Previous research has suggested that race and gender influence physicians' decisions to use invasive coronary procedures for patients with coronary artery disease. However, most studies have not sufficiently accounted for clinical characteristics that might explain treatment decisions. The purpose of this study was to evaluate race and gender patterns in use of coronary artery bypass surgery (CABG) after accounting for the clinical appropriateness of this procedure as rated by commonly accepted guidelines (developed by the RAND Corporation).

A weighted random sample of eight of the fifty-six New York State hospitals that provide angiography was selected for this study; hospitals that had more African American and Hispanic patients receiving angiograms for coronary artery disease in 1991 were given a higher weight (were more likely to be selected) in order to ensure that a sufficient number of minority patients were recruited for the study. From this hospital sample, all African American and Hispanic patients, 15% of the non-Hispanic white men, and 33% of the non-Hispanic white women were selected. Patients who had undergone previous CABG procedures or were emergency admissions were excluded. Data were collected from patient charts, telephone and mail surveys, and the New York Cardiac Surgery Reporting system for the period from November 1994 to June 1996. Each patient was rated in terms of appropriateness for each of three procedures: CABG, percutaneous transluminal coronary angioplasty (PTCA), and medical therapy. Then each patient was rated as to whether these procedures were necessary. This study assessed racial patterns in use of CABG within three months of angiography among those who were rated as being appropriate for only CABG (and not for the other procedures), and, among these patients, for whom CABG was considered necessary.

Of the 4,905 patients who received angiograms for coronary artery disease (CAD), 1,261 were found to be appropriate for CABG and inappropriate for PTCA and medical therapy.

Of the 680 non-Hispanic whites appropriate for CABG, 57% received surgery; of 314 African Americans appropriate for CABG, 45% received surgery; and of 267 white Hispanics appropriate for CABG, 46% received surgery. These patterns were statistically significant (p<0.001). Baseline characteristics of each racial group were evaluated. African American and Hispanic patients were younger, more likely to be women, and more likely to be Medicaid-insured or uninsured. Additionally, African American and Hispanic patients were less likely to have left main disease and to be “high risk,” but more likely to have 3-vessel disease and to have low ejection fractions. Logistic regression analyses showed that having 3-vessel disease and left main disease increased the odds of receiving CABG among those appropriate for CABG; and being of age greater than 80 years, having Medicaid or no insurance, and being of African American and Hispanic ethnicity significantly decreased the odds of receiving CABG. With this model, the odds ratio (OR) for African Americans versus non-Hispanic whites was 0.64 (95% confidence interval was 0.47 to 0.87), and the OR for Hispanics versus non-Hispanic whites was 0.60 (95% confidence interval was 0.43 to 0.84). It is noteworthy that gender was not predictive of CABG in this model.

CABG surgery was necessary for a total of 702 patients. Of this group, 63% of the non-Hispanic whites, 49% of the African Americans and 57% of the Hispanics received surgery. (It is noteworthy that the percentage of patients who received surgery among those for whom it was appropriate was similar to the percentage of patients who received surgery among those for whom it was deemed necessary for each racial group.) Logistic regression analyses among the subgroup for whom CABG surgery was necessary showed that having 3-vessel disease and left main disease increased the odds of CABG; and being of age greater than 80 years, having Medicaid or no insurance, and being of African American ethnicity significantly decreased the odds of CABG. The OR for African American versus non-Hispanic white in this subgroup was 0.63 (95% confidence interval was 0.42 to 0.94), and the OR for Hispanic versus non-Hispanic white was 0.87 (95% confidence interval was 0.56 to 1.34). Gender and Hispanic ethnicity were not predictive of CABG in this model.

The authors further evaluated physician decision-making by conducting a mail survey of physicians who treated a sample of the patients who were deemed appropriate for CABG but did not receive surgery. For this analysis, 950 of the original 4,905 patients were selected, oversampling patients appropriate for CABG who did not undergo surgery. Physicians were asked if they had recommended surgery and, if not, why not. Ninety-two percent of these physicians were cardiologists, 3.2% were family practitioners, and the rest were other specialties.

Responses were obtained from physicians of 717 patients. Among these patients, 464 were appropriate for CABG only, 268 were appropriate for CABG and did not receive surgery, and 241 of these 268 had complete physician responses. Regarding these 241 patients, 90% of the physicians did not recommend surgery, varying slightly by race and gender: 85% for white males and 92% for white females, 95% for African American males and 88% for African American females, and 87% for Hispanic males and 97% for Hispanic males. (These differences were not statistically significant.)

For 19 of the 217 patients who were not recommended for surgery, the physician said he/she preferred to try an intervention other than CABG first, and, for 18 of the 217 patients, the physician said he/she preferred to try PTCA specifically first. For 150 patients, the physician said the patient was not reasonable for CABG surgery. The most frequent explanations for "not reasonable" were "coronary anatomy" (87%) and "left ventricular function" (20%). No reasons were given for the other 30 patients.

The authors conclude that "this study provided the most conclusive evidence to date of significant difference in access to CABG surgery according to race/ethnicity." Due to possible racial differences in access to cardiac catheterization, the authors expect that the observed differences in access to CABG would be exacerbated if the patients had been identified before cardiac catheterization, which is a precursor to CABG surgery. Additionally, this study controlled for payer status even though it might be argued that adjustments should not be made for payer status if the primary focus is to examine utilization as a function of race/ethnicity. Thus, these results are conservative estimates of the race/CABG utilization association. Since most of the patients appropriate for but not undergoing surgery were not recommended surgery by their physicians (90%), it is unlikely that the racial patterns in receipt of surgery were due to patient refusals. The authors argue that "the compelling findings suggest the need for more research into the incentives for providing CABG surgery and the psychosocial and encounter factors than may have a bearing on patient referrals."

The authors additionally note that only 52% of those appropriate for surgery and 58% of those for whom surgery was necessary received CABG surgery. It is important to consider why the surgery rates were low. The authors suggest that this may be due to angiographic unsuitability resulting from distal or diffuse disease or small vessels that cannot be bypassed, conditions that were not measured in this study. It is possible that African Americans "were more angiographically unsuitable than white patients."

Several limitations were identified by the authors, including representativeness of the sample and the potential bias due to missing patients who had CABG surgery out of state.