Differences in Medical Care and Disease Outcomes Among Black and White Women With Heart Disease
Background— The risk of cardiovascular mortality is higher among black women than white women, and the reasons for this disparity are largely unexplored. We sought to evaluate differences in medical care and clinical outcomes among black and white women with established coronary artery disease.
Methods and Results— Among the 2699 women enrolled in the Heart and Estrogen/progestin Replacement Study (HERS), we used Cox proportional hazards models to determine the association of race with risk of coronary heart disease (CHD) events independent of major cardiovascular risk factors or medical therapies. During an average of 4.1 years of follow-up, CHD events were twice as likely in black compared with white women (6.4 versus 3.1 per 100 person-years, hazard ratio, 2.1; 95% confidence interval, 1.5 to 2.8; P<0.001). Black women had higher rates of hypertension, diabetes, and hypercholesterolemia, yet were less likely to receive aspirin or statins. Black women less often had optimal blood pressure (56% versus 63%; P=0.01) and LDL cholesterol (30% versus 38%; P=0.04) control at baseline and during follow-up. After adjusting for these and other differences, black women still had >50% higher CHD event risk (hazard ratio, 1.52; 95% confidence interval, 1.1 to 2.1; P=0.03).
Conclusions— In a large cohort of women with heart disease, black women less often received appropriate preventive therapy and adequate risk factor control despite a greater CHD event risk. Interventions to improve appropriate therapy and risk factor control in all women, and especially black women, are needed.
Received December 31, 2002; de novo received April 18, 2003; revision received May 28, 2003; accepted June 6, 2003.
Black Americans are more likely to die from heart disease than white Americans and these racial disparities are more pronounced among women than among men. Whereas age-adjusted mortality due to heart disease is 19% greater for black men compared with white men, black women are one third more likely to die from heart disease than white women.1 The Institute of Medicine reports extensive racial disparities in quality of care and outcomes in heart disease, but acknowledges that the reasons for disparities are not well delineated.2 Using standardized patients of different gender and race presenting with identical complaints of chest pain, Schulman and colleagues found that black women, compared with black men and white men and women, were less often referred for cardiac catherization despite identical clinical presentations.3 Thus, black women may be particularly vulnerable to inferior care and poor outcomes due to heart disease.
Most studies of racial disparities relating to coronary heart disease have focused on either clinical management4–6 or outcomes,7 but rarely both within the same study.8 We are unaware of any study that has simultaneously evaluated medical treatment, risk factor control and clinical outcomes longitudinally in women with heart disease. The Heart and Estrogen/progestin Replacement Study (HERS) followed women with heart disease to determine predictors of clinical outcomes. Detailed clinical data on risk factors, treatment, and outcomes were collected in this cohort of women with an average follow-up of >4 years. Because the results of the initial trial were negative, we combined the two arms of the trial to create one cohort of women with heart disease. Among these women, we evaluated racial differences in appropriate use of preventive therapies, risk factor control, and clinical outcomes and explored whether differences in treatment appeared to contribute to differences in clinical outcomes.
The details of the methods and results of the HERS trial, a randomized, placebo-controlled trial of the effect of daily postmenopausal hormone therapy have been published elsewhere.9 Participants were women age <80 years with established coronary artery disease who had not had a hysterectomy. Women with a prior myocardial infarction (MI), coronary artery bypass graft surgery (CABG), percutaneous coronary revascularization (PTCA), or angiographic evidence of a 50% occlusion of one or more major coronary arteries were eligible for the study. A total of 2763 subjects were enrolled at 20 clinical centers in the United States and followed for a mean of 4.1 years. Although the HERS cohort included Latinas (n=53), Asian-Americans (n=21), and women of other ethnic groups (n=20), we excluded them from our analyses due to small numbers.
Demographic characteristics included age, education (years), marital status, and race. Race was self-reported and recorded as white, black, Latina, Asian-American, and “other” ethnic groups. Baseline health characteristics included body-mass index, waist-hip ratio, history of hypertension, diabetes, prior MI, congestive heart failure, CABG, PTCA, and self-reported physical activity. Smoking and alcohol consumption were also recorded. Baseline angina symptoms and depressive symptoms were assessed using validated, standardized tools.10,11 Angiogram results were available for approximately 90% of enrollees and women with high-grade occlusion of 2 or more major epicardial arteries were considered to have multivessel disease. Baseline medications, blood pressure, fasting glucose, and low-density lipoprotein (LDL) cholesterol levels were ascertained. Use of all study and nonstudy medications was obtained every 4 months. Blood pressure was measured annually and serum fasting glucose and lipid levels were ascertained at enrollment, 1 year of follow-up and study closeout. There were 20 clinical centers involved in the HERS study, each of which was assigned to one of four geographic census regions.12
The primary outcome was coronary heart disease (CHD) events, a composite of nonfatal myocardial infarction and CHD death. CHD death was defined as sudden death within one hour of symptoms, unobserved death that occurred out of the hospital in the absence of other known cause, or death due to coronary revascularization procedure or heart failure. Diagnosis of myocardial infarction was based on an algorithm that incorporated clinical symptoms, electrocardiographic abnormalities, and elevated cardiac enzyme levels.9
We evaluated the use of appropriate therapy in patients with specific risk factors13 and adequate control of cardiovascular risk factors both at baseline and during follow-up. For patients with prior myocardial infarction, we assessed rates of aspirin and β-blocker use. For patients with LDL cholesterol ≥100 mg/dL, we assessed treatment with statins and other lipid lowering therapies. And, for patients with congestive heart failure or advanced renal dysfunction (creatinine clearance ≤40 mg/dL), we determined ACE-inhibitors use. We used both unadjusted and multivariate adjusted analyses to compare use of medications in these specific risk groups. Because we did not have detailed information on all contraindications to these therapies, we could not assume that 100% utilization with any of these therapies would be possible. To evaluate patient adherence to prescribed therapies, we assessed the rate of compliance with study drug (hormone therapy or placebo) during the HERS trial.
We compared the characteristics of black and white women using unpaired t tests, chi-squared tests, or the Mann-Whitney rank-sum test as appropriate. We used generalized estimating equations to determine the association of race with the repeated measures of blood pressure and serum levels of glucose and LDL cholesterol over time. We present the mean levels with 95% confidence intervals of these measurements both unadjusted, and adjusted for differences in demographic and clinical characteristics. We used logistic regression models to determine the association of race with use of medical therapies, including models adjusted for baseline differences between black and white participants.
We used Kaplan-Meier survival analysis to assess racial differences in event-free survival. Cox proportional hazards models stratified on clinical center were used to determine the independent associations of race with coronary events. We confirmed that the log relative hazards were approximately linear for continuous predictors by examining the relative hazard by quartiles.14 The validity of the assumptions of proportional hazards over time was confirmed with tests using Schoenfeld residuals.15
In multivariate analyses, we treated medication use, blood pressure recordings, and serum levels of glucose and LDL cholesterol as time-dependent covariates. We conducted the multivariate proportional hazard models in stages. The first models adjusted for differences in baseline characteristics including demographic features and preexisting comorbidities. The second set of multivariate models also adjusted for differences in treatment and risk factor control. We tested whether the effect of race on CHD events varied by subgroups and for each potential interaction, we examined both the magnitude of the two-way interaction term and its probability value. For categorical variables, we tested for linear trends and for the overall contribution of the interaction terms using Wald chi-squared tests and report all interactions with P<0.10. All analyses were performed using Stata 7.0. With the exception of tests for interaction, two-sided probability values of P≤0.05 were considered statistically significant.
The 218 black women comprised 8.1% of the participants in the HERS trial. Compared with white women, black women were younger, less educated, less likely to be married, and more likely to live in the South (Table 1). Black participants were less likely to be active or drink alcohol but there was no difference in current smoking between black and white participants. Black participants were more obese and more likely to have a history of renal insufficiency, congestive heart failure, hypertension, and diabetes.
At baseline, black women less often received a daily aspirin or an HMG-CoA reductase inhibitor (statin). The lower rates of statin use were also present among those women with the greatest proven benefit from statins: women with LDL cholesterol >100 mg/dL. The rates of β-blocker use were generally low without racial differences in their use (Table 2), including among patients most likely to benefit from β-blockers, women with prior myocardial infarctions. Black women had higher rates of ACE inhibitors use. Among women most likely to benefit from ACE inhibitors (eg, women with diabetes or CHF), black women had nonsignificantly higher rates of use. Finally, calcium channel antagonists were more likely to be used by black women than by white women, including among those with prior MI. After adjusting for baseline differences, we found that black women had significantly lower odds of receiving statins (OR=0.62, 95% CI 0.44 to 0.87; P=0.006), nonsignificantly lower odds of receiving aspirin (OR=0.84, 95% CI 0.58 to 1.19; P=0.33) and β-blockers (OR=0.83, 95% CI 0.61 to 1.16; P=0.30), but had higher odds of getting an ACE inhibitor (OR=1.32, 95% CI 0.92 to 1.90; P=0.12). Rates of adherence to the study drug at the close of the study, as a marker of compliance, were similar for black and white women (68% versus 72%; P=0.26).
At enrollment, black women were more likely to have poor blood pressure and lipid control. Only 10% of white women had systolic blood pressure >160 mm Hg, compared with 17% of black women (P=0.01). Similarly, black women were more likely to have LDL cholesterol >160 mg/dL compared with white women (38% versus 30%; P=0.04). At baseline and during follow-up, black women had higher levels of blood pressure and LDL cholesterol (Table 3). Adjusted for baseline differences, comorbidities, and medication use, black women still had higher levels of systolic blood pressure and serum LDL cholesterol during follow-up (Table 3). There were no racial differences in fasting glucose levels among women with diabetes.
Black women had higher mortality than white women, 16% (34 deaths) versus 8% (205 deaths) (P<0.001; Table 4), with 2-fold greater risk of coronary deaths and nonfatal myocardial infarctions (Table 4 and Figure 1). After adjusting for demographic factors, geographic location, and all comorbid conditions, black women had a 60% greater risk of CHD events (hazard ratio 1.60, 95% confidence interval 1.11 to 2.32; P=0.01). After adjusting for medical therapy and extent of risk factor control, black women still had an elevated CHD event risk (hazard ratio 1.52; 95% confidence interval, 1.04 to 2.21; P=0.03).
Of potential interactions evaluated, we found one: among women with prior MI, black women had a 2.6-fold relative risk of the primary event, whereas among women without prior MI, black women had a 1.5-fold risk (probability value for interaction, 0.07). Because this may represent a chance finding, and because the association was in the same direction in both subgroups, we present the unified hazard ratio in addition to the stratified estimates.
We performed stratified, multivariate adjusted analyses among multiple subgroups and found that black women were at increased risk of CHD events in nearly all subgroups examined (Figure 2). The one exception was among women with chronic renal insufficiency, where black women had slightly lower rates of CHD events compared with white women. However, the numbers in this subgroup were small and the 95% confidence intervals were wide. These associations did not vary meaningfully in any of the subgroups (for all interactions, P>0.2) except for prior MI.
We found that black women with coronary artery disease were twice as likely to sustain a CHD event compared with white women in the HERS study. Black participants had poorer blood pressure and lipid control and were less likely to receive a daily aspirin or statin. Although adjusting for these differences partly attenuated the effect, the increased risk for black women persisted, particularly for nonfatal MI. Given their increased CHD risk, the lower rates of treatment for black women are concerning.
Several possible factors may explain higher CHD event rates for black women. They were sicker at enrollment, with higher rates of hypertension, diabetes, congestive heart failure, and renal dysfunction. Despite adjusting for and stratifying by these factors, black women had worse outcomes. Less aggressive risk factor control and preventive therapy may explain worse outcomes, but racial differences persisted despite adjustment for these factors. Adjusting for the differences in disease severity, risk factor control, and use of appropriate preventive therapy accounted for about 50% of the excess risk for black women and therefore, other, unmeasured factors must have contributed to their worse outcomes. Whether these other factors have a greater association with nonfatal MI than CHD death is a possibility raised by our findings.
Several nonclinical factors might have contributed to the differences in outcomes between black and white women. Socioeconomic status may have been lower in black women, but we adjusted for education level, a close correlate, and found similar relative risks for black women with and without some college education. Differences in access to healthcare may have been a contributor16; however, all women enrolled in HERS had primary care physicians and received regular care for their heart disease. Differences in mental health may have been a factor, but the differences in outcomes persisted despite adjustment for depression. Better understanding of these increased risks for black women could lead to targeted interventions to improve their outcomes.
Because of their higher CHD risk, lower rates of some treatments and risk factor control among black women are troubling. Black women had lower use of aspirin and statins and slightly lower rates of β-blocker use, although higher rates of ACE-inhibitor use. Differences in their ability to afford prescribed medications is plausible but unlikely given that black women had greater use of expensive drugs including ACE inhibitors and calcium channel blockers. The observed “undertreatment” could represent patient preferences, given that black patients more often refuse major procedures than white patients.17 We believe that this is unlikely given that black women in HERS took more of some medications and were equally adherent with clinical follow-up and study medications. Therefore, observed differences in treatment are unlikely to represent patient preference or noncompliance. Unmeasured confounders might account for some of the observed differences in treatment, although we had detailed information on most known indications for these therapies. One remaining important explanation for these findings is physician preference. We lack information to directly evaluate the role of physician bias in prescribing preventive medications to black and white women with CHD but strongly believe that these troubling findings warrant further investigation.
There are several strengths to this study. This is one of the largest cohorts of women with heart disease followed longitudinally with rigorous adjudication and near complete ascertainment of all cardiovascular and noncardiovascular hospitalizations and clinical outcomes. Racial differences in access-to-care and medication adherence were minimized. Unlike most studies of cardiovascular outcomes and quality of care, we longitudinally and simultaneously studied preventive therapies, risk factor control, and clinical outcomes.
This study has important limitations. Participants were recruited for a clinical trial and may be different from women in the general population. However, HERS recruited women from 20 different centers from around the United States, and their outcome rates were consistent with population-based statistics.1 Because women who enroll in clinical trials may be more health conscious than other women, disparities in this cohort may portend larger differences in the general population. Although we adjusted for multiple demographic and clinical characteristics, we do not know which unmeasured factors accounted for the differences in outcomes; however, we hope future studies can elucidate them. Finally, any research on race has important limitations. Though race may have unclear biological significance,18 it remains a relevant social construct with important implications for how patients perceive the healthcare system and, unfortunately, how they are perceived by medical providers.3 Therefore, an improved understanding of the reasons behind racial disparity should foster improvements in the quality of care and outcomes for all Americans.
In summary, we found that black women with heart disease had 50% higher adjusted risk for myocardial infarction and CHD death compared with white women. We demonstrated less aggressive treatment of blood pressure and LDL cholesterol and the lower use of aspirin and statins among black women. Further investigations should focus on the mechanisms behind the poor outcomes of black women, and reasons why they receive less aggressive therapy. Most importantly, interventions are needed to improve and equalize preventive therapy use in all racial groups.
The HERS study was sponsored by Wyeth-Ayerst.
Minino AM, Arias E, Kochanek KD, et al. Deaths: final data for 2000. National Vital Statistics Reports, vol 50, number 15. Hyattsville, Md: National Center for Health Statistics; 2002.
Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington, DC: National Academy Press; 2002.
Rathore SS, Berger AK, Weinfurt KP, et al. Race, sex, poverty, and the medical treatment of acute myocardial infarction in the elderly. Circulation. 2000; 102: 642–648.
Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika. 1982; 69: 239–241.