Abstract
BACKGROUND
Clinical trial evidence suggests poorer outcomes in blacks compared to whites when treated with angiotensin-converting enzyme (ACE) inhibitor-based regimen, but this has not been evaluated in clinical practice.
OBJECTIVES
We evaluated the comparative effectiveness of an ACE inhibitor-based regimen on a composite outcome of all-cause mortality, stroke, and acute myocardial infarction (AMI) in hypertensive blacks compared to whites.
METHODS
We conducted a retrospective cohort study of 434,646 patients in a municipal health care system. Four exposure groups (Black-ACE, Black-NoACE, White-ACE, White-NoACE) were created based on race and treatment exposure (ACE or NoACE). Risk of the composite outcome and its components was compared across treatment groups and race using weighted Cox proportional hazard models.
RESULTS
Our analysis included 59,316 new users of ACE inhibitors, 47% of whom were black. Baseline characteristics were comparable for all groups after inverse probability weighting adjustment. For the composite outcome, the race treatment interaction was significant (p = 0.04); ACE use in blacks was associated with poorer cardiovascular outcomes (ACE vs. NoACE: 8.69% vs. 7.74%; p = 0.05) but not in whites (6.40% vs. 6.74%; p = 0.37). Similarly, the Black-ACE group had higher rates of AMI (0.46% vs. 0.26%; p = 0.04), stroke (2.43% vs. 1.93%; p = 0.05) and chronic heart failure (3.75% vs. 2.25%; p < 0.0001) than the Black-NoACE group. However, the Black-ACE group was no more likely to develop adverse effects than the White-ACE group.
CONCLUSIONS
ACE inhibitor-based therapy was associated with poorer cardiovascular outcomes in hypertensive blacks but not in whites. These findings confirm clinical trial evidence that hypertensive blacks have poorer outcomes than whites when treated with an ACE inhibitor-based regimen.
Keywords: antihypertensive medications, cardiovascular disease, electronic health record, race
In the United States, blacks have disproportionately higher hypertension-related morbidity and mortality than other racial/ethnic groups (1); plus, hypertension explains much of the variance in mortality between blacks and whites (2). Despite the higher rates of cardiovascular disease (CVD), blacks are underrepresented in randomized controlled trials of therapeutic medications, with a participation rate <30% in heart failure trials (3).
Angiotensin-converting enzyme (ACE) inhibitors are commonly prescribed for treatment of hypertension; however, despite their proven efficacy on blood pressure (BP) reduction (4), their relative effectiveness on cardiovascular (CV) outcomes in hypertensive blacks remains uncertain (5). Clinical trial evidence suggests that ACE inhibitors may not provide the same benefits in blacks compared to whites and, in fact, may cause harm (6–9). One retrospective study of 2,225 patients found a 19% rate of ACE-inhibitor discontinuation due to adverse events (10). Among 15,100 blacks enrolled in ALLHAT (the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial), those treated with ACE inhibitors had poorer CV outcomes and lower BP reduction than those randomized to a thiazide-type diuretic, chlorthalidone (5,9,11). The SOLVD (Study Of Left Ventricular Dysfunction) trial found a significant reduction in hospitalizations for congestive heart failure (CHF) among whites on an ACE inhibitor, but no such reduction was found in blacks (12). Despite the relatively lower clinical effectiveness of ACE inhibitor-based treatment in hypertensive blacks compared to whites enrolled in clinical trials, there are limited data on the comparative effectiveness of ACE inhibitor-based regimens on important health outcomes in hypertensive blacks compared to whites in clinical practice settings.
In this study, we evaluated racial differences in the comparative effectiveness and safety of ACE inhibitor-based regimens in hypertensive blacks compared to whites, using a longitudinal dataset derived from electronic health records (EHRs) of hypertensive patients who received care within New York City’s Health and Hospital Corporation (HHC). We hypothesized that an ACE inhibitor-based regimen would be less effective and lead to higher rates of serious adverse effects (hyperkalemia) in blacks compared to whites.
METHODS
STUDY DESIGN, SETTING, AND POPULATION
This study was conducted in New York City’s HHC, which oversees the city’s public health care system in all 5 boroughs. The corporation consists of 11 acute care hospitals, 6 diagnostic and treatment centers, 4 long-term care facilities, a certified home health care agency, and more than 80 community health clinics. It is the largest municipal hospital and health care system in the country: a $5.4 billion public benefit corporation that serves 1.8 million New Yorkers. HHC provides care for approximately 20% of all general hospital discharges and more than 30% of all emergency department and hospital-based clinic visits in New York City (NYC). Approximately 35% of patients seen in the HHC system are black and 7% are white.
Using a retrospective cohort design, we extracted EHR data (BP measurements, weight, prescription refills, laboratory test results, clinical diagnoses, encounter diagnoses for outpatient visits, diagnostic imaging tests, and health care utilization) from HHC’s clinical data warehouse. The study population was comprised of adult hypertensive patients (age ≥18 years), who received care between January 1, 2004, and December 31, 2009, and who met the following criteria: hypertension diagnosis (based on hypertension ICD-9 code on at least 2 clinic visits) and prescribed an ACE inhibitor, β-blocker, thiazide-type diuretic, or calcium channel blocker (CCB) for at least 6 months after their first date of entry into the HHC system. We excluded patients who were not self-identified as African American, black, or Caucasian, and those with prior history of nonfatal AMI, nonfatal stroke, or CHF, because these medications are compelling indications for ACE inhibitor use. The study was approved by the institutional review boards of both the New York University School of Medicine and the HHC.
STUDY MEASURES AND OUTCOMES
The primary outcome was a composite of all-cause mortality, nonfatal AMI, and nonfatal stroke. Secondary outcomes included individual components of the composite outcome, CHF, kidney failure, and safety outcomes, which included severe side effects (hyperkalemia, defined as serum potassium >5.5 mEq/l, and hypokalemia, defined as a serum potassium of 2 to 3.5 mEq/l). All outcomes were extracted from the EHR using the corresponding ICD-9 codes from the patient’s problem list and lab values. For these analyses, we followed patients for up to 2,000 days, with an average follow-up time of 4.5 years. For each adverse event, the outcome was measured as a time-to-event, starting from the index date and ending when the outcome was observed or when the study ended (resulting in right censoring). A patient’s index date was based on the start of a drug regimen (either ACE or non-ACE regimen), and was required to be at least 180 days following the patient’s entry into the study. A patient’s exposure status to prescribed antihypertensive medications was assessed for the entire study period (Figure 1).
FIGURE 1. Relationship of Antihypertensive Medication Exposures to Cardiovascular Outcomes.
This figure depicts the timeframe for study eligibility for adult hypertensive patients (age >18 years), who received care between January 1, 2004 and December 31, 2009; and who meet the following criteria: hypertension diagnosis (based on hypertension ICD-9 code on at least 2 clinic visits); and prescribed an ACE inhibitor, β-blockers, thiazide-type diuretics, or calcium channel blockers for at least 6 months after their first date of entry into the HHC system. Outcomes through December 31, 2009 were eligible for analysis.
In the EHR, patient race is typically recorded based on self-reporting, the gold standard for such classification (13,14). Race data were collapsed into African American or black, Caucasian or white, Asian, Native American/Alaskan, and Native Hawaiian/Other Pacific Islander, following the U.S. Office of Management and Budget classifications, with only blacks and whites included in the analyses.
We identified new users (an inception cohort) for each antihypertensive medication, in order to prevent “healthy user” effects that result from studying patients who are not treatment naïve (15). A 6-month period after the first date of entry into the EHR was required in order to ensure that patients were new users of the relevant medication for both ACE and non-ACE groups. Patients were grouped based on the first antihypertensive medication they were prescribed. Patients whose first antihypertensive medication was an ACE inhibitor were categorized as the ACE group and those whose first antihypertensive medications were a β-blocker, thiazide-type diuretic, or a calcium channel blocker were categorized as the NoACE group. In the NoACE group, 46.6% were on β-blockers, 51% were taking thiazide-type diuretics, and 33.6% were on CCBs. For the ACE group, during the course of follow-up, 4.6% were also on β-blockers, 8.4% were taking thiazide-type diuretics, and 19.8% were on a CCB.
We included laboratory tests, clinical diagnoses, or encounter diagnosis-derived events that occurred at least 28 days after the prescription-documented time that the patient was on that antihypertensive. Adverse events were attributed to the medication only after its initiation. If an event occurred multiple times (e.g., hyperkalemia), we included only the first occurrence.
COVARIATES USED FOR RISK ADJUSTMENT
We considered the following variables to adjust for potential confounding or as potential effect modifiers: age, sex, calendar year of entry into the system, number of clinic visits in the previous year, baseline systolic and diastolic BP, and comorbidity using the Charlson Comorbidity Index (CCI) that includes diabetes mellitus, renal disease, chronic pulmonary disease, connective tissue disease, cerebrovascular accident, liver disease, and cancer. The CCI is a well-validated tool that has been used in clinical trials to adjust for the confounding effect of comorbid conditions on outcomes and mortality (16). We also included baseline creatinine values, cholesterol values, and potassium; medications that can cause hyperkalemia or hypokalemia (e.g., nonsteroidal anti-inflammatory drugs, statin); and other antihypertensive medications (e.g., angiotensin receptor blocker, CCB, etc.) as covariates.
STATISTICAL ANALYSIS
Because patients were not randomly assigned to treatment in this study, we applied inverse probability of treatment weights (IPTW) (17,18) to our analyses. This ensured that all treatment and nontreatment groups were balanced across potentially confounding covariates and minimized bias due to confounding by indication (19). The weights were calculated by dividing the probability of group membership by the probability of group membership conditional on the covariates (i.e., the propensity score). We estimated the probabilities using multinomial logistic regression models. The covariates in the propensity score model included risk factors that were considered potential confounders of the association of race and hypertension treatment group (ACE versus NoACE) with the composite outcome. We then compared the distribution of potential confounders after applying the weights to confirm that the groups were adequately balanced.
For each primary and secondary outcome, we estimated IPTW-weighted incident rates for all 4 race/treatment groups. We used weighted Cox proportional hazard models to conduct a within-race comparison of hypertension treatment regimens (ACE vs. NoACE). The Cox regression used days as the unit of time. For each adverse event, we estimated a hazard ratio (HR) for each group (blacks and whites), providing a measure of risk for blacks and whites on ACE inhibitors compared to blacks and whites receiving other types of antihypertensive medications, respectively. Finally, we compared the HRs for blacks and whites using ratios of hazard ratios. We reported all ratios along with their 95% confidence intervals (CI). Analyses were conducted using SAS software, version 9.3 (SAS Institute, Cary, North Carolina).
RESULTS
A total of 434,646 patients met inclusion criteria (Figure 2); of these, 76,546 were prescribed medications from 1 of the 4 antihypertensive drug classes (ACE inhibitor, β-blocker, CCB, or thiazide-type diuretic) during the study period. Patients were excluded for a prior diagnosis of nonfatal AMI, nonfatal stroke, CHF, or kidney failure before prescription, first antihypertensive prescription <180 days of entry into the system, prescriptions with no refills, multiple visits for the same event, dying prior to prescription date, having a follow-up date earlier than the index date, and having an IPTW >10. This left a final sample of 59,316 patients (47% black). The baseline characteristics of all 4 study groups (Black-ACE, Black-NoACE, White-ACE, and White-NoACE) prior to IPTW matching are summarized in Tables 1 and 2. There were significant differences between the 4 groups. However, after IPTW matching, the baseline characteristics were well matched for most variables (nearly all of the standardized differences were <10%), with the exception of very low and high-density lipoprotein and triglyceride levels.
FIGURE 2. Patient Flow Chart.
The process used to select the final study sample for analyses. At baseline, 434,646 hypertensive patients met the initial study criteria, of which 359,499 met the second inclusion criteria including age and race. Of this, 275,217 were on one of the 4 study anthihypetensives for at least 6 months after their first date of entry into the HHC system. We eliminated patients who were prescribed ACE only once with no refill, and those with multiple duplicate visits, leaving a sample size of 76,546. Our final sample of 59,316 excluded additional patients including those with study outcomes of interest. AMI = acute myocardial infarction; ARB = angiotensin receptor blocker; CCB = calcium channel blocker; CHF = congestive heart failure; TTE = transthoracic echocardiography; other abbreviations as in Figure 1.
TABLE 1.
Baseline Characteristics Before Inverse Probability Weights Adjustment
| IPW Unadjusted | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||||
| Variables | White-NoACS (1) | White-ACE (2) | Black-NoACE (3) | Black-ACE (4) | Absolute Differences Between Groups* | ||||||
| (n = 15,214) | (n = 16,015) | (n = 15,150) | (n = 12,937) | 1–2 | 1–3 | 1–4 | |||||
| Age, years (mean ± SD) | 51.8 | 13.99 | 52.90 | 13.14 | 49.10 | 13.51 | 50.91 | 12.95 | 1.52 | 2.28 | 0.47 |
| Female | 63.2 | 55.9 | 65.8 | 58.2 | 7.3 | 2.6 | 5 | ||||
| BMI, kg/m2 (mean ± SD) | 33.03 | 11.83 | 32.85 | 11.09 | 32.79 | 11.00 | 32.45 | 10.98 | 0.18 | 0.24 | 0.58 |
| Systolic BP, mm Hg (mean ± SD) | 140.81 | 20.07 | 139.37 | 20.04 | 148.40 | 21.03 | 144.34 | 21.35 | 1.44 | 7.59 | 3.53 |
| Diastolic BP, mm Hg (mean ± SD) | 81.44 | 12.11 | 80.19 | 11.87 | 84.98 | 12.96 | 82.53 | 13.09 | 1.25 | 3.54 | 1.09 |
| Pulse (mean ± SD) | 78.6 3 | 13.79 | 77.98 | 13.30 | 80.17 | 13.83 | 80.22 | 14.21 | 0.65 | 1.54 | 1.59 |
| Times hospitalized within 1 year, n | 0.42 | 0.88 | 0.48 | 0.91 | 0.40 | 0.86 | 0.57 | 1.06 | 0.06 | 0.02 | 0.15 |
| Days hospitalized within 1 year, n | 2.48 | 9.76 | 2.55 | 9.65 | 2.83 | 16.33 | 3.69 | 16.35 | 0.07 | 0.35 | 1.21 |
| Clinic Visits within 1 year, n | 10.68 | 17.00 | 11.30 | 17.21 | 9.84 | 17.39 | 10.70 | 20.07 | 0.62 | 0.84 | 0.02 |
| Has insurance | 85.6 | 85.3 | 79.6 | 80.9 | 0.3 | 6.0 | 4.7 | ||||
| Number of additional antihypertensive agents | |||||||||||
| 0 | 87.3 | 86.3 | 90.2 | 83.6 | 1.0 | 2.9 | 3.7 | ||||
| 1 | 10.4 | 12.3 | 7.9 | 15.0 | 1.9 | 2.5 | 4.6 | ||||
| 2 | 1.9 | 1.0 | 1.5 | 0.9 | 0.9 | 0.4 | 1.0 | ||||
| ≥3 | 0.3 | 0.4 | 0.3 | 0.5 | 0.1 | 0.0 | 0.2 | ||||
| Comorbidities | |||||||||||
| Atrial fibrillation | 1.6 | 1.9 | 0.9 | 1.4 | 0.3 | 0.7 | 0.2 | ||||
| AMI | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Angina | 1.5 | 2.1 | 0.8 | 1.4 | 0.6 | 0.7 | 0.1 | ||||
| Cancer | 3.4 | 2.9 | 4.4 | 3.7 | 0.5 | 1.0 | 0.3 | ||||
| CAD | 3.1 | 6.1 | 1.7 | 3.9 | 3.0 | 1.4 | 0.8 | ||||
| Cardiac dysrhythmias | 3.4 | 3.7 | 2.5 | 3.1 | 0.3 | 0.9 | 0.3 | ||||
| CHF | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Chronic pulmonary disease | 0.2 | 0.2 | 0.2 | 0.2 | 0.0 | 0.0 | 0.0 | ||||
| Connective tissue disease | 0.7 | 0.8 | 0.7 | 1.1 | 0.1 | 0.0 | 0.4 | ||||
| Cerebrovascular accident | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Diabetes | 13.5 | 43.7 | 14.4 | 43.6 | 30.2 | 0.9 | 30.1 | ||||
| Diseases of mitral valve | 0.1 | 0.1 | 0.1 | 0.1 | 0.0 | 0.0 | 0.0 | ||||
| Endocardial disease | 0.0 | 0.1 | 0.0 | 0.0 | 0.1 | 0.0 | 0.0 | ||||
| Hemorrhagic stroke | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Hypertensive kidney disease | 0.1 | 0.1 | 0.1 | 0.2 | 0.0 | 0.0 | 0.1 | ||||
| Ischemic stroke | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Renal disease | 0.9 | 0.5 | 1.7 | 2.1 | 0.4 | 0.8 | 1.2 | ||||
|
| |||||||||||
| Charlson score category | |||||||||||
| 0 | 82.1 | 55.5 | 80.3 | 53.7 | 26.6 | 1.8 | 28.4 | ||||
| 1–3 | 17.5 | 44.2 | 19.2 | 45.9 | 26.7 | 1.7 | 28.4 | ||||
| ≥4 | 0.5 | 0.3 | 0.5 | 0.4 | 0.2 | 0.0 | 0.1 | ||||
| Laboratory measures | |||||||||||
| BUN, mg/dl | 15.72 | 7.46 | 16.36 | 6.70 | 14.88 | 7.73 | 15.50 | 7.76 | 0.64 | 0.84 | 0.22 |
| LDL, mg/dl | 113.24 | 35.65 | 110.27 | 36.67 | 113.70 | 37.00 | 111.03 | 38.83 | 2.97 | 0.46 | 2.21 |
| HDL, mg/dl | 50.83 | 15.01 | 49.06 | 14.30 | 55.54 | 16.67 | 53.90 | 16.89 | 1.77 | 4.71 | 3.07 |
| Creatinine, mg/dl | 0.90 | 0.55 | 0.89 | 0.36 | 1.03 | 0.88 | 1.06 | 0.89 | 0.01 | 0.13 | 0.01 |
| GFR, ml/min/1.73 m2 | 66.09 | 15.06 | 64.15 | 13.89 | 65.42 | 16.18 | 62.79 | 15.54 | 1.94 | 0.67 | 3.30 |
| HbA1c, % | 6.60 | 1.69 | 7.86 | 2.56 | 6.73 | 1.76 | 8.00 | 2.59 | 1.26 | 0.13 | 1.40 |
| Triglycerides, mg/dl | 146.63 | 90.85 | 161.63 | 102.71 | 113.42 | 69.88 | 121.16 | 79.25 | 15.00 | 33.21 | 25.47 |
| Potassium, mEq/l | 4.27 | 0.48 | 4.31 | 0.45 | 4.15 | 0.47 | 4.19 | 0.47 | 0.04 | 0.12 | 0.08 |
| Sodium, mEq/l | 139.59 | 2.87 | 139.38 | 2.98 | 139.65 | 2.96 | 139.27 | 3.40 | 0.21 | 0.06 | 0.32 |
| VLDL, mg/dl | 30.16 | 18.51 | 32.57 | 21.11 | 23.22 | 13.57 | 24.16 | 14.60 | 2.41 | 6.94 | 6.00 |
Values are mean SD or % unless otherwise indicated.
Difference between group means and proportions.
ACE = angiotensin-converting enzyme; AMI = acute myocardial infarction; BMI = body mass index; BP = blood pressure; BUN = blood urea nitrogen; CAD = coronary artery disease; CHF = congestive heart failure; GFR = glomerular filtration rate; HbA1c = glycosylated hemoglobin; HDL = high-density lipoprotein; LDL = low-density lipoprotein; VLDL = very low-density lipoprotein.
TABLE 2.
Baseline Characteristics after Inverse Probability Weights Adjustment
| IPW Adjusted | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||||
| Variables | White-NoACE (1) | White-ACE (2) | Black-NoACE (3) | Black-ACE (4) | Absolute Differences Between Groups* | ||||||
| (n = 15,214) | (n = 16,015) | (n = 15,150) | (n = 12,937) | 1–2 | 1–3 | 1–4 | |||||
| Age, years | 51.48 | 12.97 | 51.32 | 12.19 | 51.07 | 13.30 | 50.99 | 12.86 | 0.16 | 0.41 | 0.49 |
| Female | 59.6 | 58.0 | 60.4 | 59.7 | 1.6 | 0.8 | 0.1 | ||||
| BMI, kg/m2 | 32.88 | 10.62 | 32.72 | 10.23 | 32.65 | 10.74 | 32.60 | 11.08 | 0.16 | 0.23 | 0.28 |
| Systolic BP, mm Hg | 142.44 | 19.64 | 141.60 | 19.12 | 143.94 | 20.28 | 143.75 | 21.06 | 0.84 | 1.50 | 1.31 |
| Diastolic BP, mm Hg | 81.98 | 11.88 | 81.67 | 11.65 | 82.66 | 12.36 | 82.62 | 12.97 | 0.31 | 0.68 | 0.64 |
| Pulse | 78.86 | 13.21 | 78.81 | 12.59 | 79.65 | 13.32 | 79.34 | 14.02 | 0.05 | 0.79 | 0.48 |
| Times hospitalized within 1 year, n | 0.49 | 0.87 | 0.49 | 0.86 | 0.49 | 0.90 | 0.48 | 0.95 | 0.00 | 0.00 | 0.01 |
| Days | 2.77 | 9.76 | 2.78 | 10.2 | 3.05 | 12.1 | 2.90 | 13.0 | 0.0 | 0.2 | 0.1 |
| hospitalized within 1 year, n Clinic Visits within 1 year, n | 11.28 | 17.76 | 11.25 | 118.42 | 11.12 | 919.84 | 11.07 | 520.61 | 10.03 | 80.16 | 30.21 |
| Has insurance | 86.9 | 85.3 | 84.0 | 83.2 | 1.6 | 2.9 | 3.7 | ||||
| Number of additional antihypertensi ve agents | |||||||||||
| 0 | 85.0 | 86.4 | 86.4 | 86.4 | 1.4 | 1.4 | 1.4 | ||||
| 1 | 12.8 | 11.6 | 11.7 | 11.5 | 1.2 | 1.1 | 1.3 | ||||
| 2 | 1.8 | 1.5 | 1.4 | 1.7 | 0.3 | 0.4 | 0.1 | ||||
| ≥ | 0.4 | 0.4 | 0.5 | 0.4 | 0.0 | 0.1 | 0.0 | ||||
| Comorbidities | |||||||||||
| Atrial fibrillation | 1.8 | 1.6 | 1.4 | 1.6 | 0.2 | 0.4 | 0.2 | ||||
| AMI | 0.0 | 0.0 | 0.0 | 0.0 | 23 | ||||||
| Angina | 1.7 | 1.6 | 1.4 | 1.4 | 0.1 | 0.3 | 0.3 | ||||
| Cancer | 3.4 | 3.4 | 3.9 | 3.6 | 0.0 | 0.5 | 0.2 | ||||
| CAD | 4.5 | 4.3 | 3.2 | 3.9 | 0.2 | 1.3 | 0.6 | ||||
| Cardiac dysrhythmias | 3.7 | 3.3 | 3.3 | 3.2 | 0.4 | 0.4 | 0.5 | ||||
| CHF | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Chronic pulmonary disease | 0.2 | 0.2 | 0.2 | 0.2 | 0.0 | 0.0 | 0.0 | ||||
| Connective tissue disease | 0.8 | 0.9 | 0.9 | 0.9 | 0.1 | 0.1 | 0.1 | ||||
| Cerebrovascula r accident | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Diabetes | 27.6 | 30.8 | 26.8 | 29.2 | 3.2 | 0.8 | 1.6 | ||||
| Diseases of mitral valve | 0.1 | 0.1 | 0.1 | 0.2 | 0.0 | 0.0 | 0.1 | ||||
| Endocardial disease | 0.0 | 0.1 | 0.0 | 0.0 | 0.1 | 0.0 | 0.0 | ||||
| Hemorrhagic stroke | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Hypertensive kidney disease | 0.1 | 0.1 | 0.1 | 0.3 | 0.0 | 0.0 | 0.2 | ||||
| Ischemic stroke | 0.0 | 0.0 | 0.0 | 0.0 | |||||||
| Renal disease | 1.3 | 1.1 | 1.9 | 1.7 | 0.2 | 0.6 | 0.4 | ||||
|
| |||||||||||
| Charlson score category | |||||||||||
| 0 | 70.1 | 65.9 | 69.8 | 67.0 | 4.2 | 0.3 | 3.1 | ||||
| 1–3 | 29.3 | 33.7 | 29.4 | 32.4 | 4.4 | 0.1 | 3.1 | ||||
| ≥4 | 0.6 | 0.4 | 0.7 | 0.5 | 0.4 | 0.1 | 0.1 | ||||
| Laboratory measures | |||||||||||
| BUN, mg/dl | 16.14 | 8.01 | 15.85 | 6.31 | 15.74 | 7.85 | 16.06 | 7.99 | 8.13 | 0.29 | 9.83 |
| LDL, mg/dl | 110.94 | 34.62 | 111.62 | 32.61 | 111.97 | 37.81 | 110.93 | 37.01 | 0.68 | 1.03 | 0.01 |
| HDL, mg/dl | 50.99 | 15.13 | 50.96 | 14.04 | 52.78 | 15.93 | 52.02 | 15.53 | 0.03 | 1.79 | 1.03 |
| Creatinine, mg/dl | 0.97 | 0.78 | 0.91 | 0.48 | 1.02 | 0.67 | 1.03 | 0.69 | 0.06 | 0.05 | 0.06 |
| GFR, ml/min/1.73 m2 | 64.21 | 15.16 | 64.86 | 13.23 | 64.01 | 16.71 | 63.99 | 16.54 | 0.65 | 0.2 | 0.22 |
| HbA1c, % | 7.38 | 2.60 | 7.49 | 2.03 | 7.29 | 2.21 | 7.43 | 2.04 | 0.11 | 0.09 | 0.05 |
| Triglycerides, mg/dl | 143.13 | 85.45 | 145.76 | 79.74 | 130.19 | 87.96 | 139.44 | 103.35 | 2.63 | 13.11 | 3.69 |
| Potassium, mEq/l | 4.26 | 0.46 | 4.26 | 0.42 | 4.22 | 0.47 | 4.23 | 0.51 | 0.00 | 0.04 | 0.03 |
| Sodium, mEq/l | 139.41 | 2.89 | 139.41 | 2.72 | 139.57 | 2.91 | 139.44 | 3.18 | 0.00 | 0.16 | 0.03 |
| VLDL, mg/dl | 30.56 | 22.01 | 30.22 | 18.54 | 24.75 | 12.41 | 25.82 | 13.63 | 0.34 | 5.81 | 4.74 |
Values are mean SD or % unless otherwise indicated.
Difference between group means and proportions. Abbreviations as in Table 1.
As seen in the results of the primary and secondary analyses (Table 3), incident rates for CV outcomes were higher for the Black-ACE group compared to the Black-NoACE group for most of the outcomes. Specifically, the Black-ACE group had higher rates of AMI (0.46% vs. 0.26%), stroke (2.43% vs. 1.93%), and CHF (3.75% vs. 2.25%) than the Black-NoACE group. However, the hazard ratios indicate that these differences were statistically significant only for AMI (HR: 1.76; 95% CI: 1.02 to 3.01) and CHF (HR: 1.64; 95% CI: 1.34 to 2.02). The differences in incident rates of CV outcomes for the White-ACE group versus the White-NoACE group did not have any apparent pattern. All-cause mortality was the only outcome for whites with a statistically significant difference (HR: 0.87; 95% CI: 0.76 to 1.00), suggesting that the ACE-based regimen may be protective with respect to death for white patients. For all other outcomes, there were no apparent differences between ACE and NoACE groups among whites.
TABLE 3.
Incident Rates and HRs for Black and White Patients and ACE Inhibitor Use
| Outcomes | Blacks | Whites | Ratio of HRs | ||||
|---|---|---|---|---|---|---|---|
| Incident Rates Over Study Duration | Incident Rates Over Study Duration | ||||||
| NoACE | ACE | HR (CI) | NoACE | ACE | HR (CI) | (Blacks vs.Whites) | |
| Composite of AMI, stroke, and all-cause mortality | 7.7 | 8.7 | 1.11 (0.99, 1.25) | 6.7 | 6.4 | 0.94 (0.84, 1.06) | 1.18 (1.00, 1.40)* |
|
| |||||||
| AMI | 0.3 | 0.5 | 1.76 (1.02, 3.01)* | 0.3 | 0.5 | 1.43 (0.86, 2.37) | 1.23 (0.59, 2.58) |
|
| |||||||
| Stroke | 1.9 | 2.4 | 1.25 (0.99, 1.57) | 1.9 | 2.0 | 1.02 (0.80, 1.29) | 1.22 (0.88, 1.71) |
|
| |||||||
| All-cause mortality | 6.0 | 6.2 | 1.02 (0.89, 1.18) | 4.9 | 4.4 | 0.87 (0.76, 1.00)* | 1.17 (0.96, 1.42) |
|
| |||||||
| Hyperkalemia | 0.9 | 1.1 | 1.21 (0.88, 1.68) | 0.8 | 0.8 | 0.92 (0.61, 1.39) | 1.32 (0.78, 2.23) |
|
| |||||||
| Hypokalemia | 0.2 | 0.2 | 0.68 (0.35, 1.31) | 0.2 | 0.2 | 0.90 (0.26, 3.07) | 0.75 (0.19, 3.04) |
|
| |||||||
| Renal disease | 0.2 | 0.2 | 1.15 (0.63, 2.12) | 0.3 | 0.2 | 0.56 (0.26, 1.18) | 2.06 (0.79, 5.42) |
|
| |||||||
| CHF | 2.3 | 3.8 | 1.64 (1.34, 2.02)† | 2.4 | 2.7 | 1.12 (0.93, 1.35) | 1.47 (1.11, 1.94)† |
Values are %.
p < 0.05
p < 0.01
CI = confidence interval; HR = hazard ratio; other abbreviations as in Table 1.
As shown in Table 2, for the composite outcome, ACE-inhibitor use was associated with poorer outcomes in blacks (ACE vs. NoACE; 8.7% vs. 7.7%; p = 0.05) but not in whites (ACE vs. NoACE; 6.40% vs. 6.74%; p = 0.37). A comparison of the HRs of blacks versus whites indicates that the relative risk for the composite outcome for Black-ACE versus Black-NoACE was higher than the relative risk for whites on an ACE inhibitor compared to NoACE. The test for race treatment interaction was significant for the composite outcome with blacks having a higher event rate than whites (blacks vs. whites; HR: 1.18; 95% CI: 1.00 to 1.40; p < 0.05). However, the Black-ACE group was no more likely to develop adverse effects (i.e., hyperkalemia, hypokalemia) than the White-ACE group.
Because clinical trial results suggest that ACE-inhibitor treatment leads to a smaller decrease in systolic blood pressure (SBP) in blacks compared to whites (9,12), we adjusted for the black-white differences in achieved SBP in all analysis models. For this purpose, we defined the adjusted achieved SBP as the average SBP for each patient closest to the study endpoint. Our results show that blacks have higher achieved SBP levels (137.9 ± 20.2 mm Hg) compared to whites (135.1 ± 19.5 mm Hg). After adjustment for the black-white differences in achieved SBP, the primary outcomes in the study remained essentially unchanged.
DISCUSSION
This is the first study to use a comparative effectiveness research design for a longitudinal dataset of hypertensive patients who receive care in a municipal health care system to evaluate racial differences in the effectiveness and safety of ACE inhibitor-based regimen in blacks compared to whites. Our findings demonstrate that hypertensive blacks on an ACE inhibitor-based regimen had higher rates of events and were at higher risk of the composite outcome of all-cause mortality, nonfatal AMI, or nonfatal stroke than whites on an ACE inhibitor-based regimen. Additionally, blacks on ACE inhibitors were more likely to develop CHF compared to whites. There were no differences in rates of serious adverse effects for both groups.
Our findings support clinical trial evidence from ALLHAT, in which blacks on the ACE inhibitor lisinopril had a 19% higher risk of combined CVD, 30% higher risk of heart failure, and 40% higher risk of stroke than those randomized to a thiazide-type diuretic (chlorthalidone) (9). The magnitude of the risk of combined CVD and CHF noted in ALLHAT was similar to that in our study. However with respect to stroke, although blacks on an ACE inhibitor-based regimen had a 25% higher risk of stroke than their counterparts not on ACE inhibitors, in our study, this finding did not reach statistical significance.
Although the reasons for the noted disparities in clinical effectiveness of ACE inhibitor-based regimens between blacks and whites remain unclear, the prevailing theory is that blacks are less responsive to ACE inhibitors compared to whites (12), particularly with respect to BP reduction. For example, in ALLHAT, blacks had higher average follow-up BP than whites (9). The time-dependent BP adjustment did not significantly alter differences in outcome for the lisinopril versus chlorthalidone comparison among blacks in ALLHAT (9). Similarly, in our study, we controlled for baseline differences in BP between both groups using propensity score matching, and we also adjusted for the noted black-white difference in achieved SBP in the outcomes model; as such we cannot attribute the noted differences to relatively less responsiveness (with respect to SBP reduction) of blacks to ACE inhibitors. Another reason may be that blacks are inherently at higher risk for CV events, such as stroke, for the same level of elevated systolic BP. For example, a recent analysis of the REGARDS (Reasons for Geographic And Racial Differences in Stroke study demonstrated that, for the same level of SBP, blacks suffered 3 times more strokes than whites (20). The use of propensity score matching of clinical characteristics of all 4 exposure groups in our study allowed us to control for the inherent differences in CVD risk due to race (being black) in addition to controlling for baseline SBP. Thus, we could not ascertain that our findings are driven by the inherent risk of CVD attributed to being black. Our findings, in addition to those of ALLHAT (9), make a solid case for the comparatively lesser beneficial effects of ACE inhibitors in prevention of cardiovascular outcomes in hypertensive black patients compared to their white counterparts. These findings lend credence to the 2014 evidence-based guideline for the management of hypertension in U.S. adults (21), in which initiation of antihypertensive therapy in black patients precludes the use of ACE inhibitors.
In addition to providing real-world, practice-based clinical evidence on the role of ACE inhibitors in treating hypertension in blacks, findings from this study have the following strengths. First, this study includes the largest number of blacks in a comparative effectiveness research study that we are aware of; such a large sample size with practice-based clinical care data are not usually observed in randomized trials, and the complementary statistical models applied ensure our findings’ internal validity. Second, the racial diversity of the study population in the NYC HHC clinical data warehouse makes inferences drawn from our findings generalizable to hypertensive black patients who receive care in municipal health care settings across the United States. Third, access to care or health insurance coverage does not pose a barrier to patients who receive care within the HHC system, thus minimizing the bias from lack of access to care. Finally, time trends in treatment of hypertension and adoption of treatment guidelines were likely to have been consistent over the study period for which the data were collected for both racial groups.
STUDY LIMITATIONS
We should note the following potential sources of bias that are inherent to analysis of nonrandomized studies and the use of a retrospective cohort study design of ambulatory and hospital databases that may create challenges in establishing causal relationships (22). First is the issue of confounding by indication (19); we employed propensity scores and an incident cohort of patients using IPTW to address this issue in our data analysis. Propensity score models are useful to mitigate selection bias in comparative effectiveness studies using observational data (23). Although we included a wide range of potential confounders in the propensity score model, the possibility remains that unmeasured confounding could bias the results. Second is the uncertainty about the quality of medical record data given that such databases are not designed for research purposes (24). To mitigate this problem, we conducted quality assurance testing with a random sample of 200 participant charts, focusing on major outcome variables. Data from the charts were compared against those obtained from the EHRs and 90% of the data were valid. Finally, our conclusions are somewhat limited because prescription refill data were based on EHR reporting, which makes it difficult to confirm whether or not the prescription was filled. For confirmation of prescription refill, we will have to match these data to a pharmaceutical database, to which we do not have access. This makes it difficult for us to ascertain the differences in medication adherence between blacks and whites or even have a sense of the overall rate of medication adherence within the health care system. Similarly, the mortality data we employed may be incomplete given that some patients may have died outside the hospital system, which can lead to underestimation of mortality rates reported for the various groups.
CONCLUSIONS
In conclusion, ACE inhibitor-based therapy was associated with poorer CV outcomes in hypertensive blacks but not in whites. These findings confirm clinical trial evidence that hypertensive black patients have poorer outcomes than whites when treated with an ACE inhibitor-based regimen.
CENTRAL ILLUSTRATION. Hazard Ratios of Effects of ACE Inhibitor-based therapy versus NonACE-based therapy on CV outcomes in patients with hypertension.
ACE inhibitor use was associated with poorer outcomes in blacks than in whites (8.7% vs. 6.40%). A comparison of the hazard ratios of blacks versus whites indicates that the relative risk for the composite outcome for blacks on ACE compared to NoACE was higher than the relative risk for whites on ACE compared to NoACE. The test for race treatment interaction was significant for the composite outcome with blacks having a higher event rate than whites (blacks versus whites; HR 1.18, 95% CI: 1.00–1.40, p <0.05). Black-ACE group experienced higher rates of AMI, stroke, and CHF compared to white-ACE group. Abbreviations: ACE = angiotensin-converting enzyme.
PERSPECTIVES.
COMPETENCY IN PATIENT CARE
Black patients with hypertension respond less favorably to ACE inhibitors than white patients as reflected in blood pressure lowering and clinical cardiovascular outcomes.
TRANSLATIONAL OUTLOOK
Additional research is needed to understand the mechanisms underlying the disparate clinical effectiveness of ACE-inhibitor medication between blacks and whites.
Acknowledgments
We thank Louis Capponi, MD, FACP Health and Hospital Corporation and his staff (Lisa Elkind and Yan Rosentveg) for their assistance with data retrieval from the data warehouse.
Relationship with industry: Dr. Bangalore received honoraria from Abbott, Boehringher Ingelheim, Daiichi Sankyo, Merck, Gilead and Pfizer. Dr. Ogedegbe was also supported by a grant from the National Heart Lung and Blood Institutes, K24HL111315 in addition to the AHRQ grant R01HS018589. Aside from receiving a grant from AHRQ (grant # R01HS018589; MPI Ogedegbe & Bangalore), to conduct this study, all other authors have reported that they have no relationships relevant to the contents of this paper to disclose. AHRQ did not have any role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, and in the preparation, review, or approval of this manuscript.
ABBREVIATIONS AND ACRONYMS
- ACE inhibitor
angiotensin-converting enzyme inhibitor
- AMI
acute myocardial infarction
- CV
cardiovascular
- CVD
cardiovascular disease
- CHF
congestive heart failure
- EHRs
electronic health records
- HHC
Health and Hospital Corporation
- IPTW
inverse probability of treatment weights
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Mensah GA, Mokdad AH, Ford ES, et al. State of disparities in cardiovascular health in the United States. Circulation. 2005;111:1233–41. doi: 10.1161/01.CIR.0000158136.76824.04. [DOI] [PubMed] [Google Scholar]
- 2.Wong MD, Shapiro MF, Boscardin WJ, Ettner SL. Contribution of major diseases to disparities in mortality. N Engl J Med. 2002;347:1585–92. doi: 10.1056/NEJMsa012979. [DOI] [PubMed] [Google Scholar]
- 3.Heiat A, Gross CP, Krumholz HM. Representation of the elderly, women, and minorities in heart failure clinical trials. Arch Intern Med. 2002;162:1682–8. doi: 10.1001/archinte.162.15.1682. [DOI] [PubMed] [Google Scholar]
- 4.Yusuf S, Sleight P, Pogue J, et al. Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med. 2000;342:145–53. doi: 10.1056/NEJM200001203420301. [DOI] [PubMed] [Google Scholar]
- 5.Taylor AL, Wright JT., Jr Should ethnicity serve as the basis for clinical trial design? Importance of race/ethnicity in clinical trials: lessons from the African-American Heart Failure Trial (A-HeFT), the African-American Study of Kidney Disease and Hypertension (AASK), and the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) Circulation. 2005;112:3654–60. doi: 10.1161/CIRCULATIONAHA.105.540443. discussion 3666. [DOI] [PubMed] [Google Scholar]
- 6.Julius S, Alderman MH, Beevers G, et al. Cardiovascular risk reduction in hypertensive Black patients with left ventricular hypertrophy: the LIFE study. J Am Coll Cardiol. 2004;43:1047–55. doi: 10.1016/j.jacc.2003.11.029. [DOI] [PubMed] [Google Scholar]
- 7.Norris K, Bourgoigne J, Gassman J, et al. Cardiovascular outcomes in the African American Study of Kidney Disease and Hypertension (AASK) Trial. Am J Kidney Dis. 2006;48:739–51. doi: 10.1053/j.ajkd.2006.08.004. [DOI] [PubMed] [Google Scholar]
- 8.Wright JT, Jr, Bakris G, Greene T, et al. Effect of blood pressure lowering and antihypertensive drug class on progression of hypertensive kidney disease: results from the AASK trial. JAMA. 2002;288:2421–31. doi: 10.1001/jama.288.19.2421. [DOI] [PubMed] [Google Scholar]
- 9.Wright JT, Jr, Dunn JK, Cutler JA, et al. Outcomes in hypertensive Black and nonBlack patients treated with chlorthalidone, amlodipine, and lisinopril. JAMA. 2005;293:1595–608. doi: 10.1001/jama.293.13.1595. [DOI] [PubMed] [Google Scholar]
- 10.Morimoto T, Gandhi TK, Fiskio JM, et al. An evaluation of risk factors for adverse drug events associated with angiotensin-converting enzyme inhibitors. J Eval Clin Pract. 2004;10:499–509. doi: 10.1111/j.1365-2753.2003.00484.x. [DOI] [PubMed] [Google Scholar]
- 11.Officers A, Coordinators for the ACRGTA, Lipid-Lowering Treatment to Prevent Heart Attack T Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) JAMA. 2002;288:2981–97. doi: 10.1001/jama.288.23.2981. [DOI] [PubMed] [Google Scholar]
- 12.Exner DV, Dries DL, Domanski MJ, Cohn JN. Lesser response to angiotensin-converting-enzyme inhibitor therapy in Black as compared with White patients with left ventricular dysfunction. N Engl J Med. 2001;344:1351–7. doi: 10.1056/NEJM200105033441802. [DOI] [PubMed] [Google Scholar]
- 13.Hahn RA, Stroup DF. Race and ethnicity in public health surveillance: criteria for the scientific use of social categories. Public Health Rep. 1994;109:7–15. [PMC free article] [PubMed] [Google Scholar]
- 14.Williams DR. Race/ethnicity and socioeconomic status: measurement and methodological issues. Int J Health Serv. 1996;26:483–505. doi: 10.2190/U9QT-7B7Y-HQ15-JT14. [DOI] [PubMed] [Google Scholar]
- 15.Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158:915–20. doi: 10.1093/aje/kwg231. [DOI] [PubMed] [Google Scholar]
- 16.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–83. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 17.Hirano K, Imbens CW. Estimation of causal effects using propensity score weighting: An application to data on right heart catheterization. Health Serv Outcomes Res Methodol. 2001;2:259–78. [Google Scholar]
- 18.Imbens GW. The role of the propensity score in estimating dose-response functions. Biometrika. 2000;87:706–10. [Google Scholar]
- 19.Salas M, Hofman A, Stricker BH. Confounding by indication: an example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981–3. doi: 10.1093/oxfordjournals.aje.a009758. [DOI] [PubMed] [Google Scholar]
- 20.Howard G, Lackland D, Kleindorfer D, et al. Racial differences in the impact of elevated systolic blood pressure on stroke risk. JAMA Intern Med. 2013;173:46–51. doi: 10.1001/2013.jamainternmed.857. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.James P, Oparil S, Carter B, et al. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8) JAMA. 2014;311:507–20. doi: 10.1001/jama.2013.284427. [DOI] [PubMed] [Google Scholar]
- 22.Shah NR. The promise of evidence-based medicine for clinical decision making. Am Health Drug Benefits. 2008;1:45–52. [Google Scholar]
- 23.Pizer SD. An intuitive review of methods for observational studies of comparative effectiveness. Health Serv Outcomes Res Method. 2009;9:54–68. [Google Scholar]
- 24.Logan JR, Gorman PN, Middleton B. Measuring the quality of medical records: a method for comparing completeness and correctness of clinical encounter data. Proc AMIA Symp. 2001:408–12. [PMC free article] [PubMed] [Google Scholar]