Evidence of disparity in the application of quality improvement efforts for the treatment of acute myocardial infarction: the American College of Cardiology's Guidelines Applied in Practice Initiative in Michigan.

Abstract

Racial disparities exist in the management of patients with cardiovascular disease in the United States. The aim of the study was to evaluate if a structured initiative for improving care of patients with acute myocardial infarction (Guidelines Applied in Practice [GAP]) led to comparable care of white and nonwhite patients admitted to GAP hospitals in Michigan.
Medicare patients comprised 2 cohorts: (1) those admitted before GAP implementation (n = 1,368) and (2) those admitted after GAP implementation (n = 1,489). The main outcome measure was adherence to guideline-based medications/recommendations and use of the GAP discharge tool. chi(2) and Fisher exact tests were used to determine differences between white patients (n = 2,367) and nonwhite patients (n = 490).
In-hospital GAP tool and aspirin use significantly improved for white and nonwhite patients. beta-Blocker use in hospital improved significantly for nonwhite patients only (66% vs 83.3%; P = .04). At discharge, nonwhite patients were 28% and 64% less likely than white patients to have had the GAP discharge tool used (P = .004) and receive smoking cessation counseling (P < .001), respectively. Among white patients, GAP improved discharge prescription rates for aspirin by 10.8% (P < .001) and beta-blockers by 7.0% (P = .047). Nonwhite patients' aspirin prescriptions increased by 1.0% and beta-blocker prescriptions decreased by 6.0% (both P values nonsignificant).
The GAP program led to significant increases in rates of evidence-based care in both white and nonwhite Medicare patients. However, nonwhite patients received less quality improvement discharge tool and smoking cessation counseling. Policies designed to reduce racial disparities in health care must address disparity in the delivery of quality improvement programs.

Full text

Page 1

Acute Ischemic Heart Disease
Evidence of disparity in the application of quality
improvement efforts for the treatment of acute
myocardial infarction: The American College of
Cardiology's Guidelines Applied in Practice Initiative
in Michigan
Adesuwa B. Olomu, MD, MS,aMary Grzybowski, PhD, MPH,b,cVijay S. Ramanath, MD,dAdam M. Rogers, MD,e
Bonnie Motyka Vautaw, MS,c,fBenrong Chen, PhD,cCanopy Roychoudhury, PhD,cElizabeth A. Jackson, MD,
MPH,eand Kim A. Eagle, MDefor the American College of Cardiology Foundation Guidelines Applied in Practice
Steering Committee East Lansing, Detroit, Farmington Hills, Ann Arbor, and Dearborn, MI; and Lebanon, NH
Background Racial disparities exist in the management of patients with cardiovascular disease in the United States. The
aim of the study was to evaluate if a structured initiative for improving care of patients with acute myocardial infarction
(Guidelines Applied in Practice [GAP]) led to comparable care of white and nonwhite patients admitted to GAP hospitals
in Michigan.
Methods Medicare patients comprised 2 cohorts: (1) those admitted before GAP implementation (n = 1,368) and (2)
those admitted after GAP implementation (n = 1,489). The main outcome measure was adherence to guideline-based
medications/recommendations and use of the GAP discharge tool. χ2and Fisher exact tests were used to determine
differences between white patients (n = 2,367) and nonwhite patients (n = 490).
Results In-hospital GAP tool and aspirin use significantly improved for white and nonwhite patients. β-Blocker use in
hospital improved significantly for nonwhite patients only (66% vs 83.3%; P = .04). At discharge, nonwhite patients were 28%
and 64% less likely than white patients to have had the GAP discharge tool used (P = .004) and receive smoking cessation
counseling (P b .001), respectively. Among white patients, GAP improved discharge prescription rates for aspirin by 10.8%
(P b .001) and β-blockers by 7.0% (P = .047). Nonwhite patients' aspirin prescriptions increased by 1.0% and β-blocker
prescriptions decreased by 6.0% (both P values nonsignificant).
Conclusions The GAP program led to significant increases in rates of evidence-based care in both white and nonwhite
Medicare patients. However, nonwhite patients received less quality improvement discharge tool and smoking cessation
counseling. Policies designed to reduce racial disparities in health care must address disparity in the delivery of quality
improvement programs. (Am Heart J 2010;159:377-84.)
Racial disparities in the treatment of patients with
cardiovascular disease in the United States have been
extensively documented.1Studies in various clinical
populations using both administrative and clinically
detailed registry data indicate that black patients are less
likely than white patients to receive coronary angiogra-
phy, percutaneous coronary interventions (PCIs), and
coronary artery bypass graft (CABG) after acute myocar-
dial infarction (AMI).2Blacks are also less likely to receive
thrombolytic therapy and other evidence-based medica-
tions such as aspirin, β-blockers, and lipid-lowering
therapy, after AMI.3,4
It has been suggested that the disparities in care may
be due to differential access to medical care, financial
constraints, communication gaps between patients and
health care providers, or medical decision making. All
of these could negatively affect the use of evidence-
From the
bDepartment of Emergency Medicine, Wayne State University, Detroit, MI,cMichigan Peer
Review Organization, Farmington Hills, MI,dSection of Cardiology, Dartmouth-Hitchcock
Medical Center, Lebanon, NH,eDepartment of Cardiology, University of Michigan Health
System, Ann Arbor, MI, andfGlobal Risk Management Group, Dearborn, MI.
Submitted December 23, 2008; accepted December 18, 2009.
Reprint requests: Adesuwa B. Olomu, MD, B329 Clinical Center, East Lansing, MI 48824.
E-mail: ade.olomu@hc.msu.edu
0002-8703/$ - see front matter
Published by Mosby, Inc.
doi:10.1016/j.ahj.2009.12.014
aDepartment of Medicine, Michigan State University, East Lansing, MI,

Page 2

based therapies and outcomes.5These differences
persist despite adjustments for potential confounding
variables such as age, sex, clinical factors, socioeco-
nomic status, and insurance status.6Recent studies have
focused on systematic differences in the quality of care
delivered by hospitals and health care providers where
white and nonwhite patients receive their care.7These
studies suggest that limited access to high-quality
providers and institutions may be an important cause
of disparities.
The American College of Cardiology's (ACC's) Guide-
lines Applied in Practice (GAP) initiative was designed to
reduce variation in patient care and to improve the care
of AMI patients. It brings the ACC/American Heart
Association practice guidelines to the point of care in
cooperation with the Michigan Peer Review Organiza-
tion, the Greater Detroit Area Health Council, and the
Greater Flint Health Coalition, a comprehensive hospital-
based quality improvement initiative that incorporates
ACC/American Heart Association national AMI guidelines
into daily practice.8-10The GAP program focuses on
implementation of clinical tools (eg, standard admission
orders and a standard discharge contract) while targeting
patients, physicians, and nurses with educational sessions
and a system for measurement.11The GAP program has
been shown to increase the use of evidence-based
medications at hospital arrival and discharge, as well as
smoking cessation counseling in participating hospitals in
Michigan.8,9,11
The GAP discharge tool helps to ensure that patients
are educated about their condition and the treatment/
follow-up plan after they leave the hospital. It is
typically completed and reviewed with the patient by
the discharging nurse or physician. It includes instruc-
tions on taking medications, goals for smoking cessa-
tion, goals for lipid treatment, instructions on diet and
exercise, heart disease education, recommendations for
rehabilitation programs, instructions for follow-up with
the patient's primary care physician, and what to do in
the event of recurring symptoms. The patient and the
physician or discharging nurse must sign the docu-
ment, confirming the patient's understanding of the
materials. The use of the discharge tool has been
shown to drive higher adherence to most discharge
quality indicator rates including aspirin use, angioten-
sin-converting enzyme inhibitor use, and smoking
cessation.12Furthermore, data have revealed that GAP
discharge tool use was associated with a significant
reduction in 1-year mortality in Medicare patients
with AMI.12
The success of the discharge tool lies in its ability to
ensure that evidence-based processes occur at discharge.
The tool helps to foster long-term adherence to evidence-
based medications and ensure a follow-up plan at
discharge.12Lappe et al13also found that a relatively
simple quality improvement discharge medication pro-
gram led to a relative risk reduction in death and
readmission at 30 days and 1 year after AMI.
The objectives of this study were (a) to evaluate
whether GAP would lead to comparable care for white
and nonwhite patients admitted to participating hospitals
in Michigan and (b) to determine whether hospital use of
the standard discharge contract differed by race after the
initiation of GAP.
Methods
The GAP projects
After the initial development of the AMI GAP program at the
University of Michigan Health System,9the subsequent GAP
projects included a 10-hospital study in Southeast Michigan in
1999,11a 5-hospital project in Flint and Saginaw in 2001,10and
an 18-hospital project in Southeast Michigan in 2002.8A rapid-
cycle quality improvement model was created, emphasizing a
collaborative culture of learning, sharing, and problem solving
among hospitals.8Care processes were designed to assure
clinical tool use.8The methods used in the GAP projects have
been described previously.8,11,12The clinical care tool tem-
plates, including standardized orders, a pocket guideline, and a
standardized discharge tool, are available at the ACC Web site
(http://www.acc.org/gap/gap.htm).
The Medicare sample
We examined the impact of GAP on Medicare beneficiaries
by studying patients treated in each hospital before and after
the implementation of GAP. Baseline samples were created
using a 50% random sample with at least 20 cases per hospital
of Medicare AMI patients (principal diagnosis code: 410.xx)
from patients treated in the year preceding GAP implementa-
tion. The post-GAP sample included a 95% to 100% sampling of
all Medicare AMI patients in the 4 months immediately after
GAP implementation at each hospital. Hospital records for
each patient were copied and forwarded to DynKePRO's
Centers for Medicare and Medicaid Services (CMS) Clinical
Data Abstraction Center. Each record was screened to ensure
that AMI was the principal diagnosis. Data regarding patient
history, presenting symptoms and signs, rates of comorbidity,
diagnostic studies, therapies, complications, outcomes, and
use of standard orders and/or the discharge tool were
extracted. To assure quality, data were reabstracted for a
random sample of baseline and post-GAP records by the CMS
Clinical Data Abstraction Center. Reliability (94%) and accura-
cy (97%) were high.
Variables
Study variables included demographics (age, sex, race),
medical history (heart failure, stroke, hypertension, diabetes,
chronic obstructive pulmonary disease, prior AMI, prior CABG
surgery, prior PCI), presenting clinical characteristics (smok-
ing status, chest pain, heart rate, systolic and diastolic blood
pressure, body mass index [BMI]), in-patient treatments
(admission tool use, echocardiography, cardiac catheteriza-
tion, PCI, CABG), and complications (hypotension, shock,
heart failure, stroke, renal insufficiency, hemorrhage), as
shown in Tables I to II. Age and systolic and diastolic blood
378 Olomu et al
American Heart Journal
March 2010

Page 3

pressures were analyzed as continuous variables; BMI and
heart rate were analyzed as continuous and categorical
variables, while BMI was assessed dichotomously (normal/
underweight|overweight/obese) and trichotomously (normal/
underweight|overweight|obese). Implementation rates of the
GAP admission and discharge tools were determined. Mea-
sured early and late indicators included aspirin and β-blocker
administration on admission, and aspirin, β-blocker, lipid-
lowering agents, and angiotensin-converting enzyme inhibitor
prescriptions, and smoking cessation counseling, at discharge,
respectively.
To evaluate differences in evidence-based indicators, the
indicator-specific analyses were limited to ideal patients. The
ideal patient was defined as one without absolute or relative
contraindications to respective indicators that were formu-
lated by CMS.14Patients with missing indicator data were
also excluded.
Statistical analysis
Point estimates for all variables were calculated by race (white|
nonwhite) and GAP status (pre-and post-GAP). Two comparisons
were conducted: (1) differences in characteristics between pre-
and post-GAP groups by race and (2) differences in characteristics
between races among post-GAP patients. Differences in mean
values of continuous variables were assessed between each of the
2 groups described above using either the Student t test for
parametric data or the Wilcoxon rank sum test for nonparametric
data. Categorical data were analyzed using χ2with Yate's
correction for continuity as indicated.15Fisher exact test was
used if the expected number of observations in any cell was b5.
Odds ratios and 95% CI were calculated for categorical bivariate
associations. StatXact 6.0 (Cambridge, MA) was used to calculate
exact CIs.16
A post hoc analysis was done to explore if differences in
quality of care were based on race. Therefore, we calculated the
overall proportion of post-GAP discharge contract use (a proxy
for quality performance) for the 33 participating hospitals.
Discharge contract use rates (determined by dividing the
number of AMI patients who received the discharge contract
by the total number of included AMI patients in each
participating hospital) were subdivided into tertiles for 2
reasons: (1) tertiles allow for easier interpretation of results
and (2) there is no literature reporting definitions of discharge
contract use. Tertile 1 included hospitals with a discharge
contract use rate b8.4%, tertile 2 consisted of hospitals between
8.41% and 38.0%, and tertile 3 included hospitals with a
discharge contract use rate of N38.0%. Within each tertile, we
used the χ2test to assess proportional racial differences. The
Cochran-Armitage trend test was used to determine if a linear
association existed between nonwhite patients and white
patients and discharge contract use tertile. Except where
indicated, the analyses were performed using SAS v 9.1.3 (SAS
Institute, Cary, NC). For all statistical analyses, a 0.05 level of
significance was used, and all P values were 2 sided.
Results
Patients' characteristics
The baseline and post-GAP characteristics and medica-
tion use of the 2,367 white patients and 490 nonwhite
Table I. Selected characteristics of nonwhite and white patients pre- and post-GAP⁎
Characteristic
White patient†
Nonwhite patient†
Post-GAP patients†
Pre-GAP,
n = 1158
(n [%])
Post-GAP,
n = 1209
(n [%]) P
Pre-GAP,
n = 210
(n [%])
Post-GAP,
n = 280
(n [%])P
Odds ratio
(95% CI)‡
Nonwhite vs
white patients, P
Demographics
Age, mean (SD), y
Male sex
Medical history
AMI
CABG surgery
lPCI
Heart failure
Stroke
Hypertension
Diabetes mellitus
COPD
Presentation characteristics
Current smoker
Chest pain
Heart rate, mean (SD), beats/min 88.9 (24.8)
SBP, mean (SD), mm Hg
DBP, mean (SD), mm Hg
Heart failure
BMI N 25 (overweight/obese)
77.0 (9.1)
610 (52.7)
77.0 (9.2)
635 (52.5) .9401
.9855 73.5 (11.5)
102 (48.6)
73.0 (11.2) .6412
138 (49.3) .8756 0.88 (0.68-1.14)
–
≤.001
.3286
482 (41.6)
240 (20.7)
190 (16.4)
403 (34.8)
240 (20.7)
845 (73.0)
418 (36.1)
308 (26.6)
539 (44.6) .1462
259 (21.4) .6776
223 (18.4) .1917
419 (34.7) .9411
250 (20.7) .9774
935 (77.3) .0139
417 (34.5) .4139
346 (28.6) .2717
92 (43.8)
29 (13.8)
24 (11.4)
81 (38.6)
40 (19.1)
178 (84.8)
88 (41.9)
48 (22.9)
141 (50.4) .1509 1.26 (0.97-1.64)
39 (13.9) .9699 0.59 (0.41-0.86)
43 (15.4) .2103 0.80 (0.56-1.15)
116 (41.4) .5233 1.33 (1.02-1.74)
76 (27.1) .0370 1.43 (1.06-1.92)
237 (84.6) .9711 1.62 (1.14-2.30)
132 (47.1) .2487 1.69 (1.30-2.20)
83 (29.6) .0943 1.05 (0.79-1.40)
.0805
.0047
.2242
.0333
.0184
.0071
≤.001
.7331
207 (17.9)
836 (73.7)
221 (18.3) .7985
886 (75.3) .3718
86.8 (24.0) .0423
48 (22.9)
135 (65.2)
90.2 (23.8)
65 (23.2) .9260 1.35 (0.99-1.85)
183 (68.3) .4812 0.71 (0.53-0.94)
91.9 (24.7) .7032
.0589
.0185
.0084
.0044
.0142
.8598
.6455
–
–
–
140 (31.4) 139.5 (323)
72.9 (21.2)
214 (18.5)
626 (60.5)
.6985 148.5 (39.1) 146.9 (39.8) .6543
73.4 (19.5) .5566 81.5 (22.8)
217 (18.0) .737749 (23.3)
699 (62.9) .2581103 (61.0)
77.1 (22.9) .0372
49 (17.5) .1101 0.97 (0.69-1.36)
140 (61.1) .9423 0.93 (0.70-1.25)
COPD, Chronic obstructive pulmonary disease; SBP, systolic blood pressure; DBP, diastolic blood pressure.
⁎Data are presented as number (percentage) unless otherwise indicated. Percentages are based only on nonmissing data.
†Percentage is based on ideal patients.
‡Represents the unadjusted magnitude of association between predictor variable (characteristic) and race where nonwhite is the reference variable.
Olomu et al 379
American Heart Journal
Volume 159, Number 3

Page 4

patients are shown in Tables I and II. Compared with
white patients, nonwhite patients were younger (P b
.001), more likely to have hypertension (P = .007),
diabetes (P b .001), stroke (P = .018), and a history of
congestive heart failure (P = .033) but less likely to have
had prior CABG surgery (P = .005), at baseline.
In the post-GAP cohort, nonwhite patients were on
average 4 years younger than white patients and were
less likely to have a history of CABG surgery and present
with chest pain. Moreover, nonwhite patients were
more likely to have histories of heart failure, hyperten-
sion, stroke, and diabetes and presented with higher
mean heart rates and systolic and diastolic blood
pressures than their white counterparts. Post-GAP
nonwhite patients, compared to pre-GAP nonwhite
patients, more often had a history of stroke and had
lower mean diastolic blood pressures.
Upon presentation, nonwhite patients had a higher
mean systolic blood pressure (P = .004) and diastolic
blood pressure (P = .014) and were less likely to have
chest pain (P b .019). There was a trend toward a higher
proportion of current smokers among the nonwhite
patients compared to white patients (P = .059). There
were no statistically significant differences in the
presenting electrocardiogram and left ventricular ejec-
tion fraction between nonwhite and white patients
(Tables I and II).
In-hospital procedures/quality indicator rates by race
and GAP status
Post-GAP white patients more frequently were admin-
istered the admission tool and aspirin and underwent
cardiac catheterization and PCI than those in the pre-
GAP cohort. Compared to white patients post-GAP,
nonwhite patients post-GAP more frequently received
the admission tool, aspirin, and β-blockers. In treatment,
post-GAP nonwhite patients were 44%, 56%, and 55%
less likely to have had a cardiac catheterization, PCI, and
CABG surgery than post-GAP white patients. Post-GAP
nonwhite patients were also almost twice as likely to
experience renal complications and 38% more likely to
have complications related to bleeding than post-GAP
white patients.
Quality indicator prevalences and tool use at discharge
by race and GAP status
Table III displays the results of quality indicator
prevalences at discharge. At discharge, among white
patients, post-GAP patients were more frequently pre-
scribed aspirin and β-blockers and received smoking
cessation counseling more often than their pre-GAP
counterparts.Amongpost-GAPpatients,nonwhitepatients
were28%and64%lesslikelytohavehadthedischargetool
administered and be given counseling for smoking
Table II. Selected procedures, treatment, and outcomes of nonwhite and white patients pre- and post-GAP⁎
Characteristic
White patient†
Nonwhite patient†
Post-GAP patients†
Pre-GAP,
n = 1158
(n [%])
Post-GAP,
n = 1209
(n [%])P
Pre-GAP,
n = 210
(n [%])
Post-GAP,
n = 280
(n [%])P
Odds ratio
(95% CI)‡
Nonwhite vs
white patients, P
In-hospital procedures/treatment
Admission tool used
Echocardiogram
LVEF
LVEF N40%
LVEF N25% to b40%
LVEF b25%
Cardiac catheterization
PCI
CABG
Aspirin‡
β-Blocker‡
Cardiovascular complications
Hypotension
Shock
Heart failure or pulmonary edema
Stroke
Renal insufficiency
Noncardiovascular complications
Hemorrhage/bleeding
102 (8.8)
715 (61.7)
551 (45.6)
715 (59.1)
≤.001
.1952
.0925
13 (6.2)
128 (61.0)
120 (42.9)
171 (61.1)
≤.001
.9787
.8860
0.90 (0.69-1.16)
1.08 (0.83-1.41)
.4102
.5530
5.35 (55.0)
297 (30.5)
141 (14.5)
527 (45.5)
229 (19.8)
101 (8.7)
425 (85.7)
192 (76.8)
661 (59.6)
289 (28.2)
125 (12.2)
614 (50.8)
309 (25.6)
127 (10.5)
419 (92.9)
192 (82.4)
104 (61.5)
39 (23.1)
26 (15.4)
94 (44.8)
33 (15.7)
10 (4.8)
73 (81.1)
35 (66.0)
143 (63.5)
51 (22.7)
31 (13.8)
103 (36.8)
37 (13.2)
14 (5.0)
115 (90.6)
45 (83.3)
1.0 (Reference)
0.75 (0.53-1.07)
1.06 (0.69-1.63)
0.56 (0.43-0.74)
0.44 (0.31-0.64)
0.45 (0.25-0.79)
0.73 (0.37-1.47)
1.07 (0.48-2.36)
.1120
.7933
≤.001
≤.001
.0046
.3770
.8710
.0102
.0008
.1417
.0004
.1274
.0747
.4338
.9038
.0411
.0395
350 (30.2)
14 (1.2)
548 (47.3)
51 (4.4)
266 (23.1)
410 (33.9)
16 (1.3)
529 (43.8)
56 (4.6)
259 (21.6)
.0547
.8035
.0814
.7933
.3684
68 (32.4)
2 (1.0)
104 (49.5)
16 (7.6)
73 (34.9)
87 (31.1)
1 (0.7)
137 (48.9)
20 (7.1)
97 (35.3)
.7577
–
.8962
.8415
.9373
0.88 (0.66-1.16)
0.54 (0.12-2.35)
1.23 (0.95-1.60)
1.58 (0.93-2.69)
1.98 (1.49-2.63)
.3637
.4007
.1167
.0854
≤.001
271 (23.4)291 (24.1).7030 67 (31.9)85 (30.3) .7140 1.38 (1.03-1.83) .0291
LVEF, Left ventricular ejection fraction; ACE, angiotensin-converting enzyme.
⁎Data are presented as number (percentage) unless otherwise indicated. Percentages are based only on nonmissing data.
†Percentage is based on ideal patients.
‡Represents the unadjusted magnitude of association between predictor variable (characteristic) and race where nonwhite is the reference variable.
380 Olomu et al
American Heart Journal
March 2010

Page 5

cessation, respectively, than white patients (Figures 1 and
2). Furthermore, nonwhite patients were less likely to
attend a high-quality performing hospital (hospital with a
discharge contract use rate N38.0%) (Figure 3).
Discussion
Implementation of the ACC's GAP program and its
associated discharge tool were associated with significant
increase in rates of evidence-based care in both white and
nonwhite Medicare patients. Admission tool and in-
hospital aspirin use significantly improved for both
races. In-hospital use of β-blockers improved significantly
for nonwhite patients only (66% vs 83.3%) showing that
at least this aspect of in-patient care improved as much or
even more for nonwhite patients regarding medication
use in GAP hospitals.
However, GAP discharge tools improved discharge
prescription rates of aspirin by 10.8% (P b .001) and
β-blockers by 7.0% (P = .047) for white patients; among
nonwhite patients, aspirin prescriptions increased by
only 1% and β-blocker prescriptions decreased by 6%
(both P values nonsignificant). Furthermore, nonwhite
patients were 28% and 64% less likely to have had the
GAP discharge tool used and receive smoking cessation
counseling, respectively.
Unfortunately, most patients involved in our study did
not receive the benefits of the discharge tool because
overall use of the tools was suboptimal. Changing
practice was hard for some hospitals and some centers
did not devote sufficient human and/or financial
resources to ensure tool use. Institutions and caregivers
should adopt standard tools and processes to optimize
quality of care for all patients with AMI.
We can only speculate as to the reasons for disparity in
the application of the discharge tool. Potential explaina-
tions include (1) communication difficulties between the
patient and health care providers that hinder the use of
the discharge tool, (2) attitudes/beliefs of health care
providers in delivery of care, (3) a higher percentage of
nonwhite patients receiving care from lower quality
performing hospitals, and (4) and/or nonwhite patients
attending busy hospitals with inadequate staff to admin-
ister the discharge tool.
Black patients have been shown to live in regions with
different local practice patterns and hospitals,17fewer
cardiac surgeons, and primary care providers that differ
systematically in quality and resources. Barnato et al18
found that use of different hospitals by black patients
and white patients contributed substantially to observed
treatment disparities. Initiatives targeted at hospitals that
disproportionally serve minority patients could simulta-
neously address quality deficiencies for all patients in the
hospitals and potentially decrease national health care
disparities. Such an effort is currently under investigation
in a number of hospitals (“Expecting Success” Project of
the Robert Wood Johnson Foundation).19
Although we found evidence of disparity in the
application of GAP tools, Sabatine et al20in their study
of the influence of race on death in non-ST elevation
acute coronary syndrome (ACS) patients did not observe
any evidence of disparity in their protocol-guided in-
hospital therapy. They found important differences only
in the non–protocol-guided cardiac care between white
patients and nonwhite patients. In-contrast to our study,
they did not use any discharge documents or tools, and
nonwhite patients in their study had a worse prognosis
than white patients. Yancy et al21in their study of
Table III. Quality indicator rates at discharge according to the GAP discharge tool use (Medicare population, n = 2,857)⁎
Characteristic
Pre-GAP†(n = 1368)Post-GAP†(n = 1498)
Post-GAP patients only,
OR: Nonwhite vs white
patients (n = 1489)
White patient,
n = 1158
(n [%])
Nonwhite
patient,
n = 210 (n [%])P
White patient,
n = 1209
(n [%])
Nonwhite
patient,
n = 280 (n [%])P
Odds ratio
(95% CI)‡
Nonwhite vs
white
patients, P
Discharge tool used
Aspirin prescribed
Beta-blocker
prescribed
Anticholesterolemia
prescribed
ACE-inhibitor
prescribed
Counseling for
smoking cessation
21 (1.8)
296 (83.4)
102 (87.9)
2 (1.0)
58 (89.2)
31 (96.9)
.3719
.2334
.1378
364 (30.1)
393 (94.2)
129 (94.9)
66 (23.6)
74 (90.2)
20 (90.9)
.0297 0.72 (0.53-0.97)
.1764 0.56 (0.24-1.31)
.4590 0.54 (0.11-2.80)
.0297
.1764
.4590
163 (80.3) 24 (77.4).7097230 (83.0) 45 (79.0).4614 0.77 (0.38-1.56).4614
164 (82.8) 29 (93.6).1273159 (86.9)31 (93.9) .2517 2.34 (0.53-10.41).2517
48 (39.4) 13 (34.2).5694 116 (73.4)24 (50.0) .0023 0.36 (0.19-0.71).0043
GAP, Guidelines applied in practice; CI, Confidence interval.
⁎Data are presented as number (percentage) unless otherwise indicated. Percentages are based only on nonmissing data.
†Percentage is based on ideal patients.
‡Represents the unadjusted magnitude of association between predictor variable (characteristic) and race where white race is the reference variable.
Olomu et al 381
American Heart Journal
Volume 159, Number 3