mean rates and median times for measures
requiring personal interaction with patients. For patients with
heart failure, rates for providing discharge instructions varied
by race/ethnicity (White, 36.8%, Asian, 30.8%, Black, 26.1%
and Hispanic, 24.3%), as did rates for smoking cessation
counseling (White, 54.3%, Asian, 47.9%, Black, 47.8% and
Hispanic, 39.2%). For acute myocardial infarction, times and
rates varied for the following measures: time to thrombolysis
in minutes (Asian 68, White 105, Hispanic 220, and Black
476); time to PTCA in minutes (White 515, Asian 678, Blacks
878 and Hispanic 913); and rates for smoking cessation
counseling (White, 77.4%, Asian 68.6%, Black, 64.8% and
Hispanic 63.4%). For community acquired pneumonia, rates
for administration of pneumococcal vaccination varied by
race/ethnicity (White 26.3%, Hispanic 21.8%, Black 20.9% and
Asian 15.4%). Differences in rates of prescribing essential
medications (e.g., beta blockers for myocardial infarction)
were similar across racial/ethnic groups, in part because
compliance with these measures was near 100%.
Conclusions: This is the first study to look at differences in
quality of care by race/ethnicity for conditions measured by
the CMS Quality Initiative. Additional analyses are needed to
determine the statistical significance of these findings after
adjustment for potential confounders and to understand
whether these differences are explained by racial/ethnic
differences in where patients receive care or whether there are
disparities within institutions.
Implications for Policy, Delivery, or Practice: Hospitals’
efforts to reduce disparities should focus on procedures and
patient education and counseling. Hospitals should routinely
monitor disparities in quality of care for these indicators and
initiate interventions to reduce disparities. Standardizing care
plans to ensure high levels of compliance with recommended
processes of care may be the best strategy to eliminate
disparities.
Primary Funding Source: CWF
●Racial Disparities Remain As Quality Improves in
Medicare Managed Care
Sarah Scholle, MPH, DrPH, Beth Virnig, Ph.D., Sarah Shih,
MPH, Russ Mardon, Ph.D.
Presented By: Sarah Scholle, MPH, DrPH, Asst Vice
President, Research and Analysis, National Committee for
Quality Assurance, 2000 L Street NW, Suite 500, Washington,
DC 22031; Tel: (202) 955 1726; Fax: (202) 955 3599; Email:
scholle@ncqa.org
Research Objective: To determine whether secular trends in
quality of care for Medicare managed care enrollees between
1998 to 2003 differed for whites versus blacks. Previous
studies of Medicare managed care demonstrated significant
disparities in quality of care between black and white
beneficiaries. While managed care performance on key
indicators has improved since these reports, the impact on
racial disparities is unclear.
Study Design: This longitudinal panel study uses member
level data on HEDIS effectiveness of care measures submitted
to NCQA. These data were linked with CMS enrollment data
on race/ethnicity and region. Trends in performance on health
care quality indicators were analyzed using direct
standardization methods to control for patient age and sex.
Multivariate regression methods were used to control for
region. Performance rates adjusted for age, sex, and region
are reported.
Population Studied: Enrollees in Medicare managed care
organizations on performance measures for calendar years
1998 to 2003. The number of health plans available for
analyses ranged from 346 in 1998 to 170 in 2003 with
Medicare managed care enrollment between 5.4 to 7.5 million
members per each year. The number of patients varied by
measure; for example, analyses on diabetes care included
approximately 120,115 members in 2003.
Principal Findings: Quality of care improved for both Black
and white Medicare beneficiaries during the time period
studied, however, the pattern of improvement varied by the
type of measures. In measures addressing preventive and
screening services, quality generally improved and the gap in
performance between whites and blacks sometimes closed.
For example, hemoglobin A1c screening rates for diabetics
improved from an adjusted rate of 0.78 in 1999 to 0.82 in
2003 among whites and from 0.73 to 0.82 for blacks during
the same time period. For measures addressing control of
hemoglobin A1c and lipid levels, quality improved but the
racial disparity remained. For example, the rate of control for
lipid levels (LDL <130) increased from 0.58 to 0.62 for whites
and from 0.44 to 0.57 for blacks. For behavioral health
measures, improvement over time was negligible, and the
disparity between whites and blacks remained. For example,
the rate of outpatient follow-up within 30 days of a mental
health hospitalization changed from 0.53 to 0.56 for whites
and 0.37 to 0.40 for blacks. Further analyses will address
additional measures and control for socioeconomic status.
Conclusions: Quality performance in Medicare managed care
plans has improved for blacks and whites, however, the
disparity in performance remains, particularly for measures
assessing control of clinical outcomes and behavioral health.
Implications for Policy, Delivery, or Practice: Despite high
level attention and efforts to address health care disparities
between blacks and whites, disparities persist. A better
understanding of the source of disparities and new
approaches are needed to reduce these inequities in care.
Primary Funding Source: The California Endowment
●Solving Racial Disparities in Quality of Medicare
Managed Care: Geography Reconsidered
Sarah Scholle, DrPH, MPH, Beth Virnig, Ph.D., Russell
Mardon, Ph.D., Rich Mierzejewski, MS, MBA, Sarah Shih,
MPH
Presented By: Sarah Scholle, DrPH, MPH, Assistant VP,
Research and Analysis, NCQA, 2000 L Street NW, Suite 500,
Washington, DC 20036; Tel: 202-955-1726; Fax: 202-955-3599;
Email: scholle@ncqa.org
Research Objective: To examine whether the variation in
quality of care for Medicare managed care enrollees observed
in earlier reporting years persists in the most current 2004
reporting year and whether geographic disparity is greater or
less than racial disparity.
Study Design: This cross-sectional study uses individual-level
data for seven HEDIS effectiveness of care measures that is
linked with Medicare demographic information. Estimates of
rates of preventive care are calculated separately for blacks
and whites, overall and for each of nine census regions using
direct standardization controlling for age and sex.
Population Studied: The individual-level HEDIS data set
contains over 6 million records. We were able to link over 95
percent of records submitted by managed care plans to
Medicare information. The analysis is limited to persons
whose race is black or white. The number of individuals
included in analysis varies from measure to measure; for
example, there were 17,292 individuals included in analysis of
beta blocker use after heart attack.
Principal Findings: For all seven preventive services
measures, we observed overall racial disparity with blacks
receiving less care than whites. For example, the rate of beta
blocker after heart attack—BBH-- for whites was 90.2 percent
compared to 83.3 percent for blacks. Similarly, the disparities
in rate of 30-day follow-up after hospitalization for mental
illness were 55.2 percent for whites and 37.2 percent for blacks.
In addition to a global racial difference, we observed racial
differences within each census region. We also observed
considerable region-to-region variation. For example, for BBH,
the overall racial effect was a difference of 6.9 percent. The
lowest performing region had a rate of 75.4 percent while the
highest performing region had a rate of 93.9 percent. The OR
comparing the lowest and highest performing regions was
1.26 while the OR for blacks compared to whites was 1.08.
Likewise, for 30-day follow-up, the OR for race was 1.48 while
the OR comparing highest and lowest regions was 1.84. In
both cases, the regional level of performance was inversely
correlated with the percentage of the managed care
population that was black. Further analyses will examine
additional HEDIS effectiveness of care measures and control
for additional patient and plan factors.
Conclusions: Racial disparity in quality of care continues to
be observed in Medicare managed care plans and is
consistently found in measures of effectiveness of care.
However, there is also considerable regional variation in
quality that is independent of race but is correlated with higher
black populations.
Implications for Policy, Delivery, or Practice: While overall
and within region racial differences cannot be ignored, the
large region-to region variation and its correlation with
minority population suggest that equalizing regions may
prove to be an important strategy for improving preventive
services received by elderly black managed care
Primary Funding Source: California Endoment Foundation
●The Effect of Quality Improvement on Racial Disparities
in Diabetes Care
Thomas Sequist, M.D., MPH, Alyce Adams, Ph.D., Fang
Zhang, MS, Dennis Ross-Degnan, Sc.D., John Ayanian, M.D.,
MPP
Presented By: Thomas Sequist, M.D., MPH, Instructor in
Medicine and Health Care Policy, Department of Health Care
Policy, Harvard Medical School, 180 Longwood Avenue,
Boston, MA 02459; Tel: (617)432-3447; Email:
tsequist@partners.org
Research Objective: Since 1997 Harvard Vanguard Medical
Associates (HVMA) has implemented quality improvement
(QI) programs for diabetes including a new electronic medical
record (EMR), electronic physician reminders, disease
registries, and patient mailings. We evaluated whether these
QI programs were associated with changes in racial disparities
in care.
Study Design: We identified diabetic patients > 18 years old
using insurance claims and EMR data. We assessed 5 quality
measures based on American Diabetes Association
guidelines: annual retinopathy screening, annual HbA1c
testing, annual LDL cholesterol testing, and rates of
appropriate HbA1c control (< 7.0%) and LDL cholesterol
control (< 130 mg/dL). To assess changes in racial disparities
for each indicator, we used logistic regression to adjust for
race, year, race-year interactions, age, and sex, with general
estimating equations to control for repeated measures among
patients.
Population Studied: We identified adult diabetic patients
receiving primary care during 1997 to 2001 within HVMA, a
large multi-speciality group practice in Massachusetts.
Among 7,216 diabetic patients, 5,180 were white and 2,036
were black, with a mean age of 58 years and 51% female.
Principal Findings: Two quality measures demonstrated
relatively low overall baseline scores and subsequent large
improvement. The proportion of all patients receiving annual
LDL cholesterol testing increased from 40% in 1997 to 64% in
2001, while the white-black disparity decreased from 13% to
3% (p<0.001 for change in disparity). The proportion of all
patients with an LDL cholesterol < 130 mg/dL increased from
15% in 1997 to 43% in 2001, and the white-black disparity
decreased from 9% to 6% (p<0.001 for change in disparity).
The proportion of all patients with a HbA1c < 7.0% remained
low from 1997 (31%) to 2001 (33%), and the white-black
disparity remained constant at 10% (p=0.34 for the change in
disparity). Two quality measures demonstrated relatively high
baseline scores, and blacks were not less likely than whites to
receive services at baseline. The proportion of all patients
receiving annual HbA1c testing remained stable from 1997
(76%) to 2001 (77%), and there was no change in the whiteblack difference (p=0.34). The proportion of patients receiving
annual retinopathy screening exams decreased slightly from
1997 (74%) to 2001 (70%), and there was no change in the
white-black difference (p=0.83).
Conclusions: Racial disparities were diminished following
successful QI efforts in cholesterol management, but
persisted following comparable QI efforts in glycemic control
that had limited impact.
Implications for Policy, Delivery, or Practice: Generalized
quality improvement programs can result in reduced
disparities when overall quality of care improves rapidly, but
reducing other persistent disparities may require a specific
focus on minority health.
Primary Funding Source: Harvard Medical School Center of
Excellence in Minority Health and Health Disparities
Call for Papers
Disparities in Cardiac Care - New Lessons
Chair: David Nerenz, Henry Ford Health System
Monday, June 27 • 11:00 am – 12:30 pm
●Information and Quality Sorting by Ability to Pay in the
Market for Heart Surgeons
Rachel Kreier, Ph.D.
Presented By: Rachel Kreier, Ph.D., Professor, Economics,
Hofstra University, Barnard Hall, Hempstead, NY 11549; Tel:
(516)463-5659; Email: ecorek@hofstra.edu
Research Objective: This paper examines disparities in the
effects of New York's coronary artery bypass graft (CABG)
mortality rate reporting program.
Study Design: The investigation uses the 1988-1999 SPARCS
(Statewide Planning and Research Cooperative System)
administratively releasable in-patient data set, which contains
demographic, medical, and insurance coverage information
for patients discharged from New York hospitals. A strength of
the investigation is that it makes use of the relatively rich
information about primary and secondary insurance coverage
in the SPARCS data set. There are approximately 200,000
observations. Three separate econometric approaches are
applied to the data (OLS, binary logit using patient-surgeon
match as the unit of observation, and multinomial logit). The
investigation is informed by a theoretical model of quality
sorting (stratification) by ability to pay.
Population Studied: Patients undergoing CABG procedures
in New York State during the 1988-1999 period.
Principal Findings: Quality sorting of surgeons' services by
income group and type of insurance coverage increased
during the decade following the reporting program's
inception. Being African American also is a significant
predictor of seeing a surgeon with a poorer mortality rate in
the later years. The basic results are robust across
econometric specifications. Furthermore, the average
performance of the surgeons treating those in the lowest
income group shows negligible improvement, while the
average performance of surgeons treating upper income
groups shows substantial improvement.
Conclusions: The findings strongly suggest that the reporting
program has facilitated quality sorting by ability-to-pay. In
addition, there is cause for concern that the poor have not
shared in across the board quality improvements over time,
and that the reporting program may promote the continued
widening of the quality gap across income groups and groups
with different insurance coverage.
Implications for Policy, Delivery, or Practice: At least in this
market, performance reports appear to exert significantly
disparate effects on patients with different incomes, racial
backgrounds, and insurance coverage. Payment policies, such
as quality adjusted disproportionate share payments, or
Medicaid payments might offset this tendency if appropriately
designed, although they would also greatly strengthen the
incentives for providers to attempt to game the rating system
through patient selection.
Furthermore, the theoretical model is relevant in a variety of
contexts. These include, in addition to quality reporting
programs, managed competition approaches to delivering
health care, such as the Medicare + CHOICE/Medicare
Advantage program. These programs rely on market
mechanisms and consumer choice to implement quality/cost
trade offs. There is strong theoretical reason to expect them to
lead to quality stratification by ability to pay, as well.
Primary Funding Source: No Funding
●Do Physicians Discriminate Among Patients of Different
Race when Referring to High Quality Cardiac Surgeons?
Dana Mukamel, Ph.D., David Weimer, Ph.D., Alvin Mushlin,
M.D.
Presented By: Dana Mukamel, Ph.D., Associate Professor &
Senior Fellow, Center for Health Policy Research, University of
California, Irvine, 111 Academy Way, Irvine, CA 92697; Tel:
(949) 824-8873; Fax: (949) 824-3388; Email:
dmukamel@uci.edu
Research Objective: Prior studies found that Blacks in New
York State receive CABG surgery from cardiac surgeons of
lower quality, when quality is measured by the surgeon’s riskadjusted mortality rate (RAMR). The objective of this study
was to investigate whether this phenomenon can be explained
by 1) the fact that Blacks are seen by physicians whose
referrals in general are to lower quality surgeons or 2) whether
referring physicians refer their white and black patients to
different surgeons.
Study Design: Medicare claims data were used to identify the
surgeon performing the CABG surgery and the referring
physician. RAMRs for each surgeon were obtained from the
New York State Cardiac Surgery Report. A comparison of the
average surgeons’ RAMRs for referring physician who referred
only white patients and those who referred both blacks and
whites was made to answer the first question. To answer the
second question, hierarchical linear models were estimated to
predict the surgeon’s RAMR, allowing for both a random
intercept and a random race slope. The random race
coefficient was conditioned on characteristics of the referring
physician.
Population Studied: 22,571 New York State fee-for-service
Medicare enrollees having CBAG surgery in 1997-1999.
Principal Findings: The findings confirm both hypotheses. 1)
Physicians with patients of both races refer all their patients to
surgeons with higher RAMR compared to physicians who
have only white patients (p<0.000). The difference is about
10% of the state average mortality rate. 2) Physicians with
patients of both races refer their black patients to surgeons
with higher RAMR (p<0.002) compared with the surgeons to
which they refer their white patients. This disparity is not
modified by the age or specialty of the referring physician. It
does, however, decrease (i.e. less disparity relative to white
referrals) with the number of black patients in the practice of
the referring physician and if the surgeon practices in the
hospital to which the referring physician makes most of her
cardiac referrals (not only CABG). These analyses controlled
for other potential explanations for differential referrals,
including patient gender, age, co-morbidities, emergency
procedure, distance, education, and income.
Conclusions: The disparity in access of Blacks to high quality
cardiac surgeons is explained by several aspects of the referral
process. About half of the disparity can be explained by who
the referring physician is and the other half is due to
differential treatment of minority patients by same physician.
Implications for Policy, Delivery, or Practice: Because racial
disparities in access to high quality surgeons are caused by
several factors, policy interventions should be multifaceted.
Interventions, such as education or feedback of information
about referral patterns, should be aimed at referring
physicians who refer their white and black patients differently.
Minority patients should be educated about the referral
patterns of their physicians. A “referral report card” might
serve as a vehicle for disseminating information about referral
patterns. Further research is required to understand why
physicians with a racially mixed practice refer all their patients
to lower quality surgeons. In particular, geographic
distribution of patients, referring physicians and surgeons
should be examined.
Primary Funding Source: NIA, Commonwealth Fund &
NCMHD
●Surgeons' Experience with a new Cardiac Surgery
Technique and Patient Race
Dana Mukamel, Ph.D., Laurent Glance, M.D., David Weimer,
Ph.D., James Jackson, Ph.D., Thomas Pearson, M.D., Ph.D.,
Todd Massey, M.D., Jeffrey Gold, M.D., Sheldon Greenfield,
M.D., Alvin Mushlin, M.D., M.Sc.
Presented By: Dana Mukamel, Ph.D., Associate Professor &
Senior Fellow, Center for Health Policy Research, University of
California, Irvine, 111 Academy Way, Irvine, CA 92697; Tel:
(949) 824-8873; Fax: (949) 824-3388; Email:
dmukamel@uci.edu
Research Objective: CABG surgery performed on a beating
heart without the use of cardiopulmonary bypass (off-pump
CABG) has been recently been introduced as an alternative to
conventional CABG surgery. The objective of this study is to
determine whether non-white patients treated by surgeons
who are inexperienced with this new, off-pump, technique are
more likely to undergo off-pump CABG compared with white
patients treated by the same surgeons.
Study Design: Retrospective analysis of the New York State
Cardiac Surgery Reporting System data for 1999. These data
include type of surgery, clinical risk factors that are likely to
influence the choice of off-pump vs. on-pump surgery (e.g.
degree of occlusion by vessel type), patient race and ethnicity,
and surgeon identifier. We estimated logistic models
predicting the use of off-pump vs. on-pump procedure. These
models controlled for clinical risk factors, patient race and
ethnicity, and surgeon volume of off-pump cases. To account
for clustering of patients within surgeons we estimated mixed
models with random physician effects. Patient race and
ethnicity were interacted with a variable indicating the
experience of the surgeon, measured by the number of offpump surgeries the surgeon performed.
Population Studied: All patients undergoing isolated CABG
procedures in NYS in 1999.
Principal Findings: Black and white patients treated by
experienced surgeons, with a high volume of off-pump
surgeries, had the same probability of undergoing off-pump
surgery. However, Blacks treated by surgeons with low
volumes of off-pump procedures were 1.9 times (p<0.01)
more likely to have off-pump surgery compared with Whites
treated by the same surgeons, after controlling for patient risk
factors. Similar trends were observed for Hispanics.
Controlling for the number of Black patients treated by the
surgeon did not change the study findings.
Conclusions: The finding that Blacks that are treated by
surgeons with low off-pump volume are more likely to
undergo off-pump surgery than Whites cannot be explained by
patient risks or the racial composition of the surgeon practice.
Implications for Policy, Delivery, or Practice: This study
suggests that surgeons who are inexperienced with the offpump procedure preferentially perform this new procedure on
non-White patients. Since less experienced surgeons are likely
to have worse outcomes, especially given that off-pump CABG
surgery is more technically demanding than on-pump CABG
surgery, this practice may increase the risk of poor CABG
outcomes for non-White patients. This practice raises ethical
concerns about disparities that should be addressed in further
research to understand its causes and to allow development
of appropriate policy interventions.
Primary Funding Source: NIA, Commonwealth Fund, and
NCMHD
●Regional Patterns of Cardiac Procedure Use Following
Acute Myocardial Infarction Among American Indians
Thomas Sequist, M.D., MPH, Alan M. Zaslavsky, Ph.D., James
M. Galloway, M.D., John Z. Ayanian, M.D., MPP
Presented By: Thomas Sequist, M.D., MPH, Instructor in
Medicine and Health Care Policy, Department of Health Care
Policy, Harvard Medical School, 180 Longwood Avenue,
Boston, MA 02115; Tel: (617)432-3447; Email:
tsequist@partners.org
Research Objective: The prevalence of coronary heart disease
is rising among American Indians (AI), but there is limited
evidence describing processes of care for AI with acute
myocardial infarction (AMI). We compared rates of cardiac
catheterization, percutaneous coronary intervention (PCI), and
coronary artery bypass graft surgery (CABG) between AI and
whites with AMI.
Study Design: We combined data from the Nationwide
Inpatient Sample and the Indian Health Service National
Patient Information Reporting System to identify AI receiving
care in both federal and non-federal hospitals during 1998 to
2001. Admissions for AMI were defined by a primary
International Classification of Diseases 9th Revision (ICD-9)
diagnostic code of AMI. Cardiac catheterization, PCI, and
CABG were identified using ICD-9 major procedure codes. We
fit multivariate logistic regression models to compare rates of
cardiac procedure use between AI and whites after adjusting
for age, gender, and Charlson comorbidity index. We
constructed similar models stratified by geographic region
and by the presence of diabetes, as well as restricted the
analysis of rates of PCI and CABG to patients undergoing
cardiac catheterization.
Population Studied: We identified 2,511 AI and 316,526 whites
over the age of 30 years admitted with AMI during the study
period. There were substantial admissions for AI in four
geographic regions, including the Eastern region (n = 662),
West North Central region (n = 447), West South Central
region (n = 562), and the Mountain region (n=840).
Principal Findings: AI were less likely than whites to undergo
cardiac catheterization in 3 of 4 geographic regions, including
the West North Central region [odds ratio (OR) 0.57, 95%
confidence interval (CI) 0.44-0.76], the West South Central
region (OR 0.32, 95% CI 0.24-0.43), and the Mountain region
(OR 0.52, 95% CI 0.37-0.72). AI were less likely than whites to
undergo PCI in these same regions, with odds ratios ranging
from 0.43 (95% CI 0.31-0.57) in the West South Central region
to 0.69 (95% CI 0.53-0.89) in the West North Central region.
Nationally, AI were less likely than whites to undergo CABG
(OR 0.76, 95% CI 0.60-0.95), however there was no regional
variation in this difference. The presence of diabetes did not
modify differences in cardiac catheterization or PCI, however
AI were less likely than whites to undergo CABG among
diabetic patients (OR 0.48, 95% CI 0.32-0.73), but not among
non-diabetic patients (OR 0.90, 95% CI 0.72-1.12). There were
no differences in rates of PCI and CABG between AI and
whites among those receiving cardiac catheterization.
Conclusions: Differences in cardiac procedure use are
concentrated in western regions of the country and are
especially related to access to cardiac catheterization. Future
research is needed to elucidate the mechanisms of these
differences.
Implications for Policy, Delivery, or Practice: Health
disparities research has often excluded American Indians due
to data limitations. This study demonstrates the importance
of evaluating this population, as well as highlights access to
cardiac catheterization as a significant mediator of differences
in access to coronary revascularization among American
Indians. Future research is needed to further understand the
mechanisms of this disparity and its impact on health
outcomes.
Primary Funding Source: No Funding Source
●Impact of Hospital Volume on Racial Disparities in
Cardiovascular Procedure Mortality
Amal Trivedi, M.D., MPH, Thomas D. Sequist, M.D., MPH,
John Z. Ayanian, M.D., MPP
Presented By: Amal Trivedi, M.D., MPH, Research Fellow,
Health Care Policy, Harvard Medical School, 180 Longwood
Avenue, Boston, MA 02139; Tel: (617)945-0887; Email:
trivedi@hcp.med.harvard.edu
Research Objective: Hospital volume is a powerful predictor
of mortality for many procedures. We assessed the relative
contribution of hospital volume to overall disparities in
cardiovascular procedure mortality.
Study Design: We analyzed 719,679 hospital discharges from
the Nationwide Inpatient Sample from 1998 to 2001 with
principal procedure codes for abdominal aortic aneurysm
(AAA) repair, coronary artery bypass graft (CABG),
percutaneous transluminal coronary angioplasty (PTCA), or
carotid endarterectomy (CEA). For each procedure we
compared in-hospital postoperative mortality rates for Blacks,
Hispanics, and Whites, adjusting for comorbid illness using
the Elixhauser method. The adjusted relation of race to
receiving care in low-volume hospitals(below median) was
assessed in logistic regression models controlling for age, sex,
median income, location (urban/rural), payer, admission type
(elective/urgent/ emergent), and comorbidity. We analyzed
the impact of hospital volume on racial disparities in riskadjusted outcomes by predicting mortality with and without
adjusting for hospital volume. All analyses used SAS-callable
SUDAAN to account for the complex sampling and clustering
of patients by hospital.
Population Studied: The Nationwide Inpatient Sample (NIS)
is a population-based sample of nearly 1000 nongovernmental acute care hospitals in the U.S.
Principal Findings: Blacks had greater risk-adjusted mortality
than Whites following elective AAA repair (OR 1.84, 95%CI
1.20-2.84), CABG (OR 1.19, 95%CI 1.06-1.33), and CEA (OR
1.56, 95%CI 1.07-2.27), but not PTCA. Hispanics had greater
mortality following AAA repair (OR 1.43, 95%CI 1.08-1.89), but
not other procedures. After adjusting for multiple
demographic and clinical characteristics, Blacks were
significantly more likely than Whites to receive AAA repair (OR
1.78, 95%CI 1.47-2.17), CEA (OR 1.77, 95%CI 1.60-1.97) and
CABG (OR 1.18 95%CI 1.12-1.24) in low-volume hospitals.
Blacks were not more likely to receive PTCA in low-volume
hospitals. Hispanics had significantly increased odds of
receiving all four procedures in low-volume hospitals relative
to Whites with adjusted ORs ranging from 1.72 (95%CI 1.671.78) for PTCA to 2.64 (95%CI 2.40-2.90) for CEA. Asians had
the highest use of low-volume hospitals with an increased
odds relative to Whites ranging from 1.96 (95%CI 1.84-2.08)
for PTCA to 3.28 (95%CI 2.76-3.89) for CEA. Hospital volume
was a significant predictor of mortality with hospitals in the
lowest volume quartile having a 28% (CABG) to 130% (AAA)
increase in risk-adjusted mortality relative to high-volume
hospitals. Adjusting for hospital volume, however, had little
effect (<6% change) on the relative odds of death for Blacks
and Hispanics versus Whites.
Conclusions: Though minority patients are more likely to
undergo cardiovascular procedures in low-volume hospitals,
this racial difference in hospital volume is unlikely to explain a
large portion of the racial and ethnic gap in post-procedure
outcomes.
Implications for Policy, Delivery, or Practice: Additional
research is needed to determine why minorities are more likely
to receive care in low-volume hospitals and experience higher
mortality independent of hospital volume.
Primary Funding Source: NRSA Institutional Training Grant
Call for Panels
Crossing the Cultural Chasm to Improve Quality of Care &
Reduce Health Disparities
Chair: Lisa Cooper, Johns Hopkins University
Monday, June 27 • 2:30 pm – 4:00 pm
●Understanding Barriers to Healthcare Among Latinos:
the Effects of Ethnicity, Culture Change and
Discrimination
Debra Joy Pérez, MA, MPA, Ph.D.
Presented By: Debra Joy Pérez, MA, MPA, Ph.D., Program
Officer/WKKELLOGG Fellow in Health Policy
Research/Multicultural Mental Health Research Fellow/NIMH
mentee, Research & Evaluation Unit/Interfaculty Program in
Health Policy at Harvard, The Robert Wood Johnson
Foundation/Harvard University, PO Box 3311, Princeton, NJ
08543; Tel: (609)627 5966; Email: djperez@fas.harvard.edu
Research Objective: This study tests the hypothesis that
changes due to acculturaion and discrimination are associated
with variation in reporting barriers to care among Latinos and
Non-Hispanic whites. Healthcare barriers are defined as
unmet healthcare needs and postponement of needed care. A
financial barrier is defined as trouble paying medical bills. The
cultural change factors assessed are components of
acculturation: language, identity, nativity, recency (generation)
and duration in the US. The discrimination factors include
perceived personal or group experience of discrimination.
Logistic Regression Models are used to test the key
relationships between cultural factors, discrimination and
three barriers to care.
Study Design: The data for this study was collected by the
Kaiser Family Foundation/The Pew Hispanic Center in their
2002 National Survey of Latinos from April 4 to June 11, 2002.
A nationally representative sample of 4,214 adults, 18 years
and older were selected at random for this phone survey. The
Random Digit Dial sample includes sub-samples of Mexicans,
Puerto Ricans, Salvadorans, Dominicans, Colombians and
Cubans. Of the total sample, 2,929 were self-identified Latino
or Hispanic. The sample was weighted to reflect the
distribution of adults in the US by country of origin, age,
gender and region. There are 1008 non-Hispanic whites in the
sample and 171 African Americans. Samples were stratified
into three groups: total sample of whites and Latinos (3937
cases), total Latino sample (2929 cases) and Immigrant
sample (2014). Analyses were conducted using chi square and
logistic regression. A complex multiple imputation procedure
was used to impute income values for 255 missing cases.
Population Studied: Latinos: Immigrants and nonimmigrants. Substantial subsamples of Puerto Ricans,
Cubans, Mexicans, Dominicans, Central Americans, South
Americans and Whites.
Principal Findings: Being Latino was not associated with an
increased likelihood of reporting financial barriers but it was
associated with a 1.86 likelihood of reporting unmet healthcare
needs and a .86 likelihood of reporting postponement of care
compared to whites. Latinos as a whole were no more likely
to report financial barriers as compared to whites (p=.37444.)
However, stratification by immigrant status revealed that nonimmigrant Latinos were statistically more likely to report
financial barriers compared to Whites (p=0.05). As a group
Latinos were more likely to report having unmet healthcare
needs compared to whites (p=.0100.) Perceptions of
postponement of care had an peculiar relationship to
race/ethnicity. Whites were more likely than Latinos to report
postponement of care and Non-immigrant Latinos were more
likely to report postponement of care compared to
immigrants. Latinos who have been in the US for more
generations report higher rates of all three barriers. Latinos
with a college degree were more likely than Whites with a
college degree to report unmet healthcare needs but they were
as likely to report financial barriers and postponement of care
as Whites. Latino rates of barriers increase with each
generation. Latinos in the 1st generation reported less barriers
to care than Latinos in older generations. First generation
Latinos reported a lower rate of financial barriers, unmet
healthcare needs and postponement of care compared to
Latinos in 2nd generation and beyond. Twenty percent of 1st
generation Latinos reported financial barriers compared to
28% of 3rd/4th generation (p= .0172.) Thirteen percent of 1st
generation Latinos compared to 20% of 3rd/4th generation
reported having unmet healthcare needs (p=.0093.)
Seventeen percent of 1st generation Latinos reported
postponing care compared to 32% of 3rd/4th generation (p<
.000.) Logistic regression showed that discrimination is a
significant predictor of reporting healthcare barriers. Persons
who reported discrimination were three times as likely to
report an unmet healthcare need.
Conclusions: This study reveals the importance of cultural
components such as immigrant status and number of
generations in the US in reporting barriers to care. The
generation results suggest that perceptions of barriers in the
US change overtime. The increase in barriers among nonimmigrants was associated with increased exposure and
acculturation/integration to American mainstream. This
suggests that early immigrant status is protective of barriers.
The reduced reporting of barriers among immigrants may be
an indication of fewer attempts at receiving care rather than
fewer actual barriers. These findings suggest that
organizations interested in servicing Latinos may need to
develop different interventions to improve healthcare delivery
and quality of care for different generations of Latinos and at
different stages of their acculturation process.
Implications for Policy, Delivery, or Practice: Expansion of
healthcare access through insurance has been the preferred
policy solution to reducing racial disparities in barriers to care
between Latinos and Non-Hispanic Whites (Hargraves and
Hadley, 2003; Zuvekas and Taliaferro, 2003). Yet, insurance
only explains part of disparities. Other non-financial and
“unobserved factors” such as culture and discrimination are
significant predictors of access to and satisfaction with care
(Sver et. al.; Zuvekas and Taliaferro, 2003; IOM, 2002). The
findings suggest that policy orientation may need to adapt to
the mult-dimensional acculuration process of the Latino
population. First generation Latinos may need access-oriented
policies and interventions. Second generation Latinos may be
out of alignment compared to whites in identifying need and
seeking appropriate medical care. They may require consumer
activation policies and interventions. Third generation and
beyond may benefit from efforts to improved the provider
patient relationship and policies that reduce discrimination in
the healthcare setting. The mechanisms explored in this
research could help identify avenues for policy change to
eliminate racial/ethnic disparities in healthcare.
Primary Funding Source: WKK, NIMH
●Cultural Concordance Between Patient and Primary Care
Provider and Cervical Cancer Screening
Ninez Ponce, MPP, Ph.D., Melissa Gatchell, MPH, Jan Liu,
MPH, Juanita Dimas, Ph.D., Kelvin Quan, MPH, JD
Presented By: Ninez Ponce, MPP, Ph.D., Assistant Professor,
Health Services, UCLA School of Public Health, 31-254 B CHS,
Los Angeles, CA 90095; Tel: (310)206-4021; Email:
nponce@ucla.edu
Research Objective: This study examines the impact of
cultural concordance, measured by language, racial and
gender matches between patients and their providers on
cervical cancer screening rates. We hypothesize that a
language, racial and/or gender match will increase screening
rates, resulting in fewer missed opportunities for detecting a
virtually preventable cancer. We examine this question using
Medicaid managed care data from one of the most racially
diverse counties in California.
Study Design: We fit multivariate logistic models to test the
effect of language, ethnic, and/or gender match on Pap tests
for the most commonly spoken Medicaid beneficiary
languages: Vietnamese, Spanish, Farsi and English. We
adjusted for beneficiary characteristics: age, medically needy
designation; and provider characteristics: international
medical graduate, years worked at site, medical school
graduation year, specialty, and city of practice. We also
estimated random effects models with PCPs as the second
level, but found low intraclass correlation at the PCP-level.
Relative risks and predicted probabilities from the one-level
models were computed to evaluate effect sizes.
Population Studied: Female members of Alameda County’s
local Medicaid managed care plan, ages 25 to 64 with at least
11 months coverage in 1999 and 2000, and whose primary
care physicians (PCPs) were office-based (n=1140
beneficiaries, 21 PCPs). Beneficiary profiles were linked to
laboratory data on Pap test screening and to provider-level
information. For provider level data, we drew on two sources:
proficiency in languages other than English was assessed from
a provider language survey conducted by the Medicaid
managed care plan; medical training institution, specialty, and
year of graduation was derived from the American Medical
Association master files.
Principal Findings: A language match improved the
likelihood of receiving a Pap test among Spanish speaking
women (RR 2.17; 95% CI 1.93, 2.43), decreased the likelihood
for Vietnamese speakers (RR 0.88; 95% CI 0.84, 0.93), and
had no effect for Farsi-speakers. Having a female doctor was
important for Spanish speaking women, but only if that doctor
also spoke Spanish: the predicted probability for a Pap test
increased from 50% (gender match) to 85% (gender match
and language match). Racial concordance had the strongest
effect for African American women (RR 1.36; 95% CI 1.30,
1.41). There was no racial concordance effect for Latinas.
Conclusions: Our findings suggest that contracting with
Spanish bilingual providers benefits Spanish-speaking women.
While language match conferred a negative or nil benefit to
Vietnamese and Farsi speakers, a majority of these women
had PCPs who spoke their primary languages. This suggests
that efforts to raise screening rates for this group need to
target their bilingual providers. Racial concordance had the
most profound effect for African American women, inferring a
strong cultural component in the patient-physician encounter.
The lack of a measured effect on racial concordance for Latino
women resulted from the scarcity of Latino providers in the
sample, reflecting the under-representation of Latino
physicians in California.
Implications for Policy, Delivery, or Practice: Our study
provides evidence that improving provider-patient cultural
concordance can benefit some groups, and identifies training,
reimbursement and educational areas needed to raise the
quality of provider-patient encounters.
Primary Funding Source: California Cancer Research
Program
●Treatment Nihilism: Exploring Attitudinal Factors that
may Contribute to Disparities in Health Care
Dana Gelb Safran, Sc.D., Sara L. Thier, MPH, Ira B. Wilson,
M.D., SM, William Bender, MPH, Theresa Sommers, BA,
William H. Rogers, Ph.D.
Presented By: Dana Gelb Safran, Sc.D., Director, The Health
Institute, Institute for Clinical Research and Health Policy
Studies, Tufts-New England Medical Center, 750 Washington
Street, Box 345, Boston, MA 02458; Tel: (617)636-8611; Fax:
(617)636-8351; Email: dsafran@tufts-nemc.org
Research Objective: To explore attitudes of “treatment
nihilism” expressed by Medicare beneficiaries aged 65 and
older, and to examine whether these attitudes differ by race
and ethnicity. Extensive research documents treatment
differences that are associated with patients’ socioeconomic
characteristics – most notably, race and ethnicity. However,
little is understood about the mechanisms by which treatment
differences arise, and in particular, the extent to which they
represent clinicians’ deference to patient preferences as
opposed to other factors.
Study Design: A longitudinal observational study of Medicare
beneficiaries aged 65 and older from 13 states nationwide
(MA, NY, PA, FL, IL, MN, TX, NM, AZ, CO, CA, OR, WA).
From 1998 to 2002 we monitored participants’ primary care
relationships and health outcomes using a combination
annual surveys and administrative data obtained from the
Centers for Medicare and Medicaid Services (CMS). The
2002 questionnaire asked a series of questions concerning the
subject’s likely actions if told by their primary physician that
they had a “serious medical illness such as cancer or heart
disease.” Analyses presented here evaluate responses to an
item asserting: “I would seek as little medical care as possible
and let nature take its course.” We term this treatment
nihilism (TN).
Population Studied: Respondents to the 2002 questionnaire
(n=3449) are included in analyses that model “treatment
nihilism” as a function of patients’ age, sex, race (white,
African-American, Asian, other), Hispanic ethnicity, years of
education, household income, functional health status,
number of chronic conditions, and trust in their primary
physician.
Principal Findings: Overall, 10.5% of seniors revealed
attitudes of TN. African-American and Hispanic elderly were
more likely to report TN than non-Hispanic whites (23.9%,
22.5%, 9.7%, respectively, p<.001). Among those with less
than a high school education, 24.0% expressed TN, compared
with 8.9% among high school graduates (p<.001). In
multivariate models controlling only for age, sex and
race/ethnicity, TN was significantly associated with older age
(OR=1.07 per year, p=.001), African-American race (OR=3.60,
p=.001) and Hispanic ethnicity (OR=3.06, p=.001). However,
after controlling for socioeconomic status (education,
income), health (functional status, disease count), and patient
trust, the effects of race and ethnicity were mitigated
substantially (p>.06). Low educational attainment, low
household income, low mental health status and older age
were significantly associated with TN (p<.001). Patients with
more chronic conditions were less likely to express TN
(p<.01), as were those with more trusting primary care
relationships (p<.001).
Conclusions: Attitudes of TN are not rare among elderly
adults, and are significantly higher in select subgroups.
Socioeconomic factors (education, income), mental health
and age emerged as the principal characteristics associated
with preferring to “letting nature take its course” rather than
seeking medical treatment. When these variables are masked
or ignored, the attitudes appear to be attributes of AfricanAmerican race and Hispanic ethnicity.
Implications for Policy, Delivery, or Practice: The findings
highlight that, with extensive evidence of treatment disparities
in health care, attention must turn to understanding and
addressing the mechanisms by which these occur. For
clinicians and health care organizations, the findings
underscore that there is no short cut to knowing the values
and preferences that patients hold, but rather, these must be
uncovered, discussed and sometimes negotiated with each
individual patient and each episode of illness.
Primary Funding Source: AHRQ, Robert Wood Johnson
Foundation
●Disparities in Access to Care and Health Care Utilization:
Does Provider-Patient Race/Ethnicity Mix Matter?
Evidence from MEPS 2002
Jaeun Shin, Ph.D., Sangho Moon, Ph.D.
Presented By: Jaeun Shin, Ph.D., Professor, KDI School of
Public Policy and Management, 207-43 Cheongnyangri2-dong,
Dongdaemoon-gu, Seoul, 130-868; Tel: (822) 3299-1037; Fax:
(822) 3299-1240; Email: jshin@kdischool.ac.kr
Research Objective: Racial and ethnic disparities in access to
care and health care utilization are well documented. Health
insurance status and socioeconomic characteristics are often
known to produce barriers to care and use of health care
services among racial minorities. However, these factors do
not explain racial disparities. Though previous research has
raised the possibility that provider-patient race/ethnicity mix
may cause this disparity, few studies examine this issue
empirically. The purpose of this study is to understand
differences in access to care and health care utilization
between individuals with providers of identical race/ethnicity
and those with providers of different race/ethnicity, and to
assess the health policy implications for reducing racial/ethnic
disparities in the health care provision.
Study Design: Race/ethnicity mix is identified for respondents
to the 2002 Medical Expenditure Panel Survey (MEPS) and
their usual source of care (USC) provider. The access-to-care
variables are measured by its distinguishing attributes –
confidence in the USC provider (total 4 indicators) and
satisfaction with the USC provider (total 6 indicators). The
health care utilization variables include office-based service
use, hospital outpatient and inpatient services use,
ambulatory care use, dental service use, home health service
use and prescription drug use (total 11 service types).
Statistical significance of the differences in means is
calculated using t-tests.
Population Studied: The population studied includes
respondents of all ages from the 2002 MEPS who identify a
USC provider. The respondents are divided into two groups
according to the provider-patient racial/ethnic mix (PPMIX):
identical race/ethnicity (Hispanic-Hispanic, Black-Black, and
White-White, I-PPMIX) and diverse race/ethnicity (Hispanicnon Hispanic, Black-non Black, and White-non White, DPPMIX) for further analysis.
Principal Findings: Compared with respondents with DPPMIX, those with I-PPMIX are significantly (p<0.05) more
likely to have confidence in the provider (3 out of 4 indicators):
whether the person would go to the USC provider (1) for new
health problems, (2) for preventive health care, and (3) for
ongoing health problems. The satisfaction with the provider
also is positively and significantly (p<0.01) associated with
having I-PPMIX (all 6 indicators): (1) whether the provider
generally listens to the person and seek person’s advice when
choosing between treatments, (2) whether the provider asks
about and show respect for treatments other doctors may give
the person, (3) whether the provider asks the person to help
make decisions, (4) whether the provider gives the person
some control over treatment, (5) whether the provider explains
options to the person, and (6) whether the provider speaks
the person’s language or provide translator services.
Having I-PPMIX has no significant relationship with the level
of hospital outpatient total visits and total doctor visits.
Compared with respondents with D-PPMIX, hospital inpatient
total number of discharge (p<0.01) and total night stays
(p<0.1), total number of emergency room visits (p<0.01), and
home health total days and total agency-sponsored care days
(p<0.05) are significantly lower for those with I-PPMIX, while
office-based total visits and total doctor visits, total dental
visits, and total number of prescription drug refills are
significantly (p<0.01) higher for those with I-PPMIX.
Conclusions: This study provides preliminary evidence that
diverse race/ethnicity provider-patient mix may produce
barriers to care, represented as weak confidence in and low
satisfaction with the provider. With the identical providerpatient race/ethnicity mix, individuals may use more health
care services in the settings where the provision of medical
treatments involves the direct interaction between the provider
and patients, such as office-based services, dental service and
drug prescription service. Higher use of ambulatory service
and hospital inpatient services among those with D-PPMIX
may be contributed by the delayed delivery of necessary
medical care. The reluctance of patients to encounter the
provider due to the cultural discomfort in personal interaction
may lead to lack of the usual source of care, consequently
forcing the patients to depend on intensive and emergent
care. Also, the potential perceived racism on the provider side
against patients from different cultural background may
possibly be the reason for the disparity in health care.
Implications for Policy, Delivery, or Practice: Policy makers
can use results from this study in developing and improving
health care provision environments to reduce racial/ethnic
disparity in access to care and health care utilization. Further
research should pay attention to examine the influence of
provider-patient race/ethnicity mix and health outcomes of the
patients.
Primary Funding Source: KDI School of Public Policy and
Management; Sungkyunkwan University, Korea (South)
●Are Resident Physicians Prepared to Deliver Quality Care
to Diverse Population?
Joel Weissman, Ph.D., Joseph Betancourt, M.D., Elyse Park,
Ph.D., Minah Kim, Ph.D., Eric G. Campbell, Ph.D., Brian
Clarridge, Ph.D.
Presented By: Joel Weissman, Ph.D., Associate Professor,
Institute for Health Policy, MGH/Harvard, 50 Staniford Street,
9th floor, Boston, MA 02114; Tel: (617) 724-4731; Email:
jweissman@partners.org
Research Objective: The Institute of Medicine Report
“Unequal Treatment” cited cross-cultural training as a
mechanism to address racial and ethnic disparities in health
care, but little is known about residents' preparedness to
provide quality care to diverse populations or the barriers they
face in practice. This study explores residents' perceptions of
1) their preparedness to deliver care to diverse patients; 2) the
barriers they face in delivering such care; and 3) the
educational climate for cross-cultural training at their
institution.
Study Design: National survey of over 2000 residents in
seven specialties in their final year of training.
Population Studied: Residents in six specialties in their final
years of training.
Principal Findings: 97% of residents felt that it is
“moderately” or “very important” for physicians in their
specialty “to consider the patient’s culture when providing
care”. Residents reported that cross-cultural patient issues
“often” resulted in longer than average visits (41% of
residents), non-compliance with treatment (20%), and delays
obtaining consent (18%). Although only 8% of respondents
felt unprepared to care for different cultures generally, many
felt unprepared to deliver specific components of crosscultural care, including caring for patients with health beliefs
at odds w/ western medicine (25%), who were new
immigrants (25%), or whose religious beliefs affect treatment
(19%). In addition, 24% felt they lacked the skills to identify
relevant cultural customs. Major barriers to delivering crosscultural care included lack of time (56% saying it was a
moderate or big problem), inadequate training (31%), and lack
of role models (31%). 1/3 to 1/2 of Residents received little or
no instruction in specific areas of cross-cultural care beyond
what was learned in med school, and 41% (family medicine)
to 83% (surgery and OB/GYN) received little or no evaluation
on cross-cultural issues during their residencies.
Conclusions: Resident physicians preparedness to deliver
cross-cultural care lags well behind preparedness to provide
other key components of health care. Furthermore, our
results suggest that residents received mixed educational
messages. Although the topic of cross-cultural care was
perceived to be important, there was little clinical time to
address cultural issues, little training, and little to no formal
evaluation or role modeling of cross-cultural skills.
Implications for Policy, Delivery, or Practice: If eliminating
racial/ethnic disparities is in part contingent on cross-cultural
education, there is significant room for improvement in this
area. Training environments need to augment formal and
informal training mechanisms to enhance residents’
preparedness to deliver cross-cultural care to achieve the
goals set by the IOM.
Primary Funding Source: The California Endowment
Call for Papers
Communication & Care: The Role of Language
& Health Literacy
Chair: Anne Beal, The Commonwealth Fund
Monday, June 27 • 4:30 pm – 6:00 pm
●Understanding Adverse Events in Patients with Limited
English Proficiency
Chandrika Divi, MPH, Andrew Chang, JD, MPH, Jerod M.
Loeb, Ph.D.
Presented By: Chandrika Divi, MPH, Associate Project
Director, Center for Patient Safety, JCAHO, One Renaissance
Boulevard, Oakbrook Terrace, IL 60181; Tel: (630)792-5918;
Fax: (630)-792-4918; Email: cdivi@jcaho.org
Research Objective: To examine the epidemiology of adverse
events attributable to patient-provider communication
problems in limited English proficient (LEP) patients.
Study Design: Adverse event data were collected from
hospital incident reporting systems in 6 JCAHO accredited
hospitals from March 1, 2004 to August 31, 2004, and
stratified by English-speaking and LEP patients. Each hospital
provided de-identified incident reports pertaining to all LEP
patients and a random sample of 30 incident reports
pertaining to English-speaking patients. Each incident report
was examined and the adverse event was deconstructed using
the JCAHO Patient Safety Event Taxonomy (PSET), and
classified according to the primary, secondary, tertiary and
quaternary nodes of the classification scheme. Each classified
incident report was reviewed by three project team members
for concurrence on the classification categories used.
Classified incident reports that team members disagreed over
or required clinical interpretation, were further reviewed and
classified by a physician. All classified adverse events
pertaining to limited English proficient patients were
examined for language related causative or contributive
factors. The frequency of incidents categorized by the
quaternary, tertiary and secondary nodes of the PSET were
tabulated and compared between LEP and English-speaking
patients.
Population Studied: Adverse events related to Englishspeaking and limited English proficient adult patients in the
hospital setting.
Principal Findings: A total of 146 reported incidents met the
inclusion criteria for data review, and were classified and
recorded. Of these incidents, 38%(56)pertained to LEP
patients. For LEP patients, under the primary classification
category of Impact, “no harm” was recorded for 20%(11)
incidents and “some harm” was recorded for 80%(45)
incidents. Comparatively, for English speaking patients, “no
harm” was recorded for 26%(23) incidents and “some harm”
was recorded for 74%(67) incidents. Under the primary
classification of Type, “communication” was most commonly
recorded for incidents pertaining to LEP patients and Englishspeaking patients, 42%(24) and 45%(41), respectively. Under
the primary classification of Cause, for incidents pertaining to
LEP patients, 42%(24) incidents were due to organizational
factors and 39%(22) incidents were due to practitioner related
factors. For incidents pertaining to English speaking patients,
47%(42) incidents were due to organizational factors and
33%(30) incidents were due to practitioner related factors.
Conclusions: Results indicate that language issues can
impact patient-provider communication and contribute to
adverse events that harm hospitalized patients. However,
more data are required to validate the association between the
degree of harm and LEP status.This impact, nonetheless,
contributes towards the disparities in patient safety for
vulnerable patient populations experiencing language barriers
in the health care setting.
Implications for Policy, Delivery, or Practice: A review of the
literature has shown that that very few studies address the role
of language barriers in the occurrence of adverse events in
LEP patient populations. This project provides information on
the impact of LEP status on the nature and occurrence of
adverse events in hospitalized patients and identifies potential
preventive strategies. These findings will inform, guide and
facilitate the enhancement of accreditation standards directly
related to culturally and linguistically appropriate health care
services.
Primary Funding Source: CWF
●Is What We Have Here A Failure to Communicate? A
Statewide Evaluation of the Adequacy of Hospital
Interpreter Services for Patients with Limited English
Proficiency
Glenn Flores, M.D., Sylvia Torres, Linda Holmes, Debbie
Salas-Lopez, M.D., Sandy Tomany-Korman, MS, Mara
Youdelman, JD, LLM
Presented By: Glenn Flores, M.D., Director, Center for the
Advancement of Underserved Children, Pediatrics, Medical
College of Wisconsin/Children's Hospital of Wisconsin, 8701
Watertown Plank Road, Milwaukee, WI 53226; Tel: (414)4564454; Fax: (414) 456-6385; Email: gflores@mail.mcw.edu
Research Objective: 47 million Americans speak a nonEnglish language at home and 21 million (including 4 million
children) are limited in English proficiency (LEP). With 11% of
its population LEP, New Jersey (NJ) is a model state for
studying interpreter issues. The study aim was to assess the
adequacy of interpreter services (IS) in NJ hospitals in
meeting LEP patients’ needs, and to make actionable policy
recommendations on how best to meet the needs of NJ’s LEP
patients.
Study Design: Cross-sectional survey in which representatives
at all 122 NJ hospitals were contacted to answer 37 questions
on hospital and patient features, IS, and resources/policies
needed to provide quality interpreter services to LEP patients.
Population Studied: Hospitals in the state of New Jersey.
Principal Findings: 67 hospitals completed surveys (55%
response rate). A median of 9% of staff and 33% of MDs are
bilingual. A median of 93 patients per hospital need
interpreters (range=0-15,000), and the median IS budget is
$10,063 (range=$0-458,000). 97% of hospitals use phone IS.
87% of hospitals have no IS department, 19% offer no written
translation services, and 31% lack multilingual signs. Only 3%
of hospitals have a full-time interpreter, for a ratio of 1
interpreter per 235,769 LEP persons in NJ. 80% of hospitals
offer no staff training on working with interpreters. The
proportion of non-white patients is associated with providing
written translation (P<.01) and with staff training on working
with interpreters (P<.02), and the proportion of Medicare
patients is associated with having multilingual signs (P<.04).
Three hospitals had model programs with more extensive IS.
Hospitals stated the following would improve IS at their
institutions: more funding; federal/state interpreter
certification and guidelines; more MD/staff education on
interpreter use; listing locally available interpreters; and
identifying bilingual staff. Most said 3rd-party reimbursement
for IS would be beneficial, by reducing costs, adding full-time
interpreters, freeing staff for other services, meeting future
population growth, and improving communication and
education.
Conclusions: Most NJ hospitals have no formal interpreter
services department, 97% have no full-time interpreter, 80%
provide no staff training on working with interpreters, and
deficiencies exist in hospital signage and translation services.
Most NJ hospitals said 3rd-party reimbursement for
interpreter services would benefit their hospitals.
Implications for Policy, Delivery, or Practice: Based on
these findings, we suggest NJ provide third-party
reimbursement of IS statewide. If restrictive state budgetary
challenges preclude full implementation of third-party
reimbursement, we suggest considering one or more of the
following: 1) Provide Medicaid/SCHIP reimbursement for IS in
the emergency department; 2) Use federal Medicaid and
SCHIP matching reimbursement funds to train more qualified
medical interpreters; 3) Negotiate bulk discounts with
telephone interpreter services; 4) Consider using funds from
the state’s Office of Refugee Resettlement and/or English as a
Second Language Program; 5) Utilize employer pressure to
encourage greater access to healthcare services for LEP
employees; 6) Utilize innovative, cost effective technologies to
improve access to trained medical interpreters, such as
remote simultaneous interpretation and telemedicine
networks; 7) Consider implementing licensure requirements
and state regulations mandating adequate access to IS for LEP
patients; and 8) If other measures fail, consider legal options
to enforce compliance with Title VI of the Civil Rights Act.
Primary Funding Source: Office of Minority and Multicultural
Health, New Jersey Department of Health and Senior Services
●Pap Test Use Among English, Spanish, Cantonese,
Mandarin, Korean, Vietnamese, and Khmer Speaking
Women: Is there a Language Divide?
Ninez Ponce, MPP, Ph.D., Neetu Chawla, MPH, Susan Babey,
Ph.D., E. Richard Brown, Ph.D., Nancy Breen, Ph.D.
Presented By: Ninez Ponce, MPP, Ph.D., Assistant Professor,
Health Services, UCLA School of Public Health, 31-254B CHS,
Los Angeles, CA, CA 90095; Tel: (310)206-4021; Fax: (310)4559475; Email: nponce@ucla.edu
Research Objective: Racial and ethnic disparities exist in
cervical cancer screening . We examined whether primary
language use, measured by language of interview, contributes
to disparities in Pap test rates using a population-based
sample of California women.
Study Design: This study used data from the 2001 California
Health Interview Survey (CHIS). CHIS 2001 is a random-digit
dial (RDD) telephone survey of 55,428 households drawn from
every county in California and is the largest statewide health
survey conducted in the United States. The sample was
designed to provide estimates for California's overall
population, its major racial and ethnic groups, and a number
of ethnic subgroups. CHIS 2001 was conducted in six
languages: English, Spanish, Chinese (Cantonese and
Mandarin dialects), Korean, Vietnamese, and Khmer.
Multivariate logistic regression analyses were used to
determine factors associated with three-year Pap test use.
Factors studied include race/ethnicity and language of
interview, marital status, income, educational attainment, level
of acculturation, insurance status, usual source of care,
smoking status, area of residence, and health status.
Population Studied: The population studied was California
women, ages 25 to 64, who did not report a cervical cancer
diagnosis or a hysterectomy (n = 23,387).
Principal Findings: Adjusting for all other covariates, we
found that Asian women speaking English (RR 0.91; 95% CI
0.87, 0.95), Cantonese (RR 0.82; 95% CI 0.67, 0.93), Mandarin
(RR 0.78; 95% CI 0.62, 0.91), Khmer (RR 0.73; 0.33, 0.99), and
Korean (RR 0.92; 0.81, 1.00) were significantly less likely to
have had a Pap test in the past three years than Englishspeaking white women. In addition, multivariate results
showed that Spanish-speaking White women (RR 1.09; 95% CI
1.04, 1.10) and Spanish-speaking Latinas (RR 1.06; 95% CI
1.05, 1.07) were the most likely to be screened.
Conclusions: Large language disparities in cervical cancer
screening are present in California, where Asian women
speaking Cantonese, Mandarin, Khmer, and Korean had the
lowest rates of Pap tests compared to English-speaking
Whites. However, even English-speaking Asians had
considerably lower screening rates, suggesting that both
cultural and language barriers exist. Spanish-speaking women
are the most likely to report three-year Pap test use and
Vietnamese-speakers had screening rates comparable to
English-speaking White women.
Implications for Policy, Delivery, or Practice: Although
Federal civil rights law and policy require that most health care
providers make interpretation services available to patients
who need it, such laws and policies are difficult to enforce so
that barriers due to language may still exist. Our findings
suggest that in addition to English and Spanish, cancer
screening services should be available in multiple languages,
particularly Asian languages. Furthermore, patient and
provider awareness campaigns need to consider both
linguistic and cultural bridging strategies to raise the
screening participation among Asian women. Spanish and
Vietnamese language access efforts may have broader
coverage and comprehensiveness in promoting Pap test use
than similar efforts in other languages. These efforts could
serve as a model for improving screening rates among
Cantonese, Mandarin, Khmer, and Korean speakers in
California and in the United States.
Primary Funding Source: NCI, The California Endowment
●Reducing Disparities in Mental Health Care by Targeting
Language and Insurance Barriers
Tetine Sentell, Ph.D., Martha Shumway, Ph.D., Lonnie
Snowden, Ph.D.
Presented By: Tetine Sentell, Ph.D., Postdoctoral Fellow,
Psychiatry, UCSF, 2727 Mariposa Street, Suite 100, San
Francisco, CA 94110; Tel: (415)437-3075; Fax: (415)437-3020;
Email: tsentell@itsa.ucsf.edu
Research Objective: Untreated mental illness is a significant
policy problem with high fiscal and social costs, particularly
among racial/ethnic minorities. This study compared four
potential policy actions to address disparities in mental health
treatment: (1) providing health insurance, (2) providing health
insurance and minimizing language barriers, (3) providing
health insurance with mental health coverage, and (4)
providing health insurance with mental health coverage and
minimizing language barriers.
Study Design: Using the 2001 California Health Interview
Survey (CHIS), a representative sample of 55,428 California
adults, logistic regressions were estimated to predict unmet
need and to compare the statewide impact of the four policy
actions.
Population Studied: The sample included adults aged 18 to
64 who self-reported a mental health need (n=6,994; 16% of
non-elderly adult respondents). To maximize impact,
predictive models were estimated specifically for Californians
most likely to need, and least likely to receive, mental health
care (e.g., poor, limited education, limited English
proficiency).
Principal Findings: Forty-two percent of respondents had an
unmet mental health need. Race/ethnicity, lack of English
proficiency, and lack of insurance were significant barriers to
care. Fifty-three percent of Whites, 42% of African Americans,
31% of Asians, and 24% of Latinos received needed care. Fifty
percent of proficient English speakers received needed
services versus 9% of those with limited English proficiency.
Insured adults were twice as likely to have received needed
care than the uninsured when other variables were controlled.
Policy action 1— providing health insurance— increased
receipt of needed services dramatically among African
Americans, but did not greatly benefit Latinos and Asian/PIs.
Policy action 2 —providing health insurance and minimizing
language barriers — did increase receipt of needed services
among Latinos and Asian/PIs. Policy action 3— providing
insurance with mental health coverage— was particularly
beneficial to Asian/PIs, increasing receipt of needed services
to 93%. But policy action 4— providing health insurance with
mental health coverage and minimizing language barriers—
had the greatest benefit for Asians/PIs and Latinos
specifically, as well as overall, increasing receipt of services by
all disadvantaged Californians to 67%.
Conclusions: A high level of unmet need for mental health
care persists, particularly among minority groups with limited
English proficiency. Policy action appears to decrease
disparities among those who need care, though actions do
not equally benefit all groups. Tested policy actions appear to
particularly benefit Asian/PIs. Only 31% of Asian/PI CHIS
respondents received needed mental health services, but
under the final hypothetical policy action an estimated 97%
would receive needed care. Although tested policy actions
almost double receipt of needed services among Latinos from
24% (in the CHIS) to 45% (with the final policy action),
almost half of Latinos would still have unmet mental health
needs, suggesting the presence of other barriers that
perpetuate disparities among Latinos.
Implications for Policy, Delivery, or Practice: Improving
access to insurance and decreasing language barriers are
policy actions that appear to reduce racial/ethnic disparities in
unmet need for mental health services, indicating that seeking
mental health treatment, though a deeply personal choice, can
be influenced by large-scale, systemic changes amenable to
policy action.
Primary Funding Source: NIMH
●Improving Patient Safety through Informed Consent in
Populations with Limited Health Literacy
Helen Wu, M.Sc., Robyn Y. Nishimi, Ph.D., Christine M. PageLopez, Kenneth W. Kizer, M.D., MPH
Presented By: Helen Wu, M.Sc., Program Director, National
Quality Forum, 601 13th Street NW, Suite 500N, Washington,
DC 20005; Tel: (202)783-1300; Fax: (202)783-3434; Email:
hwu@qualityforum.org
Research Objective: In 2003, the National Quality Forum,
NQF, endorsed as voluntary consensus standards a set of 30
safe practices that have been demonstrated to improve patient
safety. This project evaluated number 10 of those safe
practices, which relates to informed consent and which is
particularly important in populations with limited health
literacy. This project sought to identify strategies to facilitate
widespread adoption of Safe Practice 10.
Study Design: The main component of Safe Practice 10 calls
for providers to ask each patient, or his/her legal surrogate, to
recount what he or she was told during the informed consent
discussion, with the objective of ensuring the patient
understands the proposed treatment and its potential
consequences. Also known as teach back, this practice has
been shown to improve patient recall and understanding, and
it allows providers to gauge patient understanding and
whether informed consent was actually achieved. Additional
specifications of Safe Practice 10 address other components
of the informed consent process, including consent forms,
verbal discussion, and the use of interpreters.
The project had three components: 1. an in-depth assessment
of the experiences of four hospitals that had implemented
teach back as a standard practice; 2. phone interviews with
providers that had not adopted Safe Practice 10 so as to
identify major barriers to implementation of the practice; and
3. a multi-stakeholder workshop to review and refine
preliminary project findings.
Population Studied: Patients undergoing invasive, noninvestigational, non-emergent procedures, with a special focus
on patients with limited health literacy - defined as those with
limited English proficiency, low literacy, and other difficulties
understanding
healthcare terminology and concepts.
Principal Findings: Providers who used teach back
demonstrated numerous benefits of the practice: surgical
errors and serious adverse events were avoided, efficiency was
improved and costs were lowered due to substantially lower
rates of surgical cancellations/delays, and trust was improved
between providers and patients. Teach back was found to be
an important practice to improve patient safety, quality, and
communication. Overall, these benefits far outweighed the
perceived time burden of asking patients to recount
information related to the informed consent. Healthcare
providers may face challenges in successfully adopting and
implementing Safe Practice 10, such as buy-in from physicians
about the importance of the practice, but those challenges can
be overcome if a sufficient case is made for the importance of
proper informed consent and the benefits of teach back.
Conclusions: Safe Practice 10 provides an evidence-based,
feasible, and usable strategy to enhance the safety of the
caregiving process.
Implications for Policy, Delivery, or Practice: Providing
informed consent is an ethical, professional, and legal duty of
physicians, but one that is often overlooked or given short
shrift. Patients who are well informed are more satisfied with
their care, more likely to have a good outcome, more trusting
of their providers, and more able to make decisions that
reflect their personal preferences and values. Effective
communication between providers and patients is central to
informed consent. Use of Safe Practice 10 increases the
effectiveness of provider-patient communication and may
reduce potentially adverse events.
Primary Funding Source: CWF
Call for Papers
Can the Appropriate Consumption of Health Care Reduce
Disparities?
Chair: Marsha Lillie-Blanton, The Henry J. Kaiser
Family Foundation
Tuesday, June 28 • 8:30 am – 10:00 am
●Racial/Ethnic Disparities in Behavioral Counseling: the
Case of Hispanic/White Difference in the Receipt of
Smoking Cessation Advice
Yuhua Bao, Ph.D.
Presented By: Yuhua Bao, Ph.D., Division of General Internal
Medicine and Health Services Research, David Geffen School
of Medicine, UCLA, 1100 Glendon Avenue, Suite 2010, Los
Angeles, CA 90024; Tel: (310)794-3081; Email:
yuhuabao@mednet.ucla.edu
Research Objective: Racial/ethnic differences in health
behaviors and lifestyles likely account for a substantial amount
of the racial/ethnic disparities in health. There is strong
evidence that provider behavioral counseling can be effective
in helping patients change certain health behaviors.
Disparities in behavioral counseling, if any, can thus aggravate
racial/ethnic disparities in the prevalence and burden of
diseases. This paper discusses some of the issues in studying
racial/ethnic disparities in behavioral counseling by using the
example of Hispanic/white difference in the receipt of
smoking cessation counseling.
Study Design: We define primary (secondary) behavioral
counseling as counseling delivered to patients who do not yet
have (already have) behavior-related health conditions. We
first conceptually discuss why overall disparity (or the lack of
it) could be a misleading depiction of possible disparities in
primary or secondary counseling. To provide an empirical
example, we then estimate the Hispanic/white difference in
the receipt of smoking cessation advice from their regular
healthcare providers during the past 12 months. We use the
2001 National Health Interview Survey Sample Adult File and
stratify the logistic regression analysis of receiving some
provider advice to quit smoking by the presence of smokingrelated health conditions. These conditions include myocardial
infarction/coronary heart disease/stroke, less serious heart
conditions, emphysema/chronic bronchitis/asthma, and
some types of cancer.
Population Studied: Adult patients (Hispanic or nonHispanic white according to self identification) who were
current smokers and who had some contact in the past 12
months with the health care providers or facilities they most
often went to for acute or preventive care.
Principal Findings: Overall, the rate of receiving some
smoking cessation advice was lower among Hispanic patients
than among white patients (52.6% vs. 69.5%). After adjusting
for essential demographic information, the intensity of
smoking and the presence of smoking-related health
conditions, the difference remained at 9 percentage points
(p<0.001). However, results of stratified analysis indicated
that the overall disparity was almost entirely accounted for by
the disparity in primary counseling: the adjusted
Hispanic/white ratio in the probability of primary smoking
cessation advice was 0.77, reflecting an absolute difference of
15 percentage points (p<0.001). Meanwhile, the difference
among patients with smoking-related conditions was
negligible and not statistically significant.
Conclusions: The Hispanic/white disparity in smoking
cessation counseling was mainly a result of the much lower
rate of counseling among Hispanic patients before the onset
of smoking-related conditions. Since primary behavioral
counseling is more discretionary and often receives less
priority than diagnostic or therapeutic services, factors
pertaining to the care process such as provider stereotyping
and the uncertainty in assessing minority patients’ preventive
care need could have given rise to the disparity.
Implications for Policy, Delivery, or Practice: Racial/ethnic
disparities in primary behavioral counseling could eventually
lead to or aggravate the disparities in disease prevalence;
disparities in secondary counseling may mean
disproportionate burden of morbidity and mortality associated
with the conditions. Future studies will need to investigate
behavioral counseling in the primary vs. secondary context
separately to provide relevant implications for closing the
racial/ethnic disparities in health and healthcare.
Primary Funding Source: NIH EXPORT Center
●Identifying Racial and Ethnic Disparities in Ambulatory
Care Sensitive Conditions among Medicare Beneficiaries
Arthur Bonito, Ph.D., Celia Eicheldinger, MS, Linda Greenberg,
Ph.D., Arthur Meltzer, Ph.D.
Presented By: Arthur Bonito, Ph.D., Research Triangle
Institute, Box 12194, 3040 Cornwallis Road, Research Triangle
Park, NC 27709; Tel: (919)541-6377; Fax: (919)990-8454; Email:
ajb@rti.org
Research Objective: Hospital and emergency room treatment
for ambulatory care sensitive conditions, or ACSCs, may serve
as a marker for inadequate access to or use of standard
ambulatory health care services. The objective of this study
was to more accurately assess the occurrence of racial and
ethnic disparities in the proportion of Medicare beneficiaries
with selected ACSCs in 2002.
Study Design: As a precursor to this study, we developed an
algorithm employing surname, geographic location, common
first names, and information on the Medicare Enrollment
Database to improve the race and ethnicity coding of Hispanic
and Asian and Pacific Island Medicare beneficiaries. Using
the improved race and ethnicity code, we selected a stratified
random sample of Medicare beneficiaries. Stratification was
disproportionate by race and ethnicity to include large enough
numbers to allow reliable investigation of disparities. We
used Medicare claims to identify hospital and emergency
room discharges for persons having principal diagnoses for
five chronic ACSCs: diabetes mellitus, chronic lung disease,
congestive heart failure, hypertension, and seizures.
Population Studied: The study sample of nearly two million
beneficiaries was weighted in our analyses to represent the
approximately 28 million Medicare beneficiaries in the
traditional fee-for-service plan for the entire CY 2002,
including Whites, Blacks, Hispanics, and Asians and Pacific
Islanders.
Principal Findings: Using White beneficiaries’ rates as the
basis for comparison, the rates of treatment for ACSCs for
Blacks ranged from 10 percent greater than Whites for chronic
lung disease to almost four times higher for diabetes and
hypertension. For Hispanics, the rates ranged from no
different on three ACSCs -- seizures, congestive heart failure,
and chronic lung disease -- to twice as high or more for
hypertension and diabetes. For Asians and Pacific Islanders,
the rates were comparable to Whites for diabetes, about 20
percent higher for hypertension, and about half as much for
chronic lung disease, congestive heart failure, and seizures. In
summary, Blacks had the highest rates for all five ACSCs.
Whites had the lowest rate for only one of the ACSCs,
hypertension. Whites and Asians and Pacific Islanders had
virtually the same lowest rate for diabetes mellitus. Asians and
Pacific Islanders alone had the lowest rates for the other three
ACSCs.
Conclusions: Our analyses found patterns of differences
among racial and ethnic groups for five chronic ACSCs. Of
the four racial and ethnic groups, Blacks have the worst access
to ambulatory care. Asians and Pacific Islanders have as good
or better access to services as Whites. Hispanics’ access to
primary care is better than Blacks, but worse than Whites.
Implications for Policy, Delivery, or Practice: We have
identified disparities in ACSCs among four racial and ethnic
groups based on a more complete identification of Medicare
beneficiaries by race and ethnicity. Study findings hold
promise for more valid and reliable assessment of health
disparities among the racial and ethnic minorities enrolled in
Medicare. Also, they provide policy makers with information
to more effectively target efforts to racial and ethnic groups
most in need, and to identify, monitor, and assess progress
toward elimination of health disparities.
Primary Funding Source: CMS
●To Voice or to Exit: Racial Differences in Consumer
Responses to Problematic Health Care Experiences
Lisa C. Gary, MS, MPH, Mark Schlesinger, Ph.D.
Presented By: Lisa C. Gary, MS, MPH, Doctoral Candidate,
Health Policy, Yale University, 60 College Street P.O. Box
208034, New Haven, CT 06510; Tel: (203)787-9943; Email:
lisa.gary@yale.edu
Research Objective: To identify health system characteristics
(e.g.- managed care) and patient characteristics (e.g.-level of
patient trust) that are associated with active consumer
empowerment.
Study Design: Background: Managed care has a significant
impact on the way health care is delivered in the US. In fact
most consumers with employer sponsored health insurance
are enrolled in a health plan with some type of managed care
practice such as prior authorization rules, utilization review,
closed panels of providers, and primary care provider
requirements. Current research shows that managed care has
led to diminished consumer trust in health systems and low
levels of patient satisfaction. These findings are particularly
salient in light of the vast racial health care disparities we see.
The data are overwhelming that minorities in the U.S. receive
significantly lower quality care across a range of diseases e.g.cancer, cardiac care, pain management, and some preventive
care. Low patient trust may be indicative of feelings of
disempowerment and alienation in the health care system
especially for minorities and other vulnerable groups. The
consumer empowerment behaviors under study are voicing
and exit. Voicing refers to instances when consumers express
their concern over negative health care encounters. The
targets for voicing may be informal or formal such as talking
with family and friends or registering complaints with the
health plan or a government authority. Exit refers to a
consumer’s decision to change health plans or physicians in
response to unsatisfactory health care episodes.
Data Source: Data are from the Yale/New York Academy of
Medicine Survey of Consumer Experiences in Health Care, a
telephone survey of approximately 5000 individuals
representing the US non-institutionalized population. The
racial/ethnic groups represented include African Americans,
Hispanics, Whites, Asians, and other racial groups.
Population Studied: Methods: We use logistic regression,
ordered logistic regression, and bivariate probit models with
selection (the dichotomous version of Heckman selection
models) to determine if race is a significant factor in the
degree of trust in physicians and health plans, likelihood of
experiencing negative health care episodes, and type of
consumer self-advocacy controlling for various psychosocial
and health system factors including perceived vulnerability,
informational social support, health system knowledge, and
several managed care health plan practices .
Principal Findings: For the analysis on consumer
empowerment behaviors, blacks were less likely to voice
dissatisfaction about problematic health care encounters
compared to whites. We examined voicing in three domains 1)
voicing to family, friends, or an employer, 2) voicing to the
health plan, and 3) voicing to one’s physician. Blacks were less
likely to voice to any of these entities compared to whites after
controlling for educational status, health plan characteristics,
perceived vulnerability, and informational social support.
Blacks were significantly less likely to voice dissatisfaction to
their personal physicians. Interestingly, blacks were more likely
to “exit” as a result of negative health care experiences than
whites. Blacks tend to exert control or exhibit agency by simply
leaving these bad situations; however, it is unclear whether
these exit decisions will lead to better care in the new health
plan or physician relationship. This type of empowerment
behavior may be futile and silent in many ways and in fact it
indicates that effective consumer voice may not be a real
option for many African Americans. In light of the low levels of
trust in physicians and health plans, these minority
consumers may fear retribution, limited system response to
their concerns, or be subjected to disrespect and alienation for
voicing their opinions so they simply exit.
Implications for Policy, Delivery, or Practice: The
development of culturally relevant consumer empowerment
health promotion tools could be very useful for improving
clinical and administrative interactions for minority patients.
Primary Funding Source: NIMH, Aspen Institute
●Mediators of Race Effects on Risk of Potentially
Avoidable Maternity Complications among MedicaidInsured Mothers
Sarah Laditka, Ph.D., MA, MBA, James N. Laditka, D.A., Ph.D.,
Janice C. Probst, Ph.D.
Presented By: Sarah Laditka, Ph.D., MA, MBA, Associate
Professor and MHA Program Director, Health Services Policy
and Management, University of South Carolina, 800 Sumter
Street, HESC Building, Room 116F, Columbia, SC 29201; Tel:
(803)777-1496; Fax: (803) 777-1826; Email:
sladitka@gwm.sc.edu
Research Objective: Studies have found that African
American women receive less prenatal care than do white
women, and are more likely to have pregnancy complications.
We examined both the risks of pregnancy complications and
factors associated with these risks focusing on differences by
race and ethnicity, using state-level data from South Carolina.
The data provide detailed individual-level information such as
marital status, income, and education. Such information is
not commonly available in administrative datasets. We used
the Potentially Avoidable Maternity Complication (PAMC)
indicator to evaluate pregnancy outcomes associated with
access to prenatal care of reasonable quality. PAMCs, defined
by a panel of obstetricians and health services researchers, are
identified through combinations of primary and secondary
hospital diagnoses at delivery, such as both a urinary tract
infection and a complication known to be associated with
such infections.
Study Design: Delivery hospitalization data for the year 2000
for Medicaid-insured women were obtained from the South
Carolina Budget and Control Board. We supplemented these
data with information from the 2000 Area Resource File and
the 2000 annual survey of the American Hospital Association.
Analyses included chi-square and multilevel logistic
regression. Controls included age, educational attainment,
marital status, income, comorbidities, and characteristics of
mothers’ residence counties and delivery hospitals.
Population Studied: 26,869 delivery hospital discharges for
Medicaid-insured women in South Carolina, year 2000.
Principal Findings: Not controlling for other factors, African
American mothers had higher odds of having a PAMC than
did white women (OR 1.26, p<.001). In the multivariate
model, the odds of experiencing a PAMC did not differ
between African American and white mothers. Notable PAMC
predictors were (odds ratios in parentheses): being single
(1.58, p<.0001), having income below the poverty threshold
(1.30, p<.001), being disabled (1.51, p=.05), having diabetes
(2.29, p<.001), and living in a rural county (1.96, p=.05).
Bivariate analyses showed that, compared with whites, African
Americans were more likely to (odds ratios in parentheses, all
p<.0001): be single (2.45), have income below the poverty
threshold (1.66), be disabled (1.54), have diabetes (1.82), and
live in a rural county (1.29). Thus, risk factors
disproportionately affecting African Americans contributed
notably to PAMC risk at the population level.
Conclusions: The greater PAMC risk of African Americans
mothers at the population level is notably attributable to
specific measurable risk factors. Single African American
Medicaid beneficiaries and those with diabetes are at
particularly high risk of avoidable pregnancy complications.
Implications for Policy, Delivery, or Practice: PAMC
analyses can suggest efficient approaches to addressing
disparities in pregnancy outcomes associated with the
accessibility and qualities of prenatal care. Healthy Start
programs, community health centers, and rural health clinics
may be particularly useful for addressing pregnancy risks
associated with poverty, single parent status, and chronic
disease. Particularly in rural areas, poor, disabled, and single
pregnant women may benefit from transportation and other
forms of proactive support to attend prenatal care. Prenatal
care for African American Medicaid beneficiaries should stress
diabetes control.
Primary Funding Source: HRSA
●Racial/Ethnic Differences in the Use of Health
Information to Self-Advocate During the Medical
Encounter: Is Having Health Information Enough?
Jacqueline Wiltshire, MPH, Ph.D., Kate Cronin, MPH
Presented By: Jacqueline Wiltshire, MPH, Ph.D., PostDoctoral Fellow, Center for Women's Health Research,
University of Wisconsin Medical School, 202 South Park
Street, 6W, Madison, WI 53715; Tel: (608)-267-5566; Fax:
(608)-267-5575; Email: wiltshire@wisc.edu
Research Objective: Patients with self-advocacy orientation
gain maximum benefit from the medical encounter. However,
little research has been conducted on how patients’
knowledge and use of health information influence the
medical encounter. Moreover, little is known about the
racial/ethnic differences in the use of health information to
self-advocate during the medical encounter. The purpose of
this study is to examine (1) whether health information
empowers midlife women to self-advocate during the medical
encounter, and (2) whether self-advocacy differs by
race/ethnicity.
Study Design: Study data were drawn from the 2000-01
Household Component of the Community Tracking Study, a
nationally representative survey. To operationalize health
information, sources of acquiring health information in the
past 12 months (internet, friends, TV or radio,
books/magazine/other source, and health care
professional/organization) was aggregated. Self-advocacy was
operationalized through yes/no responses to the following
questions: “Did the doctor order a test, procedure, or
prescription for you mainly because of information that you
mentioned or showed to him or her?; During the past 12
months, have you mentioned information to a doctor about a
medical condition or treatment for you that you found out
about yourself or were told by others?” All regression analyses
accounted for complex survey design and were adjusted for
race/ethnicity, age, marital status, rural/urban, education,
health status, employment, federal poverty level, and
insurance type.
Population Studied: The sample included 6,234 Whites, 858
African American, 507 Hispanic, and 207
Native/Asian/Pacific/Other women aged 45 to 64 with at least
one doctor visit.
Principal Findings: Overall, 47.5% of the women reported
getting medical information (49.3% of Whites, 39.6 of African
Americans, 44.1% of Native/Asian/Pacific/Other and 41.4 of
Hispanics). Seventeen percent of all women mentioned
medical information to doctor and 8.1% reported that the
doctor used the information to order a test, procedure, or
prescription. After multivariable adjustment using logistic
regression, women who had obtained health information were
5 times more likely to mention information to their doctor
(OR=5.00, 95% CI: 4.25-5.89) and 4 times more likely to get a
test/procedure/prescription from their doctor (OR=4.23,
95%CI: 3.36-5.33). African Americans were less likely to
mention information to a doctor (OR=0.55, 95%CI: 0.39-0.79)
and there were no significant differences in race/ethnicity with
regards to doctor ordering a test. Moreover, when Whites
were removed from the analysis (n=1,635),
Native/Asian/Pacific/Other and Hispanics were more likely to
mention information to a doctor (OR=2.05, 95%CI: 2.05, 1.263.32, OR=2.05, 95%CI: 1.61, 1.07-2.42, respectively); and
Hispanics were more likely to get a
test/procedure/prescription ordered from their doctor
(OR=1.82, 95%CI:1.05-3.17).
Conclusions: While health information empowers midlife
women to self-advocate, African American women are least
likely to self-advocate using previously acquired health
information. Factors or barriers that may impede the ability of
African American women to self-advocate need to be
identified.
Implications for Policy, Delivery, or Practice: Every health
care encounter should be seen as an opportunity to foster an
environment which promotes self-advocacy among patients.
Primary Funding Source: NIA
Related Posters
Poster Session B
Monday, June 27 • 6:15 pm – 7:30 pm
●Medicare: An Investigation of Inequalities in Access
Jean Accius, Master's in Aging Studies
Presented By: Jean Accius, Master's in Aging Studies,
Doctoral Student, School of Public Affairs, American
University, 4400 Massachusetts Avenue NW, Washington, DC
20016; Tel: (850) 321-7917; Email: acciusj@hotmail.com
Research Objective: Given well-documented racial healthdisparities, access to care within the healthcare system
amongst racial groups is a primary concern. Medicare, one of
the largest government-sponsored health programs, has the
potential to reduce these health disparities. However, whether
racial minorities receive the care they need through Medicare
is unknown as previous research on the topic has focused on
the relationship between access to care and health status, and
not race.
Study Design: This research paper contributes to the
literature by examining the disparities that exist within the
Medicare Program between black and non-black beneficiaries
within the area of access to care. More specifically, is the
Medicare program relatively inaccessible to the needs of
blacks in comparison to non-blacks in terms of access to care?
My hypothesis is that the Medicare Program is not sufficient
in meeting the needs of blacks relative to their counterparts.
Population Studied: This paper uses data from the 1993
Medicare Beneficiary Survey. The MCBS, collected by the
Office of Strategic Planning of the Center for Medicare and
Medicaid Services, is a continuous, multipurpose survey of a
nationally representative sample of aged, disabled, and
institutionalized Medicare beneficiaries. Beneficiaries are
sampled from the Medicare enrollment files and interviewed
in person three times a year using computer-assisted personal
interviewing. Newly admitted beneficiaries are added to the
sample once a year and deaths or refusals are depleted from
the sample. The response rate for the 1993 survey was 87
percent, yielding a sample size of 12,895. The analysis sample
constituted of 11, 177 observations.
Principal Findings: Preliminary bivariate results suggest that
blacks were more satisfied with the availability of care than
non-blacks.
Primary Funding Source: No Funding Source
●Patient Preferences for Participatory Decision-making:
Patient Race, Education and Physician Perceptions
Gbemisola Adeseun, BS, MPH, Lisa A. Cooper, M.D., MPH,
Presented By: Gbemisola Adeseun, BS, MPH, Medical
Student, Welch Center for Prevention, Epidemiology & Clinical
Research, Johns Hopkins University, 2024 East Monument
Street, Suite 2-500, Baltimore, MD 21287; Tel: (410) 303-9428;
Email: gadeseun@jhmi.edu
Research Objective: Participatory decision-making is linked to
positive patient outcomes, including satisfaction, adherence,
and continuity of care. Previous studies suggest that ethnic
minorities and less educated patients experience less
participatory visits with physicians. However, little work
examines whether patient preferences for participatory
decision-making (PDM) and physician perceptions of those
preferences differ across patients’ race/ethnicity or education.
We examined whether: 1) patients’ race and educational level
are associated with their preferences for PDM and 2)
physicians’ perceptions of patient preferences for PDM are
influenced by patients’ race and education.
Study Design: We conducted a brief cohort study of patients
and physicians in primary care practices serving managed care
and fee-for-service patients in Baltimore, MD, Washington DC,
and Northern VA areas in 2002. Our main outcomes of
interest, collected from pre- and post-visit questionnaires,
were: 1) patient preferences for PDM, measured before the
visit using an item from the Autonomy Preference Index and
2) physicians’ predictions of individual patient’s preferences
for PDM measured after the visit. We used logistic regression
with generalized estimating equations to account for the
clustering of patients by physicians and adjusted for patient,
age, gender, and health status and physician race in
multivariate models.
Population Studied: Two-hundred and six patients seeing 30
physicians participated in the study. Patients had an average
age of 51 years; they were 65% women; 55% African American
and 45% white; and 38% had less than high school (HS)
education, 32% were HS graduates, and 31% had some
college education.
Principal Findings: The probability of preferring PDM was
similar for white and African-American patients (72% vs. 62%,
P=0.09). Individuals with less than high school education,
high school graduates, and some college education and
beyond were equally likely to desire PDM (Pr 0.65, 0.61, 0.73,
respectively; P >0.05). Physicians perceived that: 1) white and
African-American patients were equally likely to desire
participatory decision-making (Pr 0.65 and 0.67, respectively;
P=0.42) and 2) individuals without a high school diploma
were less likely to desire PDM compared to individuals with
some college education and beyond (Pr 0.54 vs. 0.76,
respectively; P =0.03).
Conclusions: Patient preferences for participatory decisionmaking do not differ by race or education. While physicians
do not perceive ethnic minorities as preferring less
participation in care, they do misperceive individuals with less
than high school education as desiring less active roles in
decision-making.
Implications for Policy, Delivery, or Practice: Interventions
to increase physicians’ PDM behaviors with patients
regardless of educational level may better meet less educated
patients’ needs to play active roles in medical decisionmaking.
Primary Funding Source: CWF, National Center or Minority
Health Disparities
●Differences in Neonatal Mortality among Whites and
Asian Subgroups: Evidence from California 1991-2001
Laurence Baker, Ph.D., Christopher Afendulis, Ph.D., Amitabh
Chandra, Ph.D., Elena Fuentes-Afflick, M.D., Shannon
McConnville, MA, Ciaran Phibbs, Ph.D.
Presented By: Laurence Baker, Ph.D., Associate Professor,
Health Research and Policy, Stanford University School of
Medicine, HRP Redwood Building, Room 110, Stanford, CA
94305-5405; Tel: (650)723-4098; Fax: (650)725-6951; Email:
laurence.baker@stanford.edu
Research Objective: The number of Asian-American
newborns has grown rapidly over the past 20 years, but there
is relatively little evidence about health outcomes for AsianAmericans as a group or for national origin subgroups of the
larger Asian-American population. Data limitations have left
previous studies able to study only small numbers of AsianAmerican newborns and precluded detailed comparisons
across a large number of subgroups. The objective of this
study is to provide new data on Asian-American births and
outcomes, comparing most Asian-American subgroups with
large and detailed data.
Study Design: Cross-sectional and longitudinal comparisons
of birth outcomes among nine Asian-American subgroups
(Cambodian, Chinese, Filipino, Indian, Japanese, Korean,
Laotian, Thai, and Vietnamese) and Whites. We use data
linking birth certificates, death certificates, and hospital
discharge records for newborns and mothers. We compare
birthweights, newborn complications, comorbidities, and
other risk factors across the various groups. We use logistic
regression to examine differences in neonatal mortality. For
the mortality models, the dependent variable was 28 day
mortality and maternal race/ethnicity was the primarily
independent variable. Some models include only basic
controls. Others include extensive controls for maternal age,
education, nativity, and prenatal care use; newborn gender,
birth weight and neonatal and maternal risk factors.
Comparing results across models with and without the
additional controls allows us to assess the impact of including
the risk factors on the estimated mortality differences across
the various groups.
Population Studied: 2.4 million births in California for 19912001 for non-Latina White and Asian women from the nine
subgroups.
Principal Findings: Birthweights and risk factors vary
noticeably across the various groups. Mortality rates also
vary. In unadjusted comparisons, infants born to Chinese,
Japanese, and Korean women had significantly lower neonatal
mortality rates than infants born to White women, while Thai
infants had significantly higher rates. Observable risk factors
account for some of the differences, but not all. After
controlling for risk factors, the mortality risk for Japanese
newborns was about one-third lower than the risk for Whites,
and the risk for Filipino infants was about one-sixth lower
(both p<0.05). After adjustment, Thai infants had mortality
risk nearly twice as high as Whites (p<0.05). Some differential
trends over time were also observed, with infants born to
Filipina women experiencing the greatest decline.
Conclusions: : In California, the risk of neonatal mortality
varies by Asian national origin subgroup. Observable risk
factors explain some, but not all of the differences across
groups.
Implications for Policy, Delivery, or Practice: Further efforts
to reduce neonatal mortality among Asians should identify
and address the needs of high-risk subgroups. Much current
work groups all Asian Americans together, but efforts to
collect more detailed data and include more detailed
categories in analyses could be valuable.
Primary Funding Source: National Institute on Child Health
and Development
●Access to Neonatal Intensive Care and the Black-White
Newborn Outcomes Difference
Laurence Baker, Ph.D., Christopher Afendulis, Ph.D., Amitabh
Chandra, Ph.D., Elena Fuentes-Afflick, M.D., Shannon
McConnville, MA, Ciaran Phibbs, Ph.D.
Presented By: Laurence Baker, Ph.D., Associate Professor,
Health Research and Policy, Stanford University School of
Medicine, HRP Redwood Building, Room 110, Stanford, CA
94305-5405; Tel: (650)723-4098; Fax: (650)725-6951; Email:
laurence.baker@stanford.edu
Research Objective: Among low birthweight newborns, Black
newborns have better survival than White newborns for a
given birthweight. This birthweight-specific survival advantage
has never been fully explained. One previously unexplored
possibility is differential access to high-level neonatal intensive
care units (NICUs), which have been shown to contribute to
improved outcomes. This paper compares the use of NICUs
by race/ethnicity and examines relationships between access
to NICUs and differences in birthweight-specific mortality.
Study Design: The first part of the study consists of crosssectional and longitudinal comparisons of access to NICUs
among Asian, Black, Hispanic and White newborns. We use
data linking birth certificates, death certificates, and hospital
discharge records for newborns and mothers. We calculate
the distance from residence zip code to the nearest hospitals
with high-level and lower-level NICUs, and compare distances
across race/ethnicity groups and over time. We use
multinomial logistic models to examine the probability of
being born in hospitals with different NICU levels. The
independent variables in these models were race/ethnicity,
maternal and newborn demographics, and maternal and
neonatal risk factors. In the second part of the study, we use
the same data to model infant mortality. We estimate logistic
regression models of 28-day mortality, as a function of
race/ethnicity, controlling for a range of risk factors including
birthweight, maternal characteristics, and a detailed set of
controls for health status. We investigate the impact of
controlling for use of a NICU on the estimated variations in
mortality rates by race and ethnicity.
Population Studied: All births in California between 1990 and
2000 in work to date. We expect to add data from Wisconsin,
Missouri, and New Jersey in the near future.
Principal Findings: Black newborns are more likely to live
close to, and be born in, hospitals with large tertiary NICUs
than White infants. 43% of Black newborns under 2500g are
born in high level NICUs, compared with 31% of Whites.
However, despite evidence from other studies that higher
NICU use rates reduce mortality for high-risk newborns, the
higher rate of use of high-level NICUs by Black newborns
explains only a small portion of the observed Black-White
birthweight-specific mortality difference. We found only small
variations in use of NICUs among Asian, Hispanic and White
newborns.
Conclusions: There are large racial/ethnic disparities in use of
NICUs. Opposite much of the disparities literature, in which
Blacks are less likely to have access to advanced therapies,
Black infants were more likely to be born in hospitals with high
level NICUs than White infants. Differential access to
technology does not, however, explain the favorable outcomes
among Black infants.
Implications for Policy, Delivery, or Practice: Differential
access to technologies is an important policy consideration,
but can be complex. Findings from some settings that Black
patients receive less intensive care do not extrapolate
uniformly to all settings. Assuming that differential
technology use will necessarily drive outcome differences may
also be unwarranted.
Primary Funding Source: National Institute on Child Health
and Development
●Disparities in Perinatal and Infant Health Among Rural
AI/ANs and Rural Whites: Is the Gap Narrowing?
Laura-Mae Baldwin, M.D., MPH, Eric Larson, Ph.D., Elise
Murowchick, Ph.D., Holly Andrilla, M.S., L. Gary Hart, Ph.D.
Presented By: Laura-Mae Baldwin, M.D., MPH, Professor,
Family Medicine, University of Washington School of
Medicine, Box 354982, Seattle, WA 98195; Tel: (206) 685-4799;
Fax: (206) 616-4768; Email: lmb@u.washington.edu
Research Objective: To examine whether disparities in
perinatal care and birth outcomes among rural American
Indians/Alaska Natives (AI/ANs) and rural whites have
decreased during the decade and a half of national and statelevel policy change to improve access to prenatal and infant
care.
Study Design: Cross-sectional study using the National
Linked Birth-Death Database to compare prenatal care receipt,
low birthweight rates, neonatal and postneonatal death rates,
and causes of death among rural AI/ANs and whites in three
time periods: 1985-1987, 1989-1991, 1995-1997. Multivariate
logistic regression examined whether the AI/AN-white
differences diminished over the three time periods.
Population Studied: 217,064 AI/AN and 5,032,533 white, nonHispanic singleton births to women living in rural U.S.
counties in the three time periods.
Principal Findings: Prenatal care use increased substantially
and comparably for both AI/ANs and whites over the three
time periods, with the greatest improvement between 19891991 and 1995-1997 (AI/AN rates of inadequate care 33.9% in
1989-91, 26.3% in 1995-97; whites 13.5% in 1989-91, 10.0% in
1995-97). The adjusted 2.5 times greater risk of receiving an
inadequate pattern of prenatal care among AI/ANs compared
to whites did not diminish over time. Postneonatal death
rates of AI/ANs were higher than whites in each of the time
periods, although the adjusted odds decreased from 1.76 in
1985-1987 to 1.60 in 1995-1997. The most common causes of
postneonatal death for rural AI/ANs and whites were SIDS,
congenital anomalies, infectious disease, and unintentional
injuries and accidents, with AI/ANs having significantly higher
rates of SIDS, infections, and injuries/accidents throughout
the study period. In 1995-1997, there were over three times as
many infectious disease-related postneonatal deaths among
rural AI/ANs compared to whites (54 AI/AN deaths, but only
16 white deaths in a same sized population).
Conclusions: Significant improvements in access to prenatal
care and postneonatal death rates among both rural AI/ANs
and whites between the mid-1980s and late-1990s are
encouraging. There were minimal gains in closing the gaps
between AI/ANs and whites in these measures, however.
Three of the four most common causes of postneonatal death
include preventable components.
Implications for Policy, Delivery, or Practice: The fact that
perinatal outcomes of both AI/ANs and whites improved
during a time of expanding Medicaid eligibility and
enrollment, increased provision of support services and case
management, and efforts such as the Back to Sleep campaign
suggests that these types of programs can be successfully
applied in diverse populations. Changes over the past two
decades in the organization of AI/AN health services, with
increased tribal autonomy, provide excellent opportunities for
culturally appropriate interventions to further improve
perinatal outcomes. Persistent disparity in the per capita
annual health care funds available to the Indian Health Service
compared to that spent on the average U.S. citizen threatens
these opportunities, however. Redoubled efforts and
adequate funding are needed to ensure that populations with
higher rates of adverse perinatal outcomes, such as AI/ANs,
continue to experience improvements. Without these efforts
and funding, it will be difficult for AI/ANs to meet the Healthy
People 2010 objectives, in which AI/ANs and whites are
expected to reach the same goals.
Primary Funding Source: No Funding Source
●Physician and Patient Characteristics in Outpatient
Breast Cancer Screening
Rajesh Balkrishnan, Ph.D., Stephen V. Samuel, M.D.
Presented By: Rajesh Balkrishnan, Ph.D., Merrell Dow
Professor, College of Pharmacy and School of Public Health,
500 West 12th Avenue, Columbus, OH 43201; Tel: (614)2926415; Fax: (614)292-1335; Email: rbgws@sbcglobal.net
Research Objective: Breast cancer is the most common
cancer among women and is the second leading cause of
cancer death in the United States. However, death rates from
breast cancer have been declining in recent years and these
decreases are believed to be the result of earlier detection and
improved treatment. Earlier detection of breast cancer is
primarily achieved through screening methods. Typically,
screening for breast cancer consists of breast self examination
(BSE), clinical breast exam (CBE), and mammography. The
main objective of this study was to examine the factors that
are associated with physician recommendations for breast
cancer screening in United States outpatient settings.
Study Design: Analysis of cross sectional nationally
representative survey.
Population Studied: Data from the National Ambulatory
Medical Care Survey (NAMCS) database, a national
probability sample administered by the Division of Health
Care Statistics, National Center for Health Statistics (NCHS),
Center for Disease Control and Prevention (CDC) were used.
This study sample consisted of 2.04 billion female patients
aged 18 and older who visited physicians based in outpatient
settings in the US from 1996 to 2001. Multiple logistic
regression analyses were used to examine predictors of breast
self exam, mammography, and clinical breast exams in
outpatient settings during the 6 year time period.
Principal Findings: Women who paid for their office visit with
public insurance (Medicare/Medicaid), or other insurance, or
had no insurance were less likely to be offered breast self
exam counseling, clinical breast exam, or mammography than
women who had private insurance. In addition, primary care
physicians (family practice, general practice, and internal
medicine only) were less likely to offer breast self exam
counseling and less likely to perform a clinical breast exam
than their non-primary care counterparts. Furthermore,
women who were 65 years or older were significantly less likely
to receive the three screening methods when compared to
women between the ages of 35-49.
Conclusions: The findings of the study reveal several factors
that play a role in the physician recommendation of breast
cancer screening. These findings also indicate that there are
certain disparities regarding the recommendation of breast
cancer screening and further research needs to be conducted
to determine the best methods to bridge these gaps in patient
care.
Implications for Policy, Delivery, or Practice: These findings
contribute to the existing information about breast cancer
screening and highlight the physician and patient factors that
affect breast cancer screening in outpatient settings in the
United States. Policies designed to target factors associated
with higher risk of non-screening in primary care settings may
eventually lead to earlier detection and treatment of breast
cancer resulting in decreased mortality from breast cancer.
Primary Funding Source: No Funding Source
●Health Literacy and Metabolic Syndrome Risks: A Pilot
Study with Deaf Adults who Communicate with American
Sign Language
Steven Barnett, M.D., Robert Pollard, Ph.D.
Presented By: Steven Barnett, M.D., Assistant Professor,
Department of Family Medicine, University of Rochester, 1381
South Ave, Rochester, NY 14620; Tel: (585)506-9484 x110; Fax:
(585)473-2245; Email: steven_barnett@urmc.rochester.edu
Research Objective: To examine the health-related knowledge
and risk factors of deaf adults who communicate with
American Sign Language.
Study Design: We worked with a convenience sample of deaf
adults attending a social activity in Rochester NY. ASL-fluent
researchers (hearing and deaf) reviewed the research project
with potential participants and obtained consent. Each adult
completed a single page low English literacy questionnaire;
ASL-fluent researchers answered any questions. The
questionnaire asked demographic and health information.
Participants also completed a modified REALM, which asked
them to circle words they recognized and understood. After
participants completed the survey we measured blood
pressure, height, weight and waist circumference.
Population Studied: Deaf people who communicate in
American Sign Language (ASL) comprise an understudied and
medically underserved language minority community. Deaf
people who communicate in ASL experience health disparities,
including poorer health and less access to health information
and healthcare services than those in the general population.
The low literacy of many adults deaf since childhood
contributes to disparities. Limited access to health
information and healthcare services may put deaf adults at
risk for health problems related to lifestyle choices such as
diet and physical activity. There is little research on health and
deaf people who communicate in ASL. Health literacy
measures such as the REALM use pronunciation to assess
health literacy. It is unlikely that adults deaf since childhood
have heard these words pronounced.
Principal Findings: Sixty-one deaf adults participated, 30
(49%) were women. Fifty-one knew they became deaf before
age 3 years and 47 (77%) reported ASL as their one best
language. The median age was 46 (range 21-89; SD=13).
Forty-eight (79%) graduated college (associate degree or
higher) but 38 (69%) scored in the highest REALM category
(high school reading level). Only 39 (64%) rated their health
as “good.” Seven (13%) were current smokers. Twenty-six
(43%) thought they were “too heavy.” Seventeen (28%) were
obese with a body mass index (BMI) > 30 and another 26
(43%) were overweight (BMI 25-29.9). Seventeen women
(57%) and 11 men (38%) had abdominal obesity by waist
circumference. Thirteen participants (21%) thought their
blood pressure was “too high,” 12 (20%) knew they took
medicine for blood pressure, while 25 (46%) participants’
blood pressure was elevated (systolic > 124 or diastolic > 84).
The presentation will include other data as well as national
general population data for comparison.
Conclusions: We should learn more about health literacy,
health knowledge, health and risk behaviors of deaf people
who communicate in ASL. This is essential in order to
promote health and prevent disease in deaf people and their
family members. In order to do this we need to adapt and
evaluate existing research tools for use in ASL.
Implications for Policy, Delivery, or Practice: Many adults
deaf since childhood became deaf in utero during the rubella
pandemic of the early 1960s and they, like other adults their
age, will have increasing age-appropriate preventive services
needs. One of the goals of Healthy People 2010 is to improve
health and access to health research for people with
disabilities.
Primary Funding Source: CDC
●Relationship Between Local Unemployment Rate and
Hospital-Level Outcome
Raj Behal, M.D., MPH, Sarah Kinsella
Presented By: Raj Behal, M.D., MPH, Medical Director,
Clinical Effectiveness & System Redesign, University
HealthSystem Consortium, 2001 Spring Road Suite 700, Oak
Brook, IL 60523; Tel: (630)954-4892; Email: rbehal@uhc.edu
Research Objective: To assess whether the proportion of
hospital admissions that are length of stay outliers is affected
by the socioeconomic status of local population.
Study Design: A binary logistic regression model was
developed to determine the independent effect of population
characteristics on the proportion of LOS outliers after
controlling for case mix. Variables obtained from 2000 census
data included proportion of local population that was Black,
Hispanic, over age 65, under age 18, with Bachelor's degree,
foreign born, below poverty level; median household income;
hospital-level variables included annual case volume, %
patients dying within 2 days of admission, and mean
predicted LOS (using RDRG grouper to assign severity of
illness to each patient).
Population Studied: Discharges from 59 hospitals in the
state of Massachusetts in year 2000 were included in the
study.
Principal Findings: Local unemployment rate was an
independent predictor of proportion of cases that were LOS
outliers (c-index 0.764, Hosmer Lemeshow p value = 0.97).
Conclusions: After adjusting for various social and
demographic characteristics of the local population, and case
mix for each hospital, unemployment rate appears to be
positively correlated with hospital-level outcome of LOS
outliers.
Implications for Policy, Delivery, or Practice: Further
research is required to fully elucidate the effect of prevalent
unemployment rate in the community on disparities in
hospital-level outcomes.
Primary Funding Source: No Funding Source
●Should We Continue to Blame Obesity on Urban Sprawl?
Stephanie Bernell, Ph.D., Andrew Plantinga, Ph.D.
Presented By: Stephanie Bernell, Ph.D., Assistant Professor,
Public Heatlh, Oregon State University, 319 Waldo Hall,
Corvallis, OR 97331-6406; Tel: (541)737-9162; Fax: (541)7374001; Email: Stephanie.Bernell@oregonstate.edu
Research Objective: Recent studies emphasize the
connection between obesity and development patterns
associated with urban sprawl. Empirical analyses have found
higher obesity rates among residents of sprawling areas.
Some researchers have proposed changes in urban design to
reduce obesity and other health problems linked to physical
inactivity. An important assumption implicit to this line of
reasoning is that urban form is exogenous to an individual’s
weight. In particular, it is assumed that when urban
development patterns produce low levels of physical activity,
individuals have a greater tendency to become obese. In this
study, we consider the possibility that weight affects an
individual’s choice of the type of neighborhood to reside in.
Specifically, we test the hypothesis that an individual’s weight
and the urban form of the area where they live are determined
simultaneously. If true, the assumption that urban form is
exogenous to weight is incorrect. This would imply that earlier
empirical results are invalidated by simultaneity bias and
challenges the idea that obesity can be addressed through
urban design modifications.
Study Design: Our econometric model is a two-equation
system with BMI and a latent variable determining residence
in a high or low sprawl county as endogenous variables. The
exogenous variables are the other individual attributes and the
region of residence. Parameter estimates are derived using a
simultaneous equations probit estimator.
Population Studied: Our main data source is the National
Longitudinal Survey of Youth 1979 (NLSY79), which tracks a
nationally representative sample of individuals who were ages
14 to 22 years in 1979. We use data from the NLSY79 for the
year 2000 to measure body mass index (BMI), other
individual attributes (race, gender, smoking and marital
status, income, number of children, age, educational
attainment), and county and region of residence in 2000. An
individual is considered obese if their BMI exceeds 30. We
categorize counties as having a high or low degree of sprawl.
Our final sample includes 3,867 respondents to the NLSY79
who reside in counties for which the sprawl index was
measured and for whom we have complete observations of
the individual attributes.
Principal Findings: Consistent with earlier studies,
individuals with higher probability of living in high sprawl
counties are found to have greater BMI. However, we also
find that greater BMI increases the probability that individuals
live in high sprawl counties.
Conclusions: Uur results suggest that causality between
weight and urban form runs in both directions. That is, we
find evidence for the simultaneity of BMI and urban form,
implying that parameter estimates in earlier studies are
biased.
Implications for Policy, Delivery, or Practice: Our finding
that higher BMI increases the probability of residence in a
high sprawl county raises doubts about whether obesity can
be reduced through modifications of urban form . Building
compact, walkable communities may simply attract
individuals who are already physically active. Alternatively,
modifying existing communities to promote exercise—for
example, by building parks and installing sidewalks—may
have little effect if current residents are predisposed to
physical inactivity.
Primary Funding Source: No Funding Source
●Why are Some Individuals Food Insecure?: The Role of
County Level Factors
Stephanie Bernell, Ph.D., Bruce Weber, Ph.D., Mark Edwards,
Ph.D.
Presented By: Stephanie Bernell, Ph.D., Assistant Professor,
Public Health, Oregon State University, 319 Waldo Hall,
Corvallis, OR 97331-6406; Tel: (541)737-9162; Fax: (541)7378001; Email: Stephanie.Bernell@oregonstate.edu
Research Objective: This study attempts to untangle the
multiple sources of food insecurity by examining the roles of
household demographics, local economic and social
conditions and federal food security programs in explaining
the likelihood of household food insecurity in Oregon, the
state with the highest hunger rate in the nation in four of the
last five assessments. We believe this is the first study that has
examined the role of local (county-level) contextual factors in
explaining food insecurity.
Study Design: We use a multivariate logit model to estimate
the probability of household food insecurity. Maximum
likelihood coefficients and marginal effects are presented. In
addition, based on the multivariate model results, we calculate
and present predicted probabilities of food insecurity for
composite individuals living in different settings.
Population Studied: In this study, food insecurity is modeled
as a function of personal characteristics and county-level
contextual variables. The analyses are based on the 2000
Oregon Population Survey data, supplemented by data from
the 2000 Census and other sources identified in Table 2. The
Oregon Population Survey contains information on food
insecurity and on personal and household characteristics of
Oregon residents. The supplemental data contains county
specific information on wages, unemployment, program use,
and community affiliations and participation.
Principal Findings: Controlling for individual and household
demographics, local economic conditions (local wages and
unemployment) have a significant effect on the likelihood that
a given household will be food insecure. Also, whether one
lives in a rural or urban environment matters greatly. Moving
from a completely urban environment to a completely rural
environment reduces the probability of food insecurity by
about 5 percentage points, holding all else constant. The
county food stamp level is also significant determinant of food
insecurity. The empirical analysis does not suggest that high
county-level rent is associated with a significantly higher
probability of food insecurity. Evidence does suggest, however,
that very low-income households located in high-rent counties
have an increased probability of food insecurity.
Conclusions: Our results suggest food insecurity is much
more than a problem with individual choices. It is the case
that that the local community food security infrastructure
(food stamp outreach, wages and employment opportunities
and housing costs) significantly affects the likelihood that
families will be food insecure.
Implications for Policy, Delivery, or Practice: Local actions
that strengthen this infrastructure may be expected to have
stronger effects in states like Oregon where the influence of
contextual factors is relatively strong.
Primary Funding Source: No Funding Source
●Identifying New Target Areas for Reducing Racial/Ethnic
and Gender Disparities in Health Care for the Medicare
Population
Sarita Bhalotra, M.D., Ph.D., Mathilda Ruwe, M.D., MPH,
John Capitman, Ph.D., Elise Desjardins, MBA
Presented By: Sarita Bhalotra, M.D., Ph.D., Assistant
Professor, Heller School, Brandeis University, 415 South
Street, Waltham, MA 02454; Tel: (781)736-3960; Email:
bhalotra@brandeis.edu
Research Objective: To provide a framework for integrating
health disparity perspectives with current clinical care and
public health practice, we reviewed the evidence for
racial/ethnic and gender disparities in access, prevention and
treatment for five health conditions: depression, colorectal
cancer, stroke, coronary heart disease and major joint
replacement. We determined availability of evidence-based
treatment and existence of interventions to reduce
racial/ethnic or gender disparities.
Study Design: We conducted a thorough search from
multiple scholarly sources; eligible reports were: studies
involving the target populations; published between 19992003, involving population aged 65 years and over; addressing
target health condition in the target populations, or access to
preventive care or treatment, and possible interventions to
reduce disparities in colon or rectal cancer preventive services;
studies meeting minimal methodology standards; a stepwise
methodology that eliminated reports systematically yielded a
total of 327 qualified studies.
Population Studied: Elders 65 years and over who were:
Asian-Americans, African-American or Hispanic/Latina,
American-Indians, Pacific Islanders, Rural residents, Disabled,
Low Income or Dual Eligible
Principal Findings: We found racial/ethnic and gender
differences in access, prevention and treatment for all five
conditions. Racial/ethnic disparities for depression were for
both receiving a diagnosis and treatment. Although the data
on gender differences in access to invasive cardiac procedures
was somewhat inconsistent the weight of the evidence
showed that women and racial/ethnic minorities tend to
underuse and have access to these procedures at a later stage,
while white males tend to overuse the procedures. Gender
disparities also resulted from differences in clinical
presentation. Although colorectal cancer is a largely
preventable disease, racial/ethnic minorities are less likely to
be screened. Racial/ethnic minority elders were less likely to
be offered and to participate in primary preventive activities
for stroke, such as weight loss/diet improvements, smoking
cessation, and physical activity. Females and African American
elders were less likely to receive carotid endartectomy (CEA).
There are racial/ethnic differences in rates of joint
replacement despite small differences in osteoarthritis and
musculoskeletal conditions. Women tend to undergo joint
replacement surgeries more frequently, experience more
disability at the time of total hip arthroplasty, and do not
perform as well after one year as compared to men. We found
no evidence of interventions to reduce gender and
racial/ethnic disparities in access or treatment of these
conditions. There were not enough studies rural/urban, dual
eligible to allow conclusions on disparities.
Conclusions: Women and racial/ethnic minority elders can
benefit from interventions to reduce underuse of available
primary and secondary preventative services for colorectal
cancer, stroke, and depression and coronary heart disease.
However, national guidelines and interventions for reducing
disparities are currently not available for these conditions.
More studies on rural/urban and dual eligible differentials are
required.
Implications for Policy, Delivery, or Practice: Race/ethnicity
and gender encompass social and contextual variables that
are associated with lifetime exposure to behavioral and
contextual/genetic risks that can influence health outcome
disparities. Current health care system attributes also
influence health disparities. Health outcome disparities can
arise in each phase of prevention and treatment. If Healthy
People 2010 disparity goals are to be achieved, interventions
must be multifaceted and address the complex causation of
disparities
Primary Funding Source: CMS
●Trends in Geographic Disparities in the Context of
Technology Adoption and Diffusion.
Mythreyi Bhargavan, Ph.D., Jonathan Sunshine, Ph.D.
Presented By: Mythreyi Bhargavan, Ph.D., Director of
Research, American College of Radiology, 1891 Preston White
Drive, Reston, VA 20191; Tel: (703)715-4394; Fax: (703)2642443; Email: mythreyib@acr.org
Research Objective: To measure disparities in the utilization
of diagnostic imaging across states, to examine how these
disparities vary across imaging modalities (such as X-ray, CT,
MR) – specifically whether the disparities are greater among
new technologies -- and to examine whether these gaps
narrow over time. Further, we measure if state demographic
characteristics and physician supply affect the adoption and
rate of spread of technologies.
Study Design: We use Medicare summarized physician
claims data (Part B claims for non managed care enrollees)
for 1997-2003. We group the diagnostic imaging claims by
imaging modality (X-ray, ultrasound, CT, MRI, etc.), and by
state. We obtain state level statistics (from a variety of
sources), i.e., counts of radiologists, population, Medicare
enrollees, average income and per capita health spending by
state etc. We use the coefficient of variation for a simple
comparison of variability across states by modality and
observe trends over time. To measure the effect of the
adoption of new technology, we examine if this variability
across states is significantly different for newer technologies
(CT, MRI) than it is for older technologies (X-ray), and we
measure if these differences persist when we control for state
demographics and state physician supply. To measure the
effect of the diffusion of technology, we test if the variation in
per capita utilization of each imaging modality decreases over
time, overall and with controls for state characteristics that
affect demand for imaging state physician supply.
Population Studied: We study Medicare Part B fee-for-service
enrollees during the period 1997-2003.
Principal Findings: Preliminary estimates show that for all
the years examined, the coefficient of variation was higher for
MR and CT than for mammography and general radiology. In
virtually all modalities, the coefficient of variation decreased
during the period 1995-2001. We expect to find similar trends
for the period 1997-2003.
Conclusions: Small area variation needs to be viewed in the
light of the technology diffusion and adoption literature.
Though small area variation has been studied for over 30
years, this important perspective has been missing.
Implications for Policy, Delivery, or Practice: Very
widespread geographic variation in the use of many individual
health care services and in the total volume of services has
been extensively documented for over 30 years, beginning with
pioneering studies by Wennberg. However, there has been
relatively little explanation of the variation, and attempts to
explain the variation have not focused upon technology
adoption and diffusion, although that is relevant to a fast
changing field like health care. Past studies specifically of the
utilization of diagnostic imaging showed wide variation in per
capita imaging use across states, particularly for new imaging
modalities. Because technology change is particularly
prominent in imaging, a study of imaging is especially
promising for investigating the role of technology adaptation
and diffusion in geographic variation.
Primary Funding Source: No Funding Source
●The Relationship of Acculturation to Disparities in
Perceived Health Between Latinos and non-Latino Whites
Jeanne Black, MBA, Ninez Ponce, Ph.D.
Presented By: Jeanne Black, MBA, Doctoral Candidate, Health
Services, UCLA, 4396 Keystone Avenue, Culver City, CA 90232;
Tel: (310)836-8876; Email: jtblack@ucla.edu
Research Objective: Self-reported health status is widely used
as an indicator of population health, based on the question,
“In general, would you say your health is excellent, very good,
good, fair, or poor?” Using this measure, Latinos report
significantly worse health than non-Latino Whites. The
objective of this study is to examine whether acculturationrelated factors explain this disparity in addition to the
previously demonstrated effect of socioeconomic status
(SES).
Study Design: This cross-sectional study used data from the
random-digit-dial California Health Interview Survey 2001. We
used weighted logistic regression to examine the impact of
health conditions, SES, and acculturation on the odds of
reporting fair or poor health. Acculturation-related factors were
English proficiency (LEP), language(s) spoken at home, years
in the US, citizenship status, and immigrant generation.
Regression coefficients were compared using the Wald test.
Population Studied: 10,942 Latino and 25,679 non-Latino
White non-elderly adults residing in California.
Principal Findings: Latinos had 4 times the odds of
perceiving their health to be fair or poor, adjusting for age and
gender (p<0.0001). These odds remained unchanged when
BMI and chronic conditions were added to the regression
model, although being obese, or having been diagnosed with
asthma, arthritis, diabetes, hypertension, heart disease, or
cancer each had an odds ratio of at least 2.0 (p<0.0001 for
all). Adding acculturation-related variables reduced the odds
from 4.1 to 1.98. Increased odds were associated with LEP
(OR=3.7, p<0.0001); not speaking English at home (OR=1.3,
p<0.0001); and not being a US citizen (OR=1.3, p=0.012). The
odds also increased with length of residence. Speaking
Spanish and English at home, being a naturalized citizen, and
immigrant generation were not significant. When education
and poverty status were added to the model, the SES variables
had a greater effect than acculturation (F=60.1 vs. F= 7.25),
and the OR for Latinos was reduced further to 1.4. The only
acculturation-related variable to remain significant was LEP;
its associated odds decreased from 3.7 to 2.0 (p<0.0001).
Conclusions: Some factors associated with lower levels of US
acculturation appear to increase Latinos’ odds of reporting
fair/poor health status. However, acculturation level also is
associated with SES factors such as income and education.
The only acculturation-related factor to remain robust when
SES was added to the model is LEP. Despite the significant
contributions of health conditions and SES to perceived
health, there remains considerable disparity between Latinos
and non-Latino whites. This finding and the robustness of LEP
may indicate that self-reported health is a culturally-mediated
perception. Alternatively, since other acculturation-related
variables had diminished effects when SES variables were
added, the LEP effect may signal language barriers: Latinos
with LEP may have unreported chronic conditions because
they lack language access to the medical care that would
diagnose these conditions.
Implications for Policy, Delivery, or Practice: English
language proficiency explains a significant portion of the
difference in perceived health between Latinos and non-Latino
Whites, independent of socioeconomic status. Therefore,
efforts to reduce this disparity should focus not only on
health insurance coverage and tax credits, but should also
address language access in health care.
Primary Funding Source: No Funding Source
●A Method to Improve Medicare Coding of Race and
Ethnicity: Implications for Health Disparities among
Medicare Beneficiaries
Arthur Bonito, Ph.D., Celia Eicheldinger, MS, Linda Greenberg,
Ph.D., Arthur Meltzer, Ph.D.
Presented By: Arthur Bonito, Ph.D., Program on Health Care
Org, Del, and Access, RTI International, Box 12194, Research
Triangle Park, NC 27709; Tel: (919) 541-6377; Fax: (919) 9908454; Email: AJB@RTI.ORG
Research Objective: Historically, analyses of racial and ethnic
groups generally have been limited to comparisons of White,
Black, and All Other Medicare beneficiaries combined. The
objective of this study was to assess the validity and accuracy
of race and ethnicity coding on the Medicare Enrollment Data
Base, or EDB, and develop a method to improve coding of
Hispanics and Asians or Pacific Islanders so these groups
may be more validly represented in analyses identifying health
disparities.
Study Design: We pooled existing random samples of
Medicare fee-for-service and managed care beneficiaries who
responded to three annual CAHPS Medicare Satisfaction
Surveys for 2000 through 2002. The pooled sample consisted
of 830,728 unique beneficiaries who self-reported their race
and ethnicity. These self-reports constituted the gold standard
measure against which we assessed the existing EDB race and
ethnicity coding as well as our own effort to improve EDB
coding. Our approach to improve EDB coding was based in
part on algorithms we developed using US Census lists
indicating how frequently specific surnames belonged to
persons identifying themselves as Hispanic or Asian or Pacific
Islander.
Population Studied: Medicare beneficiaries enrolled in
traditional fee-for-service or managed care plans with
particular focus on Hispanics and Asians or Pacific Islanders.
Principal Findings: We developed an algorithm employing
surname, geographic location, common first names, and
information on the EDB that dramatically improved upon
existing EDB race and ethnicity coding. Sensitivity increased
from 29.5 to 76.6 percent for Hispanics and from 54.7 to 79.2
percent for Asians and Pacific Islanders, with no loss of
specificity. Kappa coefficients reached 0.80. The impacts of
improved race and ethnic coding on health service utilization
differed by condition, type of service, and whether estimating a
count, a rate, or an average. We illustrate the impact of the
improved race and ethnicity coding on identification of
diabetics and the use of diabetes related services.
Conclusions: Our study dramatically improved the validity of
Medicare race and ethnicity coding for Hispanics and Asians
or Pacific Islanders, increasing the number of Hispanics by
300 percent and of Asians or Pacific islanders by more than 30
percent. This improvement results in an enhancement of the
analytic capability and policy potential of Medicare
administrative data to provide more valid estimates of
potential disparities involving racial and ethnic minorities’ use
of and access to health services.
Implications for Policy, Delivery, or Practice: This study
developed an algorithm to recode the Medicare EDB race and
ethnicity variable that more accurately identified Hispanic and
Asian or Pacific Islander beneficiaries. Our results have
important implications for research on health disparities
among the Medicare beneficiaries and increasing our
knowledge of service use by minority groups. Our recoded
race and ethnicity variable can be substituted for the existing
EDB variable to allow more valid assessment of potential
health disparities among Hispanics and Asians or Pacific
Islanders.
Primary Funding Source: CMS
●A Sign of Our Times: Creating Effective Multilingual
Signage in a Children's Hospital by Integrating Input from
Hospital Staff, Administrators, and Families
Jane Brotanek, M.D., MPH, Glenn Flores, M.D., Suzanne
Strugalla
Presented By: Jane Brotanek, M.D., MPH, Assistant
Professor, Community Pediatrics, Medical College of
Wisconsin, 8701Watertown Plank Road, MS 756, Milwaukee,
WI 53226; Tel: (414) 456-5778; Fax: (414) 456-6385; Email:
jbrotane@mail.mcw.edu
Research Objective: 4 million US children are limited in
English proficiency (LEP), but research has not examined
barriers posed to LEP patients and their families by
monolingual English hospital signs, nor informed processes
for creating multilingual signage in a children's hospital. The
study objective was to determine the barriers and priorities in
developing multilingual hospital signage as identified by key
hospital personnel, administrative staff, and parents.
Study Design: The Children's Hospital of Wisconsin recently
approved conversion of English signage to multilingual
signage for the 2 major non-English language groups, Spanish
and Hmong. The Interpreter Services Advisory Committee
(ISAC) met to discuss barriers and priorities in implementing
multilingual signage. The ISAC is a group of 13 hospital staff,
including the Director of Public Relations, an outreach
specialist, a clinical learning specialist, and 2 pediatricians.
The ISAC, together with Latino and Hmong families,
conducted several hospital walk-throughs to evaluate current
signage and priorities for multilingual signs. Participant
observation techniques were used to record themes noted in
the discussions and walk-throughs, and a summary of themes
was compiled using grounded theory.
Population Studied: Key hospital and administrative staff of
the Children's Hospital of Wisconsin, along with Latino and
Hmong families
Principal Findings: The ISAC and families initially considered
comprehensively translating all hospital signs. Consensus was
reached on prioritizing multilingual sign placement in hightraffic areas using a comprehensive approach spanning from
the parking garage to multiple end destinations, with maps
provided in different languages. The group agreed that the
most efficient use of resources would be to post separate
signs in Hmong and Spanish next to/below existing signs,
rather than changing current signs. The importance of
responding to future needs also was raised, including
consideration of electronic systems, computer kiosks,
removable signs, and consultation with signage experts from
other industries.
Conclusions: Children's Hospital staff, administrators, and
families identified the following priorities in creating
multilingual signage in a children's hospital: multilingual sign
placement in high-traffic areas using a comprehensive
approach; multilingual maps on each hospital floor; separate
signs in Hmong and Spanish alongside/below existing signs;
and the importance of creating signage systems responsive to
future changes and needs.
Implications for Policy, Delivery, or Practice: This
comprehensive approach integrating the perspectives of key
stakeholders could serve as a model for other children's
hospitals considering implementing a multilingual signage
system.
Primary Funding Source: Medical College of Wisconsin,
Department of Pediatrics
●Urban African American Men with HIV/AIDS: Exploring
Association of Stigmatization, Health Systems Barriers
with Overall Life Functioning and Life Satisfaction
Dr. Aaron Buseh, Ph.D., MPH, MSN, Kelber, S., MSc, Stevens,
PE, Ph.D., RN, FAAN, Park, C. Gi., Ph.D.
Presented By: Dr. Aaron Buseh, Ph.D., MPH, MSN, Assistant
Professor, College of Nursing, University of WisconsinMilwaukee, 1921 East Hartford Avenue, Room 569, Milwaukee,
WI 53211; Tel: (414) 229-5462; Fax: (414) 229-5504; Email:
aaronbg@uwm.edu
Research Objective: In 2000, the death rate for black males
attributed to AIDS was 5 times that of white men. African
American men whose lives are extended with antiretroviral
therapies face an array of challenges: lack of finances, lack of
access to medical care, and social stigma due to HIV/AIDS. In
order to develop effective intervention programs for black
males living with HIV, it is essential to identify the factors
associated with the overall functioning and quality of life of
this population. The purpose of this study was to examine the
association and relationships between socio-demographic
variables, stigmatization, health systems barriers, and life
satisfaction and overall functioning as self-reported reported
among an urban sample of HIV infected black males.
Study Design: Cross-sectional descriptive survey design was
used to examine selected socio-demographic variables, their
association and influence on the quality of life among a
sample of urban African American males. Data were analyzed
using descriptive statistics, followed by correlations and
hierarchical multiple regressions statistics.
Population Studied: Participants included a convenience
sample of black males (18 years or older) who self-reported as
having been diagnosed with HIV infection (n = 55) and were
recruited through a local African American community-based
organization in a Midwest metropolitan area. A survey
questionnaire pertaining to socio-demographic indicators,
subjective health status, stigmatization, health systems
barriers, provider engagement and trust, life satisfaction and
overall functioning were combined into one survey instrument
and was administered to the participants.
Principal Findings: Barriers to health care was inversely
associated with the participants’ perception of their health; the
poorer the health of the participants the greater their
perception of the degree of barriers to obtaining health care.
Provider trust was positively associated with education; the
higher the level of education, the higher the level of trust in the
provider. All of the variables measuring stigma and health
system barriers were inversely associated with quality of life;
as barriers to health care and perceived stigma increased,
quality of life decreased. Stigmatization and health systems
barriers were associated. After adjusting for the demographic
variables of age, income, education, length of diagnosis,
sexual orientation, and health status, the factors significantly
impacting quality of life in our model were: personalized
stigma and barriers to health care- accounting for 55% of the
variance in quality of life. Also, after adjusting for length of
diagnosis, age, income and clinical status, the factors of
engagement with health provider and barriers to health care
accounted for 39% of variance in overall functioning.
Conclusions: African Americans are disproportionately
represented among existing and new cases of HIV infection.
The importance of factors examined in this study cannot be
understated. Understanding the experiences of African
American males living with HIV/AIDS will assist health care
professionals in working with this population in maximizing
the management of their disease and improving overall quality
of life.
Implications for Policy, Delivery, or Practice: Assessing
social stigma and health systems barriers in relation to overall
functioning and health-related quality of life among African
American males has important implications in developing
programs that will enable this population to live well. The
levels of social stigmatization and health system barriers can
be modified and are important influences on adaptation
processes that may enhance overall quality of life of urban
African American men living with HIV/AIDS.
Primary Funding Source: Sigma Theta Tau International
Honor Society of Nursing, Eta Nu Chapter
●Ethnicity and Employer-Sponsored Insurance as
Predictive Factors of Uninsurance Among Hispanics in
Nebraska
Erin Carlson, MPH, Jane Meza, Ph.D., Katherine Jones, Ph.D.,
PT
Presented By: Erin Carlson, MPH, Health Data Analyst,
Preventive and Societal Medicine, University of Nebraska
Medical Center, 984350 Nebraska Medical Center, Omaha, NE
68198-4350; Tel: (402) 559-8406; Fax: (402) 559-7259; Email:
ekcarlso@unmc.edu
Research Objective: To provide information to policymakers
for expanding coverage to Nebraska’s growing Hispanic
population by determining whether Nebraska’s Hispanics are
less likely to have employer-sponsored insurance (ESI) than
are other racial/ethnic groups in Nebraska and to identify
whether a correlation exists between Hispanic ethnicity, ESI,
and uninsurance in Nebraska.
Study Design: The 2004 Nebraska State Planning Grant
(NSPG) random-digit dialed telephone survey identified
characteristics of Nebraska’s uninsured and sources of
coverage for the insured. A translated survey was
administered to Spanish-speaking respondents. Descriptive
and bivariate analyses described differences between
Hispanics and non-Hispanics in employment status and
predictive factors of uninsurance; multivariate regression
analysis detected association between uninsurance, ethnicity,
and ESI.
Four of 13 NSPG focus groups also provided data. Key
contacts recruited participants for all groups and comoderated the Hispanic groups in Spanish. Group
proceedings were audio-recorded and transcribed for coding
and analysis.
Population Studied: The telephone survey interviewed 3750
adults under age 64 across the state. Hispanics were over-
sampled. The 448 surveys of Hispanics were compared to
non-Hispanics. Most of the two Hispanic groups’ 27
participants were uninsured and employed. The two employer
groups included 12 participants with one group conducted in a
rural community and the other in an urban community.
Principal Findings: Hispanics were more likely than nonHispanics to be employed and to have ESI available. Ten
percent fewer Hispanics than non-Hispanics were enrolled in
available ESI. Only .2% of Hispanics whose employer offered
health insurance said they were not enrolled because they “did
not need/want” health insurance. The most common reason
(23%) given for why Hispanics were not enrolled in ESI was
“don’t know.” Hispanics (21.7%) and non-Hispanics (20.0%)
responded similarly that they did not enroll in ESI because
contributions were “too expensive.”
When controlling for age, marital status, and poverty level,
ethnicity was not found to be a significant predictor of ESI
(p=.275). Uninsurance was more strongly associated with
poverty status (p=.001) than ethnicity or ESI. Hispanics
whose jobs did not offer ESI were more likely to be uninsured
than were those with ESI available (OR=1.11).
Hispanic focus group participants valued health insurance but
said they could not afford the employee contributions. Both
groups offered they would pay 10% of their paycheck for family
coverage. Many participants believed ESI was their only
insurance option, but said it was difficult to obtain and use
primarily due to lack of information and cost, in addition to
other factors.
Employer participants indicated that ESI was not a factor in
recruiting and retaining Hispanic employees because
Hispanic workers preferred money over benefits and thus
seemed to not value ESI.
Conclusions: Uninsurance among Nebraska’s Hispanics is
due to poverty rather than ethnicity or availability of ESI.
Implications for Policy, Delivery, or Practice: Policies to
mitigate uninsurance among Nebraska’s Hispanic population
should make ESI affordable for low-wage workers.
Information about how Nebraska’s employers explain ESI and
enroll Hispanic workers is necessary to better inform
employees about ESI. Research is needed to define the
economic contribution the Hispanic workforce makes to
Nebraska’s economy to support policies that improve health
access for this population.
Primary Funding Source: HRSA
growing trend towards the utilization of paid caregivers in
hospice, such as aids or various nursing staff. Studies show
that terminally-ill minorities with low income are more in need
of paid caregivers. The objective in this study was to
determine whether and how two variables in hospice care, the
race of the patient and the type of caregiver (paid vs. unpaid),
were related to the trend of shorter stays in hospice.
Study Design: To model the risk of dying within one month
after admission to hospice care, we used the Cox proportional
hazard because it allows for both non-normal distributions (to
account for the well-known highly skewed length of stay) and
censoring (to account for patients discharged alive). Patient
characteristics (age, gender, and death place), clinical
characteristics (referral source and diagnostics at discharge)
and hospice characteristics (profit/nonprofit) were factored in
to the model.
Population Studied: In this retrospective study, we looked at
a sample of patients over the age of 65 who died or were
discharged alive (N=3,142) from hospice care from National
Home and Hospice Care Survey (pooled data for 1998-2000).
Principal Findings: There were no main effects of race and
type of caregiver, but there was a significant racial effect,
depending on the type of caregiver. Minorities with paid
caregivers were 39% more likely than non-Hispanic whites
with paid caregivers to die within one month (p=0.0206). In
contrast, racial disparity in the risk was not found when both
groups were taken care of by their relatives/ unpaid caregivers
(p=0.2088).
Conclusions: Racial disparity in outcomes between hospice
patients was found in only one subgroup: between minorities
and whites taken care of by paid caregivers, but not found
when both groups were taken care of by their relatives/ unpaid
caregivers. This suggests that the dying minority elderly may
only achieve similar timely access to hospice care as
compared to whites if they are cared for by their relatives.
Implications for Policy, Delivery, or Practice: We need to
examine policy interventions that might benefit minority
hospice patients that have to rely on paid caregivers. We
should also explore why elderly minority hospice patients are
more susceptible as a group to delayed access to hospice care
than non-Hispanic whites when they both have to rely on paid
caregivers. These questions are important, considering our
finding that there was no racial disparity when taken care of by
relatives/unpaid caregivers.
Primary Funding Source: Retirement Research Foundation
●Disparities for Minority Elderly in Receipt of Timely
Hospice Care
Kyusuk Chung, Ph.D., Ross Mullner, Ph.D., Audrey K.
Gordon, Ph.D., Duck-Hye Yang, Ph.D., Joseph Lee, M.D.
●Effect of Retirement on Body Mass Index among the
Near Elderly
Sukyung Chung, MPH
Presented By: Kyusuk Chung, Ph.D., University Professor,
Health Administration, Governors State University, One
University Parkway, University Park, IL 60466; Tel: (708) 5344047; Fax: (708) 534-8041; Email: k-chung@govst.edu
Research Objective: Although the use of hospice care among
the terminally-ill elderly has tripled over the last decade, there
has been a growing concern about the steady increase in the
proportion of hospice patients with short stays (less than one
month). Short stays in hospice may be troublesome as it can
prevent the full benefit of hospice care. Though the majority
of hospice patients are still taken care of by unpaid primary
caregivers (such as family or friends), there has been a
Presented By: Sukyung Chung, MPH, Graduate Student,
Health Policy and Administration, University of North Carolina
at Chapel Hill, 1103B McGavran-Greenberg Hall, CB#7411,
Chapel Hill, NC 27599-7411; Tel: (408)746-0916; Email:
sukyungc@email.unc.edu
Research Objective: Retirement changes the availability of
financial resources and leisure time, and could act as either an
opportunity or a barrier to the maintenance of healthy body
weight. This study investigates the causal effect of retirement
on BMI to help determine the population of priority for
interventions to halt the growth of obesity and related health
care costs. The role of wealth and the relative importance of
quantity and content of leisure time in the BMI response to
retirement are also explored.
Study Design: Retirement in this study is defined as currently
not working and self-considered as retired. To get consistent
estimate of retirement, not induced by health conditions, the
fixed effects with instrumental variable method is employed.
Eligibilities for defined-benefit pension and Social Security
benefit were used as instruments for retirement. Subsamples
based on total household wealth were compared to examine
the role of wealth in the BMI response to retirement. The
relative importance of retirement-related changes in quantity
and content of leisure time is examined by estimating the
effect of reduced working hour on BMI among workers, and
the effects of retirement on BMI as compared to working in
physical jobs and sedentary jobs.
Population Studied: This study uses six panels of the Health
and Retirement Study (HRS) conducted every two years from
1992 to 2002 with age ranged from 51 to 71 (n=41,725). The
HRS is a national survey including all genders and racial
groups.
Principal Findings: The mean BMI of the sample was 28.5
and retirement increases BMI by 0.4 points on average
compared to currently working status either full time or part
time. The BMI response to retirement was greater, 1.0 in BMI,
among those with lower wealth, but disappeared among the
wealthier group. Retirement significantly increased BMI as
compared to working in sedentary jobs or physical jobs.
However, reduced working hour or increasing leisure time did
not change BMI for current workers. The weight gain effect
was persistent in the model using a broader definition of
retirement.
Conclusions: Retirement causes the greatest gain in BMI for
those with less wealth. Two pieces of evidence – gain in BMI
with retirement compared with working, regardless of job
types, and no effect on BMI from reduced working-hours
among workers – suggest that change in content of leisure
time with retirement lead to BMI gain. Changing leisure time
health behavior, such as physical activity and diet, with
retirement, and differences in these responses between people
with different socioeconomic status should be further
explored.
Implications for Policy, Delivery, or Practice: The modest
marginal weight change with retirement has a sizable impact
on the population obesity rate because the majority of people
in their fifties and sixties are overweight prior to retirement.
The impact of BMI gain with retirement on public health and
Medicare costs will accelerate with the expanding retired
population. Interventions to improve leisure time health
productivity of retired people in low wealth status should be
among the highest priorities in public health policy.
Primary Funding Source: No Funding Source
●The Utilization of Dental Services by Diabetics in
Colonias on the Texas/Mexico Border
Martha Conkling, MSPH
Presented By: Martha Conkling, MSPH, Graduate Assistant,
Health Policy and Management, School of Rural Public
Health, 3000 Briarcrest, Suite 300, Bryan, TX 77802; Tel:
(979)458-3031; Email: mctromp@srph.tamhsc.edu
Research Objective: One-third of people with diabetes have
severe periodontal disease. Recent evidence suggests a two
way association between chronic periodontal inflammation
and diabetes. This study was conducted to investigate if
individuals with diabetes in the colonias near McAllen, Texas
have lower realized access to dental services than those
individuals without diabetes. It is hypothesized that diabetics
are more likely to have seen a health care provider than to
have seen a dentist in the last year.
Study Design: Baseline household survey data collected from
December 2002 to April 2003 in the colonias near McAllen,
Texas was used for this study. Multiple logistic regression
was performed to assess whether diabetics in this population
have lower realized access to dental services than those
individuals without diabetes.
Population Studied: Four hundred and forty-six adults were
interviewed. A blood sugar test or screening had been received
by 231 participants. These people were included in the study.
Eighty four of the adult participants reported having diabetes.
As a population residents of colonias are known to be poor,
uneducated, and overweight.
Principal Findings: Seventy-eight percent of people with
diabetes reported that they had an education at the 8th grade
level or less and 72 percent live in households with an income
less than 8,000 dollars per year. People with diabetes were 86
percent less likely to have seen a dentist in the last year, than
those without diabetes (p-value 0.024). People who had seen
a health provider were 11 times more likely to have seen a
dentist (p-value 0.033). Participants over 65 years were 96
percent less likely to have seen a dental health provider in the
last year. Participants reporting a low perception of their
health were 85 percent less likely to have seen a dentist in the
past year (p-value 0.006).
Conclusions: A clear relationship has been demonstrated
between those who have realized a dental provider visit in the
last year and diabetes status. This disparity is caused to some
degree by the lack of dental insurance in this population and
must also be affected by the extreme poverty of the
households that diabetics reside in. A minimum of education
is a factor since an ability to obtain access to health services is
contingent on basic literary skills.
Implications for Policy, Delivery, or Practice: The
interrelationship between diabetes and periodontal
inflammation suggests that preventive dental care may be
important in preventing complications of both diseases. The
Centers for Disease Control recommends that diabetic
patients see a dentist at least once every 6 months, with more
frequent dental visits for those with periodontal disease. The
American Diabetes Association’s standards for treating
diabetic patients include an examination of the oral cavity as
part of the patient’s initial visit. However, periodic oral
examinations are not included as a standard of continuing
care. It is hoped that new strategies to control diabetes by
improving dental care in the colonias can be devised. An
inability to consume foods due to unhealthy gums and lost
teeth could result in untimely deaths.
Primary Funding Source: RWJF, HRSA
●A Cultural Competence Organizational Model: Lessons
from the Safety Net
Linda Cummings, Ph.D., Donna Sickler, MPH, Ed Martinez,
MPA, Marsha Regenstein, PhD, Jennifer Cromwell
Presented By: Linda Cummings, Ph.D., Director of Research, ,
National Association of Public Hospitals and Health Systems,
1301 Pennsylvania Avenue, NW, Suite 950, Washington, DC
20004; Tel: (202) 585-0100; Fax: (202) 585-0101; Email:
lcummings@naph.org
Research Objective: The largest concentration of LEP
patients in the American health care setting is in the public
hospital system. These hospital systems care for a highly
diverse patient population. Some of the nation’s largest safety
net hospitals and health systems receive interpreter services
requests in over 60 languages. A variety of cultural
competence practices have been developed to better serve
culturally and ethnically diverse patients who often are also
low-income and uninsured. Much has been written about the
goals of culturally and linguistically appropriate practices, but
few models exist that offer practical examples to guide work in
this area. This study focused on cultural competence
practices at Woodhull Medical and Mental Health Center, part
of the largest public hospital system, the New York City Health
and Hospital Corporation, in the most diverse city in the
nation.
Study Design: Woodhull Medical Center was selected for a
comprehensive case study because it is a mature, complex
health care system that has integrated cultural competence
practices throughout the organization, operationalizing these
practices into quality improvement initiatives on a sustained
basis. Researchers conducted in-depth interviews during an
extensive site visit at Woodhull with the chief executive officer,
chief medical officers, human resource director, director of
communications, patient care coordinators, and other key
staff. Researchers visited programs and reviewed materials
developed to support cultural and linguistic services at the
medical center.
Population Studied: Major safety net hospital in New York
City.
Principal Findings: Researchers found cultural competence
programs and policies at Woodhull to be integrated into key
aspects of hospital operations including
administrative/medical leadership, human resources,
interpreter services, customer service, infrastructure, and
community relations. Cultural competence practices and
policies at Woodhull are closely linked with quality
improvement efforts. Despite the considerable resource
constraints affecting safety net organizations, leadership
commitment and ingenuity are instrumental in building a
culturally competent organization and effective approaches,
especially with new or difficult-to-reach patient populations.
Conclusions: This study found that cultural competence
practices at one large safety net hospital can be systematically
described to provide guidance to other health care
organizations as they begin implementation of these
practices. This study offers a practical model for
implementing culturally and linguistically appropriate
practices that identifies unique features of the safety net
setting and the challenges or barriers to implementation.
Implications for Policy, Delivery, or Practice: For safety net
providers, this case study offers a model for implementing
significant cultural competence practices and embedding
them in hospital practices, despite considerable resource
constraints. The study offers valuable lessons for the health
care system as a whole, given the nation’s changing
demographics. These findings have relevance for a variety of
hospitals and health care systems that are experiencing an
influx of new immigrant groups or retooling practices to better
serve a diverse patient population.
Primary Funding Source: Other, National Association of
Public Hospitals and Health Systems
●Use of Adjuvant Chemotherapy and Radiation Therapy
for Colorectal Cancer among Asian Americans and Pacific
Islanders
James W. Davis, Ph.D., Todd B. Seto, M.D., MPH
Presented By: James W. Davis, Ph.D., General Internal
Medicine Fellow, Internal Medicine, Harvard Medical School,
Massachusetts Veterans Epidemiology Research and
Information Center; 790 Boylston Street, 24F, Boston, MA
02199; Tel: (808)382-7767;
Research Objective: Consensus guidelines and randomized
trials support the use of chemotherapy for stage III colon
cancer and chemotherapy with radiation therapy for stages II
and III rectal cancer to improve survival. While the use of
these treatments may be partly related to patient ethnicity, the
management of Asian Americans and Pacific Islanders
(AAPIs) with colorectal cancer remains largely unknown.
Objective: To evaluate the use of chemotherapy and radiation
therapy for colorectal cancer among AAPIs.
Study Design: Retrospective analysis using data from the
Hawaii Tumor Registry, a Surveillance, Epidemiology, and End
Results (SEER) Program population-based registry, linked to
administrative data from Hawaii´s largest health insurer.
Population Studied: We studies the first course of therapy for
602 patients (336 Japanese, 46 Chinese, 74 Filipino, 48
Hawaiian, 98 white; mean age 67+/-13 yrs) diagnosed with
stage III colon cancer (n=347) or stages II (n=109) or III
(n=146) rectal cancer in Hawaii from 1995-2001. We
evaluated the use of standard-of-care treatments as defined by
NCI/NIH guidelines: (1) chemotherapy for stage III colon
cancer and stages II or III rectal cancer, and (2) radiation
therapy for stages II or III rectal cancer. We used logistic
regression to examine the association between ethnicity and
quality of care.
Principal Findings: Overall, 71.6% (n=431) of patients
received chemotherapy. Only 57.6% (n=147) of patients with
rectal cancer received radiation therapy. Chemotherapy rates
were similar among all ethnicities (from 67.6% of Filipinos to
77.1% of Hawaiians, P=0.8). Adjusting for tumor stage and
site, age, gender, year of diagnosis, rural residence, cancer
history, grade, Charlson comorbidity index, health plan (HMO,
fee-for-service, Medicare), and income, chemotherapy was
used significantly less often among older patients (OR, 0.02;
95%CI, 0.01-0.06, for age >/=85 vs. <55) and those with prior
cancer (OR, 0.53; 95%CI, 0.29-0.99) or stage II rectal cancer
(OR, 95%CI: 0.37, 0.21-0.65, for stage II rectal; 1.66, 0.92-3.01,
for stage III rectal vs. stage III colon). Radiation rates were low
among all ethnic groups (from 52.9% of Chinese to 77.8% of
Hawaiians, P=0.26). Hawaiians were more likely than whites
to receive radiation in adjusted analyses (OR, 95%CI: 4.61,
1.25-16.99, for Hawaiians; 1.30, 0.54-3.12, for Japanese; 0.91,
0.29-2.82, for Filipinos; 0.95, 0.25-3.57, for Chinese).
Significant predictors of lower radiation use included
advanced age (OR, 0.09; 95%CI, 0.02-0.47, for age >/=85 vs.
<55) and prior cancer (OR, 0.28; 95%CI, 0.11-0.74).
Conclusions: We found low rates of standard treatments for
colorectal cancer among all ethnic groups in our population,
without significant disparities among AAPIs.
Implications for Policy, Delivery, or Practice: Further study
is needed to determine the reasons for the observed low use
of chemotherapy and radiation therapy in our population.
Primary Funding Source: AHRQ
●Disparities Changes in the United States: Findings from
the National Healthcare Disparities Report
Elizabeth Dayton, MA, Dwight McNeill, Ph.D., MPH, Ernest
Moy, M.D., MPH, Ed Kelley, Ph. D.
Presented By: Elizabeth Dayton, M.A., Junior Service Fellow,
Department of Health and Human Services, Agency for
Healthcare Research and Quality, 540 Gaither Road, Rockville,
MD 20850; Tel: (301)427-1320; Fax: (301)427-1341; Email:
edayton@ahrq.gov
Research Objective: Our objective is to begin describing
changes in disparities in American healthcare and to draw
interpretations and make recommendations for improvement.
Study Design: This study is based on research from the
National Healthcare Disparities Report (NHDR), a
Congressional mandated report that annually analyses
healthcare differences by race/ethnicity and socioeconomic
status across about 250 measures from three-dozen sources.
Sources include 7 databases sampling the civilian,
noninstitutionalized population; 5 sampling healthcare
facilities and providers; 7 extracting from data systems of
healthcare organizations; 4 surveillance and vital statistics
systems databases. Data for minority and poorer populations
are compared with data for whites and higher-income
populations, respectively. Annual rates of change are
determined for measures with multi-year data. Disparities are
considered meaningful if relative differences are at least 10%
and statistically significant with p<0.05, assessed using ztests.
Population Studied: The national U.S. population.
Principal Findings: The 2004 NHDR reports some disparities
changes, though overwhelmingly disparities persist, with
minorities and poorer populations receiving worse care than
whites and higher income groups, respectively.
Some disparities exhibit improvement. For example, in 1998
elderly Asian and Pacific Islander Medicare beneficiaries had
lower rates of influenza vaccination than whites; in 2000 no
significant difference was observed. In 1999 blacks and
Asians were less likely than whites to report a source of
ongoing care; in 2000 a significant difference was not evident.
In 2000 black children age 19-35 months had lower rates of
measles-mumps-rubella vaccination than whites, and
Hispanic children had lower rates of Haemophilus influenzae
vaccination than non-Hispanic whites; a significant difference
was not evident in 2002. In 1992 black women had a higher
rate of late stage breast cancer than whites; 2001 rates were
similar due to falling rates among blacks coupled with rising
rates among whites.
However, overall disparities changes from 2000 to 2001
observed in the NHDR are moderate. About 3% of measures
for blacks improved; about 5% deteriorated. About 14% of
measures for Hispanics improved; about 3% deteriorated.
About 6% of measures for Asians improved; none
deteriorated. American Indians and Alaskan Natives saw no
measures improve; about 5% deteriorated. In sum,
minorities’ healthcare quality improved for about 6% and
deteriorated for about 3% of measures. The poor were the
only income group with any significant change, experiencing
improvement for about 5% of measures.
Conclusions: The national average rate of change for
healthcare disparities is slow, but these analyses illustrate that
improvement is possible. Some changes parallel specific
improvement efforts. For example, elimination of some
vaccination disparities for minority children may be partly
related to the CDC National Immunization Program.
Elimination of the disparity for late stage breast cancer
between black and white women may be partly attributable to
the CDC National Breast and Cervical Cancer Early Detection
Program and improving rates of mammography among
blacks.
Implications for Policy, Delivery, or Practice: Improvement,
deterioration, and disparities highlight best practices and
components and populations in healthcare where targeted
improvement efforts are most needed and would garner the
greatest results; understanding trends in healthcare quality
and disparities may play a key role in improving healthcare.
Primary Funding Source: AHRQ
●Racial/Ethnic Differences in Tumor Stage at the Time of
Diagnosis in Managed Care Settings for Cancers
Detectable by Screening.
Chyke Doubeni, M.D., MPH, Terry S. Field, D.Sc., Diana Buist,
Ph.D., Shelley Enger, Ph.D., Eli Korner, Ph.D., Lois Lamerato,
Ph.D.
Presented By: Chyke Doubeni, M.D., MPH, Assistant
Professor, Family Medicine and Community Health, Meyers
Primary Care Institute/University of Massachusetts, 55 North
Lake Avenue, Worcester, MA 01655; Tel: (508)334-7772; Fax:
(508)595-2200; Email: chyke.doubeni@umassmed.edu
Research Objective: Racial/ethnic differences in the stage of
tumor at the time of diagnosis contribute to disparities in
cancer survival rates. Limited data are available examining the
association between race/ethnicity and cancer stage at
diagnosis in racially diverse populations in settings where
health insurance is not a barrier. We examined this
association among patients who were diagnosed with breast,
cervical, prostate or colon/rectum cancer while enrolled in one
of six geographically diverse managed care organizations
(HMO).
Study Design: This is a repeated cross-sectional study.
Population Studied: This population-based study was
conducted on patients diagnosed with breast, cervical,
prostate or colon/rectum cancer between 01/01/93 and
12/31/98, who were identified from tumor registries associated
with six HMOs. Patients with cancers of unknown tumor
stage were excluded. We further restricted our sample to
patients identified as non-Hispanic white (“white”), nonHispanic black (“black”), Hispanic, or Asian.
Principal Findings: This study was comprised of a total of
24,225 patients with breast, 1,396 with cervical, 22,456 with
prostate and 13,741 with colon/rectum cancer. In
multivariable logistic regression models that adjusted for age
at diagnosis, year of diagnosis, study site and duration of
enrollment with the health plan prior to diagnosis, black
women were much more likely to have advanced breast
((adjusted odds ratio [OR] (95% confidence interval [CI]): 1.62
(1.42-1.84)) or cervical (OR (95% CI): 1.35 (0.90-2.01) cancer at
the time of diagnosis compared to white women. Black men
were also more likely to have advanced prostate cancer (OR
(95% CI): 1.15 (1.01-1.31)) compared to white men. Hispanic
women were slightly more likely to have advanced breast (OR
(95% CI): 1.09 (0.94-1.26)) or cervical (OR (95% CI): 1.11
(0.78-1.57)) cancer at the time of diagnosis compared to white
women but these associations were not statistically
significant. Asian women were less likely to have advanced
breast cancer (OR (95% CI): 0.87 (0.73-1.03)), but were more
likely to have advanced cervical cancer (OR (95% CI): 1.51
(1.01-2.26)) compared to white women. In logistic regression
models that also adjusted for sex, blacks, Hispanics and
Asians were slightly more likely to have advanced
colon/rectum cancer at diagnosis compared to whites,
although the associations did not reach statistical significance.
However, Black men with colon/rectum cancer were
significantly more likely to have advanced disease at diagnosis
compared to white men.
Conclusions: Our study found pervasive racial/ethnic
differences in stage of cancer at diagnosis even in settings
where financial barriers to healthcare access may not exist.
More research is needed to better understand the causes of
these differences.
Implications for Policy, Delivery, or Practice: Elimination of
financial barriers alone may not eliminate racial/ethnic
disparities in cancer stage at diagnosis.
Primary Funding Source: NCI
●The Socio-Economic Gradient in Health and Functioning
Among Children and Youth
Lisa Dubay, ScM
Presented By: Lisa Dubay, ScM, Principal Research Associate,
Health Policy Center, The Urban Institute, 2100 M Street NW,
Washington, DC 20037; Tel: (202)261-5667; Fax: (202)2231149; Email: ldubay@ui.urban.org
Research Objective: The socio-economic gradient in
morbidity, mortality, and overall health status at various
points across the lifespan has been well documented in the
medical, public health, economic, and sociology literatures.
There is also a body of literature that argues that a life-course
perspective is critical to understanding disparities associated
with socio-economic position (SEP). This view derives from
evidence that children’s health and functioning and the
physical, social and psychosocial environments in which they
reside at critical developmental junctures, particularly early
childhood and adolescence, are associated with health status
in adulthood. Surprisingly, there is scant evidence of a socioeconomic gradient in health and functioning in childhood and
adolescence in the US. Perhaps more important to the
development of policies to reduce disparities, or gradients,
there is a paucity of research that examines the pathways
through which SEP affects child health and functioning or
examines multiple dimensions of health and functioning
within the same study. This paper attempts to close this
critical gap in the literature.
Study Design: Data for this paper are derived from the
recently released 1997 and 2002 Child Development
Supplement to the Panel Survey of Income Dynamics (PSID).
The Child Development Supplement (CDS) is uniquely suited
to examine these issues because of the survey’s breadth and
depth of content. Using the PSID, a new measure of socioeconomic position, based on Britain’s National Statistics-Socio-economic Classification (NS-SEC), is used to examine
the extent of the SEP gradient in child and adolescent health
and physical, cognitive, and emotional functioning and to
assess whether this gradient increases as children age. The
paper documents the extent to which SEP is negatively
associated with exposures to family, school, and
neighborhood environments that threaten children’s health
and functioning and positively associated with potential
mediating and adaptive mechanisms. Descriptive,
multivariate, and factor analyses are used in this paper.
Population Studied: A nationally representative sample of
children ages 0-12 were drawn from participants in the PSID in
1997 (n=2907) and were interviewed and re-interviewed five
years later in 2002.
Principal Findings: The results suggest that a SEP gradient
exists in child health and physical, cognitive, and emotional
functioning and that these gradients grow as children age.
Furthermore, children of lower SEP are more likely to live in
family, school, and neighborhood environments that threaten
rather than promote child health and functioning.
Conclusions: Understanding the pathways through which
SEP affects child health and functioning is critical to
developing effective primary and secondary interventions that
can reduce SEP disparities.
Implications for Policy, Delivery, or Practice: The current
policy focus on reducing health disparities through the health
care system may be necessary but is an insufficient
mechanism for eliminating health disparities. Unless the
fundamental disparities in children’s family, school, and
neighborhood environments are ameliorated, the gradient in
health disparities will continue to exist and to grow over the
life-course of these children.
Primary Funding Source: Other
●Willingness to Participate in Clinical Trials Among Elderly
Whites and African Americans
Raegan Durant, M.D., Roger B. Davis, Sc.D., Edward
Marcantonio, M.D., MSc, Marcie B. Freeman, MEd, Bruce E.
Landon, M.D., MBA
Presented By: Raegan Durant, M.D., General Medicine
Fellow, Division of General Medicine and Primary Care, Beth
Israel Deaconess Medical Center, 330 Brookline Avenue, Rose
111, Boston, MA 02130; Tel: (617)667-0487; Fax: (617)667-2751;
Email: rdurant@bidmc.harvard.edu
Research Objective: Despite National Institutes of Health
mandates for minority inclusion in clinical research, African
Americans are underrepresented in clinical trials. Many have
hypothesized that African Americans are less willing to
participate in clinical trials partially due to higher levels of
distrust in the health care system. We sought to determine
the differences in willingness to participate in clinical trials
among elderly whites and African Americans and the factors
that influence their willingness to participate.
Study Design: We assessed willingness to participate in 3
hypothetical clinical trials with increasing levels of risks (diet
trial for obesity, antihypertensive trial for HTN, chemotherapy
trial for cancer) on an ordinal scale from 1 (“not willing”) to 5
(“very willing”). We created a mean willingness to participate
scale calculating the mean score for each respondent based
on responses to the 3 vignettes (Cronbach’s alpha 0.74). For
analysis, each respondent was categorized as less (mean
score < 4) or more (mean score > or = to 4) willing. We
assessed trust in one’s primary care provider (PCP) and trust
in clinical research separately using two previously validated
scales. We also assessed sociodemographic factors,
comorbidities, health status, previous exposure to clinical
research, personal experiences with discrimination, and
attitudes toward potential barriers and incentives associated
with trial participation. We performed bivariable analyses to
examine the associations between these factors and
willingness to participate. Those factors significantly
associated with the outcome were used to create a
multivariable logistic regression model.
Population Studied: We surveyed, via mail, 1412 eligible
community dwelling elderly (age >50) whites and African
Americans in the Boston metropolitan area.
Principal Findings: We received responses from 755 eligible
persons (53% response rate). The study population was 63%
white and 37% African American. African Americans were not
significantly more distrustful of their care providers (46.3% vs.
44.6%, p=0.66) or clinical research (3.2% vs. 3.0%, p=0.84).
In unadjusted analyses, whites were slightly more likely to be
more willing (32.8% vs. 29.9% p=0.13) to participate in clinical
trials compared to African Americans. In the multivariable
model, male gender (OR 2.40, 95% CI 1.58,3.64), having
dependent care responsibilities (OR 1.96, 95% CI 1.16,3.32),
previous trial participation (OR 1.84, 95% CI 1.24,2.74), and
the endorsement of the importance of participant benefit from
trial participation (OR 2.16, 95% CI 1.24,3.76) were all
associated with an increased willingness to participate.
Respondents who thought it would be important for their own
PCP to ask them to be in a trial were less willing to participate
(OR 0.60, 95% CI 0.37,0.96). Neither race nor distrust was
independently related to willingness to participate in the final
multivariable model.
Conclusions: African Americans were not significantly less
willing to participate in clinical trials compared to whites.
Distrust was not an independent predictor of willingness to
participate in clinical trials. Low representation of African
Americans in clinical trials may be due to factors other than a
lower willingness to participate.
Primary Funding Source: NIA
●Effects of Medicare Beneficiary Race/Ethnicity, SocioEconomic Status, and Coverage Type on Mammography
Screening: Comparison of a Diverse Group of Medicare
Advantage and Fee-for-Service Enrollees
Kelle Eason, MPH, Ashley Antler, BA, Michelle Fernandez,
MPH, Susan Merrill, PhD, MPH
Presented By: Kelle Eason, MPH, Healthcare Data Analyst,
Scientific Affairs, Lumetra, One Sansome Street, Suite 600,
San Francisco, CA 94104; Tel: (415)677-2168; Fax: (415)6778436; Email: keason@caqio.sdps.org
Research Objective: Evidence suggests patients enrolled in
Medicare Advantage (MA) (Medicare managed care, formerly
known as Medicare+ Choice) may be more likely to receive
regular preventive care than their counterparts with Fee-forService (FFS) coverage. Furthermore, older women, non-white
women, and women in lower socio- economic strata are less
likely to receive mammograms. The primary objective of this
study is to identify patterns in mammography screening
among diverse populations in both MA and FFS settings. The
aim is to compare mammography rates and patterns in each
setting while controlling for demographic factors associated
with receipt of mammography; race/ethnicity, age, dualeligibility (enrollment in both Medicare and Medicaid), and
socio-economic factors such as income and education, in
order to determine if these factors alone or in combination
account for variance in mammography screening rates
between settings with different funding and organizational
structures. We hypothesized that the MA population would
have consistently higher mammography rates than women
enrolled in FFS, and that mammography screening patterns
would differ between the MA and FFS populations.
Study Design: Health Plan Employer Data and Information
Set (HEDIS) mammography population data (2001-2002) and
Centers for Medicare and Medicaid Services (CMS) (20012002) Medicare administrative claims data were used to
assess screening status of female MA and FFS Medicare
enrollees in California. Demographic variables for both
populations were obtained from CMS enrollment data, and
2000 Census data were used to estimate socio-economic
factors of the beneficiary neighborhood by zip code. Analyses
examined individual demographic characteristics that were
associated with the receipt of biennial mammography in each
population, and those that had an inverse effect. Combined
and interactive effects were also evaluated, and differences in
mammography patterns between the two populations were
identified. This study does not include comparisons between
managed care plans or physician offices.
Population Studied: Study population includes women aged
50-67 as of January 2001 and their mammography screening
status during January 2001-December 2002.
Principal Findings: Overall, MA mammography rates were
higher (76.4%) than those in FFS (45.8%). In the MA
population, dual eligibility and/or Hispanic race/ethnicity were
both independently associated with not receiving a
mammogram. In the FFS population, dual eligibility was not
as significant, but any non-white race/ethnicity (Black,
Hispanic, Asian, Native American, and Other) was associated
with not having received a mammogram during the two-year
measurement period.
Conclusions: Contrary to our initial hypothesis that the MA
population would have consistently higher mammography
rates compared to women enrolled in FFS, our findings
suggest that this pattern may not hold true across all
demographic groups. Dually eligible women had a lower
mammography rate in both settings, a finding consistent with
previous studies indicating that lower income women are less
likely to receive a mammogram. However, the disparity in
rates between dually eligible and non-dually eligible women in
MA was twice that observed in the FFS population, suggesting
that the pattern of better preventive services in MA does not
necessarily apply to lower income women.
When analyzed by race/ethnicity, similar patterns were
observed for MA and FFS women. Non-White women in both
settings experienced lower rates of mammography screening
than their White counterparts in the same setting.
Comparison of rates between settings revealed that White,
Black and Asian women in MA were more likely to receive a
mammogram than their counterparts in FFS. Surprisingly, a
different pattern was observed for Hispanic women, who were
no more likely to have received a mammogram in the MA
versus FFS setting.
Implications for Policy, Delivery, or Practice: When patterns
within the MA and FFS settings are identified and compared, a
more complex picture of the effects of demographics begins
to emerge. Providers and administrators may consider special
breast cancer education and mammography reminder
programs with particular attention to coverage type and
population diversity. Providers in both settings may share best
practices to improve mammography screening among their
respective patients. The patterns observed in this study may
have larger implications for the future of Medicare
mammography screening, as the diversity of this study
population mirrors predicted changes in the Medicare
population nationwide.
Primary Funding Source: CMS
●Differences In Symptom Presentations and Screening of
African Americans and Whites for Benign Prostatic
Hyperplasia (BPH)
Bennett Edwards, Ph.D., Daniel Howard, Ph.D., Kimberly
Whitehead, Ph.D., Ahinee Amamoo, MS, Yhennekko Jallah,
MS, Paul Godley, M.D., Ph.D., MPH
Presented By: Bennett Edwards, Ph.D., Associate Professor,
Allied Health Professions, Shaw University, 118 East South
Street, Raleigh, NC 27601; Tel: (919)231-2501; Fax: (919)7434693; Email: bedwardssu@aol.com
Research Objective: To determine whether there are racial
differences over time in symptoms, screening, and health
care-seeking behaviors for BPH
Study Design: Four common urinary symptoms evaluated
were: pushed or strained to urinate; urinated again shortly
after urination; stopped and started several times when
urinating; and dribbled urine after initial urination period.
Severe urinary symptoms (incontinence, and prostate problem
prevented urination) were also evaluated. Screening was
measured by whether the respondents received regular digital
rectal exams (DRE) by a physician (MD). Respondents' were
asked if they regularly go to a MD and if they put-off care quite
often, which were used to measure the health care-seeking
behaviors of the respondents. In addition to these factors,
information on respondents' socioeconomic factors, health
conditions, and MD characteristics were also collected.
Comparisons between AA and W elders' responces to
common and severe symptons, screening behaviors, and
health care seeking behaviors were compared using Fisher's
exact tests. Generalized Estimating Equations (GEE) were
used to determine the racial effects over time between urinary
symptoms and BPH, adjusting for socioeconomic factors,
health conditions and MD characteristics.
Population Studied: The study population included male
respondents, ages 65-85, from Waves 3 (1994-African
Americans (AA)=482; Whites (W)=407 and 4 (1998-AA=253;
W=237) of the Established Populations for the Epidemiologicic
Studies of the Elderly (EPESE).
Principal Findings: W reported all four common symptoms
more frequently than AA in 1994 and 1998, with stopped and
started several times when urinating (p=0.002); and dribbled
urine after initial urination period (p=0.034) being significant.
However, AA were more likely to report severe symptoms,
with incontinence being statistically significnat over time
(p=0.002). AA were less likely than W to have a regular DRE
in both 1994 and 1998 (p=0.008 and p<0.001, respectively)
and less likely to see an MD in 1994 (p<0.014).Futhermore,
over time AA were more likely to not have a regular rectal
exam because they were not bothered by their symptoms
(p=0.016). GEE statistics for AA over time demonstrated
significantly more No Named Physician responses than W.
Those that regularly had a DRE by an MD were less likely to
live in a rural area(OR=0.301;p<0.004) and less likely less
likely to put off health care (OR=0.295; p=0.004), when
adjusted for other socioeconomic, health conditions and MD
characteristics. AA had significantly more incidences of
diabetes than W in 1998 (p=0.023).
Conclusions: Whites reported common urinary symptoms
more often although, African Americans appear to present
severe symptoms differently and are less likely to see an MD.
The lack of access to a MD or lack of bother may contribute to
their advanced BPH symptoms. In addition, Diabetes may
confound and/or mask BPH symptoms.
Implications for Policy, Delivery, or Practice: A continued
effort is needed to increase early MD visits with African
American males males to decrease the number of those
presenting with severe BPH symptoms.
Primary Funding Source: AHRQ, NIH National Center on
Minority Health and Health Disparities/R24 MD00167/P60
MD000239
●Racial Disparities in Lipid Management in Patients with
Diabetes
Jennifer Elston Lafata, Ph.D., Manel Pladevall, MPH, M.D., Jan
Simpkins, MA, Kaan Tunceli, Ph.D., George Devine, Ph.D., L.
Keoki Williams, MPH, M.D.
Presented By: Jennifer Elston Lafata, Ph.D., Director, Center
for Health Services Research, Henry Ford Health System, 1
Ford Place, Suite 3A, Detroit, MI 48202; Tel: (313)874-5454;
Fax: (313)874-7137; Email: jlafata1@hfhs.org
Research Objective: To describe lipid management over time
in a cohort of patients with diabetes (DM) and evaluate
whether care receipt differed between African American and
White populations in an equal access environment.
Study Design: Automated claims and clinical databases were
used to identify a cohort of patients with DM in 1997/1998
that was retrospectively followed through 2002 (mean follow-
up = 42.1 months). Overall and race stratified rates of
hypercholesterolemia screening, treatment and goal
achievement were estimated in each follow-up year.
Treatment was determined by a claim for lipid lowering agents
and goal attainment was defined as low density lipoprotein
cholesterol (LDL-C) less than 100 and 130 mg/dL.
Population Studied: Retrospective cohort of 11,411 HMO
enrollees aged 18+ years (50.8% female; 53.2% White, 43.1%
African American, and 3.7% other), with DM who were
continuously enrolled during 1997/1998.
Principal Findings: During follow-up, rates of testing,
treatment and goal attainment improved over time for both
races. Racial disparities favoring the White cohort were
evident for all rates in each year. Rates of testing increased
from 60.7% in 1999 to 76.8% in 2002 for Whites and 48.2%
to 71.1%, respectively for African Americans. Rates of
treatment increased from 34.6% in 1999 to 53.4% in 2002 for
Whites and 26.1% to 45.7%, respectively for African
Americans. Rates of goal achievement at LDL-C <100 mg/dL
increased from 34.9% in 1999 to 42.5% in 2002 for Whites
and 23.9% to 30.8%, respectively for African Americans. Rates
of goal achievement at LDL-C <130 mg/dL increased from
71.2% in 1999 to 79.7% in 2002 for Whites and 59.1% to
67.6%, respectively for African Americans. Among patients
treated with lipid lowering agents, rates of goal achievement
over the same period improved from 67.3% to 75.5% when
using a goal of LDL-C <130 mg/dL but only 34.1% to 40.3%
when using the currently recommended goal of LDL-C <100
mg/dL.
Conclusions: Our preliminary findings show that racial
disparities in rates of testing tended to decrease over time,
while those associated with LDL-C goal achievement and
treatment with lipid lowering drugs tended to persist over
time. Overall gains in all rates were achieved between 1999
and 2002 but the percentage of these high risk patients at the
current recommended LDL-C goal (i.e., LDL-C <100 mg/dL)
remains low regardless of race. We are in the process of
evaluating these racial disparities adjusting for other factors.
Implications for Policy, Delivery, or Practice: This research
shows persistent underachievement of recommended LDL-C
goal levels among those known to be at elevated risk for
cardiovascular disease, especially among African Americans.
Since appropriate use of preventive and early diagnostic
interventions is key in reducing the health and economic
burden of cardiovascular disease it is incumbent on health
care delivery systems to address these disparities in
treatment.
Primary Funding Source: Blue Cross Blue Shield of Michigan
●Reducing Disparities in Diabetes Care among the
Nation’s Most Vulnerable Patients
Christopher Forrest, M.D., Ph.D., Kaytura Felix-Aaron, M.D.,
Jinlin Wang, MS, Amisha Pandya, MHS, Elvan Daniels, M.D.,
Leiyu Shi, Dr.PH.
Presented By: Kaytura Felix-Aaron, M.D., Chief, Clinical
Quality Data Branch, Division of Clinical Quality, Bureau of
Primary Health Care, 5600 Fisher Lane, Room 17C-26,
Rockville, MD 20857; Tel: (301) 594-4306; Fax: (301) 594-5224;
Email: KFelix-Aaron@hrsa.gov
Research Objective: Low-income patients are
disproportionately affected by diabetes, one of the most
prevalent and costly chronic conditions affecting the US
population. The federal government funds health centers
(HCs) to provide services to vulnerable patients residing in
medically underserved communities. One of the primary
goals of HCs is to provide high quality services to all patients
according to their needs, rather than their insurance status,
race/ethnicity, or poverty status. This study was done to
examine the costs and quality of diabetes care in HCs to
assess whether they are fulfilling their goal of reducing
healthcare disparities.
Study Design: Cross-sectional analysis using sociodemographic, encounter, and practitioner data from 2002 and
readily available from HCs’ management or clinical
information systems. A standardized charge obtained from
the Medicare fee schedule was applied to each procedure
code in the encounter file. Morbidity burden and risk adjusted
charges were obtained using the Johns Hopkins ACG case-mix
system. We used generalized estimating equations (GEE) to
control for age, sex, morbidity burden, and delivery site effects.
Population Studied: The study was conducted in the Sentinel
Center Network (SCN), a newly formed, national network of 52
health centers that share clinical data on all their patients to
inform policy and improve practice for the health center
safety-net. Data from a total of 962,011 patients (about 10%
of all HC patients in the nation) comprise the SCN database.
The diabetes patient study sample includes 58,519 adults age
18 and older.
Principal Findings: Overall, 9.2% of the adult patient
population, 8.1 % of white, 10.8% of black, 10.0% of Asian,
and 9.2% of Hispanic had diabetes. Compared with those
without diabetes, patients with diabetes had 2-fold higher
morbidity burden (ACG risk score 2.99 v 1.46, p<.001); higher
primary care expenditures/patient ($598 v $340, p<.001); and,
higher visit rates/patient (7.81 v 3.81, p<.001). Multivariable
analysis showed charges for uninsured patients were
comparable to privately insured diabetes patients, whereas
those for Medicaid and Medicare patients were significantly
higher. In addition, charges for black and white patients were
comparable, but those for Hispanic and Asian patients with
diabetes were higher than those for whites. Charges for
patients with diabetes whose income was below the federal
poverty level (FPL) were similar to those between 101-200% of
FPL and higher than those above 200% of FPL. With respect
to quality of diabetes care overall, 57.8% of adult patients with
diabetes had 1+ Hemoglobin A1C tests, 45.4% 1+ LDL tests,
and 40.3% both tests done during 2002. The adjusted odds
of having both tests performed was highest for the uninsured
compared with those with public or private insurance, and the
odds for ethnic minorities was higher than whites.
Conclusions: We found no disparities in the costs or quality
of diabetes care by insurance payer, race/ethnicity, or family
income in the nation’s health centers.
Implications for Policy, Delivery, or Practice: Health centers
are achieving their goal of eliminating disparities by matching
services according to patients’ needs. This equitable model of
primary care service delivery is likely a result of the unique
organization and financing of health centers, their governance
and mission, including their responsiveness to local needs,
and the health professionals who staff them. The health
center model of primary care should be considered more
broadly as a policy option for eliminating disparities
throughout the nation.
Primary Funding Source: HRSA
●Racial Segregation of Nursing Home Staff and Residents
Zhanlian Feng, Ph.D., Mary Fennell, Ph.D., Vincent Mor, Ph.D.
Presented By: Zhanlian Feng, Ph.D., Project Analyst, Center
for Gerontology and Health Care Research, Brown University,
2 Stimson Avenue, Providence, RI 02912; Tel: (401) 863-9356;
Fax: (401) 863-9219; Email: Zhanlian_Feng@brown.edu
Research Objective: Previous research has documented
persistent racial and ethnic disparities in various health care
outcomes for elderly patients in long-term care settings. Yet,
until recently little attention has been given to the issue of
unequal access to nursing home care, which suggests that
minority residents, particularly African Americans, are at far
greater risk of residing in facilities with limited resources and
historically poor performance than are whites. There is a
lamentable dearth as well of available data on the racial/ethnic
mix of nursing home staff. The purpose of this study is to
describe the extent of racial segregation in the distribution of
nursing home staff vis-à-vis residents.
Study Design: The facilities under study represented a
subsample from a national sample of nursing homes which
participated in a study of organizational strategy and
performance, in which data were collected on race/ethnicity of
nurse aides. The subsample was comparable to national
averages in terms of urban location, fairly close to national
averages on percent of residents on Medicaid and percent
self-pay, and these facilities were somewhat less likely to be
for-profit or chain affiliated. Responses were matched with
concurrent Minimum Data Set (MDS) resident assessments
to obtain information on race/ethnicity of residents. The
black-white Dissimilarity Index was used to depict the patterns
of racial segregation of both facility staff and residents.
Population Studied: The sample included 19,874 long-stay
nursing home residents and 2,130 nurse aides from 145
freestanding facilities in 8 states in 1995.
Principal Findings: Nearly 36% (N=52) of the study facilities
had no black residents, whereas a few facilities (N=6) had over
50% black residents. Over 80% of all black residents resided
in just under 20% of all facilities. The Dissimilarity Index was
.648 for residents and .653 for staff, meaning about 65% of all
black residents (or staff) would have to be relocated to achieve
a perfectly even proportion of black residents (or staff) across
all facilities. The correlation between the proportions of black
residents and of black staff was fairly strong (.7), suggesting
that homes with more black residents are heavily concentrated
with black staff, and vice versa. Furthermore, facilities that
were highly concentrated with black residents were more likely
to be those with poorer resources as indicated by a
substantially higher proportion of Medicaid residents and a
lower proportion of private-paying residents.
Conclusions: Access to nursing home care is marked by
considerable racial/ethnic segregation, as evidenced by the
high concentration of black residents in a small number of
facilities. The racial/ethnic composition of nursing home staff
also exhibited a similar pattern of segregation.
Implications for Policy, Delivery, or Practice: Additional
research is needed to assess the extent to which racial
segregation of staff and residents in nursing homes mirrors
residential segregation patterns in the local community, or is
associated with other community or market characteristics.
The primary question generated by these findings is what
difference this high degree of racial segregation makes in the
quality of care provided the average non-white nursing home
resident.
Primary Funding Source: NIA
●Social Identity and the Perceived Efficacy of and
Engagement in Health Behaviors among a Low-Income,
Minority Community Sample
Tracy Finlayson, BS, Shawna Lee, MSW, MPP, Daphna
Oyserman, Ph.D.
Presented By: Tracy Finlayson, BS, Research Assistant, Health
Management and Policy, University of Michigan School of
Public Health, 109 South Observatory, Ann Arbor, MI 481092029; Tel: (734)615-5131; Email: tfinlays@umich.edu
Research Objective: This research seeks to explore new
frontiers in the determinants of health disparities by
examining why risky and protective health behaviors are
differentially related to social class. We investigate the
interface between social identity and health, and develop a
process model that accounts for when health messages and
behaviors are judged as relevant for socially disadvantaged
Americans and motivate health-related behavioral changes.
We postulate that the idea of health in this country is socially
represented by white, middle-class images, values, lifestyles,
and behaviors that are not meaningful for people in other
social groups. We aim to show that common health
messages advocating certain health-promoting behaviors in
the realms of oral health, physical health, and mental health
are not perceived as relevant to many minority groups and
socio-economically disadvantaged Americans, and are not
effective in bringing about a change in behavior.
Study Design: We held focus groups to inform the
development of a survey on personal engagement in
behaviors and perceptions about social identity and behaviors.
Following a standard priming methodology, to make salient
stigmatizing social identity membership (being reminded that
one is a poor minority) participants were randomly assigned
to prime or no prime survey conditions.
ANOVA and regression analyses were performed to examine if
making social identity salient altered efficacy ratings of various
types of health behaviors and reported engagement in those
behaviors, controlling for the individual’s level of identification
with that social identity (Group Belongingness).
Population Studied: We surveyed a community sample of
primarily low-income African-American women attending the
weekly job fair at a state social service agency in Michigan. 121
surveys were collected during Summer 2004.
Principal Findings: Strongly identifying with one’s social
group (High Group Belongingness) predicted participants’
beliefs that other members of their social group smoked often
and were not likely to eat fruits/vegetables regularly. It also
predicted a decreased likelihood of flossing in the past week.
Low Group Belongingness was a significant predictor of lower
self-rated oral health and of decreased likelihood of having
eaten fresh fruits/vegetables in the past week. Results also
indicated that being in the social identity prime condition
(reporting on social identity at the beginning of the survey)
and being high in Group Belongingness interacted to predict
whether respondents currently smoke. Additionally,
perceiving that utilization of mental health services was
relevant to one’s social group was positively correlated with
perceptions related to preventive oral health and physical
health behaviors.
Conclusions: Several behaviors were related to how strongly
individuals identified with their social identity. The overall
level of Group Belongingness influenced respondents’
endorsement of their personal engagement with health
behaviors related to smoking, oral health, and diet.
Implications for Policy, Delivery, or Practice: These findings
can inform the development of targeted, culturally sensitive
health education and intervention efforts intended to reduce
existing health disparities by extending accurate health
information and services to hard-to-reach high-risk
subpopulations.
Primary Funding Source: Walter J. McNerney Award
●Health Services Utilization in Canadian Ethnic
Populations
Andrew Fong, B.Comm., BA, Hude Quan, M.D., Ph.D., Jianli
Wang, M.D., Ph.D., William Ghali, M.D., MPH, Richard
Musto, M.D., Tom Noseworthy, M.D.
Presented By: Andrew Fong, BA, Senior Research Analyst
with the Calgary Health Region, Community Health Science,
University of Calgary - Graduate Student, 1403 - 29th Street
NW, Calgary, T2N 2T8; Tel: (403) 944-8926; Fax: (403) 9448950; Email: Andrew.Fong@calgaryhealthregion.ca
Research Objective: Racial disparities in health care
utilization and outcome have been extensively studied in the
United States. These studies found that visible minorities have
a lower likelihood of receiving effective procedures and certain
beneficial medications such as preventive care services and
referral to specialists for evaluation. Moreover, visible
minorities in the United States are more likely to receive
delayed care and/or poorer quality care. In Canada, ethnic
differences and related disparities in Canada's health care
system have not been studied extensively and is not
understood sufficiently. This study assesses the health
services utilization in ethnic populations residing in Canada.
Study Design: This study analyzed data from the Canadian
Community Health Survey (cycle 1.1). This national crosssectional telephone survey collected self-reported ethnicity and
utilization of family doctor, specialist and in-hospital stay in
the last 12 months and cancer screening test in their life time
as of 2001. The sampling weights endorsed by Statistics
Canada were used to calculate estimates.
Population Studied: Of 131,535 respondents, 3665 (2.8%)
aboriginal and 6558 (5.0%) unspecified ethnicity data were
excluded. This resulted in a data set that included 7057 visible
minority. Of visible minority, 29.9% were Chinese, 0.29%
Japanese or Korean, 20.5% South Asian, 13.8% South East
Asian, 7.2% West Asian/Arabic, 14.4% Black and 5.9% South
American.
Principal Findings: Visible minorities had similar frequencies
of visit with family doctors with 78.4% of both Caucasion and
Visible minorities visiting a family doctor within a 12 month
period. In contrast, Visible minorities had lower frequencies of
consultations with medical specialist (48.3% vs. 54.7% of
Caucasian), lower length of stay in hospital (5.5 vs 8.3 days
with Caucasian), lower psa test amongst men aged 40 years
and older (30.4% vs 44.7% of Caucasian), lower mammogram
and PAS smear test for women aged 35 years and older
(53.9% vs. 65.7% of Caucasian) and ( 78.4% vs 90.2% of
Caucasian) respectively.
Conclusions: The level of health services utilization varies by
type of service among Canadian ethnic populations. Visible
minorities appear to under-utilize health care that is accessible
through physician's referral compared with Caucasians. The
reasons for under-utilization of specialty, in-hospital care and
cancer screening test in visible minorities should further be
studied.
Implications for Policy, Delivery, or Practice: Despite
Canada's universal health care system, ethnic diversity in
health services utilization still exits.
Primary Funding Source: No Funding Source
●Heart Disease Risk Factors: Diagnosis Differences
Sheila Franco
Presented By: Sheila Franco, Staff Fellow, National Center for
Health Statistics, CDC, 3311 Toledo Road, Hyattsville, MD
20782; Tel: (301)458-4331; Fax: (301)458-4038; Email:
sfranco@cdc.gov
Research Objective: Heart disease is the leading cause of
death in the United States. Hypertension and high cholesterol
are independent risk factors for cardiovascular disease.
People with multiple risk factors are at greater risk for heart
disease than those with a single risk factor and risk increases
as the number of risk factors increase. For those without
insurance, these risk factors may go undiagnosed, and
therefore untreated, leading to increased morbidity from heart
disease. Those without insurance may also be less likely to
afford treatment, including drug therapy, even if these
conditions are diagnosed. This paper examines multiple heart
disease risk factors by age, sex, race/ethnicity, health
insurance, and poverty status.
Study Design: This analysis used data from the National
Health and Nutrition Examination Survey (NHANES).
NHANES collects data from physical exams and information
on family characteristics during interviews. Bivariate analysis
was used to examine multiple risk factors for heart disease.
The multivariate analysis is ongoing.
Population Studied: Persons age 20 and over are included. A
person with undiagnosed hypertension is defined as not being
told by a doctor they have hypertension but having measured
elevated blood pressure. A person with undiagnosed high
serum cholesterol is defined as not being told by a doctor they
have high cholesterol but having measured total serum
cholesterol greater than or equal to 240 mg/dL.
Principal Findings: In 1999-2000, 14 percent of adults aged
20 and over had either undiagnosed hypertension or
undiagnosed high serum cholesterol. Of those with high
cholesterol, 15 percent were undiagnosed. Of those with
hypertension, 30 percent were undiagnosed. Undiagnosed
hypertension was more common for those without insurance
(35%) than those with some form of coverage (29%).
Undiagnosed high cholesterol was also more common for
those without insurance (24%) than those with insurance
coverage (15%). Among those lacking health insurance
coverage, poverty status, gender, and age were not predictors
of having undiagnosed hypertension, among those with
hypertension. Among those with hypertension but without
insurance coverage, undiagnosed hypertension was more
common for Mexicans (45%) than black non-Hispanics (37%)
and white non-Hispanics (32%).
Conclusions: The lack of insurance is related to having
undiagnosed high cholesterol or hypertension. Notably,
among those with hypertension but without insurance,
Mexicans were significantly more likely to undiagnosed than
white or black non-Hispanics.
Implications for Policy, Delivery, or Practice: Screening for
hypertension and high cholesterol is effective in identifying
and improving treatment for these conditions. Those lacking
health insurance may be less likely to be screened, diagnosed,
or treated for these conditions. Treatment of heart disease
risk factors can reduce morbidity and health care utilization. If
those without health insurance are less likely to be diagnosed
with these important heart disease risk factors, then they are
also less likely to get treatment for these conditions. This may
lead to increased morbidity and health care utilization and
costs in later years.
Primary Funding Source: CDC
●Ethnic Disparities in Health and Health-Related
Behaviors among Homeless Women
Lillian Gelberg, M.D., MSPH, Austin Erika, Ph.D., Ronald M.
Andersen, Ph.D.
Presented By: Lillian Gelberg, M.D., MSPH, George F. Kneller
Professor, Department of Family Medicine, UCLA, 50-071 CHS
Box 951683, Los Angeles, CA 90095-1683; Tel: (310) 794 6092;
Fax: (310) 794 6097; Email: lgelberg@mednet.ucla.edu
Research Objective: Limited empirical research has focused
on ethnic differences in the health of homeless women. Using
a probability sample of homeless women in Los Angeles, we
employ the Behavioral Model for Vulnerable Populations to
examine the predisposing, enabling, and need factors
associated with health and health-related behaviors in this
population.
Study Design: A community-based probability Sample of 974
homeless women in 66 Los Angeles County shelters and meal
programs underwent a one hour structured face-to-face
personal interview.
Population Studied: Homeless women.
Principal Findings: Numerous ethnic differences in the
correlates of health and health-behavior exist among homeless
women. White women are especially vulnerable to poor
health, reporting more gynecological symptoms, limitations
on physical functioning, history of drug and alcohol abuse,
and history of psychiatric hospitalization. White women’s
experience of homelessness is also more severe compared to
African Americans and Hispanics; White women had been
homeless longer on average and reported more exits from
homelessness. No significant ethnic differences exist in key
sociodemographic factors including income, health insurance
coverage, work status, or receipt of food stamps.
Conclusions: The experience of homelessness differs by
ethnicity, as do the correlates of health and health-related
behaviors. Notably, White women in this population
experience poorer health than African American and Hispanic
women on a number of measures. These ethnic differences
must be considered when using the Behavioral Model for
Vulnerable Populations to examine health outcomes.
Implications for Policy, Delivery, or Practice: Ethnic
disparities may differ in various vulnerable populations. While
homeless women of all ethnic backgrounds experience poorer
health compared to housed women, our findings suggest that
white homeless women experienced poorer health relative to
Hispanics and African American women. While most
homeless women have health disparities, our findings that
white homeless women had the greatest disparity on many
measures suggest that this subgroup also needs to be
targeted by program planners and policy makers to improve
their health and access to care.
Primary Funding Source: NIDA, Robert Wood Johnson
Foundation, NIAAA, AHRQ
●Inner City Adolescents At Highest Risk of Health
Consequences of Obesity
Roy Grant, MA, Marian Larkin, M.D., Alan Shapiro, M.D.,
Sandra Goldsmith, MS RD, Basma Faris, MS RD,
Presented By: Roy Grant, MA, Director of Research, Research,
The Children's Health Fund, 317 East 64th Street, New York,
NY 10021; Tel: (212)535-9400; Fax: (212)535-7699; Email:
rgrant@chfund.org
Research Objective: Obesity (BMI => 95th percentile) more
than doubled from 1980-2000 among children 6-11 years old
and tripled for adolescents 12-19. Associated health conditions
including type 2 diabetes (T2DM) and hypertension also
increased. Black and Hispanic children and youth are
disproportionately affected. Adult obesity, T2DM and
cardiovascular disease (CVD) are more firmly associated with
adolescent than childhood obesity. Our objective was to refine
a protocol to target patients at highest risk for health
consequences of obesity for weight management
interventions. Using this protocol, we sought to describe the
characteristics of this highest risk group, for whom weight
management intervention would be medically necessary.
Study Design: An algorithm was developed for use at
comprehensive physical exams to identify these risk factors:
race-ethnicity, family history, BMI, physical findings, lab
results. Identified patients were referred for further lab testing,
nutrition counseling, and endocrinology consultation if
needed. Only patients Tanner II or higher and age 12-19 were
included to describe patients at the highest risk. Data quality
was checked by retrospective chart review.
Population Studied: Based on retrospective chart review of a
random sample of 200 patients who had a comprehensive
physical examination between June 1 and December 31, 2003,
prevalence among patients age 6-19 years in our
predominantly Hispanic South Bronx Health Center is 19%
overweight (BMI 85th to 94th percentile) and 25% obese (BMI
at or above 95th percentile). With 2,226 patients 6 to 19 years
old, this indicates that approximately 1,000 are at risk of
overweight, obesity, and related health consequences.
Principal Findings: For the first 75 identified patients:
Demographics, 71% Hispanic, 29% Black; 56% male; mean
age, 14.7 years; BMI: 91% obese, 9% overweight, mean, 33.5;
Other risk factors: 87% family history of diabetes; 70%
acanthosis nigricans; 35% dyslipidemia; 24% elevated blood
pressure (BP). On chart review, nutrition notes were found in
48% of charts. Dyslipidemia and elevated BP had been underidentified by about half.
Conclusions: In an inner city community health center,
prevalence of pediatric overweight and obesity is high. To
make better use of scarce resources, a protocol to identify
patients at the most immediate risk is useful. This protocol is
effective in getting high risk patients to a nutritionist and
identifying patients with insulin resistance and incipient CVD.
The prevalence of obesity with complications suggests that
weight management interventions are medically necessary for
these patients.
Implications for Policy, Delivery, or Practice: For practice:
primary care providers should consistently reference lab
results received after the visit in determining patient risk
status. BP readings should be consistently checked for ageappropriate norms. For policy: the profile of patients at the
highest risk of T2DM and CVD identified in this population
may serve as the basis for a diagnostic entity of “adolescent
obesity syndrome” for which nutrition counseling and other
weight management services should be reimbursed as
medically necessary. The short-term increase in health
expenditures is anticipated to be small in relation to the
eventual savings achieved by preventing early onset diabetes
and heart disease.
Primary Funding Source: Bristol-Myers Squibb
●Racial Differences in Attitudes Toward Innovative
Medical Technology
Peter Groeneveld, M.D., MS, Seema S. Sonnad, Ph.D., Anee K.
Lee, BA, David A. Asch, M.D., MBA, Judy A. Shea, Ph.D.
Presented By: Peter Groeneveld, M.D., MS, Assistant
Professor of Medicine, Center for Health Equity Research and
Promotion, Philadelphia VA Medical Center, 3900 Woodland
Ave, Philadelphia, PA 19104-4155; Tel: (215) 898-2569; Fax:
(215) 573-8779; Email: peter.groeneveld@med.va.gov
Research Objective: Patient "innovativeness," defined as the
propensity to adopt and use new technology, is likely to be an
important component of healthcare decision-making. Our
goal was to measure differences in innovativeness between
black and white patients. Pronounced racial variation in
attitudes may explain differences in both patient preferences
for particular technologies and in utilization rates.
Study Design: We constructed a self-administered survey
instrument that combined previously validated survey
components for general innovativeness (i.e. 6 questions about
patients' attitudes toward new ideas and ways of living) and
content-specific innovativeness (i.e. 5 questions regarding
patients' attitudes toward medical technology). We also
presented a brief vignette describing a hypothetical
implantable medical device, and we subsequently asked a
series of 8 questions about patients' reactions to the
technology. All questions were in Likert scale format.
Population Studied: The survey was administered to a
convenience sample of 81 primary care patients (50 black, 31
white) at the Philadelphia Veterans Affairs Medical Center.
Principal Findings: There were no significant differences in
age (mean 55, range 32-85), educational attainment, and prior
exposure to medical technology between white and black
respondents, although whites reported higher incomes.
Cronbach's alpha was 0.62 for the six general innovativeness
questions, 0.59 for the five content-specific innovativeness
questions, and 0.72 for the eight vignette questions. The
overall Cronbach's alpha was 0.82 for the entire instrument,
indicating high overall consistency. With responses scaled
from 0 (resistant to innovation) to 4 (highly innovative), there
was no racial difference in general innovativeness (blacks: 2.3,
whites: 2.3, p=0.86). However, whites had greater
innovativeness in regard to medical technology (blacks: 2.2,
whites: 2.6, p=0.01). We found no racial differences in
attitudes toward use of the hypothetical medical device
presented in the vignette (blacks: 2.5, whites: 2.4, p=0.39). The
response of black patients to the specific technology in the
vignette was significantly more innovative (p=0.04) than their
response to medical technology in general.
Conclusions: There were no measurable differences between
white and black veterans in their attitudes toward new ideas or
ways of thinking, yet white veterans were more favorably
disposed toward new medical technology. The lower level of
healthcare-technology-specific innovativeness among blacks
surprisingly did not result in a more negative response by
blacks to a hypothetical example of a new implantable medical
device.
Implications for Policy, Delivery, or Practice: Blacks and
whites respond differently to innovative medical technology.
The lack of a racial difference in the response to the vignette
might have been due to the manner in which the technology
was presented, which may have allayed concerns and
encouraged greater innovativeness among black respondents.
Primary Funding Source: VA
●Primary Care for Low-Income Populations: A Comparison
of Health-Care Delivery Systems
Ellie Grossman, M.D., Anna T.R. Legedza, Sc.D., Christina C.
Wee, M.D., MPH
Presented By: Ellie Grossman, M.D., Fellow, Div of General
Medicine and Primary Care, Beth Israel Deaconess Medical
Center, 330 Brookline Avenue, Rose 105, Boston, MA 02215;
Tel: (617) 667-1960; Fax: (617) 667-2751; Email:
egrossma@hsph.harvard.edu
Research Objective: In an effort to cut health-care costs,
some states have considered shifting primary care for lowincome populations from hospital-based practices to
community health centers (CHCs). To explore ramifications of
these proposals, we examined quality measures at different
types of primary-care sites.
Study Design: This study is a secondary analysis of the 20002002 National Health Interview Survey (NHIS). We described
low-income patients (pts) who receive primary care at hospital
clinics, physician (MD) offices or HMOs, or CHCs, and
compared 1) emergency-room (ER) visits; 2) delays in medical
care due to office/clinic constraints (appointment availability,
office wait time, and accessibility of scheduling staff), and 3)
influenza and pneumococcal vaccinations. Multivariable
logistic regression models adjusted for age, sex,
race/ethnicity, education, language, region, employment,
insurance, transportation accessibility, and illness burden
(self-reported health status, hospitalizations and office visits,
comorbidities and disabilities, mood disturbances and need
for mental health care). All analyses were performed using
SUDAAN and were appropriately weighted to account for the
complex sampling.
Population Studied: Adult NHIS participants with family
income less than 200% of the poverty threshold (n = 38,329).
Principal Findings: More than half of our sample (59%)
received primary care at an MD office or HMO, while 20%
went to a CHC and 3% to a hospital clinic; 18% did not
identify any of these three sites as a source of primary care.
There were significant demographic differences between the
pts at the three site types (p<.001 for all comparisons). The
mean pt age was 49 years (y) at MD offices / HMOs, 48y at
hospital clinics, and 41y at CHCs; 31% of MD office / HMO
pts were racial minorities, vs. 61% of hospital-clinic pts and
48% of CHC pts. Only 13% of MD office / HMO pts were
uninsured - compared to 29% of hospital-clinic pts and 33% of
CHC pts. Hospital-clinic pts were more likely to have a higher
illness burden. We found significant differences across sites
for our quality measures of interest – even after adjustment
for sociodemographic and health-status factors. Hospitalclinic pts were more likely to report > 1 ER visit in the past year
than CHC pts (OR 1.8; 95% CI 1.4, 2.3) or MD office / HMO
pts (OR 1.9; 95% CI 1.5, 2.4). However, hospital-clinic pts were
no more likely to report a delay in care than pts at CHCs (OR
1.2; 95% CI 0.9, 1.6). Hospital-clinic pts were more likely than
CHC pts to be vaccinated for influenza (OR 1.3; 95% CI 1.1,
1.7) or pneumococcus (OR 1.4; 95% CI 1.1, 1.8); vaccination
rates at MD offices / HMOs were similar to those at CHCs.
Conclusions: Our results suggest that no one type of primary
care site delivers clearly higher quality care for low-income
populations. Further studies will need to examine additional
quality indicators.
Implications for Policy, Delivery, or Practice: Quality issues
should be considered as we decide how best to deliver care to
vulnerable populations. Primary-care sites which may be
perceived as ‘low-cost’ may actually deliver mixed-quality care.
Primary Funding Source: HRSA
●Racial Differences in Hospice Use Among Non-Hispanic
Blacks and Whites
Beth Han, M.D., Ph.D., MPH, Robin Remsburg, Ph.D.
Presented By: Beth Han, M.D., Ph.D., MPH, Health Scientist,
Centers for Disease Control and Prevention, National Center
for Health Statistics, 3311 Toledo Road, Hyattsville, MD 20782;
Tel: (301)458-4137; Fax: (301)458-4032; Email: hih9@cdc.gov
Research Objective: Despite many benefits, hospice appears
to be underutilized. We examined differences in hospice use
rates among younger and older non-Hispanic Blacks and
Whites between 1991 and 2000 and investigated racial
differences in hospice patient characteristics. We tested
differences in length of hospice survival between hospice
enrollment and death in hospice among non-Hispanic Blacks
and Whites.
Study Design: Series of cross-sectional studies using the
1991-2000 annual Underlying and Multiple Cause-of-Death
Files and the 1992-2000 National Home and Hospice Care
Surveys. Chi-Square tests, linear regression methods, the
Kaplan-Meier method, and Cox proportional hazards models
were used.
Population Studied: We examined data from the 1991-2000
annual Underlying and Multiple Cause-of-Death Files,
providing cause-of-death information reported by all 50 states
and the District of Columbia from 1991 to 2000. We also
investigated 19,547 hospice patients from the 1992, 1994,
1996, 1998, and 2000 National Home and Hospice Care
Surveys, which provide nationally representative data on U.S.
home and hospice care agencies and their patients.
Principal Findings: Hospice use rates doubled for nonHispanic Whites and tripled for non-Hispanic Blacks between
1991 and 2000. In 1999/2000, differences in hospice use
rates between younger non-Hispanic Whites and Blacks
diminished, whereas differences between elderly non-Hispanic
Whites and Blacks increased. Non-Hispanic Black hospice
patients were more likely to be younger; have Medicaid-only;
have HIV/AIDS; and to be referred by hospitals than their
White counterparts. In 1999/2000, 25% of non-Hispanic
Whites and 25% of non-Hispanic Blacks died within 7 days
after hospice enrollment. After adjustment for covariates,
non-Hispanic Black and White hospice patients had similar
length of hospice survival between 1991 and 2000.
Conclusions: Hospice use rate for elderly non-Hispanic
Blacks was at least 9% lower than for elderly non-Hispanic
Whites in 1999/2000. Despite differences in hospice patient
characteristics, non-Hispanic Blacks and Whites had similar
length of hospice survival during the 1990s. Many patients
who entered hospice care did so late in the trajectory of their
terminal illnesses.
Implications for Policy, Delivery, or Practice: Future
research is needed to understand why hospice use rate for
elderly non-Hispanic Blacks was at least 9% lower than for
elderly non-Hispanic Whites in 1999/2000. Hospice agencies
should understand characteristic differences between nonHispanic Black and White hospice patients in order to address
their needs more sensitively and effectively. Strategies are
needed to improve length of hospice survival among both
non-Hispanic White and non-Hispanic Black hospice patients.
Primary Funding Source: CDC
●Racial and Ethnic Disparities in Access to Vision Care
among Children with Special Health Care Needs in the
United States
Kevin Heslin, Ph.D., Richard S. Baker, M.D., Magda Shaheen,
M.D., Ph.D., Richard Casey, M.D.
Presented By: Kevin Heslin, Ph.D., Assistant Professor,
Research Centers in Minority Institutions, Charles R. Drew
University of Medicine and Science, 1731 East 120th Street, Los
Angeles, CA 90059; Tel: (310)761-4726; Fax: (310)631-1495;
Email: keheslin@cdrewu.edu
Research Objective: All children should receive periodic
vision screening and, if indicated, vision care services to
address impaired functioning. The purpose of this study was
to examine racial and ethnic disparities in access to vision
care using secondary survey data on a representative sample
of children with special health care needs (CSHCN) in the
United States.
Study Design: The National Survey of Children with Special
Health Care Needs was a cross-sectional survey conducted
from October 2000 to April 2002. A random-digit-dial sample
of households with children under age 18 was selected from
the 50 states and the District of Columbia. Respondents were
parents or guardians who knew the most about the index
child’s health. Unmet need for vision care was assessed by
asking, “During the past 12 months, was there any time your
child needed eyeglasses or vision care?” and then: “Did your
child receive all the eyeglasses or vision care that he or she
needed?” If the respondent answered “yes” to the first
question and “no” to the second, the child was categorized as
having unmet need for vision care. We used multivariate
logistic regression to estimate the independent association of
race and ethnicity with unmet need. Control variables included
child gender, age, disability status, insurance coverage, illness
severity, mother’s educational achievement, cultural
competence of the child's primary care physician (as
perceived by the respondent), and a count variable of unmet
needs for other types of health services that ranged from 0 to
13. Odds ratios (ORs) and 95% confidence intervals (CIs) were
calculated. The Stata software program was used to estimate
population prevalence parameters and, in the multivariate
analysis, to adjust standard errors for the complex survey
sampling design.
Population Studied: 14,070 children with special health care
needs.
Principal Findings: Of the estimated 3,321,870 CSHCN
represented in this analysis, 6% had unmet need for vision
care. Approximately 5% of whites had unmet need, compared
with 9% of African Americans, 10% of Latinos, 14% of
multiracial children, and 8% of children of “other” racial and
ethnic backgrounds (p<0.0001). Multivariate analysis showed
that, compared with whites, African Americans had twice the
odds (OR = 2.01; 95% CI = 1.37, 2.96) and children of
multiracial backgrounds had three times the odds (OR = 3.06;
95% CI = 1.62, 5.75) of having unmet need for vision care.
Latino children and those of “other” racial and ethnic
backgrounds did not differ from whites in the multivariate
analysis.
Conclusions: Outreach programs should target CSHCN in
underserved racial and ethnic minority communities to reduce
disparities in access to vision care.
Primary Funding Source: AHRQ
●Rural/Urban Differences in Access to Health Care: Does
Universal Coverage Reduce Inequalities in Health?
Nicole Huang, Ph.D., Winnie Yip, Ph.D., Hong-Jen Chang,
M.D., MS, MPH, Yiing-Jenq Chou, M.D., Ph.D.
Presented By: Nicole Huang, Ph.D., Assistant professor,
Department of Health Education, National Taiwan Normal
University, 162 Heping East Road, Section 1, Taipei, 106; Tel:
886-2-23636880 x220; Fax: 886-2-23630326; Email:
nhuang@cc.ntnu.edu.tw
Research Objective: Rural-urban disparities in access to
health care have always been a focus of concern.
Considerable literature has documented that rural residents
are at especially high risk of inadequate and delayed access to
health care due to both demand-side and supply-side factors.
Some speculate that introduction of a single-payer health-care
system, which has monospony power to take care of both
demand and supply issues, may substantially reduce ruralurban health disparities. However, whether a single-payer
universal coverage eliminates disparities in access to care
among rural population remains unknown. The objective of
this paper is to compare the changes in rates of ruptured
appendicitis between rural and urban residents in Taiwan
from 1996 to 2001. As ethnic disparity in access to care under
the NHI program is also a focus of rural health policy, we also
conducted additional analyses on disparities between
aborigine and non-aborigine populations.
Study Design: Ruptured appendicitis was used as an
indicator of access-induced health outcome. The NHI
hospital discharge ICD-9-CM codes were used to define
hospitalization for ruptured (ICD-9-CM code 540.0 and 540.1)
and non-ruptured appendicitis (ICD-9-CM codes 540.9). Of
309 townships in Taiwan, 30 mountainous areas are
proclaimed “rural areas.” The patients living in these 30
mountainous towns were identified as rural residents while
the patients living in other areas were identified as urban
residents. Other variables such as age, gender, aborigine
status, socioeconomic status, health status, and hospital
characteristics were also included in the analyses.
Population Studied: The NHI enrollment files, hospital
discharge data, and the household registry were three main
data sources. We identified all 128,930 patients undergoing
appendectomy for which either a normal or perforated
appendix was the principal diagnosis in Taiwan between 1996
and 2001.
Principal Findings: During the first 3 years under the NHI
program, although the differences between the rural and
urban areas were apparent, the differences were narrowing.
As the downward trends continued, in looking at the trends
since 1999, we found very few discernible differences between
the rural and urban areas. After adjusting for demographic,
SES, health status, and hospital characteristics, the ruptured
rate among rural residents was decreasing significantly faster
(1.1%) than that among urban residents over time. More
specifically, the children showed a significantly steeper
narrowing trend (3.4%) in the rural-urban disparities under the
NHI program than adults. Furthermore, unlike the significant
gap narrowing observed between rural and urban areas, the
pace of reduction in ruptured rate among aborigine people
was only 0.5% faster than that among non-aborigine people
and the difference was not significant.
Conclusions: Our findings demonstrate the significant
narrowing of health disparities between rural and urban
populations since the implementation of the NHI program.
In particular, of all age groups, the most dramatic narrowing
of the gap was observed among children after adjusting for a
variety of patient and hospital characteristics.
Implications for Policy, Delivery, or Practice: In conclusion,
while many countries are discussing the option of single-payer
universal coverage system, our findings add new knowledge in
that effectiveness of single-payer universal insurance program
ameliorates rural-urban disparities in health and provide an
important reference to their discussion.
Primary Funding Source: Other, National Health Research
Institute and National Science Council
●Reducing Disparities in Healthcare Access: The CostEffectiveness of Non-Emergency Medical Transportation
Paul Hughes-Cromwick, BS, MA, Ph.D. Candidate, Richard R.
Wallace, BS, MS, Ph.D. Candidate, Hillary J. Mull, BA, MPP
Presented By: Paul Hughes-Cromwick, BS, MA, Ph.D.
Candidate, Senior Analyst, Health Solutions Division, Altarum
Institute, 3520 Green Court, Ann Arbor, MI 48105; Tel:
(734)302-4616; Fax: (734)302-4994; Email: paul.hughescromwick@altarum.org
Research Objective: We hypothesize that that improving
access to healthcare for the transportation-disadvantaged
population will lead to improved quality of life, potential
enhancements in life expectancy, and a decrease in health
costs nationally. This decrease may exceed the incremental
increase in transportation costs.
Study Design: Millions of Americans are considered to be
“transportation disadvantaged.” Members of this population,
due to low income, physical or mental disability, inability to
drive, or some other reason, cannot transport themselves or
are unable to purchase transportation services. This
dependency reduces access to essential healthcare services.
In some cases this access disparity leads to lost opportunities
for early detection or decreases in health status. The
Transportation Research Board of the National Academies of
Science launched the project, “Cost Benefit Analysis of
Providing Non-Emergency Medical Transportation.” Arising
from this project, the current study estimates the size of the
transportation-disadvantaged population that misses care due
to a lack of non-emergency medical transportation (NEMT),
and analyzes the cost-effectiveness of providing the missed
transportation in terms of improved health status (measured
as quality adjusted life years, or QALYs) relative to increased
transportation costs. Cost-effectiveness is demonstrated two
different ways: (1) at a macroscopic level and (2) at a
microscopic level for two critical conditions that affect the
transportation disadvantaged: asthma and heart disease.
Population Studied: This is a nationally representative study
of those we estimate to be transportation-disadvantaged
relative to accessing healthcare. We identify relevant
individuals using multiple years of the National Health
Interview Survey and the Medical Expenditure Panel Survey.
Principal Findings: Our results indicate that at least 3.6
million Americans miss medical care due to a lack of
transportation in a given year. They are disproportionately
female, poorer, and older; have less education; and are more
likely to be minorities than those who do not miss care due to
a lack of NEMT. While adults who miss care are spread
across urban and rural areas like the general population,
children lacking transportation are more concentrated in
urban areas. These 3.6 million suffer from co-morbidities at a
much higher rate than do their peers. Many of the conditions
they face, however, can be managed if appropriate care is
made available. Based on our macroscopic cost-effectiveness
analysis, we found that only small health benefits (about
1/50th of a QALY per person) are needed to balance the added
costs; furthermore, at the microscopic level, we found that
providing missing NEMT to asthma and heart disease
patients is highly cost effective.
Conclusions: Based on our findings, we see great promise for
net societal benefits from improving the quality of life of this
population by increasing its access to transportation.
Implications for Policy, Delivery, or Practice: We
recommend modifications to national healthcare and
transportation data sets to allow more direct assessment of
the magnitude of this problem and of the benefits that could
be achieved by remedying it. We believe that transportation
adds a novel dimension to health services research on
healthcare access and disparities.
Primary Funding Source: Transportation Research Board of
the National Academies of Science
●Quality of Hospital Care among Hispanics: Does primary
language make a difference?
Margarita Hurtado, Ph.D., MHS, January Angeles, MPP
Presented By: Margarita Hurtado, Ph.D., MHS, Principal
Research Scientist, American Institutes for Research, 10720
Columbia Pike, Silver Spring, MD 20901; Tel: (301) 592-2215;
Fax: (301) 593-9433; Email: mhurtado@air.org
Research Objective: To determine whether the quality of
hospital care reported by Hispanic patients differ according to
the primary language they speak, with an emphasis on
communication-related aspects of service delivery.
Study Design: We derived patients’ primary language using
two variables: (1) primary language spoken at home; and (2)
language in which the questionnaire was administered. When
language indicated for both variables was the same, that
language was chosen as the respondent’s primary language.
In cases where the language indicated differed, individuals
were designated as bilingual. We conducted regression
analyses of responses to the HCAHPS survey on quality of
hospital care controlling for potential confounders, including
age, sex, education, overall health status and admission status
(medical, surgical, and obstetrics). We also accounted for
correlated error by including the hospital as a random effect in
mixed effects models.
Population Studied: The study sample includes 1,806
Hispanics who had an inpatient hospital stay. Data were
collected in three states – Arizona, Maryland, and New York –
through the HCAHPS pilot test. Of the entire sample, 80%
were female, almost half (48%) were between the ages of 18 to
34, one-third (29%) did not graduated from high school, and
only 14% were college graduates. Forty-four percent were
identified as primarily English-speaking, 32% primarily spoke
Spanish, and 23% were bilingual.
Principal Findings: Quality of hospital care reported by
Hispanic patients did not differ significantly by language
group for any of the HCAHPS composite measures
(communication with nurses, communication with doctors,
communication about medication, nursing services, discharge
information, pain control and physical environment). Global
ratings of nurses, physicians, and the hospital also did not
show any significant differences. Analyses of individual item
analysis, however, revealed significant differences between
language groups for an important component of physicianpatient communication: providing explanations that are easy
to understand. Those who spoke Spanish as their primary
language were significantly less likely than those who primarily
spoke English to report that doctors explained things in a way
they could understand (p=0.02). Both Spanish-speaking and
bilingual Hispanics were less likely than English-speaking
Hispanics toi report having nurses explain things clearly to
them, but the difference was only significant for bilingual
Hispanics (p=0.01). There were no significant differences with
respect to having doctors or nurses listen carefully to them.
Conclusions: This study suggests that those who do not
speak English as their primary language experience greater
difficulty communicating with clinicians than their Englishspeaking counterparts. With respect to other aspects of care,
Hispanic patients report similar experiences with the quality of
their hospital care regardless of the language they
predominantly speak.
Implications for Policy, Delivery, or Practice: This study
points out an important component of care that hospitals
need to examine and improve in order to provide better care
to non-English speaking patients. Further work is needed to
define the most effective ways of doing so.
Primary Funding Source: AHRQ
●Health Related Quality of Life and Direct Medical Care
Cost of Prostate Cancer Patients
Ravishankar Jayadevappa, Ph.D., Bernard S. Bloom, Ph.D.,
Sumedha Chatre, Ph.D., Kenneth M. Fomberstein, S. Bruce
Malkowicz, M.D.
Presented By: Ravishankar Jayadevappa, Ph.D., Research
Assistant Professor, Department of Medicine, University of
Pennsylvania, Room # 224, 3615 Chestnut Street, Philadelphia,
PA 19104; Tel: (215)898-3798; Fax: (215)573-8684; Email:
jravi@mail.med.upenn.edu
Research Objective: Introduction and Objective: Multiple
factors (demographic, clinical, social, economic and
environmental) influence the medical care cost and health
related quality of life (HRQoL) of patients with prostate cancer
(PCa). It is important to assess the factors associated with
variations in health resource utilization, costs and their
relationship to outcomes for effective management and policy
decision. We evaluated generic and prostate-specific HRQoL
and direct medical care cost for men with PCa.
Study Design: Methods and Statistical Analysis:
We conducted a prospective cohort study of newly diagnosed
PCa patients. Additionally, we recruited matched controls
(cancer free) from same hospitals. All cases completed SF-36
and UCLA-PCI surveys prior to treatment, and at 3, 6, 12 and
24 months after surgery. Controls completed surveys at
baseline only. Direct medical care cost (DMC) data were
obtained from a hospital based administrative database and
clinical data were obtained via structured medical chart review.
Demographics and HRQoL were compared using T-test,
Fisher Exact and Chi-sq. Wilcoxon and log-T tests were used to
compare DMC. Multivariate regression models were used to
assess incremental cost of PCa and the predictors of 24
months PCa specific HRQoL.
Population Studied: Newly diagnosed PCa patients, recruited
from the Urology clinics of a large academic health care
system and VA hospital.
Principal Findings: Results: We recruited total of 521 PCa
patients and achieved a retention rate of 85%. PCa patients
showed significant variations in baseline characteristics,
treatment pattern and HRQoL by hospitals, age and ethnicity.
For PCa patients who had 24 months follow-up, mean annual
medical care cost was $4,160 for the treatment year and mean
length of stay was 3.5 days. Baseline physical function for
cases was 9 points higher than controls (72.6 vs.61: p =.0107)
and score on bodily pain was higher, indicating lower pain
(91.5 vs. 76.9 p=.0054). For baseline prostate specific HRQoL
(UCLA- PCI), cases had better functions on sexual function
(71.5 vs 48.7 p=.0009), bowel bother (96.9 vs 86 p=.0160) and
sexual bother (89.1 vs. 69.4 p=.0160). At 12 months post
treatment, cases had generic HRQoL similar to their baseline
values and those of controls for all domains. Urinary function
(80.1 vs.93.9: p .0019), sexual function (39.4 vs. 71.5:
p<.0001), urinary bother (83.1 vs. 94.1: p.0319) and sexual
bother (35.5 vs. 89.1: p<.0001) were lower at 12 months for
cases than baseline values, while rest of the PCI domains were
comparable. At 24 months of post treatment, generic HRQoL
of cases returned to their baseline values. Urinary function
(77.2 vs. 93.9: p=.0014), sexual function (42.7 vs. 71.5:
p=<.0001), and sexual bother (41.1 vs. 89.1: p<.0001) were
lower and rest of the PCI domains were comparable. In a
multivariate log-linear regression, marital status was a
significant predictor of sexual, urinary and bowel functions at
24 months.
Conclusions: Conclusions: PCa patients reported weaker
sexual function, urinary function and sexual bother at 24
months post treatment compared to their baseline values.
Baseline characteristics, treatment pattern and HRQoL vary
across age, ethnicity and hospital settings.
Implications for Policy, Delivery, or Practice: There exists a
significant opportunity for improving HRQoL of men with PCa
that requires multisystem approach.
Primary Funding Source: Department of Defense
●What Happens Ehen You Leave? Outcomes of Patients
who Leave Hospital Against Medical Advice
Yan Jin, MA, Donald Schopflocher, Ph.D.
Presented By: Yan Jin, MA, Project Consultant, Alberta Health
and Wellness, 22nd Floor, TPNT, 10025 Jasper Avenue,
Edmonton, T5J 2N3; Tel: (780)415-0210; Fax: (780)422-2880;
Email: yan.jin@health.gov.ab.ca
Research Objective: Although many studies have been
conducted on patients leaving hospital against medical advice
(AMA), few examined mortality as an outcome of AMA
discharge. The purposes of our retrospective study used a
large-scale dataset to identify patient’s characteristics
associated with AMA discharge and compare not only
readmission but also mortality rates between patients with
AMA and formal discharges.
Study Design: The data used in this study are from the
Canadian Institute for Health Information’s (CIHI) Inpatient
Discharge Abstract Database (DAD) for Alberta and from the
Alberta Health Care Insurance Plan (AHCIP) Registry File. A
large range of variables related to patients’ characteristics,
socio-demographics, Charlson comorbidity index, substance
abuse, and admission category were included. Data linkage
was conducted to obtain death status and the defined
variables. We applied logistic regression and Cox proportional
hazard models in the analysis.
Population Studied: We defined the study cohort as index
discharges or the first discharges in 2000/01 from all Alberta
acute hospitals for Alberta residents over age 18 who left
hospital alive.
Principal Findings: Of the 343,309 hospital discharges,
171,043 met our selection criteria. Of these, 169,163 were
formal discharges with: an average length of stay (ALOS) of
6.2 days, mean age of 50 and 63.7% female. In contrast, there
were 1,880 AMA discharges with: an ALOS of 4.6 days, mean
age of 42 and 44.9% female. The odds ratios for AMA
discharge were 1.25 for patients younger by 10 years, 1.58 for
males, 2.59 for patients on welfare, 4.05 for First Nation
patients, 1.65 for patients living alone, 2.59 for patients with
elective admission via emergency department or ambulance,
5.45 for patients with urgent/emergency admission, 3.36 for
patients with drug related diagnosis, and 2.96 for patients
with alcohol related diagnosis. Within one year of index
discharge, AMA discharged patients had 49% higher risk of
being readmitted to hospital and 26% greater risk of death,
compared to patients formally discharged.
Conclusions: Leaving hospital AMA is associated with: (1)
being younger, male, First Nations or on welfare; (2) living
alone; (3) having substance abuse problems; and, (4) having
an urgent/emergency admission. Patients leaving hospital
AMA are at greater short-term risk of adverse outcomes
including readmission and death.
Implications for Policy, Delivery, or Practice: Aappropriate
intervention strategies for patients leaving hospital AMA could
improve spending in the health care system and the health
status of individuals.
Primary Funding Source: No Funding Source
●Ethnicity and Other Social Determinants in Pap Testing
Among Canadian Immigrant Women
Arminee Kazanjian, Dr.Sc., Alice Chen, MC, MS, Arminee
Kazanjian, Dr.Sc.
Presented By: Arminee Kazanjian, Dr.Sc., Professor, Health
Care & Epidemiology, University of British Columbia, 5804
Fairview Avenue, Vancouver, BC, V6T 1Z3; Tel: (604)8224618; Fax: (604)822-4994; Email: a.kazanjian@ubc.ca
Research Objective: Many studies reported that ethnic
minorities and immigrants in North America are less likely to
receive Papanicolaou (Pap) smears, a recommended
screening test for cervical cancer. At the same time, there is
also evidence that some of these groups experience higher
rates of cervical cancer incidence and mortality. The underutilization of Pap screening by immigrants and minority
groups is of particular concern in the province of British
Columbia (BC), Canada where 26% of its population in 2001
were immigrants. Moreover, more than one-third of these
immigrants arrived within the previous ten years and the
predominant majority of them are of non-European origins.
The objectives of the present study are:
(1) To compare the rates of Pap testing among recent
immigrants of different ethnic origins in BC
(2) To examine the socio-demographic factors that affect the
probability of Pap testing among immigrant women
Study Design: This study is a retrospective analysis of
administrative databases. A database of immigrants who
landed in Canada in 1985-2000 and destined for BC was linked
to the province’s administrative health databases as part of a
national immigrant health project. The control group was
randomly selected from the general population of BC and
individually matched to the immigrants by sex, age and health
region. Receipt of Pap testing was estimated from the
physicians’ claims for service database for the years 19992001. Socio-demographic factors examined were marital
status, educational level and the ability to communicate in
English.
Population Studied: The study population consists of women
between the ages of 18 and 69 and enrolled in BC’s health
plan for any time in 1999-2001. Six subgroups of immigrants
were selected for this study based on their place of birth: those
from China, South Asia, Philippines, Western Europe, South
and Central America and the USA. A total of over 100,000
immigrant women were included.
Principal Findings: All subgroups of immigrants were less
likely than the control group to receive Pap testing, with an
overall odds ratio of 0.46. Ethnic differences were significant,
with South Asian, Chinese and Filipino women showing the
lowest probability of receiving Pap tests. All the sociodemographic factors studied were found to have an effect on
the probability of Pap testing among immigrant women.
Conclusions: Immigrant women are less likely to receive Pap
testing than the non-immigrant population. The difference is
especially profound for ethnic minority immigrant women.
Marital status, education and English ability all contribute to
the probability of Pap testing in immigrant women.
Implications for Policy, Delivery, or Practice: Given the risk
for advanced stages of cervical cancer when screening services
are not received, targeted efforts are necessary to increase the
utilization of Pap testing by Asian immigrant women.
Primary Funding Source: Canadian Population Health
Initiative
●Cancer Risk Perception and Preventive Screening by
Women of Diverse Populations
Sue Kim, Ph.D., MPH, Sabrina Wong, Ph.D., RN, Celia Kaplan,
Dr.PH, Judith Walsh, M.D., MPH, George Sawaya, M.D.,
Eliseo Pérez-Stable, M.D.
Presented By: Sue Kim, Ph.D., MPH, Postdoctoral Scholar,
Medicine, University of California, San Francisco, 3333
California Street, Suite 335, San Francisco, CA 94143; Tel:
(415)502-4078; Fax: (415)502-8291; Email:
sekim@medicine.ucsf.edu
Research Objective: Inaccurate perceptions of risk
compromise informed decision-making and may have
important behavioral and psychological consequences. Few
studies have addressed if ethnic differences in cancer risk
perception exist. This study examines the perception of
breast, cervical, and colorectal cancer risk and compare the
actual risk and screening behavior of a sample of women from
various ethnic groups.
Study Design: A sample of women, aged 50 to 80, who have
made a visit to a primary care physician in the past two years
were recruited from ambulatory practices. Trained interviewers
administered an initial screening telephone survey and a
follow-up in-person interview in English, Spanish or Chinese.
Women’s risk perception was examined using different
response formats (e.g. numerical scales, word scales, and
assessment of cancer risk relative to other risks). Multivariate
regression models examined ethnic differences in risk
perception and screening, controlling for education, age,
income, previous cancer history, and family history of cancer.
Population Studied: 536 women completed the survey: 282
non-Latino White (52.6%), 72 African-American (13.4%), 81
Latina (15.1%), and 101 Asians (18.1%).
Principal Findings: Latina perceived themselves to have the
highest risk for getting cancer (estimating their lifetime risk to
be 37.1%), compared to African-American (29.8%), White
(21.0%), and Asian (22.0%) (P<.01). Multivariate results
showed that compared to White women, Latina women
perceived higher risk of cervical (OR, 2.08; 95% CI, 1.14-4.15)
and colorectal cancer (OR, 2.49; 95% CI, 1.36-4.56). African
American perceived higher colorectal cancer risk (OR, 2.20;
95% CI, 1.21-3.98) and Asian women perceived lower breast
cancer risk compared to White women (OR, 0.48; 95%CI 0.260.89). Having a college degree was related to a lower cervical
cancer risk perception (OR, 0.45; 95% CI, 0.26-0.89) and
family history of cancer was related to higher breast (OR, 2.56;
95% CI, 1.70-3.83) and colorectal cancer risk perception (OR,
1.91; 95% CI, 1.27-2.88). In regards to cancer screening, there
were no differences in cervical and breast cancer screening
rates. Higher percentages of Latina (15.00%) and Asian
(17.82%) women never had any colorectal cancer screening
compared to White women (11.39%) (P<.01).
Conclusions: All women did not have an accurate
understanding of their cancer risk. Yet, ethnic differences
persisted in cancer risk perception even after controlling for
education, age, income, and cancer history (self and family).
Having a higher risk perception was not directly related to
higher screening rates. We are currently collecting more data
to run detailed analyses comparing the four ethnic groups.
Implications for Policy, Delivery, or Practice: It’s important
to consider ethnic differences when communicating with
patients about cancer risk. It appears that there is a difference
in cancer risk perception among women and may reflect their
screening behavior. When clinicians communicate with
patients about risk information, the traditional percentage and
numeracy information may not be as effective in all diverse
populations.
Primary Funding Source: AHRQ
●The Role of Medical Education in Reducing Health Care
Disparities: the Drew/UCLA Medical Education Program
Michelle Ko, M.D., Ronald Edelstein, EdD, Kevin Heslin,
Ph.D., LuAnn Wilkerson, EdD, Lois Colburn, Kevin Grumbach,
M.D.
Presented By: Michelle Ko, M.D., Internal Medicine Resident,
Internal Medicine, UCLA Center for Health Sciences, 16-155
CHS, 10833 LeConte Avenue., Los Angeles, CA 90095; Tel:
(310)825-8307; Email: jassmine@ucla.edu
Research Objective: The Drew/UCLA Medical Education
Program selects students for their demonstrated commitment
to underserved areas. Students receive basic science
instruction at UCLA and complete their required clinical
rotations in South Los Angeles, an impoverished urban
community. We examined the association between graduation
from Drew/UCLA and subsequent practice in a medically
underserved area, incorporating the potential confounding
predictors of student demographics, socioeconomic status
markers, and specialty choice.
Study Design: A retrospective cohort study.
Population Studied: Graduates of the UCLA School of
Medicine and the Drew/UCLA Medical Education Program
from 1985-1995, in active practice, with a self-reported mailing
address in California from the American Medical Association
(AMA) Physician Masterfile.
Principal Findings: Twenty-nine percent of all graduates
studied are practicing in medically underserved areas. Fiftythree percent of Drew/UCLA graduates are practicing in
underserved areas, in comparison to 26.1% of UCLA
graduates. In multivariate analyses, only underrepresented
minority race/ethnicity (OR: 1.572; 95% CI: 1.096-2.255) and
participation in the program (OR: 2.466; 95% CI: 1.586-3.832)
were predictive of practice in underserved areas.
Conclusions: Physicians who participated in the Drew/UCLA
Medical Education Program are more likely to practice in
underserved areas than those who completed a traditional
medical school curriculum. This may be attributed to the
selection of committed students, as confirmed by the
association of race/ethnicity and practice. Our results also
indicate an independent program effect, perhaps by providing
a supportive learning environment in a medically underserved
urban community.
Implications for Policy, Delivery, or Practice: Investment in
the development of medical education programs targeting
underserved areas can help to alleviate disparities in health
care delivery.
Primary Funding Source: National Center on Minority Health
and Health Disparities
●A Provider-Friendly Screening Tool to Identify Patients
Who Lack Physician Trust
David Kuykendall, Ph.D., MBA, Michael Kallen, Ph.D., MPH
Presented By: David Kuykendall, Ph.D., MBA, Director,
METRIC, Department of Veterans Affairs, VAMC (152), 2002
Holcombe Boulevard, Houston, TX 77030; Tel: (713) 794-8514;
Email: david.kuykendall@med.va.gov
Research Objective: To develop an easy-to-use screening tool
for the identification of primary care patients who enter a
medical encounter with low physician trust.
Study Design: Consecutive patients waiting for appointments
at outpatient clinics were asked to provide demographic
information and then to complete a battery of scales about
their health and health care. Average scores on the 11-item
Trust in Physician Scale were calculated and used to
categorize patients as high trust, i.e., average responses
favorable, or lacking trust, i.e., average responses neutral or
unfavorable. To the extent that initial levels of physician trust
are based on life experiences, groups with different life
experiences should exhibit different levels of trust. Modeled
proxies for patients’ previous life experiences included gender,
income, age, ethnicity, education, and pain. The study tested
whether the number of these predisposing factors, each of
which the literature has shown to be associated with low
physician trust, could explain physician trust as well as a
complex, multivariable model that included predisposing
factors as main effects and significant interactions involving
ethnicity.
Population Studied: Participants came from one private, one
public, and one VA clinic. One hundred fifteen African
American and 148 White American patients participated: 44%
were females; 54% had incomes <= $20,000 per year; 37%
completed high school or less; 41% experienced moderate to
severe pain during the previous 4 weeks.
Principal Findings: Logistic regression identified patient
characteristics associated with entering the clinic having low
physician trust. Similar accuracies were achieved using
number of predisposing factors present as the sole predictive
variable and the more complex multivariable model that
included multiple main effects and a significant interaction:
sensitivity=.474 vs .368; specificity=.855 vs .932; positive
predictive value=.514 vs .636. The percentage of patients who
lacked physician trust escalated as the number of
predisposing factors per patient increased. Overall, 25% of
respondents lacked physician trust. With one predisposing
factor, 32% of respondents lacked physician trust; with three
predisposing factors, 50% lacked trust; and with five
predisposing factors, 67% lacked trust.
Conclusions: Identification of patients who lack physician
trust can be achieved by considering the number of
predisposing factors pertaining to a given patient. The
approach is as accurate as the multivariable logistic model
with main and interaction effects, and is simple to implement,
calculate and interpret.
Implications for Policy, Delivery, or Practice: Disparities in
quality of care can result from a lack of physician trust existing
prior to scheduled medical encounters. Patients with a
propensity toward lack of trust possess multiple proxies for
previous life experiences that are known to influence trust, i.e.,
low income, African American, low education, poor health,
men, older. A simple screening tool to identify patients who
lack physician trust, based on readily available measures, can
help providers identify patients who may have special
concerns. Understanding these concerns is a first step toward
addressing root causes of low physician trust and toward
improving delivery processes for vulnerable populations.
Primary Funding Source: AHRQ, Department of Veterans
Affairs
●Help Seeking Among Individuals with Inadequate Health
Literacy
Shoou-Yih Lee, Ph.D., Julie A. Gazmararian, Ph.D., Ahsan M.
Arozullah, M.D., MPH
Presented By: Shoou-Yih Lee, Ph.D., Assistant Professor,
Health Policy and Administration, University of North Carolina
at Chapel Hill, 1101 McGavran-Greenberg Hall (CB# 7411),
Chapel Hill, NC 27510-7411; Tel: (919) 966-7770; Fax: (919)
966-6961; Email: sylee@email.unc.edu
Research Objective: We know very little about how
individuals with inadequate health literacy cope with literacy
problems and the effects of the coping behavior. Thus, the
study examines (1) the relationship between health literacy
and help seeking behavior and (2) how the assistance received
is related to health status and health care utilization.
Study Design: Health literacy was measured using the short
version of the Test of Functional Health Literacy in Adults (STOFHLA). The score ranged from 0-100 and inadequate
health literacy was defined as score < 67. Health status was
measured with self-reported general health status and SF-12
(physical and mental health). Health care utilization included
regular physician visits and hospitalization. Individuals were
asked if they received the following assistance: (1) help with
reading medical information from doctors and hospitals, (2)
being reminded to do things to stay healthy, and (3) tangible
social support (i.e., help when bed-ridden, help with doctor
visits, help with preparing meals, help with daily chores, and
informational help to understand situation). Individual
attributes (age, gender, ethnicity, educational attainment, and
marital status) were included as controls in multivariate
analysis.
Population Studied: The sample consisted of 3,260 Medicare
enrollees, age 65 or older, at four sites of the Prudential
HealthCare Plans (853 in Cleveland, 498 in Houston, 975 in
South Florida, and 934 in Tampa); 2,956 spoke English and
304 spoke Spanish as their native language.
Principal Findings: Compared to those with adequate health
literacy, individuals with inadequate health literacy were more
likely to seek and receive assistance with reading medical
information (20.2% vs. 3.7%) and to be reminded to do things
to stay healthy (46.4% vs. 41.3%); but they perceived less
tangible social support available to them (p<0.0001).
Multivariate analysis showed that among individuals with
inadequate health literacy skills, reading assistance and
healthful reminders were negatively related to physical and
mental health status while tangible social support was
positively related to health status. Receiving healthful
reminders was positively related to routine physician visits.
No other association was found between
information/assistance and health care utilization.
Furthermore, tangible social support appeared to reduce the
negative impact of inadequate health literacy on self-reported
general health status.
Conclusions: Individuals with inadequate health literacy seek
and receive various types of assistance in dealing with health
problems. The assistance they receive has differential impact
on health outcomes and health care utilization. Interestingly,
informational assistance appears to have negative correlations
with health status, possibly due to the quality of information
received. Tangible social support, on the other hand, is
positively associated with health status and appears to reduce
the adverse effect of inadequate health literacy on health
status.
Implications for Policy, Delivery, or Practice: Interventions
to reduce the adverse consequences of inadequate health
literacy focus mostly on providing informational and
communicative support (e.g., simplifying medical
information) and on improving individuals’ ability to use
health information. Tangible support in the individual’s social
environment is largely ignored and should be emphasized in
interventions and clinical practices.
Primary Funding Source: AHRQ
●Determinants of Diabetes Control Among African
Americans and White Veterans
Judith A. Long, M.D., Joshua P. Metlay, M.D., Ph.D.
Presented By: Judith A. Long, M.D., Physician Researcher,
Philadelphia VA CHERP, 1201 Blockley Hall, 423 Guardian
Drive, Philadelphia, PA 19104; Tel: (215)898-4311; Fax: (215)5738778; Email: jalong@mail.med.upenn.edu
Research Objective: Compared to whites, African Americans
(AA) have worse diabetes control, and more complications
from diabetes mellitus (DM). The objective of this study was
to determine if socio-demographic, self-care, or clinical factors
help explain why AAs have worse diabetes control than whites.
Study Design: This project is a clinic-based, case-control
study. Using hospital administrative and clinical records, we
identified veterans living in Philadelphia with DM diagnosed
after age 30 and having a HbA1c drawn at the VA within 3
months of the date of sampling. Cases were defined as
diabetic veterans with a HbA1c = 8% and controls were
diabetic veterans with a HbA1c < 8%. Each participant
completed a telephone survey and had their VA medical chart
reviewed. We performed bivariate and multivariate analyses
to determine predictors of diabetes control and after adjusting
for clustering by provider, sequential logistic regressions to
determine if groups of variables helped explain the association
between diabetes control and race.
Population Studied: Diabetic Veterans
Principal Findings: 433 veterans (67% of those contacted and
eligible) participated in the study (147 cases, 286 controls, 317
AAs, and 116 whites). Among AA, only 61% had good
diabetes control, compared to 81% among whites (p < 0.001).
In unadjusted analyses, the odds for AAs, compared to whites,
of having good diabetes control was 0.35 (95% CI 0.19-0.64).
In the final adjusted model, in addition to race, the following
were associated with good control: older age, an unchanged
or better financial situation compared to the previous year,
having no additional health insurance, better diabetes self-care
as determined by the Diabetes Care Profile Self-Care
Adherence Scale, better self-rated general health, a total
cholesterol under 200 mg/dl, and having had diabetes for
fewer years. After adjusting for these factors, the odds for
AAs, compared to whites, of having good control was 0.34
(95% CI 0.19-0.63).
Conclusions: In these analyses, we show that predictors of
diabetes control include socio-demographic, self-care, and
clinical factors; however, these factors did little to explain the
disparities in control between AA and white diabetic veterans.
Implications for Policy, Delivery, or Practice: Until we better
understand reasons for racial differences in diabetes control
we are unlikely to eliminate this large disparity.
Primary Funding Source: VA
●Processes and Outcomes in the Transition of Care for
Youth with Chronic Childhood Illness
Debra Lotstein, M.D., Moira Inkelas, Ph.D., Ronald Hays,
Ph.D., Neal Halfon, M.D., MPH
Presented By: Debra Lotstein, M.D., Assistant Professor of
Pediatrics, Pediatrics, UCLA, 10833 Le Conte Avenue, 12-358
MDCC, Los Angeles, CA 90095; Tel: (310)825-9346; Email:
dlotstein@mednet.ucla.edu
Research Objective: As children with chronic medical
conditions grow into adulthood, they lose support from childspecific systems, and comparable adult programs do not exist.
There is evidence that young adults with chronic conditions
have difficulty transitioning to adulthood, and that assistance
with the transition process can improve access to health care
and health-related outcomes. The purpose of this study is to
describe the health care quality and outcomes for young
adults with chronic illnesses who have aged out of eligibility
for a child-specific care coordination and payment program,
and to describe the relationship of these outcomes to prior
assistance with the transition process.
Study Design: We conducted a cross-sectional survey of
former clients of a federally-funded case management and
medical care payment program for children with serious
medical conditions. Descriptive statistics were calculated to
describe socio-demographic and medical characteristics of
this sample. To measure transition support, subjects reported
on assistance in four domains of transition (medical care,
insurance, self-care skills and school/job planning). Measured
outcomes included health–related quality of life measures and
education, vocational and social status. Access measures
included interruption in care and insurance coverage since
leaving the program. Also, we estimated bivariate
relationships between childhood factors and both transition
processes and outcomes.
Population Studied: Potential subjects were identified using
archival records from the San Bernardino County, California
Title V program. One hundred former clients of the program,
aged 21-24 (mean age=23 years), without cognitive
impairment, completed the survey.
Principal Findings: Survey respondents had a broad range of
medical conditions, of which 28% were congenital. Prior to
age 21, half had Medicaid and most received care from an
academic setting. Only 31% of respondents had discussed the
need for an adult-oriented physician with their pediatric
providers, and only 25% received assistance in finding an
adult provider. About 25% reported a current unmet need for
medical care. Over 40% have had an interruption in insurance
coverage since turning 21, and 22% were uninsured at the
time of the interview. Almost half were unemployed and 25%
had not completed high school.
Conclusions: Transition assistance for this population of
youth with chronic conditions falls short of national
recommendations. Access to appropriate adult-oriented
medical care is also sub-optimal. These youth report
substantial unmet needs for medical care.
Implications for Policy, Delivery, or Practice: An improved
transition process could decrease the discontinuties seen in
care across the life span. Public agencies, as well as specialty
and primary care providers, should improve transition
planning and assistance to youth with chronic childhood
conditions.
Primary Funding Source: RWJF
●Differences in Health Literacy Among Subgroups of
Medicare Beneficiaries
Lauren McCormack, Ph.D., MSPH, Jennifer Uhrig, Ph.D., Carla
Bann, Ph.D.
Presented By: Lauren McCormack, Ph.D., MSPH, Program
Director, Health Communication, RTI International, 3040
Cornwallis Road, RTP, NC 27709; Tel: (919) 541-6277; Fax:
(919) 990-8454; Email: Lmac@rti.org
Research Objective: To measure health literacy of
beneficiaries enrolled in the Medicare program and to
examine factors that predict health literacy levels.
Study Design: Analysis of the Medicare Current Beneficiary
Survey (MCBS). Working in collaboration with the Centers for
Medicare and Medicaid Services, we developed and tested two
types of survey questions to measure health literacy among
Medicare beneficiaries – health insurance terminology
questions (n= 10) and reading comprehension and numeracy
questions (n= 17), creating an index for each set of questions
(Cronbach’s alpha = .76 and .85, respectively). The reading
comprehension and numeracy questions were based on
excerpts from the Medicare & You 2001 handbook about
various topics including information available on the Medicare
hotline, how to interpret a Medicare claims notice, an
overview of the Medicaid program for dual eligibles, and
exercises related to choosing a health plan. The questions
were fielded in Round 36 of the MCBS using in in-person
interview format.
Population Studied: Nearly 1,000 elderly and disabled
Medicare beneficiaries residing in community settings.
Principal Findings: Terms that all beneficiaries were least
likely to correctly define include “provider network” (39
percent), “formulary” (42 percent), and “Medigap” (55
percent). Nearly nine out of every 10 respondents knew what
a “primary care doctor” is. After reading a short overview,
only about half of respondents correctly indicated that the
purpose of the Personal Plan finder is to help beneficiaries
choose a health plan. About 60 percent of beneficiaries
correctly reported that Medicaid is a joint Federal/state
program despite this being directly stated in a reading
comprehension paragraph. Respondents also had difficulty
interpreting some data in the claims notice, particularly the
amount paid to the doctor and the amount of deductible met.
Based on descriptive statistics, some of the more typical
vulnerable subgroups of Medicare beneficiaries also had
significantly lower levels of health literacy relative to their
counterparts. This includes disabled beneficiaries and those
in the oldest age group (85+), non-whites, Hispanics, those
with less than a high school education, and those in fair/poor
health. Controlling for these and other factors in our
multivariate analysis, we examined the determinants of health
literacy, modeling the two indices separately. Preliminary
results suggest that older beneficiaries and those with less
education and income had significantly lower levels of health
literacy in both indices. Non-whites had significantly lower
health terminology levels.
Conclusions: Selected subgroups of Medicare beneficiaries
appear to have lower levels of health literacy based on
questions that address health related terminology and
navigation of the Medicare program.
Implications for Policy, Delivery, or Practice: Efforts need to
continue to inform and educate Medicare beneficiaries, with
consideration given to how address lower levels of health
literacy particularly among selected subgroups. Additional
research is needed to refine this initial set of health literacy
measures and to examine the factors that affect health literacy
in the Medicare population.
Primary Funding Source: CMS
●Racial/Ethnic Differences In CAM Use For Chronic
Medical Conditions
Shannon Mitchell, Ph.D.
Presented By: Shannon Mitchell, Ph.D., Associate Research
Scientist, Epidemiology and Public Health, Yale University,
240 East 86 #15H, New York, NY 10028; Tel: (212)706-2751;
Email: shannon.mitchell@yale.edu
Research Objective: This project examines an important and
heretofore largely unexamined aspect of racial/ethnic
disparities in health services utilization--the use of
complementary and alternative medicine for chronic health
conditions. Although CAM is increasingly used by Americans,
little is known about patterns of CAM use for treatment of
common medical conditions among racial/ ethnic minorities.
Specifically the study examines whether there are racial/ethnic
disparities in1) the prevalence of overall CAM use and
provider based vs. self-administered CAM therapies for
chronic medical conditions (including cardiovascular disease,
diabetes, cancer, back pain/problems, arthritis, chronic pain,
anxiety/depression, mobility problems, and obesity), 2) CAM
use as a complement or substitute for conventional medical
care, 3) the importance of CAM therapies in maintaining
health and well-being, and 4) perceived effectiveness of CAM
therapies in treating chronic medical conditions.
Study Design: The study employs a cross-sectional design to
examine racial/ethnic disparities in CAM using data from the
2002 National Health Interview Survey (NHIS). The NHIS is
a nationwide, personal interview household survey conducted
annually through the National Center for Health Statistics,
which contains a representative sample of the civilian,
noninstitutionalized population of the United States. In
2002, the NHIS included an Alternative health/
Complementary and Alternative Medicine supplement
implemented as part of the Sample Adult Core. Respondents
were asked about their use of 17 non-conventional therapies
during the last 12 months. Interviews were conducted in
English and Spanish and had a final sample response rate of
74 percent. Logistic regression analysis was used to
determine the amount of disparities in the use of CAM for
chronic medical conditions, importance of CAM in
maintaining health and well-being, and perceived effectiveness
between Whites, African-Americans, and Hispanics controlling
for other respondent demographic characteristics including
insurance status. Separate models of CAM use were
estimated for individual chronic conditions as well as any
chronic condition.
Population Studied: The sample of patients included in the
study consists of those persons included in the 2002 NHIS
adult sample core (N=31,044).
Principal Findings: Preliminary analyses show that Whites are
more likely to report any CAM use and provider-based CAM
use for chronic medical conditions. However, patterns of use,
importance of CAM, and perceived effectiveness vary among
racial/ethnic minorities. African Americans and Hispanics
were more likely to use specific self-administered CAM
therapies for the treatment of chronic medical conditions
compared to Whites. There were racial/ethnic disparities in
the importance and perceived effectiveness of CAM therapies
in the treatment of chronic medical conditions.
Conclusions: Because the analysis is still in the preliminary
stages, we do not offer conclusions at this time.
Implications for Policy, Delivery, or Practice: The broad
goal of this study is to further our understanding of how CAM
utilization might contribute to eliminating disparities in
health. Given that racial/ethnic minorities are more likely to
suffer from many chronic medical conditions, have poorer
health status and health outcomes, understanding
racial/ethnic differences in the use of multiple healing
practices, would further serve to improve cultural competency
in health care delivery and may reduce disparities in health.
Through better understanding of racial/ethnic disparities in
CAM use, importance of CAM and perceived effectiveness in
the treatment of chronic medical conditions, the study
provides information and direction to strategies for improving
quality of care and cultural competency.
Primary Funding Source: National Center for
Complementary and Alternative Medicine, NIH
●Evaluation of the Single Item Health Literacy Screener
Nancy Morris, Ph.D., APRN, Charles D. MacLean, MDCM,
Benjamin Littenberg, M.D.
Presented By: Nancy Morris, Ph.D., APRN, Associate
Professor, Nursing, The University of Vermont, 228 Rowell,
106 Carrigan Drive, Burlington, VT 05405; Tel: (802) 656-3057;
Fax: (802) 656-8306; Email: nancy.morris@uvm.edu
Research Objective: Reading ability, an aspect of health
literacy, is important to knowledge about health care: health
information, management of one’s health, utilization of
services, and outcomes. A short, simple measure to screen for
limited health literacy would allow for targeted interventions to
enhance communication and optimize patient understanding
when appropriate. Our aim was to assess the diagnostic
accuracy of the Single Item Literacy Screener (SILS) as
compared with the Short Test of Functional Health Literacy in
Adults (STOFHLA).
Study Design: Cross-sectional survey, with in-home
interviews of patients recruited from primary care practices
participating in the Vermont Diabetes Information System.
The STOFHLA is a valid, reliable and widely used 7-minute
timed test of reading comprehension to measure health
literacy. Results are categorized into inadequate (0-16 correct
answers), marginal (17-22 correct answers), and adequate
health literacy (23-36 correct answers). The SILS is a single
brief question that can be asked by a health care professional
or clerk: “How often do you need to have someone help you
when you read instructions, pamphlets, or other written
material from your doctor or pharmacy?” Responses of
“sometimes,” “often,” or “always” are considered a positive
result and suggest further assessment of health literacy is
warranted. A “never” or “rarely” response is negative.
Population Studied: 766 outpatient adults with diabetes
confirmed by their Primary Care Provider, receiving primary
care in Vermont or northern New York.
Principal Findings: Of the 766 persons screened, 142 (18%)
had limited health literacy [marginal or inadequate (STOFHLA
score of 0-22), blind, or otherwise unable to read]. Six hundred
(78%) screened negative on the SILS and 546 of these
subjects had adequate health literacy (negative predictive
values=91% (95% confidence interval 88%, 93%)). Of the 166
who scored positive on the SILS, 73 had limited literacy
(positive predictive value = 44% (.36, .52)). The sensitivity of
the SILS was 54% (95% confidence interval 45%, 62%) and
the specificity was 85% (82%, 88%). The area under the
Receiver Operative Characteristic Curve (ROC) was 0.74 (0.69,
0.78).
Conclusions: The SILS is a simple means to discriminate
between limited and adequate health literacy in ambulatory
adults with diabetes.
Implications for Policy, Delivery, or Practice: An ideal
screening instrument has a high level of sensitivity, acceptable
specificity, ease of administration, reasonable costs, and
negligible risk and minimal burden to the patient. Although
not very sensitive, the SILS can easily exclude patients who do
not need further assessment. This allows providers to target
additional assessment of health literacy to the subset of the
population most in need.
Primary Funding Source: , National Institute of Diabetes and
Digestive and Kidney Diseases
●Providing Health Education to People with Disabilities
Jill Morrow, M.D., MBA
Presented By: Jill Morrow, M.D., MBA, Medical Director,
Public Welfare, Office of Mental Retardation, Commonwealth
of Pennsylvania, 512 Health and Welfare Building; North 7th
and Forster Street, Harrisburg, PA 17108; Tel: (717)787-5110;
Fax: (717)772-0012; Email: jmorrowgor@state.pa.us
Research Objective: People with mental retardation (MR)
often have complex medical problems, but limited ability to
understand health education materials designed for the
general public. Multiple people support them and need to
understand their medical conditions. We sought to identify
whether or not we could deliver health information to a broad
group of stakeholders across a geographically diverse state.
We also asked if that health information would impact the
health of this population. The Pennsylvania Health Care
Quality Unit (HCQU) initiative aims to address this.
Study Design: We designed a demonstration project to create
HCQUs to provide health training and gather health data.
Groups of counties collaborated to develop proposals to
create HCQUs based on a template. Surveys of need related
to health care issues guided the proposals and initial activities
of the HCQUs. Eight HCQUs were created across the
Commonwealth representing all of Pennsylvania’s 67
counties. HCQU clinical staffing includes registered nurses
and consultant primary care and psychiatric physicians.
Additional staffing includes individuals with a mental health
background, IT support, and a designated training
coordinator. HCQUs average about 10 employees each. We
measured the ability to provide training at the appropriate
level for the audience and used the health data to assess the
impact of that training.
Population Studied: We targeted the population of people
receiving licensed residential services through the Office of
Mental Retardation, Commonwealth of Pennsylvania and the
people who support them including families, case managers,
and staff providing direct care. The individuals with MR live in
group homes, Intermediate Care Facilities for the Mentally
Retarded (ICF/MR), and family living. Adults comprise over
95% of this population of 15,504
Principal Findings: HCQUs provided 2199 trainings to
people from every county in the Commonwealth in quarters 3
and 4 of fiscal year 2003-2004. This involved 3309 hours and
21,112 people. The distribution of stakeholders receiving
training included 68% direct care staff, 10% supports
coordinators, 2% family members, 5% consumers, 10% health
care workers, and 5% other. Basic, intermediate and advanced
training levels comprised 75, 22, and 3% respectively.
Training topics consisted of general health topics including
nutrition, diabetes, medications, and bipolar disorder. The
training topics also included those that are more common in
people with MR like seizures, Trisomy 21, lifting and body
mechanics, and dysphagia. Trainees indicate an overall
training satisfaction over 90%.
Preliminary impact data indicates that nutrition is one of the
most frequently delivered topics and that the number of
overweight individuals with a weight reduction program
increased by 5% from fiscal years 2001-2002 to 2003-2004.
Conclusions: We demonstrated that small units can
effectively deliver health information to a wide audience across
a large geographic area. Trainees overall demonstrate
satisfaction with training. The data shows impact on behavior
related to weight reduction for overweight individuals.
Implications for Policy, Delivery, or Practice: Policy makers
should consider this model for similar populations in other
areas and other groups of individuals with disabilities.
Definitive assessment of the impact of this approach on
knowledge of health information and change in behavior
requires additional study
Primary Funding Source: No Funding Source
●Impact of Health Education on Weight in People with
Disabilities
Jill Morrow, M.D., MBA
Presented By: Jill Morrow, M.D., MBA, Medical Director,
Public Welfare, Office of Mental Retardation, Commonwealth
of Pennsylvania, 512 Health and Welfare Building; North 7th
and Forster Street, Harrisburg, PA 17108; Tel: (717)787-5110;
Fax: (717) 772-0012; Email: jmorrowgor@state.pa.us
Research Objective: The prevalence of obesity in the general
population is known through public health surveillance
surveys. However, these surveys typically either cannot
identify those with disabilities or do not capture a large
enough sample of that population. This study identifies the
prevalence of obesity and overweight in a population of people
with mental retardation (MR) and assesses the impact of
modified health education on this condition.
Study Design: b. We gathered measurements of weight,
height, diet, and physical activity on a random sample of
people with MR. A standardized instrument called the Health
Risk Profile was used and the data gathered from residential
medical records by registered nurses. The data was gathered
in two fiscal years 2001-2002 and 2003-2004. Training about
nutrition occurred between those two years. The general
population prevalence rates were used as a control.
Population Studied: c. People with MR receiving licensed
residential services including group homes, Intermediate Care
Facilities for the Mentally Retarded (ICF/MR), and family living
through the Office of Mental Retardation (OMR),
Commonwealth of Pennsylvania make up a population of
15,504. We used a random sample of 1286 adults; 748 in
2001-2002 and 538 in 2003-2004. We excluded 58 individuals
for incomplete data.
Principal Findings: d.
We used height and weight to
calculate Body Mass Index (BMI) for all individuals in the
sample. BMI was categorized using the CDC guidelines for
obesity groups. We had complete data for 1228 people.
During the first year 30% of people fell into the overweight
category and 32% in the three CDC obesity categories making
a total of 62% in both categories. In the second year 28%
made up the overweight category and 32% the obese ones for
a total of 60%. This is comparable to existing Pennsylvania
prevalence of 60%, but somewhat higher than the 24% in the
obese categories. We noted a difference of 4% in the
proportion of people in the overweight or obese categories
who had a weight reduction program between 2001-2002 and
2003-2004. We demonstrated an increase of 8% between
those two years in the proportion of people with a weight loss
program who lost at least 5 pounds in the last year. A
difference did not exist in the proportion of people who had an
exercise program; 38% in 2001-2002 and 40% in 2003-2004.
Conclusions: e. Obesity and overweight impact individuals
with MR to the same extent as the general population. Our
data shows that individuals with MR can participate in weight
reduction programs and lose weight. Our results indicate that
training about nutrition lead to weight loss and potentially the
impact of obesity on chronic health conditions.
Implications for Policy, Delivery, or Practice: f.
Policy
makers should consider the application of this model to other
disability groups and to other health conditions. Further
research is needed to assess the longevity of the effects of the
training on weight.
Primary Funding Source: No Funding Source
●Social Cohesion, Social Support and Health Among
Latinos in the United States
Norah Mulvaney-Day, Ph.D., Margarita Alegria, Ph.D
Presented By: Norah Mulvaney-Day, Ph.D., Associate
Director, Cambridge Health Alliance, Center for Multicultural
Mental Health Research, 120 Beacon Street, 4th Floor,
Somerville, MA 02472; Tel: (617) 503-8448; Fax: (617) 5038430; Email: nmulvaney-day@charesearch.org
Research Objective: Understanding the source of health
disparities among Latinos is complicated by the heterogeneity
of this group. Social supports have been found to have a
positive influence on health in the general population. This
analysis examines the relationships between different forms of
social support (neighborhood social cohesion, family support
and friend support) and self-rated physical and mental health
in a population of Latinos.
Study Design: Two dichotomous variables were created to
represent a poor/fair self-rating of (1) general physical health
and (2) mental health as the dependent variables. Logistic
regressions using weighted data were conducted in order to
observe the main effects of the levels of social cohesion,
family support and friend support on these two dependent
variables. Control variables were age, sex, poverty, education
and marital status. The sample was stratified across nativity
(U.S. born versus non-U.S. born) in order to observe
differences.
Population Studied: The data set for this analysis is from the
National Latino and Asian American Study (NLAAS), a
nationally representative household survey of Latinos and
Asians living in the United States. The sample consists of
2554 Latinos from four distinct subgroups: 577 Cuban; 495
Puerto Rican; 868 Mexican; and 614 Other Latino.
Principal Findings: High levels of family support, social
support and social cohesion all had negative relationships
with poor/fair self-rated general health in the combined
sample. Family support and friend support were also
negatively related to poor/fair mental health, although social
cohesion did not emerge as a factor in the mental health
model. When models were stratified across nativity, family
and friend support remained significant for those born outside
the U.S. but not for those in the U.S., and social cohesion
approached significance for those born in the U.S. (p<.10) but
not for those born outside. For self-rated mental health,
family support was the only significant support variable across
both nativity groups.
Conclusions: Family support, friend support and social
cohesion represent distinct characteristics of social interaction
that impact health in different ways across nativity status. For
new immigrants, health is dependent upon family and friend
support networks, and not necessarily upon relationships of
trust and cohesion with unrelated others in one’s
neighborhood. For US-born Latinos, social cohesion is more
important, suggesting that if family and friend supports break
down over time, the level of social cohesion is more critical to
individual health.
Implications for Policy, Delivery, or Practice: This study
suggests that there may be different pathways toward health
for different generations of Latinos and also across general
versus mental health. This complexity indicates that it is
critical to avoid mental health and health care policy
development aimed at Latinos as a uniform group. Further,
the importance of community level support for Latinos born in
the U.S. should to be taken into account by primary care
clinicians. Community interventions that emphasize
neighborhood level cohesion may be a vehicle for reducing
health disparities. Finally, the relationship between family
support and self-rated mental health should be considered in
mental health treatment plans for Latinos.
Primary Funding Source: National Institute of Mental Health
●Do Racial and Ethnic Disparities Exist in Unmet Needs
for Specialty, Dental, Mental, and Allied Health Care
among Children with Special Health Care Needs?
Emmanuel Ngui, DrPH, MSc, Glenn Flores, M.D.
Presented By: Emmanuel Ngui, DrPH, MSc, Assistant
Professor of Pediatrics, Epidemiology and Health Policy,
Center for the Advancement of Underserved Children,
Department of Pediatrics, Medical College of Wisconsin, 8701
Watertown Plank Road, Milwaukee, WI 53226; Tel: (414)4564302; Fax: (414)456-6385; Email: engui@mail.mcw.edu
Research Objective: To examine racial/ethnic disparities in
reported unmet needs for specialty, dental, mental, and allied
health care, and to identify factors associated with unmet
needs among racial/ethnic CSHCN.
Study Design: We analyzed data from the 2001 National
Survey of CSHCN (NS-CSHCN). For each outcome, unmet
need was defined as parental report of the child not receiving
all needed health care in the past year. Covariates examined
included child’s age, gender, residence, maternal education,
insurance coverage, poverty status, availability of a personal
doctor or nurse, language of interview, the child’s condition
severity, and whether the condition is stable or changes all the
time.
Population Studied: A nationally representative sample of
38,886 CSHCN <18 years old participating in the 2001 NSCSHCN.
Principal Findings: Black and Hispanic CSHCN were
significantly (p<.05) more likely than white CSHCN to be poor
(29% vs. 32% vs. 9%). Black CSHCN also were significantly
more likely than white CSHCN to have no personal doctor or
nurse (14% vs. 9%) and to have conditions that change all the
time (9% vs. 6%), whereas Hispanic CSHCN were
significantly more likely than white CSHCN to be uninsured
(9.3% vs. 4.5%). The prevalence of reported unmet needs
among CSHCN ranged from 7% for specialty care to 18% for
mental health care. Compared to white CSHCN, black
CSHCN had significantly more unmet needs for specialty
(10% vs. 7%), dental (16% vs. 8.7%), and mental health care
(27% vs. 17%), and Hispanic CSHCN had more unmet dental
care needs (16% vs. 9%). Except for unmet mental health
care needs among Hispanic CSHCN (OR, 0.6; 95% CI, 0.40 0.96) all other racial/ethnic disparities in unmet needs
disappeared after multivariate adjustment. Black female
CSHCN, however, were disproportionately more likely to have
unmet mental health care needs than all other groups (41%
vs. 13-20%), and had significantly greater adjusted odds of
unmet mental health care needs than black male CSHCN (OR,
2.2; CI, 1.2-4.0). For each racial/ethnic group and for CSHCN
overall, the most significant determinants of unmet needs
included lack of insurance, poverty, lack of a personal doctor
or nurse, and having conditions that change all the time.
Conclusions: Racial/ethnic disparities exist in unmet
specialty, dental, and mental health care needs among
CSHCN, but disappear after adjustment, except for unmet
mental health care needs among black female CSHCN.
Within and across racial/ethnic groups, key factors associated
with disparities in unmet needs included lack of insurance
coverage, poverty, lack of a personal doctor or nurse, and
having conditions that change all the time.
Implications for Policy, Delivery, or Practice: Our findings
suggest the need for universal mental health screening of
black female CSHCN. In addition, because minority children
are disproportionately more likely than white children to live in
poverty, be uninsured, and lack a personal doctor or nurse,
eliminating disparities in unmet specialty, dental, and mental
health care needs for all CSHCN, and especially minority
CSHCN, will require greater efforts to reduce the number of
poor and uninsured CSHCN, and to ensure that all CSHCN
have a medical home.
Primary Funding Source: No Funding Source
●Are There Racial and Ethnic Disparities in Satisfaction
with Care Among Parents of Children with Special Health
Care Needs (CSHCN)?
Emmanuel Ngui, DrPH, MSc, Glenn Flores, M.D.
Presented By: Emmanuel Ngui, DrPH, MSc, Assistant
Professor of Pediatrics, Epidemiology and Health Policy,
Center for the Advancement of Underserved Children,
Department of Pediatrics, Medical College of Wisconsin, 8701
Watertown Plank Road, Milwaukee, WI 53226; Tel: (414)4564302; Fax: (414)456-6385; Email: engui@mail.mcw.edu
Research Objective: To examine whether racial/ethnic
disparities exist in parental satisfaction with care among
CSHCN, and to identify factors associated with these
disparities.
Study Design: Bivariate and multivariate analyses of data
from the 2001 National Survey of CSHCN (NS-CSHCN).
Covariates included sociodemographic factors, insurance,
language of interview, condition severity and stability, ease of
using services, provider-patient communication factors (PPC),
and having a personal doctor/nurse. PPC measures included
parental ratings of the healthcare provider as
never/sometimes: spending adequate time with the child
(inadequate time), listening carefully to parents (inadequate
listening), being sensitive to the family’s values/customs
(cultural insensitivity), providing needed information to
families (inadequate information), and helping parents feel
like partners in care (inadequate partnership).
Population Studied: Nationally representative sample of
38,886 CSHCN < 18 years old.
Principal Findings: Nationally, 8% of parents of CSHCN were
dissatisfied or somewhat dissatisfied with their children’s
healthcare, with Hispanic (16%) and black (13%) parents more
dissatisfied than white parents (7%). Dissatisfaction was
greater among parents of CSHCN who were poor, uninsured,
publicly insured, had no personal doctor/nurse, had a mother
who had not graduated from high school, interviewed using a
non-English survey, and with conditions that change
frequently. Minority parents were significantly more likely than
white parents to report problems in ease of using healthcare
services and PPC. Black and Hispanic parents were
significantly more likely than white parents to be dissatisfied
(odds ratio [OR], 2.3; 95% confidence interval [CI], 1.6-3.3; OR,
2.6; CI, 1.8-3.8, respectively), but these disparities disappeared
after multivariate adjustments. Among all parents of CSHCN,
inadequate listening (OR, 2.4; 95% CI, 1.6-3.5), information
(OR, 1.8; CI, 1.3-2.6), time (OR, 1.7;CI, 1.2-2.4), and partnership
(OR, 1.6; CI, 1.1-2.3), difficulty using services (OR, 3.4; CI, 2.54.8), language barriers (OR, 2.1;CI, 1.1-4.4), and condition
severity (OR, 1.1; CI, 1.06 -1.2) were significantly associated
with dissatisfaction. Within racial/ethnic groups, unique
factors associated with dissatisfaction included cultural
insensitivity among blacks (OR, 2.7; CI, 1.2-5.9), language
barriers among Hispanics (OR, 2.7; CI, 1.2-6.15), and lack of
insurance (OR, 1.8; CI, 1.1-3.1) and inadequate time (OR, 1.7;
CI, 1.1-2.6) among whites. Inadequate partnership was
significantly associated with dissatisfaction among Hispanics
(OR, 5.2; CI, 2.2-12.4) and white parents (OR, 1.7; CI, 1.1-2.6)
but not black parents.
Conclusions: Racial/ethnic disparities exist in satisfaction
with care among parents of CSHCN, but disappear after
adjustment. Across racial/ethnic groups, key factors
associated with dissatisfaction included inadequate listening
and information, condition severity, and difficulty using
services. Factors associated with dissatisfaction that were
unique to specific racial/ethnic groups included provider
cultural insensitivity for black parents, language barriers for
Hispanic parents, and lack of insurance, and inadequate time
with the provider among white parents. Inadequate
partnership with the provider was associated with
dissatisfaction among Hispanic and white but not black
parents.
Implications for Policy, Delivery, or Practice: Regardless of
race/ethnicity, satisfaction with care among the parents of
CSHCN can be improved by enhancing the ease of using
services and the quality of provider-patient communication,
particularly provider listening skills, information provision, and
a greater focus on condition severity. Greater satisfaction
among parents of Hispanic CSHCN may be achieved by
addressing language barriers, such as by providing adequate
interpreter services and bilingual staff. For all CSHCN, and
particularly minority CSHCN, greater satisfaction with care
might be achieved by implementing more family-centered
healthcare that incorporates improved provider listening skills,
adequate information and time with patients and their
families, fewer language barriers, cultural sensitivity, and
easier access to services.
Primary Funding Source: No Funding Source
●A Survey on Drug Utilization Pattern in the Health Care
System of India.
Bhagirath Patel, Ph.D., Paresh Shah, Ph.D., Rames Goyal,
Ph.D.
Presented By: Bhagirath Patel, Ph D., Pharmacologist, Shri B
M Shah College of Pharmacy, College Campus, Modasa,
383315; Tel: +91 2774 249475; Email: bk121@rediffmail.com
Research Objective: Drug utilization pattern is an important
factor determining the efficiency of a health providing set up.
An efficient set up requires usage of quality drugs at an
affordable price. For this purpose it is essential to collect data
to evaluate the drug utilization pattern and to effect necessary
interventions and policy changes to eliminate the
shortcomings present, if any.
Study Design: A model study was conducted at retail
pharmacy outlets in rural and urban areas of North Gujarat,
India. Data (1796) were collected by author and qualified
pharmacist who were trained in survey procedures, using
specially prepared and pre-tested questationnaires. The
collected data were tabulated and analyzed by using normal
proportion test.
Population Studied: Total population is around 2 millions in
the area of North Gujarat, India.
Principal Findings: Broad spectrum antibacterial, compound
with analgesics, antipyretic and anti-inflammatory effects,
antacids/antiulcer and nutritional products constituted a high
proportion of prescription in both urban and rural areas.
Based on WHO criteria, most of the drugs (57%) prescribed in
the rural areas were non-essential, compared with (48%) in
urban areas.
Conclusions: Use of fixed dose combinations, frequency use
of antibiotics, vitamin tonic and poly pharmacy was the trend
identified through this study. Indiscriminate use of antibiotics
is leading to cause resistance development.
Implications for Policy, Delivery, or Practice: The results of
this study emphasize the need for comprehensive measures,
including training, information, legislation and education at all
level of drug purchase to rationalize the drug through by
improving prescribing patterns.
Primary Funding Source: State Government
●The Prevalence of Discrimination among Latinos in the
US: Cultural and SES Correlates
Debra Joy Pérez, MA, MPA, Ph.D., Margarita Alegria, Ph.D.,
Lisa Fortuna, M.D., MPH
Presented By: Debra Joy Pérez, MA, MPA, Ph.D., Program
Officer/WKKELLOGG Fellow in Health Policy
Research/Multicultural Mental Health Research Fellow/NIMH
mentee, Research & Evaluation Unit/Interfaculty Program in
Health Policy at Harvard, The Robert Wood Johnson
Foundation/Harvard University, PO Box 3311, Princeton, NJ
08543; Tel: (609)627 5966; Email: djperez@fas.harvard.edu
Research Objective: Numerous studies established a
relationship between discrimination and physical & mental
health. Yet, we know little about the likelihoods of
discrimination among Latinos in different socio demographic
strata. Few studies have shown the prevalence or severity of
discrimination experiences among Latino and Latino
subgroups. Most of the existing literature focuses on
Black/White differences. This paper examines the prevalence
and severity of perceived discrimination among US Latinos.
This study will address the following questions: What is the
rate and severity of self-reported perceived discrimination
among Latinos and Latino subgroups in the US? Do subgroup
difference remain when accounting for difference in income
and education? How do cultural factors vary perceived
discrimination? Do Cubans, Puerto Ricans or Mexicans
attribute discrimination to the same causes? How does the
rate or severity of discrimination vary for Latinos in different
generations or different stages of acculturation?
Study Design: Data are from the National Latino and Asian
American Study (NLAAS). The NLAAS is based on a stratified
area probability sample design of persons 18 years of age and
older in the non-institutionalized population of the 50 states
and Washington D.C. The sampling frames and sample
selection procedures selected for this study are common to
the University of Michigan Survey Research Center’s (SRC)
National Sample Design. Through the following four step
sampling process, a probability sample of respondents for
screening interviews was selected; 1) primary stage sampling
of US Metropolitan Statistical Areas (MSAs) and counties, 2)
second stage sampling of area segments, 3) third stage
sampling of housing units within the selected area segments,
4) random selection of respondents from the sample housing
units. Additionally, the NLAAS sample design includes two
components; 1) an NLAAS core sample which is designed to
provide a nationally representative sample of all national
origin groups regardless of geographic residential patterns
and 2) the NLAAS-HD supplements which are targeted oversampling of geographic areas with a moderate to high density
(5 %+) of targeted Latino and Asian households. The
supplements were added to the sample plan as a cost
reduction strategy to obtain the desired sample of national
origin groups not widely dispersed in the US (e.g. Puerto
Rican, Cuban, Chinese, Filipino, and Vietnamese). The
Mexican subgroup did not require supplemental over
sampling. Individuals residing in the high-density areas had
two chances of selection, one under the NLAAS Core sample
and the other under the NLAAS-HD sample. This sample
design requires weighting corrections for joint probabilities of
selection. The NLAAS uses validated measures of sociocultural factors like acculturative distress, measures of ethnic
identity, information regarding immigrant and generational
status and language proficiency. The NLAAS is the largest
national mental health study of Latinos and ever conducted
for understanding mental health and services needs in these
communities. All NLAAS subjects are interviewed in person
regarding mental health diagnoses and the socio-cultural
variables examined in this study. The target population for this
investigation includes all Latino adults (ages 18 and over) in
the NLAAS criteria for eligibility. The average final weighted
response rates for the NLAAS first respondents was 75%. The
NLAAS battery includes a nine-item scale of self-perceived
discrimination. This discrimination scale measures the
frequency of routine experiences of unfair treatment in
general. Respondents indicate how often they experience the
following situations: 1.) being treated with less courtesy that
other people, 2.) being treated with less respect than other
people, 3.) receiving poor service than other people at
restaurants or stores, 4.) people acting as if they think the
respondent is not smart, 5.) people acting as if they are afraid
of respondent, 6.) people acting as if they think respondent is
dishonest, 7.) people acting as if they think the respondent is
not as good as they, 8.) being called names or insulted and 9.)
being threatened or harassed. These nine items were taken
from the Detroit Area Study (DAS), (Jackson & Williams, 1995;
Williams, Yu, Jackson, and Anderson, 1997). The 6 response
categories range from Almost Everyday (1) to Never (6). The
prevalence of each item is assessed separately for each of the
analysis proposed in this paper. he items in the scale had a
standardized Cronbach alpha score of .9105. This study uses
logistic regression and chi-square to test the various
hypotheses.
Population Studied: Latinos, Puerto Ricans, Mexicans,
Cubans, Other Latinos
Principal Findings: The overall rate of discrimination among
Latinos was 34%. Rates of any perceived discrimination
varied by ethnicity and demographic characteristics. Puerto
Ricans were more likely to report any discrimination than
other Latinos. Over forty percent of Puerto Ricans (43%)
reported some experience of discrimination compared to
33.6% of Mexicans and 34.8% of other Latinos (p=0.022.)
Cubans were significantly less likely than all other groups to
report any discrimination (20.3%; p=0.022).
Age and Gender: More men than women reported any
discrimination (39.4% compared to 28.8%; p<.000). Age was
inversely associated with reporting any discrimination.
Younger Latinos were significantly more likely than older
Latinos in any other age group to report any discrimination.
There is an almost perfect linear relationship between increase
in age and decrease reporting of perceived discrimination
ranging from 50.3% at 18 to 24 year olds compared to 10.9%
of Latinos age 65 and over; p<.000). Reports of any
discrimination increased with more education, just over a
quarter of Latinos with less than a high school degree
reported any discrimination compared to 43% of Latinos with
some college or better (p<.000.) Socioeconomic
Factors:Latinos in higher income ranges were also more likely
to report any discrimination. Less than a third of Latinos living
in households with less than 15K per year reported any
discrimination while approximately 40% of Latinos in
households earning more than 75K reported any
discrimination (p=0.0289.) Cultural Factors: We found that
English proficient Latinos were more likely than Spanish
dominant Latinos to report any discrimination (45.3% v.
22.5%, p <0.000). More US born Latinos reported higher
rates of discrimination compared to immigrant Latinos.
Nearly half of native-born Latinos (45.9% reported any
discrimination) compared to only a quarter of immigrants
(25.5%; p<.000). Just under a quarter of Latinos who indicated
that associations with his/her own ethnic groups were very
important reported higher rates of perceived discrimination
compared to nearly forty percent of Latinos who rated these
associations as less important (24.27% vs. 37.65%; p <.000.)
A quarter of first generation Latinos (25.47%) reported any
discrimination compared to 38.96% of Latinos in the second
generation, 47.24% of third generation Latinos and 51% of
fourth generation Latinos (p<.000.) Similarly, possessing a
strong ethnic identity and having less than a high school
diploma was also associated with a decreased mean on the
severity scale of perceived discrimination. Being English
proficient and being a 3&4 generation Latino was also
associated with an increased probability of reporting perceived
discrimination after adjusting demographic and cultural
factors.
Conclusions: The results showed that being Cuban was
associated with lower rates of perceived discrimination. One
explanation may be that Cuban immigrants have a good
infrastructure for transition into the US of any other Latino
group (Rumbaut 1994). In addition, the effects of living in an
ethnic enclave may provide protection against the perception
of discrimination if not protection against discriminatory acts.
As hypothesized increases in exposure to groups in cultural
conflict was associated with increases in perception of
discrimination. Indeed, as some studies on African Americans
have demonstrated gains in education were associated with
increases in the rate of perceived discrimination. The
proportion of Latinos with college degrees who perceived
discrimination supports this theory. Another important
conclusion is that a clear acculturation pattern emerges with
regard to generation change and perceptions of
discrimination. Third and fouth generation Latinos report
much higher rates of any discrimination than Latinos in first
generations. For all cultural factors tested, this low
acculturation-high discrimination relationship was consistent
except for ethnic identity. Latinos with the highest ethnic
identity reported the lowest rates of perceived discrimination
compared to Latinos with lower ethnic identity. Ethnic identity
may be protective against initial perception of discrimination.
Strong ethnic identity may reduce the exposure to perceived
discrimination because individuals who strongly identify are
less likely to be exposed to different ethnic group members.
They may also be less hyper-vigilant or sensitive than those
who are more ethnically isolated.
Implications for Policy, Delivery, or Practice: Latinos will
have increasing impact on the healthcare system and health
policy issues. People who care about how the health care
system operates should care about how the prevalence of
discrimination in the larger society might impact perceptions
of fairness in healthcare setting. Organizations need to be
responsive to the perception of discrimination by this
population. Reducing discrimination can improve the
patient/provider relationship and lead to a welcoming
environment that in turn can improve compliance and followup. Understanding how discrimination correlates with English
proficiency and generation have implications for service
delivery. Knowledge of the prevalence of discrimination could
lead to changes in healthcare provider training.
Primary Funding Source: WKK, NIMH
●Racial and Ethnic Disparities in Rural Mental Health
Treatment
Stephen Petterson, Ph.D., Emily J. Hauenstein, Ph.D., LCP,
APRN, Douglas Wagner, Ph.D., Virginia Rovnyak, Ph.D.,
Elizabeth Merwin, Ph.D., MSN, Barbara Heise, MSN, NP
Presented By: Stephen Petterson, Ph.D., Assistant Professor,
School of Nursing, University of Virginia, P.O. Box 800782,
Charlottesville, VA 22908; Tel: 434-243-9832; Fax: 434-9821809; Email: smp4n@virginia.edu
Research Objective: To assess racial and ethnic differences in
mental health treatment rates and pathways to care among
Mexican American, African American and Non-Hispanic white
non-elderly adults residing in rural areas.
Study Design: This study uses data from the Medical
Expenditure Panel Survey (MEPS) to estimate racial and ethnic
disparities in mental health treatment in rural areas, as well as
compare urban and rural group differences. Three calendaryear measures of mental health treatment are examined: any
medical visit related to a mental health condition, the number
of visits and any specialized mental health care visit. Several
measures of rurality are examined, including the urban-rural
continuum and the proportion of a county that is classified as
rural. Binary logistic and linear regression analyses were
performed to estimate differences in mental health treatment,
adjusting for demographic, need, and access variables.
Separate models are estimated for rural subsamples of
Mexican Americans, African Americans and non-Hispanic
whites. Variance estimation is based on the clustered survey
design of the MEPS; the weights and design variables take
multiple observations of an individual into account.
Population Studied: The MEPS is a large nationallyrepresentative panel survey of households designed to provide
estimates of the use of health services, medical expenditures
and sources of payment. Data is collected in five interviews
over a two-year period. Our sample consists of adults aged 18
to 64 in the first four panels of the MEPS, covering a period
from 1996 to 2000. The sample size is 34,299 respondents;
this figure includes 5,765 rural non-Hispanic whites, 234 rural
Mexican Americans and 745 rural African Americans.
Principal Findings: Our findings show that rural minorities
are significantly less likely than rural whites to obtain needed
mental health care despite their having significantly poorer
mental health,. For instance, multivariate results indicate that,
all else being equal, the odds that rural whites obtain any
treatment are nearly 1.7 times greater than the odds for rural
Mexican Americans and 2.6 times greater than rural African
Americans. A surprising result is that while rural minorities
are at a considerable disadvantage relative to non-Hispanic
whites, there are no urban-rural differences among either
Mexican Americans or African Americans. By contrast, the
odds of mental health treatment are 20% greater for urban
non-Hispanic whites than for rural non-Hispanic whites.
Conclusions: Minorities are underserved by the rural mental
health system, reflecting among other factors, restricted
access to mental health services, high poverty rates and low
levels of insurance coverage.
Implications for Policy, Delivery, or Practice: Rural
residents, particularly from minority groups, are at greater risk
of mental health problems. Low levels of treatment may
heighten that risk.
Primary Funding Source: NIMH
●Healthcare Barriers and Deaf People: Facilitating a
Cross-Cultural Dialogue
Brian Piotrowski, M.D., Steven Barnett, M.D.
Presented By: Brian Piotrowski, M.D., National Center for
Deaf Health Research, University of Rochester, 1331 East Victor
Road, Victor, NY 14564; Tel: (585) 924-2100; Fax: (585) 9245920; Email: bepiotrowski@hotmail.com
Research Objective: To create an environment for respectful
cross-cultural discussion in order to identify barriers to
healthcare for deaf people and their families.
Study Design: The Deaf Heath Task Force (DHTF) was a
thirty-member group that included a unique mix of
representatives of local hospitals and health systems, schools,
service agencies, clinicians, and community-based deaf
organizations. A goal of the DHTF was to identify and
prioritize barriers to health care. This information would then
be used in realistic mapping strategies for meaningful change
in improving health care access for deaf individuals. We used
various group process techniques modified for use with a
bilingual group (ASL and spoken English).
Population Studied: Members of culturally distinct minority
groups often have to deal with challenges in health care
access and utilization. Deaf people who communicate in
American Sign Language (ASL) comprise a sometimes
unrecognized cultural minority group. The city of Rochester,
NY - with the highest per capita population of deaf individuals
in the United States - has both dynamic deaf community
leaders as well as many heath care providers passionate about
improving the health of the deaf community. Despite this,
there are few settings in which hearing and deaf individuals
may meet and discuss these issues - bringing together
perspectives from both consumers and providers.
Principal Findings: The DHTF met monthly in a round-table
setting. Deaf culture and the mechanics of working with ASL
interpreters were constantly reinforced. Initially, the group
generated a list of barriers to health care access and organized
them in terms of both importance to improving health care
access and relative difficulty of affecting change in them.
These barriers included economic, cultural, and logistic
components. The DHTF later published a report on these
barriers and created a plan to share this information the local
community. In addition, the members and the work of the
DHTF were instrumental in securing funding for the NCDHR,
the first CDC-funded prevention research center focused on
health promotion and disease prevention with a disability
population.
Conclusions: We will describe the process of creating the
DHTF and negotiating the bilingual/bicultural group process.
We will share the list of barriers, their ranking, and the
blueprint for addressing those barriers. We will also discuss
observations of the dynamics of the group process in the
context of diversity (including language, race, sex, education
and experience).
Implications for Policy, Delivery, or Practice: While this
particular project deals with health care access for deaf
individuals, there is no reason that a similar technique (with
modifications) could not be used to address other issues (e.g.
schools, unemployment) in other socially/linguistically distinct
populations (e.g. immigrant communities)
Primary Funding Source: CDC, ATPM
●Does Race Matter in the Use and Outcomes of
Hematopoietic Stem Cell Transplantation (HSCT) in
Children with Cancer?
Jasmina Radeva, MA, Lesley H. Curtis, Ph.D., Philip P.
Breitfeld, M.D.
Presented By: Jasmina Radeva, MA, Clinical Research
Manager, Center for Clinical and Genetic Economics, Duke
Clinical Research Institute, 2400 Pratt Street Room 0311
Terrace Level, Durham, NC 27705; Tel: (919)668-8654; Fax:
(919)668-7124; Email: radev001@dcri.duke.edu
Research Objective: Hematopoietic stem cell transplantation
(HSCT) is an established treatment approach for many
malignant and non malignant diseases in children. Stem cells
can be derived from three sources: autologous (patient’s
own), allogeneic (unrelated donor), or syngeneic (identical
sibling). The success of unrelated donor stem cell transplant
is related to the number of donor’s and recipient’s matching
human leukocyte-associate antigens. Although the number of
donors is increasing overall, individuals from certain ethnic
and racial groups still have a lower chance of finding a
matching donor. Our objective was to determine if race is
associated with the likelihood of undergoing HSCT form any
source and autologous HSCT in children hospitalized with
cancer and whether race impacts length of stay (LOS) and
mortality during HSCT.
Study Design: Using the 2000 Kids Inpatient Database and
Clinical Classification Software, we identified all childhood
cancer encounters and those involving HSCT procedures.
Cancer encounters were defined as encounters with primary
or secondary CCS diagnosis code between 1 and 45. HSCT
was defined as primary or secondary CCS procedure code of
64. Encounters with HSCT procedures were further
categorized as allogeneic, autologous, bone marrow
transplant not otherwise specified, and multiple HSCT
procedures on the basis of primary and secondary ICD-9-CM
procedure. Comorbidities were also identified using primary
and secondary ICD-9-CM diagnosis codes. Logistic regression
was used to model the probability of undergoing HSCT from
any source or from autologous source controlling for
demographic characteristics, clinical variables, and hospital
characteristics. Ordinary least squares model was used to
model LOS and death during HSCT admission. Likelihood
ratio and Wald tests were used to evaluate the inclusion of
certain hospital characteristics in the models.
Population Studied: Cancer-related hospital encounters for
patients up to the age of 18 excluding neonatal encounters
were selected from the KID 2000 data. The KID is part of the
Healthcare Cost and Utilization Project. After excluding cases
where gender was missing, the study sample included 49,994
cancer-related encounters for patients of age 18 and younger
excluding neonatal encounters.
Principal Findings: Compared to whites, blacks and
Hispanics were less likely to undergo HSCT from any source
(Odds Ratio [OR] = 0.54, marginal effect [ME] = -0.0076,
p<0.001 for black; OR = 0.71, ME = -0.0047, p= 0.003 for
Hispanic) and less likely to undergo autologous HSCT
(OR=0.47, ME = -0.0045, p<0.001 for black; OR= 0.69, ME = 0.0026, p= 0.022 for Hispanic). Regression analysis suggests
neither black nor Hispanic race has impact on LOS during
HSCT. The positive coefficient on black however had statistical
significance when evaluating inpatient death during an HSCT
procedure (coefficient: 0.069, p=0.039).
Conclusions: This analysis supports that, compared to
whites, blacks and Hispanics are less likely to undergo any
form of HSCT, not just allogeneic procedures. Black race was
significantly associated with inpatient mortality during HSCT.
Implications for Policy, Delivery, or Practice: In the future,
these analyses should be controlled for patient level clinical
data including disease stage and severity and data on
parent/guardian perceptions of health care.
Primary Funding Source: No Funding Source
●Disparities in Glycemic Control among Patients with
Diabetes Receiving Care in Public Hospital Systems
Marsha Regenstein, Ph.D., Jennifer Huang, MS, Dean
Schillinger, M.D., Daniel Lessler, M.D., MHA, Brendan Reilly,
M.D.
Presented By: Marsha Regenstein, Ph.D., Director, National
Public Health and Hospital Institute, 1301 Pennsylvania
Avenue NW, Suite 950, Washington, DC 20004; Tel: (202)
585-0100; Fax: (202) 585-0101; Email: mregenstein@naph.org
Research Objective: Public hospitals provide care to many of
the country’s most vulnerable residents and share a
commitment to eliminate barriers to care based on race,
ethnicity, or ability to pay. While these organizations provide
significant amounts of inpatient, outpatient, emergent and
supportive services to patients, little is known about the care
of patients with diabetes across public hospitals, or the extent
to which disparities in diabetes outcomes exist within these
systems.
Study Design: In 2002-3, the National Public Health and
Hospital Institute assembled a quality improvement
consortium involving 7 public hospital systems that provide
care to over 125,000 patients with diabetes on an annual
basis. We obtained demographic, clinical, and utilization data
over a 3-year period on a random sample of 3,000 patients
with diabetes from each system.To be included, patients had
to be *18 years old and have made 2 or more outpatient visits
during the study period with an associated ICD-9 code of
diabetes (excluding gestational diabetes). Six of the 7 sites
reported race/ethnicity data on >98% of patients. Using
patients' last HbA1c value, we measured the proportion of
patients with poor glycemic control (HbA1c *9%, a standard
benchmark), and explored whether glycemic control varied by
race/ethnicity, insurance status, or use of outpatient services.
Population Studied: Adult patients who received care at 7
public hospital systems during the period Jan 1 2000 through
Dec 31 2002. Overall, 69% of patients were non-White
minorities, and 42% were uninsured.
Principal Findings: Over the 3-year period, 34% of patients
rarely made outpatient visits (2-4 visits/3 yrs), 26% made
occasional visits (5-10 visits/3 yrs) and 39% made regular
visits (>10 visits/3 yrs). Overall, 26% of patients had poor
glycemic control. Poor glycemic control was more common
among Blacks (29%) and Hispanics (31%) than Whites (19%)
or Asians (16%) (P<.001) and was greater among the
uninsured (33%) than those with Medicaid (24%), commercial
insurance (24%) or Medicare (17%) (P<.001). Adjusting for
age, sex, race/ethnicity, insurance, and outpatient utilization,
Blacks and Hispanics had worse glycemic control as
measured by mean HbA1c (8.3% and 8.2%) than Whites or
Asians (7.8% and 7.9%), as did the uninsured (8.2%)
compared to Medicaid (8.0%), commercially insured (7.9%)
and Medicare (7.5%) (P<.001). In adjusted models, there was
a graded relationship between extent of ambulatory use and
glycemic control with rare, occasional, and regular users
having mean HbA1c levels of 8.4%, 8.1% and 7.8%,
respectively (P<.001).
Conclusions: Despite limited resources and the challenges
posed by caring for diverse and vulnerable populations,
public hospital systems perform at a level comparable to other
health systems with regard to glycemic control. Our results
suggest that public hospital systems should target access for
the uninsured, continuity of care for all patients, and quality of
care for minority populations.
Implications for Policy, Delivery, or Practice: Given their
shared missions and the availability of electronic data
systems, public hospital systems have the opportunity to
collaborate in quality of care benchmarking and crafting
innovative diabetes improvement initiatives.
Primary Funding Source: CWF
●A Comparison of Cancer Patients Visits to Physician’s
Offices with Visits to Hospital Outpatient Departments
Lisa Richardson, M.D., MPH, Florence Tangka, Ph.D.
Presented By: Lisa Richardson, M.D., MPH, Medical Officer,
Division of Cancer Prevention and Control, Centers for
Disease Control and Prevention, 4770 Buford Highway NE,
MS K-55, Atlanta, GA 30341; Tel: (770) 488-4351; Fax: (770)
488-4639; Email: lfr8@cdc.gov
Research Objective: To compare the characteristics of cancer
patients who made ambulatory visits to physician’s offices
(POs) with the characteristics of those who made ambulatory
visits to hospital outpatient departments (OPDs).
Study Design: We analyzed data collected during the 2001
and 2002 National Ambulatory Medical Care Survey (NAMCS)
of POs and National Hospital Ambulatory Medical Care
Survey (NHAMCS) of OPDs. Sampling weights that
accounted for multistage sample design and nonresponse of
in-scope practitioners were used to obtain national estimates
for the number of PO or OPD visits made by patients with
cancer. We examined cancer-related and personal
characteristics of patients by the visit setting (PO versus
OPD).
Population Studied: Patients with a primary diagnosis of
cancer made a total of 3,116 visits to a PO or an OPD in 2001
or 2002.
Principal Findings: We estimated that 7,856,379 patients
made 27,456,800 cancer-related ambulatory visits in 2001 and
2002, 15% to an OPD and 85% to a PO. We also found that
the characteristics of patients varied by setting type. The
proportion of cancer-related visits for lung cancer and
leukemia/lymphoma (10% and 35%, respectively) were higher
in OPDs than in POs (8% and 11%, respectively). In contrast,
the proportion of visits for breast and prostate cancer were
higher in POs (23% and 19%, respectively) than in OPDs (14%
and 6%, respectively). Anti-cancer drugs were ordered or
administered in 11% of all visits, with similar proportions in
POs and OPDs, although patients indicated that intravenous
chemotherapy was the main reason for 10% of OPD visits but
only 2% of PO visits. Patients younger than 45 accounted for
a much higher proportion of cancer-related visits to OPDs
(19%) than to POs (2%); while those older than 45 were more
likely to be seen in POs. Blacks and Hispanics accounted for a
higher percentage of visits to OPDs (14% and 12%,
respectively) than to POs (9% and 3%, respectively), and
Medicaid paid for more visits to OPDs (17%) than to POs
(4%). Patients who had made six or more cancer-related
medical visits in the previous year accounted for 56% of all
OPD visits but only 25% of all PO visits.
Conclusions: Minority and Medicaid-insured cancer patients
are more likely than other cancer patients to make ambulatory
visits to OPDs. Although the percentage of visits associated
with anti-cancer drugs in OPD settings is similar to that in PO
settings, intravenous chemotherapy is far more likely to be
listed as the main reason for cancer-related visits to OPDs
than visits to POs.
Implications for Policy, Delivery, or Practice: These results
confirm recent findings that black and Hispanic cancer
patients and cancer patients insured by Medicaid are more
likely to be served in hospital-based clinics than are other
cancer patients (Richardson et al., accepted to AJPH).
Because little research has been done to determine whether
the location of cancer treatment is a causal factor in cancerrelated health disparities, future studies should examine the
content and quality of the cancer care delivered in each
treatment location.
Primary Funding Source: CDC
●Racial and Ethnic Disparities in Access to Meantal
Health Services
Patrick Rivers, Ph.D., MBA, Saundra Glover, Ph.D., Kai-Li Tsai,
Ph.D.
Presented By: Patrick Rivers, Ph.D., MBA, Associate Professor
& Director, Health Care Management, Southern Illinois
University Carbondale, CASA -6615, Carbondale, IL 629016615; Tel: (618)453-8842; Fax: (618)453-7020; Email:
privers@siu.edu
Research Objective: To assess and categorize the impact of
structural, financial and severity barriers on differential access
of mental health services among adults by race. This study
examined the influence of health insurance, availability of
community medical resources, transportation, and access to
care, on the racial and ethnic variations in utilization of mental
health services.
Study Design: We performed the following analyses: (1)
descriptive analysis to estimate population size and weighted
percent; (2) Chi-square analysis to examine the association
between mental health services and the explanatory variables;
(3) logistic regression analysis to show the pattern of mental
health service-seeking behavior relative to race and ethnicity;
and (4) stratified analysis to determine how factors differently
affect mental health service visits by race/ethnicity. The major
independent variable used in this model was race (i.e., White,
African American, Asian/Pacific Islander, and Hispanic). Selfreported mental health visits served as the dependent
variable.
Population Studied: The authors conducted secondary data
analysis from the 1998-1999 Community Tracking Study (CTS)
household survey.
Principal Findings: Persons from the lower income
population make fewer visits to mental health care facilities
than people with higher incomes. Significant effects of
race/ethnicity remained when the data was examined for
those individuals diagnosed with depression.
Conclusions: Results of this study support the findings of the
2003 U. S. Surgeon General report documenting the unmet
mental health needs of minorities in America. Racial/ethnic
minorities face financial and structural barriers in accessing
mental health services.
Implications for Policy, Delivery, or Practice: Policies that
will extend mental health coverage to the working poor are
needed. Additional funding is needed to expand communitybased mental health clinics to improve access and quality of
care, particularly for minority populations.
Primary Funding Source: No Funding Source
●Outpatient Diabetes Care For Homeless and Housed
Veterens: More Equal Than Not?
Stefan Kertesz, M.D., MSc, Monika M. Safford, M.D.,
Stephanie A. Rose, M.D., Katri P. Palonen, M.D., Katharine
Kirk, Ph.D., Catarina I. Kiefe, Ph.D., M.D.
Presented By: Jewell H. Halanych M.D., MSc, Division of
Preventive Medicine, University of Alabama at Birmingham,
1530 3rd Avenue South MT608, Birmingham, AL 35294; Tel:
(334)444-5396; Fax: (205)935-7959; Email:
jhalanych@mail.dopm.uab.edu
Research Objective: National data suggest limitations in
homeless persons' access to ambulatory care. No studies,
however, compare utilization of ongoing care for homeless
and housed outpatients in health systems designed to serve
both groups. We used Veterans Health Administration (VHA)
data to compare primary care entry, and utilization of ongoing
care for homeless and housed VHA users with diabetes.
Study Design: The proportion of VHA users achieving 1 or
more Primary Care Visits (PC1) reflected entry to care.
Ongoing care endpoints were proportions with 4 or more
Primary Care Visits (PC4), and 1 or more Eye Clinic Visits
(Eye). We made unadjusted comparisons between homeless
and housed veterans using Chi-squared tests. The 3
multivariable logistic regressions adjusted for age, race, illness
burden (Diagnostic Cost Groups (DCG)), mental illness,
alcohol and drug abuse.
Population Studied: All patients in 2003 national VHA data
with >1 diabetes-related diagnosis were considered to have
diabetes (n=735,897). “Homeless” patients were those using
VHA homeless services or carrying an “absence of housing”
diagnosis (n=8855, 1.2%).
Principal Findings: Homeless persons were younger (mean
(sd) age 54 (9) vs. 67 (11) years, for homeless vs. housed,
respectively), had higher illness burden (DCG score 1.9 (1.6)
vs. 1.0 (1.3)), and higher proportions with alcohol abuse (43%
vs. 3%), drug abuse (36% vs. 1%) and mental illness (71% vs.
25%) (all p<.0001). While modestly fewer homeless (vs.
housed) patients achieved primary care entry (PC1) (92% vs.
97%), more of the homeless obtained ongoing care: PC4
(59% vs. 47%), Eye (56% vs. 53%) (all p<.0001). After
multivariable adjustment, the homeless had reduced odds
ratios (OR) for PC1 (0.68, 95% CI 0.63-0.75), somewhat
increased OR for PC4 (1.05, 95% CI 1.00-1.10) and modestly
reduced OR for Eye (0.89, 95% CI 0.85-0.93). Models
including all significant interactions did not qualitatively alter
the findings.
Conclusions: Homeless and housed diabetic VHA users had
high (>90%) proportions entering primary care, although the
homeless were moderately disadvantaged for this endpoint. In
contrast, we did not find major disadvantages to homeless
patients in measures of ongoing care. Our operational
definition for homelessness probably selected for individuals
with strong connections to VHA services.
Implications for Policy, Delivery, or Practice: While national
data indicate that not all homeless veterans obtain VHA
services, our study suggests the VHA is able to achieve nearparity in utilization of ongoing care for diabetic homeless
veterans who do. These findings should reinforce efforts to
connect homeless veterans to VHA care.
Primary Funding Source: VA, National Institute on Drug
Abuse
●Rural-Urban Disparities in Access to Care for CYSHCN: A
Tale of Two States
Virginia Sharp, MS, ABD
Presented By: Virginia Sharp, MS, ABD, Senior Research
Associate, Center for Children with Special Needs, Children's
Hospital & Regional Medical Center, 1100 Olive Way, Suite
500, Seattle, WA 98101; Tel: (206) 987-5311; Fax: (206) 9875741; Email: ginny.sharp@seattlechildrens.org
Research Objective: The primary objectives of this research
project were to (1) develop and test a methodology for
identifying rural-urban differences in access to care for
children and youth with special health care needs (CYSHCN)
using the National Survey of Children with Special Health Care
Needs (NS-CSHCN), and (2) evaluate rural-urban differences
in access to care for CYSHCN between two seemingly similar
states, Oregon and Washington.
Study Design: Zip code level Rural Urban Commuting Area
(RUCA) codes were linked to zip code of residence recorded
for all Oregon and Washington respondents to the NSCSHCN and grouped into four categories--urban, suburban,
large town, small town/rural. These RUCA groupings were
used to analyze variables related to access to care and unmet
service needs in the NS-CSHCN.
Population Studied: The population included all children
identified as CYSHCN in the NS-CSHCN residing in Oregon
(n=745) and Washington (n=746). CYSHCN were identified
using the CSHCN Screener, a set of five multi-part questions
that identify children as having special health care needs
based on the consequences of their health condition, rather
than a list of specific diagnoses.
Principal Findings: The NS-CSHCN includes information on
need for and receipt of 14 types of child health services and 4
family support services. At the state level, Oregon and
Washington look very similar, with no statistically significant
differences between the states for any of these 36 variables.
However, when the RUCA codes are used to disaggregate
each state's data, different patterns of access to care and
unmet needs appear for the two states. For child health
services generally, in Oregon children living in suburban areas
tend to be less likely to receive all needed routine preventive
care, dental care and vision services. In Washington, children
residing in large towns are more likely to have unmet child
service needs, especially for dental care, specialized therapies,
mental health care, and vision services. These same patterns
also hold true for family services. In Oregon, suburban
families have more unmet needs, while in Washington,
families in large towns have more unmet needs. Comparing
RUCA categories across the two states uncovers additional
interesting differences in CYSHCN care needs and patterns of
care.
Conclusions: Even though Oregon and Washington are
generally perceived to be very similar socially, politically,
geographically and in other ways, they are not similar in terms
of rural-urban disparities in access to care for CYSHCN. While
the state sample sizes in the NS-CSHCN were too small to
produce statistically significant differences for most access to
care variables, those rural-urban disparities that were
significant exhibited distinctly different patterns in the two
states.
Implications for Policy, Delivery, or Practice: This analysis
indicates that the two states need to focus their resources for
CYSHCN services differently to reduce access disparities
related to rural-urban locational differences. More importantly,
it highlights the need for more research on both intra- and
inter-state patterns of health care resource distribution and
access to those resource.
Primary Funding Source: Oregon & Washington
Departments of Health, CSHCN Programs
●Out-of-Pocket Health Spending Between Low- And
Higher-Income Populations: What Factors Influence
Financial Burden?
Yu-Chu Shen, Ph.D., Joshua R. McFeeters, MPP
Presented By: Joshua R. McFeeters, MPP, Graduate School of
Business and Public Policy, Naval Postgraduate School, 555
Dyer Road, Code GB, Monterey, CA 93943; Tel: (831) 656-2951
Research Objective: In this paper, we explore the
characteristics of adults and their families with low, moderate,
and high financial burden from out-of-pocket health care
spending. We focus on the following questions: 1. What are
the social demographic determinants of financial burden for
out-of-pocket spending? 2. Does private or public insurance
coverage protect individuals from having a high burden of outof-pocket spending? 3. Does local health care market
infrastructure help to reduce high out-of-pocket burden?
Study Design: We obtained data on out-of-pocket health care
spending, income, and other demographic and health
insurance information primarily from the 2002 National
Survey of America’s Families (NSAF). Health care financial
burden was defined as the ratio of out-of-pocket health care
spending to family income, and grouped into low, moderate,
and high categories. Descriptive analyses compared the
distribution of financial burden and demographic
characteristics between low-income and higher-income
populations. A multinomial logit model was used to examine
the relative risk of having a high financial burden to a low
burden for given individual and area characteristics. Separate
multinomial logit models were done for low-income and
higher-income populations because of the different spending
patterns of the two groups.
Population Studied: The population studied was 38,594 noninstitutionalized adults, ages 18 to 64 and their families who
maintained the same type of health insurance coverage
(employer-sponsored, private non-group, public or other) for
the previous twelve months. The population was divided into
two groups: low-income (income less than 200 percent of the
Federal Poverty Level, FPL) and higher-income (income
greater than or equal to 200% FPL).
Principal Findings: 1. The effect of having unhealthy family
members on financial burden is larger for higher-income
adults than for low-income adults. 2. Employer-sponsored
health insurance keeps the higher-income population from
being in the high financial burden group relative to the
uninsured higher-income population. However, low-income
adults with employer-sponsored health insurance have a
higher risk of being in the high financial burden group than
uninsured, low-income adults. 3. Adults and families living in
areas with high HMO enrollment have lower odds of high
financial burden. For the low-income population, Federally
Qualified Health Centers reduce the odds of high financial
burden.
Conclusions: 1. Low-income adults with family members in
fair or poor health may be less willing and more constrained
to spend on health care relative to higher-income adults with
family members in fair or poor health. 2. Low-income adults
and families with employer-sponsored coverage or all adults
and families with private, non-group coverage have better
health outcomes compared to uninsured adults and families,
but are not protected from high financial burden. 3. Higher
HMO enrollment and the presence of Federal Qualified
Health Centers reduce the risk of high financial burden.
Implications for Policy, Delivery, or Practice: It is important
to understand factors that are associated with increasing or
decreasing risks of high out-of-pocket health burden for adults
and their families. Identifying these factors would help to
more effectively target policies that reduce the financial
burden of health care for both low-income and higher-income
families.
Primary Funding Source: RWJF
●Who are the Patients? Collection of Race, Ethnicity and
Primary Language across U.S. Hospitals
Donna Sickler, MPH, Marsha Regenstein, PhD
Presented By: Donna Sickler, MPH, Research Analyst,
National Public Health and Hospital Institute, 1301
Pennsylvania Avenue NW, Suite 950, Washington, DC 20004;
Tel: (202)585-0100; Fax: (202)585-0101; Email:
dsickler@naph.org
Research Objective: Previous studies have attempted to
determine the extent to which hospitals systematically collect
information on patients’ race and ethnicity, but low response
rates have impeded efforts to understand national practices.
This study uses a more aggressive research methodology to
survey acute care hospitals to develop estimates of data
collection practices vis-à-vis race, ethnicity, and primary
language. The study also examines operational aspects of data
collection and the ways that race/ethnicity data is currently
being used by clinical and administrative hospital staff. Finally,
it identifies common barriers to data collection efforts.
Study Design: Senior leaders from a representative sample of
500 acute care hospitals in the United States were surveyed by
telephone in January/February 2005. The survey methodology
and instrument were designed based on findings from an
informal telephone survey of two dozen hospital CEOs who
provided input into the content, proposed length, and
preferred respondent for a successful response rate. Based on
the CEO input, we identified chief financial officers and chief
information officers to be the preferred respondents for the
survey. The representative sample was drawn from the
American Hospital Association database of U.S. hospitals.
Findings were weighted to reflect the true distribution of
hospitals by location in terms of urban/rural and census
divisions.
Population Studied: Acute care hospitals in the United
States.
Principal Findings: (Findings available March, 2005) Early
evidence suggests that the majority of U.S. acute care
hospitals collect some information on the race and ethnicity of
patients, although the practices are highly variable and
inconsistent within and across organizations. Fewer hospitals
collect information on patients’ primary language, and only a
small number of hospitals use this information in quality
improvement efforts. Hospitals identified a number of barriers
that serve as impediments to data collection efforts.
Conclusions: Despite significant interest in disparities in
health care both within and across institutions, current
practices suggest that hospitals are only in their infancy in
terms of data collection practices. Clearer direction from
hospital leadership could substantially expand data collection
efforts, which are often technologically possible but are not
considered organizational priorities.
Implications for Policy, Delivery, or Practice: Opportunities
exist to improve data collection efforts to develop more robust
sources of information about the quality of care received by
individuals of different races and ethnicities. Given the
nation’s strong interest in improving health information
technology (HIT) and the overwhelming evidence of the
existence of disparities in nearly all aspects of health care
delivery, policies that encourage better and more consistent
data collection, reporting, and analysis by race/ethnicity,
within the context of improvements in HIT, could have a
direct and immediate impact on the quality of care for all
patients.
Primary Funding Source: RWJF
●Differences in Utilization and Access Among Latinos and
Asians in Alameda County
Paula Song, MHSA, MAE, Catherine G McLaughlin, Ph.D.
Presented By: Paula Song, MHSA, MAE, Doctoral Student,
Department of Health Management and Policy, University of
Michigan, 555 South Forest Street, Ann Arbor, MI 48105; Tel:
(734)647-9604; Fax: (734)998-6341; Email:
phsong@umich.edu
Research Objective: While there have been numerous studies
studying observed health disparities between Latinos and
Whites and between Asians and Whites, there has been
relatively little research investigating disparities between
Latinos and Asians, particularly among the immigrant
population. This paper helps fill that gap by examining
differences in perceived health status and access, as well as
self-reported utilization, among Latino and Cantonesespeaking Asian immigrants participating in a demonstration
program in Alameda County, California.
Study Design: We conducted a multi-wave telephone survey
of enrollees in the Alameda Alliance Family Care health plan,
part of the Robert Wood Johnson Foundation’s Communities
in Charge (CIC) program. Surveys were conducted in English,
Spanish, and Cantonese at three different times, at enrollment
and at 6 and 12 months after enrollment. We analyze these
data, comparing demographic, access, and utilization data at
baseline and again after one year in the program.
Population Studied: From August 2001 to January 2002, all
new enrollees in the FamilyCare program were contacted.
Over 80% participated. The sample for this research consists
of 302 Asian and Latino adults who responded to all three
waves of the survey. As required by the program, all adults
were low-income uninsured parents of children who
participated in Medi-Cal (S-CHIP). The two primary sources of
enrollees were an Asian and a Latino health clinic. Reflecting
this recruitment strategy, 97% of the sample population are
immigrants.
Principal Findings: Asian and Latino enrollees differed along
several demographic characteristics. On average, Asians were
older, more educated, and had higher household incomes
than Latinos. Asians were significantly less likely than Latinos
to have experienced a physician, emergency department (ED),
or inpatient visit, and receive several preventive services.
Controlling for individual characteristics, Latinos were more
likely to utilize the ED. In addition, Latinos were significantly
more likely to report more unmet needs and to have a usual
source of care, both of which were significant predictors of ED
use. Utilization levels for all services, except ED visits,
increased significantly for both groups after enrollment.
Asians continued to use the ED at significantly lower levels
than Latinos, but differences in other types of utilization were
no longer statistically significant. The percent of Asians
reporting unmet needs did not change after enrollment,
although the number of those reporting fair/poor health and
having a chronic condition did increase.
Conclusions: Prior to enrollment, Asians and Latino
immigrants had different utilization patterns and perceptions
of access to health services. After participating in a managed
care program, these differences in utilization patterns were no
longer significant, with the exception of ED visits. However,
Latinos continued to report having more unmet needs,
despite having similar utilization levels and fewer chronic
conditions than Asians, suggesting more subtle cultural
differences may account for differences in perception of
access to care.
Implications for Policy, Delivery, or Practice: Aggregating
immigrants or ethnic groups may result in ineffective
programs targeted at low-income uninsured immigrant
groups. Programs that do not take into consideration
differences in utilization and perceptions may not design the
“right” program to address the needs of the uninsured and
immigrant population.
Primary Funding Source: RWJF
●Outcomes for Rural Patients who Travel for Care
Sharon Sweeney Fee, RN, Ph.D.
Presented By: Sharon Sweeney Fee, RN, Ph.D., Assistant
Professor, College of Nursing, Montana State UniversityBozeman, P.O. Box 173560, Bozeman, MT 59717; Tel: (406)
994-2705; Fax: (406)994-6020; Email: sfee@montana.edu
Research Objective: The purpose of this research was to look
at the impact of distance on hospital outcomes for rural
patients using risk adjustment to understand the relationship
between distance and risk adjusted patient health outcomes.
Study Design: Regression analysis was used to identify
variations in outcome between groups of patients who
traveled between 30 and over 800 miles for care. The
outcomes measured included number of procedures and
length of stay. Patient characteristic used in the study included
payer, type of admit and source of admit and hospital size,
diagnosis related group (DRG), type of service, the number of
therapeutic and diagnostic services, and the number of
secondary diagnoses.
Population Studied: Using GIS technology, one year of
Arizona hospital discharge data was analyzed and resulted in
20,926 cases of rural patients who traveled over 30 miles for
health care between July of 2001 and June of 2002.
Principal Findings: When controlling for age, distance was
significant at the p = .01 level for the outcomes of number of
procedures and length of stay. The negative Beta (-.03) in the
procedures analysis demonstrates that the farther the patients
travel, the fewer number of procedures they receive and the
greater the length of stay. Significance (p<.01) was also found
between a majority of the patient characteristics and both
outcomes demonstrating the variation in type of rural dweller
that travels for health care.
Conclusions: The results of this study demonstrate that new
characteristics may need to be part of rural health research
and federally funded projects. Both outcomes were
significantly related to distance traveled. These outcomes
demonstrate new opportunities to develop more effective and
efficient interventions for rural patients.
Implications for Policy, Delivery, or Practice: In 2005, the
IOM endorsed the overriding principle that “all rural
Americans should have access to the full spectrum of highquality, appropriate health care.” This study demonstrates one
method toward an improved understanding of health
outcomes for rural Americans.
Primary Funding Source: No Funding Source
●Educational Attainment And Cancer Mortality, 1959 2001: Patterns, Trends, And Pathways
Elizabeth Tarlov, Ph.D., Robert Kaestner, Ph.D., Richard
Warnecke, Ph.D.
Presented By: Elizabeth Tarlov, Ph.D., Post-doctoral Fellow,
Midwest Center for Health Services and Policy Research,
Hines VA Hospital, PO Box 5000 (151H), Hines, IL 60141; Tel:
(708)202-2254; Email: elizabeth.tarlov@med.va.gov
Research Objective: A substantial body of empirical research
evidence supports the existence of a causal link between
educational attainment and health but little is known about
the relationship between education and cancer. This study
examined associations between educational attainment and
cancer mortality in the United States between 1959 and 2001
and interpreted findings in light of changing technological,
policy, and cultural contexts. Examining the relationship in an
historical context can provide valuable insight into the
involved pathways which are incompletely understood.
Study Design: Census and mortality data on gender-race-agespecific subpopulations within states were used to estimate
the effect of educational attainment on cancer mortality in five
decennial time periods (1959-1961 to 1999-2001). Dependent
variables were deaths due to lung, colon, breast, and prostate
cancer and to all cancers combined. Estimates were obtained
in gender-specific models using negative binomial regression
and accounted for age, income, marital status, and timevarying state effects. Robust standard errors accounted for
clustering within states.
Population Studied: U.S. White and Black adults aged 45
years and older in the last four decades of the 20th century.
Principal Findings: In 1959-1961, education had positive
effects on lung and breast cancer mortality among women
and on all cancers combined, lung, colon, and prostate
cancers among men. There were no education effects in the
following three time periods among women. Positive but nonlinear effects for lung and breast cancer re-emerged in the
1999-2001 data. Among men, education had less positive
effects in later decades and in 1999-2001 education had a
negative effect on mortality due to prostate cancer and to all
cancers combined.
Conclusions: Education effects varied by gender, cancer site,
and over time and were often nonlinear. Based on theory and
existing research evidence, it was hypothesized that time
trends in education effects would be negative and that
education effects on cancer mortality in 2000 would be
negative and graded. However, time trends in education
effects, although generally negative between 1960 and 1970,
were characterized by a variety of patterns and changes in
direction in subsequent years. Further, there was no evidence
of a gradient in education effects on cancer mortality. Trends
in the correlation between education and proximate
determinants of cancer or cancer outcomes (for example,
health behaviors and utilization of early detection services)
may explain some of the patterns observed but the
mechanisms that produced others remain unclear.
Implications for Policy, Delivery, or Practice: The results
suggest that education effects on cancer do not parallel those
found for all-cause and cardiovascular disease mortality and
for other measures of health and functioning such as selfrated health. Further refinement of theoretical pathways
linking societal context, educational attainment, and cancer
will be important to future progress in understanding these
relationships and reducing social and economic disparities in
health. Finally, research on the relationship between social
position and cancer is hindered by the absence of
socioeconomic information in large cancer datasets. An
additional contribution of this study, therefore, is its
demonstration of the advantages and limitations of using
non-cancer datasets within a sound research design to
address questions related to education and cancer.
Primary Funding Source: Pre-doctoral traineeship
●A Comparative Evaluation of the Cost Effectiveness of
Treating the Metabolic Syndrome in the African American
and General Population
Joseph Tasosa, MS, MBA, Richard Schuster, M.D., MMM,
John McAlearney, Ph.D.
Presented By: Joseph Tasosa, MS, MBA, Research Assistant,
Community Health, Health Systems Management, Wright
State University, 3139 Research Boulevard, #205, Kettering,
OH 45420; Tel: (937) 258-5555; Fax: (937) 258-5544; Email:
joseph.tasosa@wright.edu
Research Objective: To assess the cost effectiveness of early
treatment of three Metabolic Syndrome (MS) risk factors in
the African American and general population. To compare the
cost-effectiveness of early treatment of three MS risk factors in
the African American and general population.
Study Design: A cost effectiveness analysis was carried out
using a Markov decision model to compare early treatment
and late treatment of three MS risk factors (hyperlipidemia,
diabetes and hypertension) in African Americans and the
general population. The main outcome measure was the
incremental cost per Quality Adjusted Life Year (QALY).
Population Studied: African American population and the
general population in the United States.
Principal Findings: With the exception of early treatment of
hyperlipidemia in African Americans ($187,462/QALY), early
treatment of individual MS risk factors at age 30 was found to
be cost effective (<$27,000/QALY) for both African Americans
and the general population. The incremental cost of treating
hyperlipidemia, diabetes and hypertension simultaneously in
African Americans and the general population at age 30 was
$53,140/QALY and $63,926/QALY respectively. With the
exception of treatment of hyperlipidemia, early treatment
strategies targeted at African Americans were found to be
more cost effective than those targeted towards the general
population. Sensitivity analyses indicated that age and cost of
treatment were the most influential factors in the model. The
cost effectiveness of early treatment of MS risk factors in
blacks and the general population compares favorably with
similar health care interventions.
Conclusions: Early treatment of MS risk factors saves lives in
a cost effective manner. The study supports a growing body of
literature that indicates the cost effectiveness of providing
preventative services to apparently healthy individuals. Our
study also shows that disparate access to quality health care
makes African Americans especially susceptible to the adverse
effects of MS. We show that reducing these racial disparities
by providing early treatment is cost effective.
Implications for Policy, Delivery, or Practice: Proactive
health policies that provide preventative services early are
more beneficial and more cost effective than reactive policies
that provide services later. Proactive policies may help reduce
health disparities between African Americans and the general
population.
Primary Funding Source: No Funding Source
●Barriers to Vaccination Against Influenza in CommunityDwelling Veterans With Dementia: The Role of Caregiver
Depression
Joshua Thorpe, MPH, Betsy Sleath, Ph.D., RPh, Courtney
VanHoutven, Ph.D., Lawrence Landerman, Ph.D., Elizabeth
Clipp, Ph.D., RN
Presented By: Joshua Thorpe, MPH, Doctoral Candidate,
Pharmaceutical Policy and Evaluative Sciences, UNC - Chapel
Hill, 323 Columbia Place East, Chapel Hill, NC 27516; Tel:
(919) 928-8757; Email: thorpej@unc.edu
Research Objective: The primary aim of this study is to
examine the association between informal caregiver
depressive symptoms and access to annual influenza
vaccination in community-dwelling elderly with dementia. We
hypothesize that greater levels of caregiver depression are
associated with decreased likelihood of patient vaccination. A
secondary aim of this study is to determine whether
depression mediates the relationship between other
predisposing, enabling, and need variables and access to flu
shots.
Study Design: Independent variables were collected as part of
the first wave of the National Longitudinal Caregiver Sample, a
longitudinal investigation of primary informal caregivers of
elderly (age 60 and older) male U.S. veterans with dementia.
The NLCS identified a national sample of veterans with at
least one outpatient visit for dementia to a VA facility in 1997.
Independent variables, collected in early 1998, were used to
predict veterans’ receipt of flu shots during the 1998 influenza
vaccination season. We examined VA Outpatient Data Files
for visits occurring between September 1 and December 31,
1998 with a clinic stop code or CPT code indicating flu shot
administration. Our primary independent variable was
caregiver depressive symptoms as measured by the Center for
Epidemiologic Studies Depression Scale short Boston form.
The CES-D scale ranges from 0-20, with higher scores
indicating greater depressive symptomatology. The selection
of other independent variables was guided by the Andersen
behavioral model of health service use. Predisposing variables
included caregiver (CG) age, CG education, CG race, CG
relationship to veteran, and veteran age. Enabling variables
included CG perceived financial adequacy, veteran insurance
coverage, CG physical health, CG social support, and region of
country. Medical need variables included veteran behavioral
disturbance, duration of dementia symptoms, and comordid
conditions. A structural equation model was used to estimate
direct and indirect path coefficients between independent
variables, caregiver depression, and veteran vaccination
against influenza.
Population Studied: 1,424 community-dwelling male veterans
with dementia and their informal caregivers. Veterans less
than 65 years of age, those with no VA outpatient visits in the
previous year, and those who died or entered a nursing home
anytime during the flu shot season were excluded.
Principal Findings: Results support our hypothesis that
greater caregiver depressive symptoms are associated with a
decreased likelihood of dementia patient vaccination against
influenza. Further, CG depressive symptoms mediated the
relationship between several predisposing, enabling, and need
variables and the likelihood of vaccination. The following
independent variables had a significant indirect effect on
likelihood of vaccination via their influence on caregiver
depression: CG social support, CG physical health, CG
perceived financial adequacy, veteran depressive symptoms,
and veteran level of irritability/aggression.
Conclusions: We found that informal caregivers' depressive
symptoms decreased the likelihood of vaccination against
influenza in community-dwelling elderly with dementia.
Results also showed that several independent variables exert
their effects on likelihood of vaccination by influencing
caregiver depression.
Implications for Policy, Delivery, or Practice: Access to
medical care in community-dwelling dementia patients may
be improved directly by detecting and treating caregiver
depression, and indirectly by addressing pathways to caregiver
depression.
Primary Funding Source: AHRQ, American Foundation for
Pharmaceutical Education
●Do Black Patients Still Have Lower 30-day Mortality Than
White Patients Within VA?
Kevin Volpp, M.D., Ph.D., Roslyn Stone, Ph.D., Aashish Jha,
M.D., Judith Lave, Ph.D., Mark Pauly, Ph.D., Heather Klusaritz,
MSW
Presented By: Kevin Volpp, M.D., Ph.D., Assistant Professor
of Medicine and Healthcare Systems, School of Medicine and
the Wharton School, University of Pennsylvania, 1232 Blockley
Hall, 423 Guardian Drive, Philadelphia, PA 19104; Tel: (215)
573-0270; Fax: (215)573-8778; Email:
volpp70@mail.med.upenn.edu
Research Objective: Several studies have shown that Black
patients hospitalized within the VA system have lower 30 day
mortality than White patients. We examined whether these
differences (1) exist for conditions for which hospital
admission is generally not discretionary; (2) persist over time;
and (3) are due to within hospital differences in outcomes as
opposed to a non-random distribution of patients between
facilities of differing average outcome.
Study Design: We conducted an observational study using VA
data in the National Patient Care Database on 30-day mortality
for six conditions over a 7-year period. We used random
effects logistic regression to assess differences in 30-day
mortality between Black and White patients hospitalized with
each of these conditions, with hospital site as the random
effect and covariate adjustment for age, gender, fiscal year,
and 30 comorbidities. We assessed interactions involving race
and time, and assessed whether the racial disparities varied
across hospitals by including a random coefficient for race.
Population Studied: 340,982 Black or White patients
hospitalized with one of the following conditions at VA
hospitals nationwide between fiscal years 1996 and 2002:
acute myocardial infarction, congestive heart failure,
gastrointestinal bleeding, hip fracture, pneumonia, or stroke.
Principal Findings: For each condition, Black patients had
consistently lower 30-day mortality than White patients, with
an estimated odds ratio (OR) of 30-day mortality of 0.84 (95%
CI: 0.76-0.92) for acute myocardial infarction, OR=0.71 (95%
CI: 0.66-0.76) for congestive heart failure, OR=0.89 (95% CI:
0.81-0.98) for gastrointestinal bleeding, OR=0.66 (95% CI:
0.53-0.82) for hip fracture, OR= 0.93 (95% CI: 0.88-0.99) for
pneumonia, and OR=0.89 (95% CI: 0.82-0.97) for stroke
(p<0.02 for each). The ORs of black to white mortality did not
change substantially when adjustments for site were added,
indicating that the overall racial difference in outcomes is
predominantly due to within-site differences in outcomes
between Whites and Blacks. The hypothesis that the ORs of
black to white mortality were consistent in each year between
1996 and 2002 could not be rejected for any condition
(p>0.26 for each condition).
Conclusions: Black patients had lower 30-day mortality than
White patients for six common conditions treated in VA
hospitals between 1996 and 2002. The degree of reverse
disparity appears to reflect within-site differences in outcomes
by race rather that the possibility that Blacks may be treated in
hospitals with lower mortality rates and was consistent over
this time period for all conditions.
Implications for Policy, Delivery, or Practice: Lower
mortality rates exist for blacks than whites within the VA
system. Determining the degree to which this is due to
selection as opposed to treatment effects is important to
understanding how well the VA is treating health disparities.
Primary Funding Source: VA
●Marital Status and Cancer Screening Disparities among
Older Latina and Non-Latina Women
Melanie Wasserman, Ph.D, Melissa A. Clark, Ph.D., William
Rakowski, Ph.D.
Presented By: Melanie Wasserman, Ph.D, Post-doctoral
fellow, Center for Gerontology and Health Care Research,
Brown University, 2 Stimson Avenue, Box G-St311, Providence,
RI 02912; Tel: (401)863-9036; Fax: (401)863-3489; Email:
Melanie_Wasserman@brown.edu
Research Objective: Women who do not follow traditional
gender roles may be at a disadvantage in accessing preventive
health services. Unmarried status is often used as a control
variable in multivariate analyses of cancer screening, but has
rarely been studied in and of itself as a source of health care
disparities. This study examined disparities in utilization of
cervical, breast, and colorectal screening as well as availability
of a usual source of care by marital status, comparing Latinas
and non-Latinas.
Study Design: Cross-sectional analysis of 2003 Behavioral
Risk Factor Surveillance System (BRFSS) data. Bivariate
tabulations were done of marital status (married, never
married, divorced/ separated, widowed, member of unmarried
couple) against recent cervical, breast and colon cancer
screening and availability of a usual source of care.
Tabulations were stratified by ethnicity (Latina, non-Latina)
and language (Spanish, English). Multivariate analyses were
used to control for confounders (age, insurance, having a
regular doctor, delayed care due to financial considerations,
health status). Analyses were adjusted for sampling weights
and survey design.
Population Studied: Married and unmarried women between
the ages of 40 and 75 who were interviewed in English or
Spanish for the 2003 BRFSS. Analyses of cervical cancer
screening were restricted to women who had not had a
hysterectomy. Analyses of colorectal cancer screening were
restricted to women over 50.
Principal Findings: For all three populations (non-Latinas,
Latinas and Spanish-speaking Latinas), access to a primary
care doctor is highest for married women, followed by the
widowed, divorced or separated, and never married, with the
greatest disparity affecting members of unmarried couples.
There was a 30 percentage point gap between married nonLatinas and Spanish-speaking Latinas who were members of
an unmarried couple (92% vs. 62%). Cervical cancer
screening rates were high (between 80 and 85%) for married
women in all three groups, with only slightly lower rates
(between 77 and 82%) for the unmarried. Colon cancer
screening rates were low for all groups (44% for all groups
combined). Latinas, particularly Spanish speakers, remained
at a disadvantage in breast and colon cancer screening relative
to non-Latinas, and being married remained the category with
the highest utilization of cancer screenings for all three
subpopulations. Marital status operated differently for breast
and colon cancer screening, as well as for Latinas vs. nonLatinas. Members of unmarried couples had the lowest colon
cancer screening rates among all three subpopulations, but
Latina members of unmarried couples had the highest breast
cancer screening rates of any group (66%). Disparities
persisted after controlling for confounding factors.
Conclusions: Unmarried women, both Latina and non-Latina,
are less likely than their married counterparts to use cancer
screening services. Marital status operates differently among
Latinas and non-Latinas.
Implications for Policy, Delivery, or Practice: Providers of
cancer screening services should target their outreach efforts
to unmarried women. Further research is needed to better
design outreach strategies specific to never married, divorced,
widowed, or separated women. The effect of marital status on
cancer screening is different among Latinas, and should be
studied separately for this population.
Primary Funding Source: AHRQ,
●Disability and Clinical Preventive Services
Wenhui Wei, Ph.D., Patricia Findley, DrPH, MSW, Usha
Sambamoorthi, Ph.D.
Presented By: Wenhui Wei, Ph.D., Assistant Reserach
Professor, Institute for Health, Health Care Policy, and Aging
Reserach, Rutgers University, 30 College Avenue, New
Brunswick, NJ 08901; Tel: (732)932-0365; Fax: (732)932-8592;
Email: wwh@rci.rutgers.edu
Research Objective: Individuals with disabilities constitute
21% of the overall U.S. population and 53% of individuals with
disabilities are women. Low utilization of clinical preventive
services among people with disabilities may lead to poor
health and decreased independence and hamper the fullparticipation of these individuals in the society. The objective
of the study is to examine patterns of clinical preventive
services use among adults with disabilities and compare these
patterns to adults without disabilities.
Study Design: Secondary data analyses of longitudinal
household component of the Medical Expenditure Survey
(MEPS) for the calendar years 1999 through 2001. Disabled
individuals were identified as those who had difficulty in any
area of activity of daily living as a result of their impairment or
problems in body function, including psychological functions
for two calendar years. Recommended preventive health
services consisted of cervical and breast cancer screening for
women, colorectal, cholesterol screening and annual influenza
immunization for both men and women. Group differences
were tested with chi-square statistic. Multiple regressions were
used to examine patterns of clinical preventive services use.
Population Studied: Study sample included 3,067
community-dwelling individuals who were aged 51 to 64,
followed for two years, and alive at the end of follow-up
period.
Principal Findings: Overall, 21% of the study sample reported
having a functional limitation for two calendar years.
Among disabled individuals, 77% reported receiving
mammogram, 75% Papanicolaou (Pap) test, 88% received
cholesterol screening within the recommended period and
45% reported annual influenza immunization. Only a third of
the study sample reported ever receiving colorectal cancer
screening. Multivariate regressions revealed that disabled
individuals were more likely to receive annual influenza
immunization and less likely to receive Pap tests. We did not
find significant differences in receipt of other preventive
services. Among disabled individuals significant demographic
and socioeconomic differences were found.
Conclusions: Disabled women were less likely to receive
some of the cancer screening services suggesting barriers to
accessing health care services for women with disabilities.
Some of the clinical preventive services care remained below
targeted goals for Healthy People 2010.
Implications for Policy, Delivery, or Practice: Interventions
to promote cervical cancer screening among women with
disabilities are needed.
Primary Funding Source: Institute of Child Health and
Disability. Medical Rehabilitation
●Testing Theories of Discrimination in Health Care:
Evidence from New York’s CABG Report Card
Rachel Werner, M.D., Ph.D.
Presented By: Rachel Werner, M.D., Ph.D., Assistant
Professor, Department of Medicine, Univeristy of
Pennsylvania, Blockley Hall, Room 1208; 423 Guardian Drive,
Philadelphia, PA 19104; Tel: (215) 898-9278; Fax: (215) 5738778; Email: rwerner@wharton.upenn.edu
Research Objective: Evidence suggests that disparities in
health care utilization stem in part from discrimination in the
physician-patient interaction. Two theories have been
suggested to explain why this discrimination exists. The first
hypothesizes that physicians are biased against minority
patients and thus treat black and white patients differently
(also called taste-based discrimination). The second suggests
that imperfect information causes clinical uncertainty about an
individual patient’s treatment benefit. Therefore, physicians
categorize patients by racial or ethnic group and base
treatment decisions on the group’s average treatment benefit
rather than an individual’s treatment benefit. Also called
statistical discrimination, this theory suggests that disparities
exist because physicians believe that minority patients have
less treatment benefit on average compared to white patients.
The objective of this study is to test these two theories of
discrimination.
Study Design: In December 1991, New York began publishing
coronary artery bypass graft (CABG) report cards, which
publicly rate cardiac surgeons on their patients’ risk-adjusted
mortality rates following CABG surgery. Using this exogenous
shock, I test for changes in disparities in CABG use to test
these theories of discrimination. If racial and ethnic
differences in CABG use stem from clinical uncertainty, and
surgeons believe that racial and ethnic minorities will have
worse outcomes than white patients, surgeons will
preferentially treat white patients after report cards are
released, increasing racial and ethnic differences in CABG use.
Alternatively, if differences in CABG use are only due to tastebased discrimination, and clinical uncertainty plays no roll in
these differences, CABG report cards will not change racial
and ethnic differences in CABG use. To test these hypotheses,
I use multivariate regression to estimate differences in CABG
rates between white versus black and Hispanic patients in
New York before and after New York’s first CABG report card
was released, compared to rates of CABG surgery between
white versus black and Hispanic patients over the same time
period in 12 control states. I also estimate differences in
cardiac catheterization and percutaneous transluminal
coronary angioplasty (PTCA) rates and in-patient mortality
rates.
Population Studied: All patients hospitalized with a principal
diagnosis of acute myocardial infarction (AMI) were identified
in hospital discharge data from New York and the Nationwide
Inpatient Sample (HCUP-3) during the study period (1988 to
1995). A total of 928,551 patients were identified: 310,412 from
New York and 618,139 from the control states. Black and
Hispanic patients accounted for 13.7% of patients in New York
and 9.8% of patients in the control states.
Principal Findings: Racial and ethnic differences in CABG use
rose significantly in New York after New York’s CABG report
card was released compared to changes in racial and ethnic
differences in the control states. At the same time there were
no significant changes in the rates of cardiac catheterization
and PTCA. Despite the increase in racial and ethnic
differences in CABG use in New York, there was no detectable
change in in-patient mortality rates following AMI in New York
compared to the control states.
Conclusions: The increase in racial and ethnic differences in
CABG use following the release of New York’s CABG report
card is consistent with statistical discrimination as a cause of
racial and ethnic disparities in health care utilization, rather
than taste-based discrimination or bias. In this case,
statistical discrimination did not impact in-patient mortality
rates.
Implications for Policy, Delivery, or Practice: Future
interventions aimed to decrease disparities from
discrimination in the physicians-patient interaction should
address the mechanism of clinical uncertainty in statistical
discrimination.
Primary Funding Source: AHRQ
●What do the Disabled Report about their Experience with
Medicare? ADL Disability and CAHPS Measures among
Medicare Fee-for-Service Beneficiaries
Nathan West, MPA, Celia Eicheldinger, MS, Shulamit L.
Bernard, Ph.D., RN
Presented By: Nathan West, MPA, Health Research
Associate, Health Services and Social Policy Research, RTI
International, 3040 Cornwallis Road, RTP, NC 27709; Tel:
(919)541-6816; Email: nathanwest@rti.org
Research Objective: To examine patterns of reported
experiences and satisfaction with health care services among
Medicare Fee-for-Service beneficiaries with difficulty or
inability in performing activities of daily living (ADL). We
differentiate between disability as defined by Medicare
entitlement and reported disability as measured by ADL
functioning.
Study Design: Data come from the 2003 Medicare CAHPS
Fee-for-Service survey, a cross sectional, national survey of
Medicare beneficiaries that includes 107,893 survey
respondents. We exclude beneficiaries with ESRD and those
under 65 without disability as original reason for entitlement.
The regression models included two variables created from
survey responses and information provided from the
Enrollment Database: ADL limitation and Medicare
Entitlement. The ADL limitation variable has three levels: no
ADL difficulties, mild ADL difficulties, and severe ADL
difficulties. The Medicare Entitlement variable also has three
levels: greater than or equal to 65 and Aged into Medicare;
less than 65 and Medicare Entitlement by Disability; and
greater than or equal to 65 and Medicare Entitlement by
Disability. Case-mix variables race, ethnicity, insurance status,
and whether or not the beneficiary had a personal doctor were
also added to the regression models. All models were
weighted to reflect the full Medicare Fee-for-Service
population.
Population Studied: Beneficiaries enrolled in the Medicare
Fee-for-Service program during 2003.
Principal Findings: In general, we found a linear relationship
between extent of disability and CAHPS scores. Regardless of
a beneficiary’s original Medicare entitlement status, the
greater the level of difficulty with ADL functioning, the lower
Medicare beneficiaries rate the Medicare program and their
overall health care. We found this also to be true for
composite measures of beneficiaries’ experiences with health
services in Medicare, particularly whether or not they received
needed care and had good communication with their
providers.
Conclusions: Our findings suggest that the severity of a
disabling condition, as measured by ADLs, does account for
variation in expectations and satisfaction in the Medicare
program. Even among beneficiaries whose reason for
entitlement was disability, there was a linear trend between
functional disability and satisfaction and experience. While
prior work demonstrated that disabled Medicare beneficiaries
reported worse experience with their health care, this study
suggests that even among the disabled those with functional
disability report lower satisfaction and worse experience.
Implications for Policy, Delivery, or Practice: The findings
suggest that despite the insurance coverage offered by the
Medicare program, beneficiaries with ADL functional
limitations and potentially the greatest health care needs
experience more barriers and delays to obtaining health care
services.
Primary Funding Source: CMS
●Ensuring Low-Literate Adult Access to Online Health
Information: Developing a Research Program
Pamela Whitten, BSM, MA, Ph.D., Michael Mackert, BS, MA,
Brad Love, BS
Presented By: Pamela Whitten, BSM, MA, Ph.D., Associate
Professor, Telecommunication, Michigan State University,
409 Comm Arts Building, East Lansing, MI 48824; Tel:
(517)432-1332; Fax: (517)355-1292; Email: pwhitten@msu.edu
Research Objective: Researchers at Michigan State University
studied the potential for a website to deliver health
information to adults with low levels of literacy.
Study Design: MSU researchers created a website to provide
diabetes information to adults with low literacy levels. The
website was designed with Flash technology to provide the
health information primarily via images, animations, and
audio. The website performed like a slideshow, controlled
play, pause, and reverse controls. Even research participants
who had never used a computer before were able to
successfully navigate and control the website. Research
participants were tested for diabetes knowledge before and
after viewing the website to determine if the site increased
their knowledge of diabetes. Subjects also completed the
short version of the Test of Functional Health Literacy in
Adults (TOFHLA). To conclude their participation in the
study, subjects were interviewed in-depth to learn more about
their experience using the website.
Population Studied: Research subjects were recruited from a
variety of outlets to find adults with lower levels of literacy,
including community neighborhood centers and health
outreach support groups. Many of these individuals were
from lower income groups that would typically be eligible to
receive Medicaid support.
Principal Findings: Results show that subjects, in general,
improved their diabetes knowledge as a result of going
through the website. Interview results made it clear that
participants approved of and enjoyed the website, particularly
an interactive component that allowed subjects to assess their
possible risk of developing diabetes.
Conclusions: This pilot project helps to demonstrate that lowliterate populations can receive health information via
websites that are designed with this group in mind. Some
participants even said they prefer receiving information from
such a website, citing greater privacy and ease of access.
Further research to develop websites based on feedback
received from participants, as well as confirming the results
with other health information, will be a useful next step.
Implications for Policy, Delivery, or Practice: It is evident
that even adults with low levels of literacy can benefit from the
provision of health information online, and it is possible that
such material could be vastly more useful to certain adults
than traditional print materials that are used in patient
education. Further research needs to be conducted to learn
more about what works and what does not in using websites
to deliver health information to this underserved population,
but the potential cannot be ignored and could quickly become
a valuable tool for patient education.
Primary Funding Source: Medicaid Office in the State of
Michigan
●Racial Disparities in Pneumococcal Vaccination Among
Managed Care Seniors
Carla Winston, Ph.D., MA, Adrienne Mims, M.D., MPH,
Michael Blue, MHSA, Kecia Leatherwood, MS, Pascale
Wortley, M.D., MPH
Presented By: Carla Winston, Ph.D., MA, National
Immunization Program, Centers for Disease Control and
Prevention, 1600 Clifton Road NE Mailstop E-52, Atlanta, GA
30333; Tel: (404)639-6248; Fax: (404)639-8614; Email:
ctw3@cdc.gov
Research Objective: One-time pneumococcal vaccination is
recommended for all persons age 65 years and older, yet
marked racial/ethnic differences in pneumococcal vaccine
coverage exist. According to 2004 National Health Interview
Survey (NHIS) data for persons over age 64 years, 61.6% of
non-Hispanic Whites compared to 34.5% of non-Hispanic
Blacks have received pneumococcal vaccine. We surveyed
managed care patients to assess disparities in pneumococcal
vaccination and reasons for nonvaccination among older
African Americans and Whites in a managed care population.
Study Design: We randomly sampled from five general
medicine clinics 1198 Atlanta, Georgia HMO members over
age 64 years whose electronic medical records indicated they
were unvaccinated. Patients were asked, “Have you ever had
a pneumonia shot?” and if unvaccinated, “What would you
say is the main reason you have not had a pneumonia shot?”
Unvaccinated patients were told they were indicated for
pneumococcal vaccination and that vaccination was a covered
benefit of their managed care insurance with no copayment.
Unvaccinated patients were asked, “Would you like to get a
pneumonia shot?” and had the opportunity to schedule a
clinic visit for vaccination. Patients were asked to self-report
race, ethnicity, and education level, since these variables were
not in medical records.
Population Studied: 850 (71%) sampled patients were
surveyed by telephone in August 2004 regarding
pneumococcal vaccination and attitudes towards
pneumococcal vaccine. 61% of respondents were female;
48% were non-Hispanic Black; mean age was 72.2 years.
Principal Findings: 377 patients (44%) reported that they had
previously received pneumococcal vaccine. One quarter of
patients reporting prior vaccination had recently been
vaccinated after receiving a mailed reminder. Vaccinated
patients were older than the unvaccinated (73.2 v. 71.5 years, p
< 0.0001) but there was no significant difference in length of
HMO enrollment, gender, or educational attainment.
However, non-Hispanic Blacks were significantly less likely to
report prior vaccination than non-Hispanic Whites (33% v.
51%, p < 0.0001). Overall, 41% of the unvaccinated did not
know they needed a pneumococcal shot, 14% were concerned
the shot would make them sick, 14% did not want the shot,
and 9% stated they were not at risk for pneumonia, with no
significant differences by race. When the pneumonia shot was
offered, 63% of Whites stated they would like to get the shot
versus 55% of Blacks (p = 0.10). Black patients were more
likely than Whites to say they were unsure whether they would
like to get the shot (17% v. 9%, p = 0.03). The percentage of
patients who subsequently scheduled a clinic visit for
vaccination did not differ statistically by race (44% of Whites v.
37% of Blacks, p = 0.15).
Conclusions: Little is currently known about why racial
disparities exist in managed care populations with presumably
equal access to care. We will follow patients for six months to
determine which patients are vaccinated and what factors are
predictive of patients completing a scheduled clinic visit.
Vaccination outcomes through February 2005 will be
presented.
Implications for Policy, Delivery, or Practice: Data from this
study will be used to improve clinical outreach and education
to patients regarding pneumococcal vaccination.
Primary Funding Source: CDC, Kaiser Permanente, Atlanta,
GA
●What’s Behind the Observed Disparity of Health Care
Use and Expenditure?
Diana Wobus, Ph.D.
Presented By: Diana Wobus, Ph.D., Senior Analyst, Survey
Operations, Westat, 1650 Research Boulevard, Rockville, MD
20850; Tel: (240) 453-2967; Fax: (301) 294-2038; Email:
dianawobus@westat.com
Research Objective: This study is an investigation on
explanations for disparities observed in health care use and
expenses found in a previous study. In a study, the utilization
and expenditures of health care services was found different
among people of different incomes. As a follow-up, clinical
information on the study population was used to determine
whether or not the disparity results from different levels of
need for health care.
Study Design: The 2001 Medical Expenditure Panel Survey
data were used. The analysis unit is at person level. Linear
regressions were used to determine if the differences of use
and expenses among people of different incomes result from
health status or medical conditions for which health care
services were sought. The dependent variables are the
number of visits (for ambulatory care) or hospital nights (for
inpatient care) and the expenses (i.e. payments) associated
with the services. The independent variables are household
reported diagnosis (based on ICD-9 condition codes), selfreported health status in addition to a number of economic
and demographic characteristics.
Population Studied: The study population included the U.S.
uninstitutionalized population. MEPS collects data on health
care services from nationally representative samples.
Principal Findings: The null hypothesis is that health status
and conditions do not explain the disparity; thus the disparity
observed in the previous study results primarily from
economic or demographic characteristics, such as income or
race. The alternate hypothesis is that the disparity is a
function of health needs controlling for economic (resulting in
insurance disparity) or demographic characteristics. The
preliminary findings reveal that people of lower incomes used
the health care services less frequently than people of higher
incomes. However, for those people of lower incomes who
used the health care services, average payment for their
services tend to be higher for hospital services but lower for
office-based visits. Analyses of use and expenses based on
medical conditions and controlling for income levels and
insurance confirmed that a major portion of the disparity is a
direct function of the need for services as it relates to the
severity of health problems.
Conclusions: Claims on disparity of health care use and
expenditures based entirely on demographic characteristics
can be misleading if clinic information is ignored. By
controlling for need for health care, interpreted in diagnosis or
procedures, the study on disparity can reflect the real world.
Primary Funding Source: Westat
●Standardized Measurement and Assessment of Health
Disparities and Culturally and Linguistically Appropriate
Services in Managed Care – a Feasibility Study
Lok Wong, MHS, Sarah Hudson Scholle, MPH, DrPH,
Alexandra Love, MPH
Presented By: Lok Wong, MHS, Senior Health Care Analyst,
Quality Measurement, National Committee for Quality
Assurance, 2000 L Street, NW, Suite 500, Washington, DC
20036; Tel: (202)955-1784; Fax: (202)955-3599; Email:
wong@ncqa.org
Research Objective: This research study aims to explore and
identify feasible methods for monitoring and measuring
health disparities and the delivery of culturally and
linguistically appropriate services (CLAS) by managed care
organizations (MCOs) to diverse populations. The research
objective is to understand the current state of evidence
supporting common quality expectations in managed care
and inform the field on potential methods for standardized
measurement and assessment.
Study Design: We identified key concepts and areas for
potential evaluation and measurement based on a literature
review on health disparities and CLAS and through an expert
panel consensus process. We conducted qualitative key
informant interviews to understand the feasibility of proposed
evaluation areas identified and quantitative analysis of
Medicare HEDIS data by race/ethnicity and CAHPS patient
experience of care to examine the feasibility of reporting
standardized performance measures by race/ethnicity or
language.
Population Studied: Populations including commercial,
Medicare, Medicaid managed care. Interviews with managed
care plans, purchasers, researchers, and consumer/advocacy
groups.
Principal Findings: Topics for potential evaluation of
managed care organizations were identified under three broad
areas: Linguistic Access, Cultural Competence and Reducing
Health Care Disparities; specific topics included, infrastructure
for data collection and quality improvement, ability to conduct
subgroup analyses by race/ethnicity, providing effective
language access, culturally competent care at the point of
care, engaging communities and qualified interpretation
services. These topics and areas resonated with key
informants, including health plans, but operational definitions
of common quality expectations and evaluation approaches
need further refinement and flexibility to be feasible across all
plans and gain consensus in all stakeholder groups.
Conclusions: Managed care plans are interested in evaluating
the quality of services received by culturally, linguistically and
racially diverse populations and identifying opportunities for
quality improvement. Informing MCOs of approaches feasible
for monitoring the quality of health care services received by
diverse populations may furnish MCOs with needed tools to
begin to address and understand disparities and issues of
CLAS within their organization, such as understanding health
care needs of their diverse populations. Encouraging MCOs to
use standardized measurement and assessment tools to
assess potential disparities and services to diverse
populations may her facilitate our understanding of gaps in
quality of care received by diverse populations in managed
care.
Implications for Policy, Delivery, or Practice: Introduction
of standardized measurement and assessment of MCOs may
lead to accountability to the health of diverse populations,
and improve the quality of care and services delivered through
managed care.
Primary Funding Source: The California Endowment
●Improving Use of Prescription Medications
Helen Wu, M.Sc., Robyn Y. Nishimi, Ph.D., Christine M. PageLopez, Kenneth W. Kizer, M.D., MPH
Presented By: Helen Wu, M.Sc., Program Director, National
Quality Forum, 601 13th Street NW, Suite 500N, Washington,
DC 20005; Tel: (202)783-1300; Fax: (202)783-3434; Email:
hwu@qualityforum.org
Research Objective: Poor patient adherence to prescription
medication use instructions is both common and costly,
affecting the majority of U.S. healthcare consumers. Despite
attempts by many entities to address this problem, large-scale
improvement has not occurred. There is an acute need to
identify and widely deploy standardized metrics to improve
consumer use of prescription medications. This project
sought to identify a national action plan to increase consumer
adherence to recommended use of prescription medications,
focusing especially on populations with low literacy and
limited English proficiency.
Study Design: The project involved three elements: 1)
reviewing the literature to identify practices that have been
demonstrated to improve adherence; 2) developing a
comprehensive national framework for evaluating and
implementing practices to increase patient adherence; and 3)
holding a multi-stakeholder workshop to reach agreement on
a national action plan for broadly improving prescription
medication use.
Population Studied: All individuals using prescription
medications, with a special focus on those at increased risk of
non-adherence due to communication challenges with
providers - e.g., those with limited English proficiency or low
health literacy.
Principal Findings: There is a complex array of factors related
to non-adherence, and those that stem from unintentional
reasons (e.g., lack of patient understanding, language
barriers) are particularly well positioned for action and
improvement. Multiple evidence-based strategies exist to
effectively improve consumer adherence with the
recommended use of prescription medications, many of
which are centered on better communication between patients
and providers. Basic steps can be taken to address the needs
of patients with communication challenges, such as providing
medication labels and packaging inserts in multiple languages
and simplified verbiage. A wealth of data is also available that
could be used to facilitate provider follow-up to promote
improved adherence, such as prescription fill and refill
tracking. National standards can and should be promulgated
to identify a comprehensive set of practices and performance
measures for improving the communication and care
processes involving prescribers, pharmacists, and consumers,
which will in turn substantially improve adherence.
Conclusions: A combination of strategies is needed to
effectively improve the use of prescription medications at the
system, organization, and individual levels. Establishing
national consensus standards to address non-adherence,
based on existing evidence and data, will be an important first
step. The recommended national action plan calls for a broad
group of healthcare stakeholders to step forward in addressing
these critical issues.
Implications for Policy, Delivery, or Practice: Systematic
implementation of a comprehensive set of national standards
for improving medication use could have a substantial
positive impact on safety, efficiency, effectiveness, equity, and
other domains of healthcare quality.
Primary Funding Source: California Endowment
●Effects of Patient Obtained Medical Information (POMI)
on Healthcare Services Utilization in U.S. Counties
Bin Xie, Ph.D, David Dilts, Ph.D, MBA
Presented By: Bin Xie, Ph.D, Management of Technology,
Vanderbilt University, 801 Inverness Avenue Apartment B16,
Nashville, TN 37204; Tel: (615)322-8494; Email:
bin.xie@vanderbilt.edu
Research Objective: Patient obtained medical information
(POMI) has attracted considerable attention as the Internet
becomes a major source of medical information, but the
impacts of such POMI on healthcare utilization and equality
of healthcare are not clear. Would patients with POMI have
higher healthcare utilization than patients without POMI?
Would counties with higher portion of patients with POMI
have higher per capital healthcare utilization? What are the
implications for equality of healthcare? This paper addresses
these questions.
Study Design: We performed a cross-sectional analysis of
41485 individuals in 306 counties using data from the 20002001 Community Tracking Study Household Survey and the
2003 Area Resource File. Surgery rate and numbers of
physician visits are used as indicators for healthcare
utilization, and t-test, ANOVA and multiple regression are
used to analyze the impacts of POMI on healthcare utilization.
Population Studied: A national representative sample of
41485 individuals in 306 counties in the 2000-2001
Community Tracking Study Household Survey.
Principal Findings: Of 41485 individuals, 16296 (39%) have
POMI, 7190 (17%) sought POMI from the Internet. Of
patients with similar characteristics, patients with POMI have
higher surgery rate (for patients with excellent health, 0.18 vs.
0.11, p<0.001) and higher numbers of physician visits (for
patients with excellent health, 3.03 vs. 2.05, p<0.001) than
patients without POMI. In 306 counties, average portion of
patients with POMI is 37.5%, with a range from 9% to 71%. As
the portion increases in a county, surgery rate and numbers of
physician visits tend to increase as well, although not
monotonically. As the portion increases from 9% to 71%,
surgery rate increases from 0.18 to 0.29, although it remains
at the 0.22 level when the portion increases from 33% to 50%
(p<0.001). Numbers of physician visits show similar pattern,
increasing from 3.41 to 4.66 as the portion increases from 9%
to 71%, although it remains at the 4.00 level when the portion
increases from 29% to 50% (p<0.001). Surgery rate by
patients without POMI also increases in a county as the
portion of patients with POMI increases, although the trend is
not monotonic. It increases from 0.16 to 0.26 as the portion
increases from 9% to 71%, although it remains at the 0.19
level when the portion increases from 33% to 50% (p=0.07).
Numbers of physician visits by patients without POMI, on the
other hand, do not change much as the portion increases, as
it remains at the 3.40 level as the portion increase from 9% to
71% (p=0.79).
Conclusions: Per capital healthcare utilization in a county
increases as the portion of patients actively seeing medical
information increases, but the relationship between per capital
utilization and the portion of informed patients in a county is
not linear nor monotonic; per capital utilization increases
suddenly when the portion of informed patients crosses some
critical values, and remains stable otherwise. Further research
is needed to ethics, policy and practical implications of this
phenomenon.
Implications for Policy, Delivery, or Practice: Firstly, we
should pay more attention to traditional channels (such as
TV/Radio and Books/Magazines) in delivering medical
informatio to patients. Secondly, there is no evidence that
disparities in access to medical information are causing
disparities in healthcare utilization. Finally, differences in
access to medical information may contribute to the regional
variances in healthcare utilization.
Primary Funding Source: No Funding Source
●Regional Variations of Aggressive Medicare Treatments
in VA and Medicare Hospitals
Wei Yu, Ph.D., Ariel Hill, AB, Samuel Richardson, BA
Presented By: Wei Yu, Ph.D., Health Economist, Health
Economics Resource Center, Department of Veterans Affairs,
795 Willow Road, Menlo Park, CA 94025; Tel: (650) 493-5000
x23157; Fax: (650) 617-2639; Email: wyu2@stanford.edu
Research Objective: Researchers have observed a positive
relation between healthcare utilization and medical capacity,
suggesting that demand is induced by supply. However, we
know little about how much of the variation is explained by
other factors such as practice pattern. This study analyzes the
variation in aggressive medical treatments near the end of life
in the Department of Veterans Affairs (VA) and Medicare
hospitals. Because the VA budget is allocated based on the
number of patients, VA hospitals have little financial incentive
to induce demand for aggressive treatments. Thus, variations
within the VA and the difference between VA and Medicare
hospitals could reveal the impact of other factors on medical
treatments.
Study Design: We retrospectively extracted healthcare
utilization provided by the VA or Medicare hospitals during
the final 2 years of life and analyzed variations between 21 VA
integrated service networks (VISNs) in the utilization of five
aggressive medical treatments: intensive care unit (ICU),
mechanical ventilator, pulmonary artery monitor, cardiac
catheterization, and dialysis. We grouped Medicare hospitals
into the 21 VISNs by their location. We compared treatment
variations among acute hospital stays during the final 30 days,
the final year, and the second year before death with bivariate
and multivariate methods. We examined the regional effects
using a Probit regression model, controlling for
demographics, principal diagnosis, and severity of comorbid
conditions measured by the Charlson Comorbidity Index.
Population Studied: Veterans who died between October 1,
1999 and September 30, 2001, were over age 67 at death, and
used the VA system during their final 2 years of life
(N=169,314). We excluded people under 67 at death so that
all subjects were eligible for both VA and Medicare during the
study period.
Principal Findings: Use of aggressive treatments varied
considerably among the 21 VISNs in both VA and Medicare
hospitals. During hospital stays within the final 30 days of life,
ICU use varied from 23.3% to 47.5% in VA and from 31.9% to
50.5% in Medicare, ventilator use varied from 9.0% to 20.1%
in VA and from 14.3% to 24.5% in Medicare, pulmonary artery
monitor use varied from 1.4% to 6.5% in VA and from 1.5% to
2.9% in Medicare, cardiac catheterization use varied from
0.7% to 3.4% in VA and from 1.9% to 3.8% in Medicare, and
dialysis use varied from 2.5% to 5.6% in the VA and from
2.6% to 7.0% in Medicare. These variations persisted after
controlling for other factors and in all the three time periods in
regression analysis. The aggressiveness of care was not
consistent between Medicare and VA facilities in the same
region.
Conclusions: Regional practice patterns could significantly
affect the intensity of medical treatments and could be a
significant factor of the variation observed in the private
sector. The difference in practice patterns between VA and
Medicare hospitals suggests that financial incentives is also
involved in medical decisions.
Implications for Policy, Delivery, or Practice: Reducing
variation in medical treatments could enhance the quality of
care near the end of life. As practice pattern is also a
significant factor in medical decisions, evidence-based
guidelines may improve the quality of care.
Primary Funding Source: VA
●Ambulatory Care Sensitive Conditions, Community
Factors, and Urban Low-Income Adults: Adapting a
Conceptual Framework
Wanqing Zhang, M.D., Med, Keith Mueller, Ph.D.
Presented By: Wanqing Zhang, M.D., Med, Analyst,
Preventive and Societal Medicine, University of Nebraska
Medical Center, 984350 University of Nebraska Medical
Center, Omaha, NE 68198-4350; Tel: (402)559-8441; Fax:
(402)559-7259; Email: wazhang@unmc.edu
Research Objective: Preventable hospitalization for
ambulatory care sensitive conditions (ACSCs) has been the
subject of multivariate analysis efforts for a decade. However,
researchers currently lack a theoretical framework for much of
their work. Using a conceptual framework, we explored the
relationship between access outcomes and community factors
among low-income adults residing in metropolitan statistical
areas (MSAs), and we examined community-based measures
of appropriate use of the health care system among a lowincome population.
Study Design: Research has shown that individuals of lower
socio-economical levels and some community factors are
associated with a higher rate of hospitalization for ACSCs. But
how do policy and community intervention modify these
factors? The conceptual framework for the effects of
community factors on access to ambulatory care for lowincome populations has been suggested by Davidson and
colleagues. That framework divides health access and
outcomes into three categories: potential access, realized
access, and access outcomes. In this study, we examined the
relationship between access outcome (ACSCs) and
community determinants for urban low-income adults. We
used the framework as a basis for constructing community
variables hypothesized to influence access outcome. Hospital
discharge data for ACSCs was obtained from the Health Cost
and Utilization Project’s (HCUP) National Inpatient Database,
which is maintained by the Agency for Health Care Research
and Quality (AHRQ). Thirteen quality indicators for
preventable conditions, developed by AHRQ, were drawn from
PQI software. The community variables that reflect the
population difference were based primarily on data from the
U.S. Census Bureau’s Current Population Survey, and the
Bureau of Health Professional’s Area Resource File. The
financial support variable for safety-net services was from the
Center for Medicare & Medicaid Service’s Medicare cost
report.
Population Studied: We studied a large national sample of
U.S. hospitalized patients in the United States in 2002 - adults
18 or older with income <$25,000 in 33 MSAs.
Principal Findings: Hospital discharge for ACSCs varies
greatly among low-income adults from 33 selected MSAs
(from 3.4% to 45.5 %). For discharge-level factors, Black adults
had the highest odds for preventable hospitalizations due to
ACSCs after case-mix adjusting (OR=1.51, P<0.001). Being
uninsured also increased the odds for preventable
hospitalization due to ACSCs (OR= 1.20, P<0.001). As we
hypothesized, several community factors dramatically
increased the likelihood for preventable hospitalization. A
higher proportion of non-citizen immigrants and non-White
population increased the likelihood of preventable
hospitalization due to ACSCs by 5 and 10 times respectively.
Uncompensated care costs were inversely related to
preventable hospitalization due to ACSCs (OR=0.76,
P<0.001).
Conclusions: Guided by a conceptual framework, we found
that community and public policy factors were significantly
related to preventable hospitalization for ACSCs among urban
low-income adults.
Implications for Policy, Delivery, or Practice: The results
suggest that care management and other interventions (e.g.,
language assistance, cultural sensitivity and awareness) are
needed in communities with a higher proportion of noncitizen immigrants and non-White population. Meanwhile,
preventable hospitalization due to ACSCs as a measure of
health outcome is important in determining how to target
community-wide interventions for improving community
health and reducing health disparities.
Primary Funding Source: Departmental funds
Download