Racial Differences in Long-term Survival among Patients with Coronary Artery Disease

Kevin L. Thomas, MD, FACC; Emily Honeycutt, MBI; Linda K. Shaw, MS; Eric D. Peterson, MD, MPH, FACC

Disclosures

Am Heart J. 2010;160(4):744-751. 

In This Article

Methods

Patient Population

From January 1986 through December 2004, we examined the records of 22,618 patients who underwent cardiac catheterization at Duke University Medical Center for the primary indication of diagnosing the presence of obstructive CAD. This analysis was limited to patients with significant CAD (stenosis >70% in 1 or more vessels). Patients were excluded from the analysis if their race was missing or classified other than black or white; if a left ventriculogram was not performed; or if the patient had prior coronary artery bypass surgery (CABG), percutaneous coronary intervention (PCI) within the last year, congenital heart disease, or severe valvular heart disease. The final study population consisted of 19,304 (85.3%) white patients and 3,314 (14.7%) black patients (Figure 1). There were no racial differences among the patients excluded from the initial population. At the time of this data collection, racial designation was determined by the administrative staff during registration by observation, direct question to the patient, or review of the medical record. Based on the racial composition of patients receiving care in the Duke University Health System, the DDCD categorized race as black, white, American Indian, or other. The small number (<3% of DDCD) of other racial/ethnic groups including Hispanic patients undergoing cardiac catheterization at Duke University Medical Center made inclusion of this group in our analysis prohibitive.

Figure 1.

Patient flow applying exclusion criteria to study population.

Data Collection

The DDCD is a clinical databank that consists of the ongoing data retrieval of all patients who undergo diagnostic cardiac catheterization at Duke University Medical Center (Durham, NC). Data collection of baseline clinical variables for each patient is performed as part of routine patient care and was stored in the DDCD according to previously described methods.[12–16] Patients whose data are in the DDCD are followed routinely at 6 months, 1 year, and then annually using mailed questionnaires and telephone interviews for nonresponders. Patients in this study population were contacted yearly after catheterization for vital status including death, cardiovascular events, rehospitalization, and any type of revascularization procedure during the preceding year. Patients who were not successfully contacted through the aforementioned mechanism had vital status determined through a search of the National Death Index. Follow-up for this study population was complete through July 1, 2006, and was 99.2% complete for this study population. In addition to clinical data, ZIP code–based income and domicile value data were obtained from the 2000 Census report to approximate socioeconomic status. The protocol for this analysis was reviewed and approved by the Duke University Internal Review Board.

Statistical Analysis

Baseline characteristics were described by medians and interquartile ranges (IQRs) (from 25th to 75th percentile) for continuous variables and by percentages for categorical variables. The comparisons of baseline characteristics by race were analyzed using χ2 tests for categorical variables or Wilcoxon rank sum tests for continuous variables. Unadjusted survival results were examined using Kaplan-Meier methods, and comparisons between groups were made using the log-rank test. Variables were examined using Cox proportional hazards regression modeling in both the unadjusted and adjusted setting. Continuous and ordinal categorical variables were tested for linearity over the log hazard and were transformed as necessary to meet this modeling assumption. Specifically, body mass index (BMI) has a point of inflection at 26; and diastolic blood pressure has an inflection point at 76 in our data. Variables thought to have clinical importance as well as those with P < .10 in the univariable analysis were candidate variables in stepwise multivariable modeling. These variables are typical of those used in subjects with cardiovascular disease when analyzing survival. The candidate variables included the following: age, gender, race, heart rate, systolic and diastolic blood pressure, year of treatment, acute MI within 3 days before cardiac catheterization, history of peripheral vascular disease (PVD), hypertension, history of diabetes mellitus, history of smoking, hyperlipidemia, history of heart failure (HF), New York Heart Association classification of HF, family history of CAD, mild valvular heart disease, history of MI, MI within the prior 6 weeks, ejection fraction (EF), history of cerebrovascular disease, left main stenosis ≥50%, severity of cardiovascular disease (number of diseased vessels), S3 gallop, carotid bruits, BMI, renal insufficiency, chronic obstructive pulmonary disease (COPD), connective tissue disease, end-organ complications of diabetes mellitus, mild/moderate/severe liver disease, peptic ulcer disease, any tumor, metastatic cancer, aids, leukemia, lymphoma, dementia, revascularization within 30 days of index catheterization, and markers of socioeconomic position (ZIP code level median household income and house value).

Patients were assigned to PCI or CABG groups, respectively, if they underwent these procedures within 30 days of the initial cardiac catheterization. Patients who did not undergo revascularization within 30 days were designated as the medical therapy group. For those analyses stratified by revascularization, patients who died within 5 days of the initial cardiac catheterization without coronary revascularization (the median time to PCI or CABG at Duke University Medical Center) were excluded from the analysis to avoid attributing early deaths to the medically treated cohort. In addition, because coronary revascularization was not randomly assigned, we adjusted for physician propensity to select revascularization as a treatment option. Logistic regression analysis was used to determine the probability of the physician's propensity to choose coronary revascularization as treatment option for patients based on clinical variables. Clinical data available to clinicians were used as candidate variables in developing the propensity model. This propensity score was used as a covariate in the final survival model.

Given the objectives of this study, we specifically examined interactions between race and several important variables: age, gender, EF, diabetes mellitus, COPD, BMI, PVD, cerebrovascular disease, smoking, renal insufficiency, hypertension, and markers of socioeconomic position. The interactions were tested in a multivariable model that included all significant variables from the main effects model. A P value ≤ .05 was used to indicate statistical significance for all comparisons and analyses. All analyses were performed using SAS version 8 statistical software (SAS Institute, Cary, NC). No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study; all study analyses, the drafting and editing of the paper, and its final contents.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....