Racial Disparity Persists After On-Pump and Off-Pump Coronary Artery Bypass Grafting
Background— Race has been shown to be an independent risk factor for operative mortality after coronary artery bypass grafting (CABG). This study sought to determine the extent to which race is a risk factor for adverse events, long-term mortality, and whether off-pump surgery (OPCAB) modifies that risk.
Methods and Results— The Society of Thoracic Surgeons Adult Cardiac Database at Emory Healthcare affiliated hospitals was queried for all primary isolated CABG records from 1997 to 2007. A propensity score was formulated to balance the patient groups with respect to treatment assignment (OPCAB or CABG on cardiopulmonary bypass). Multivariable logistic regression was used to assess the impact of black race and OPCAB on in-hospital outcomes (death, stroke, myocardial infarction, and their composite, major adverse cardiac events). Cox proportional hazards regression model and Kaplan–Meier curves determined whether black race affected long-term all-cause mortality. Interaction terms were constructed to test whether OPCAB surgery influences surgical results differently in black patients than in white patients. There were 12 874 consecutive CABG patients, including 2033 (15.8%) blacks and 10 841 (84.2%) whites. Survival at 3, 5, and 10 years for blacks (87.5%, 81.4%, 63.8%) was significantly lower than for whites (90.7%, 85.2%, 67.1%, P<0.001). Blacks (adjusted odds ratio, 0.77; 95% CI, 0.44 to 1.36) and whites (adjusted odds ratio, 0.72; 95% CI, 0.53 to 0.99) who had OPCAB had lower risk-adjusted odds of major adverse cardiac events than their racial counterparts who had CABG on cardiopulmonary bypass.
Conclusions— Short- and long-term outcomes are significantly worse in black than in white patients undergoing primary isolated CABG. OPCAB does not narrow the disparity in outcomes between blacks and whites.
Racial disparities in clinical medicine have been noted for some time. In 1999, the United States Congress requested the Institute of Medicine to study the extent to which such disparities existed throughout health care in America. These findings along with factors contributing to the inequities and recommendations for their elimination were published in the Institute of Medicine report “Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care,” published in March of 2002.1
The report found a number of disparities in access, treatment, and outcomes for minorities related to cancer, diabetes, and heart disease. In particular, blacks were less likely to receive recommended treatment for coronary artery disease or myocardial infarction and were less likely to be referred for percutaneous coronary interventions and bypass surgery. These findings have been substantiated and highlighted in several publications before and subsequent to the Institute of Medicine report.2–8 Two studies from large database registries have confirmed that race is an independent risk factor for adverse events and mortality after coronary artery bypass grafting (CABG).3,4
As a means of reducing the morbidity and mortality associated with the use of cardiopulmonary bypass (CPB) required for performance of conventional CABG, beating heart, or off-pump coronary artery bypass (OPCAB) was introduced in the United States in the mid-1990s.9 In 2007, approximately 20% of all isolated CABG operations were performed with off-pump techniques (www.sts.org).10 OPCAB has been shown to be beneficial for patients at increased risk of stroke, diabetic patients, and women.11–15 In particular, we have previously demonstrated a significant early mortality and morbidity advantages for women and equal 10-year survival compared with men.12 Specifically, among OPCAB patients, we have previously reported the mitigation of adverse events and equalization of outcomes in women compared with men.14 The present study seeks to determine whether the disproportionate advantage of OPCAB observed in 1 high-risk demographic group (women) may also be present for another identifiable high-risk demographic group (blacks).
The Society of Thoracic Surgeons (STS) Adult Cardiac Database was searched for all patients who underwent primary isolated CABG at Emory Health Care Hospitals (Emory University Hospital and Emory Crawford Long Hospital) and our affiliate Wellstar-Kennestone Hospital between January 1, 1997, and March 30, 2007. This time frame was chosen to coincide with the entire period during which multivessel OPCAB techniques were practiced institutionally. The study cohort consisted of 12 874 consecutive patients including urgent and emergent patients. Of these, 10 841 (84.2%) were white and 2033 (15.8%) were black. Patients of other races were excluded from this study. Extracted records from this retrospective study included demographic data, preexisting comorbidities, surgeon identity, operative strategy, and clinical outcomes. The study was approved by the Emory University Institutional Review Board in compliance with HIPAA regulations and the Declaration of Helsinki. The institutional review board waived individual patient consent.
Interventions, Surgeons, and Surgical Technique
Each patient underwent a single surgical session consisting of either OPCAB or CABG on CPB, performed at the discretion of any of 17 faculty surgeons. OPCAB techniques were performed with one of several commercially available cardiac positioning and coronary artery stabilizing devices. Conventional CABG on CPB was performed with standard techniques, using roller head pumps, membrane oxygenators, cardiotomy suction, arterial filters, cold antegrade and retrograde blood cardioplegia, and medium systemic hypothermia (30° to 34°C). Patients who were converted intraoperatively from OPCAB to CABG on CPB or from CABG on CPB to OPCAB were entered into the database according to the operation they ultimately received. Intraoperative conversion only began to be recorded as part of the institutional database in 2002, when a data field for conversion was introduced into the STS national adult cardiac database.
The Social Security Death Index database of death records is extracted from the United States Social Security Administration’s Death Master File Extract. Persons with a Social Security number who have died since 1963 and whose death was reported to the Social Security Administration are listed in the Social Security Death Index. Thus, for each patient who died before the cutoff date of March 30, 2007, a mortality date was provided, allowing construction of Kaplan–Meier long-term survival curves. Cause of death is not available; thus, this study seeks to describe all-cause long-term mortality.
Variables of Interest
Before analysis, preoperative risk factors for the outcomes of interest were identified and harvested from the STS database (Table 1). Standard STS definitions of each risk factor and outcome were used. Race was dichotomized as either white or black. The outcomes examined in this study were in-hospital mortality, permanent stroke, myocardial infarction, and the composite end point for any of these major adverse cardiac events (MACE).
The institutional medical records database was populated by trained personnel devoted exclusively to this task; thus, missing data were scarce. Data were 100% complete for the critical risk factors of interest (surgery type and race) as well as for each major postoperative hospital outcome. Data were missing for the following variables; body mass index (n=782, 6.1%), body surface area (n=782, 6.1%), height (n=704, 5.5%), and weight (n=749, <5.8%). Ejection fraction (n=1622, 12.6%), STS predicted risk of mortality (n=4, <0.1%), number of diseased vessels (n=919, 7.1%), and last creatinine level (n=2495, 19.4%).
Data Management and Statistical Analysis
All data for consecutive patients were entered into a computerized cardiac surgical database, using the regimented data fields and definitions of the STS National Adult Cardiac Database. Checks for data quality are used at the institutional level and before entry into the STS national adult cardiac database.
A multiple imputation algorithm was used to impute values that reflect the uncertainty surrounding the missing data. This was not performed to recreate the true variables; rather, the goal of the imputation was to avoid selection bias that can occur by deleting cases with missing variables of interest. Ten data sets were imputed, and parameter estimates from the 10 data sets were combined using methods comprehensively described by Schafer16 and Molenberghs and Kenward.17 Data were assumed to be missing at random.
Patients were classified primarily according to their race and the surgery type (OPCAB or CABG on CPB) they received. To help control for potential selection bias, propensity scores, described by Blackstone18 and D’Agostino,19 were calculated for each patient, based on 49 risk factors available before surgery (including surgeon identity, year of surgery, and 9 indicators of missingness; Table 1). The goal of the propensity score adjustment is to “postrandomize” or “balance” the groups with respect to their preoperative risk factors so that unbiased comparisons of group effects can be estimated. For the propensity score calculation, a multiple logistic regression model was used nonparsimoniously to model OPCAB (yes or no) as a function of all 49 risk factors. The resulting conditional probability of a patient receiving OPCAB is the propensity score (PS). The PS was then used as a regression covariate in the logistic regression models (see below). The assumption of a linear relationship between the propensity score and the logit of the predicted probabilities of each outcome was verified by plotting these quantities in deciles. Additionally, the mean PS was very similar between the white and black patients (0.46 versus 0.54) and the variances were nearly identical (0.605 versus 0.606), both requirements before using the PS as a regression covariate.
To statistically evaluate the effects of race and surgery type, multivariable logistic regression models were constructed for each outcome of interest. Each model consisted of dichotomous race (black or nonblack), surgery type (OPCAB or CABG on CPB), and their interaction, adjusted for the PS. A significant interaction would indicate that surgery type produces different outcomes in the races and thus would be the focus. However, if the interaction was statistically insignificant, then that term was removed from the model and the main effects alone were evaluated. This model was primarily designed to evaluate the effects of black race and surgery type in unison and to determine whether OPCAB lessened or worsened outcomes in the presence of black race. Adjusted odds ratios (AORs) associated with OPCAB and black race, along with 95% CIs, were computed for each of the 4 adverse outcome end points. All logistic models were adjusted with the propensity score.
Kaplan–Meier product-limit estimates (unadjusted) and associated Kaplan–Meier curves were generated that provide survival estimates at postoperative points in time. Unadjusted differences between races in long-term survival were determined by log-rank tests. These estimates include operative deaths.
Long-term survival comparisons were made using adjusted Cox proportional hazards regression to model the instantaneous hazard of death as a function of race, surgery type, and their interaction (if significant), adjusted for the patient age and PS. For this analysis, a separate PS was calculated that omitted age from its formulation. The proportional hazards assumptions were verified via a correlation analysis of the Schoenfeld residuals and ranked follow-up time. Hazard ratios (HRs) were generated for each model term, along with 95% CIs.
The data were managed and analyzed using SAS Version 9.1 (SAS Institute, Inc, Cary, NC). Unadjusted comparisons were performed with χ2 tests and 2-sample t tests for categorical and numeric predictors, respectively. All statistical tests were 2-sided, using an α=0.05 level of significance. No adjustments for multiple tests were made.
Preoperative Patient Characteristics
Preoperative characteristics of the entire cohort, 12 874 patients, are listed in Table 1. There were 2033 (16%) blacks and 12 841 (84%) whites. Blacks were younger (60.0 versus 63.3, P<0.001) and had an increased preoperative risk profile relative to whites. In particular, blacks had a statistically higher prevalence of diabetes (46.8% versus 33.6%), renal failure (14.5% versus 5.5%), peripheral vascular disease (11.4% versus 8.6%), and congestive heart failure (23.1% versus 15.5%. P<0.001 for each comparison).
Among blacks, more women were treated with OPCAB than CABG on CPB (45.7 versus 38.2%, P<0.001). Additionally, among the black patients, the OPCAB group had more patients with renal failure(15.7% versus 13.2%, P=0.12), renal failure requiring dialysis (8.9% versus 5.8%, P=0.008), chronic obstructive pulmonary disease (13.4% versus 9.4%, P=0.005), hypertension (89% versus 86.3%, P=0.07), peripheral vascular disease (14.4% versus 8.0%, P<0.001), stroke (14.1% versus 9.0%, P<0.001), and heart failure (26.0% versus 19.9%, P=0.001; Table 1).
Short-Term Unadjusted Outcomes
As expected, based on increased preoperative risk profile, black patients both in the OPCAB and CPB groups had higher rates of death (OPCAB, 2.3% versus 1.27%, P=0.003; CPB, 3.38% versus 2.25%, P=0.001), stroke (OPCAB, 2.3% versus 0.95%, P<0.001; CPB, 2.85% versus 1.75%, P=0.001) and MACE (OPCAB, 4.6% versus 2.52%, P<0.001; CPB, 6.34% versus 4.29%, P<0.001) than their white counterparts. Among blacks, OPCAB patients had lower rates of death (2.3% versus 3.38%, P<0.001), stroke (2.3% versus 2.85% P=0.001), and MACE (4.6% versus 6.34%, P<0.001) than those in the CABG on CPB group. Similarly, white patients in the OPCAB group had lower rates of death, stroke, and MACE than those treated with CABG on CPB. The incidence of myocardial infarction in and among all groups was similar whether done on or off pump. Among CABG on CPB patients, female patients had higher rates of death (4.1% versus 1.8%, P<0.001), stroke (2.9% versus 1.6%, P<0.001), myocardial infarction (1.3% versus 0.5%, P<0.001), and MACE (7.4% versus 3.5%, P<0.001). However, among OPCAB patients, major complication rates were similar between the sexes—death (1.7% versus 1.3%, P=0.25), stroke (1.5% versus 1.0%, P=0.12), myocardial infarction (0.8% versus 0.5%, P=0.12)—but statistically higher for women when considering the composite end point of MACE (3.6% versus 2.5%, P=0.019; Table 2). Observed to expected STS predicted risk of mortality was similar among black and white patients (Table 3).
Propensity-Adjusted Comparisons of Short-term Outcomes
When compared with CABG on CPB, OPCAB was associated with a significantly lower adjusted rate of stroke (AOR, 0.55; 95% CI, 0.36 to 0.83; P=0.004) and MACE (AOR, 0.74; 95% CI, 0.56 to 0.97; P=0.029). Both black (AOR, 0.77; 95% CI, 0.44 to 1.36; P=0.37) and white (AOR, 0.72; 95%CI, 0.53 to 0.99; P=0.046) patients who had OPCAB had lower risk-adjusted odds of MACE than their racial counterparts who had CABG on CPB. When compared with white patients, blacks had significantly higher adjusted risk of death (AOR, 1.69; 95% CI, 1.25 to 2.28; P<0.001), stroke (AOR, 1.93; 95% CI, 1.40 to 2.66; P<0.001), and MACE (AOR, 1.68; 95%CI, 1.35 to 2.09; P<0.001). No significant interaction existed. OPCAB did not mitigate the racial disparity in outcomes after CABG (Table 4).
Estimates of all-cause and 1-, 3-, 5-, and 10-year survival are listed in Table 5 and represented as Kaplan–Meier survival curves in the Figure. The log-rank test revealed the Kaplan–Meier survival curves between black and white patients were not equal. Blacks have poorer long-term survival after CABG than whites regardless of whether CABG is performed on or off pump. Blacks had worse long-term survival than whites at 1 (92% versus 95.1%), 3 (87% versus 90.7%), 5 (81.4% versus 85.2%), and 10 years (63.8% versus 67.1%).
The Cox proportional hazards regression model was used to assess the effect of surgery type and race on long-term survival. Adjusted long-term survival was similar after OPCAB or CABG on CPB for patients of either race (OPCAB HR, 1.08; 95% CI, 0.95 to 1.22; P=0.25). Blacks had shorter adjusted survival than whites (HR, 1.50; 95% CI, 1.33 to 1.68; P<0.001).
Poorer outcome for blacks after coronary artery bypass procedures is not a novel observation. The current study validates previous reports that there remains a disparity in the presentation of and consequently the outcomes of black patients undergoing CABG.2–8 The first such comparative studies to document these differences in modern times were conducted in the latter decades of the 20th century.20–23 Oberman et al20 studied the natural history of coronary artery disease in blacks and compared outcomes between blacks and whites after coronary arteriography and CABG. Whites with 3-vessel disease were almost 3 times as likely to have CABG as blacks (42.5% versus 16.6%).19 Patients in that study over the age of 50 also had decreased survival after CABG than whites. Maynard et al21 found no significant difference in 5-year survival between blacks and whites. This study used data obtained from the Coronary Artery Surgery Study database. Higgins et al6 examined the effect of payor status on outcomes after CABG and found an operative mortality of 5.5% for blacks versus 2.5% for whites. Bridges et al4 found a small but significant difference in mortality rate for blacks (3.83%) when compared with whites (3.14%). In particular, the difference was most pronounced among lower-risk patients. Hartz et al,5 using the same STS database, corroborated these results and confirmed that race and sex independently predict adverse outcomes after bypass surgery.
Delineating why these differences exist has been much more difficult. Although the literature has suggested a number of potential confounders, including socioeconomic status, education level, and other less well-defined socio-cultural factors, none have been proven.23–24
Despite these differences, the current study sought to determine the impact of OPCAB on outcomes for blacks. Using our institutional database, we have previously found a significantly beneficial effect of OPCAB on women.12 In that study, among women treated with CABG on CPB, there was a doubling of mortality (AOR, 2.07; 95% CI, 1.24 to 3.44; P=0.005) compared with women treated off pump. Furthermore, after risk adjustment, OPCAB significantly narrowed the sex gap in outcomes. Women treated on-pump had an AOR of death of 2.31 when compared with men treated on-pump. The AOR narrowed to 1.29 (P=NS) for women treated off-pump versus men treated off-pump.
The effect of OPCAB on outcomes among blacks is notable. In general, blacks in the OPCAB group had a significantly higher risk profile, as evidenced by the increased prevalence of renal failure, dialysis, peripheral vascular disease, congestive heart failure, hypertension, and chronic obstructive pulmonary disease (Table 1). Further, blacks were treated more liberally with statin and angiotensin-converting enzyme/angiotensin receptor blocker inhibitors than white patients in this patient series; black patients had higher postoperative rates of statins (50.4% versus 39.1%, P<0.001) and statistically similar rates of angiotensin-converting enzyme/angiotensin receptor blocker inhibitor regimens (36.3% versus 34.9%, P=0.23). Despite these differences, blacks having OPCAB had a significant improvement in unadjusted outcomes compared with those having CABG on CPB (Table 2).
To our knowledge, this is the first study to examine the effect of both short- and long-term outcomes for blacks treated with OPCAB. Although we have no definitive explanation for these differences, within the constraints of a single-institution database, we examined and adjusted through a detailed propensity analysis the effect of >49 preoperative variables. It is interesting that the black patients are younger and have more hypertension, renal failure, heart failure, diabetes, stroke, and peripheral vascular disease. Could these in fact be markers of some yet-undetermined biological predisposition to the stress response of surgery? This has yet to be determined and is beyond the scope of this study. What does remain clear is that blacks are predisposed to significant untoward outcomes after coronary artery bypass surgery.
We acknowledge the limitations of our study. In particular, this was a single-institution, retrospective, registry review. We also acknowledge that despite the use of advanced, statistical methods, there may be confounders of risk that have yet to be determined. In addition, we did not use an intention-to-treat analysis. Therefore, as discussed previously, patients who were converted intraoperatively from OPCAB to CABG on CPB and vice versa were analyzed according to the operation they received in the primary analysis. We have previously noted from our institutional database no significant impact of these conversions on postoperative outcomes.12 In this study, for the patients available from 2002 to the end of the study period, the complication rates for patients converted from OPCAB to CABG on CPB do not significantly differ from contemporaneously treated CABG on CPB patients. Specifically, the rates for converted CABG on CPB patients (n=86) and nonconverted CABG on CPB patients (n=1725) since 2002 are as follows: death (2.3% versus 2.8%, P=0.80), stroke (3.5% versus 1.6%, P=0.19), myocardial infarction (0.0% versus 0.8%, P=0.42), and MACE (5.8% versus 4.6%, P=0.61). Furthermore, we examined “all-cause” long-term mortality and not necessarily surgery- or disease-specific mortality.
In conclusion, blacks undergoing coronary artery bypass procedures are at increased risk of death, stroke, and the composite of death, stroke, and myocardial infarction, compared with white patients. OPCAB significantly reduces risk of morbidity and mortality among both blacks and white patients. OPCAB does not neutralize the racial disparity in outcomes for blacks compared with whites. Future research on racial disparities should focus on preoperative risk modification and operative techniques that neutralize risk for high-risk blacks requiring coronary artery bypass surgery.
Drs Cooper, Thourani, and Puskas are Maquet Cardiovascular consultants. Drs Thourani, Lattouf, Guyton, and Puskas are Medtronic, Inc consultants.
Sources of Funding
Drs Puskas, Guyton and Thourani receive clinical research funding from Maquet and Medtronic.
Presented in part at American Heart Association Scientific Sessions 2008, November 8–12, 2008, New Orleans, La.
The Institute of Medicine. Unequal Treatment: Confronting Racial and Ethnic Disparities in Healthcare. Washington, DC: National Academies Press. March 2002.
Higgins RS, Paone G, Borzak S, Jacobsen G, Peterson E, Silverman NA. Effect of payor status on outcomes of coronary artery bypass surgery in blacks. Circ Suppl. 1998; 98: II-46–II-50.
Becker ER, Rahimi A. Disparities in race/ethnicity and gender in in-hospital mortality rates for coronary artery bypass surgery patients. J Nat Med Assn. 2006; 98: 1729–1739.
The Society of Thoracic Surgeon’s National Adult Cardiac Surgery Database. 2007 report. Available at: www.sts.org. April 2008.
Puskas JD, Kilgo, Kutner M, Pusca SV, Lattouf O, Guyton RA. Off-pump techniques disproportionately benefit women and narrow the gender disparity in outcomes after coronary artery bypass surgery. Circulation. 2007; 116 (suppl I): I-192–I-199.
Schafer JL. Analysis of Incomplete Multivariate Data. Boca Raton, FL: Chapman & Hall/CRC; 1997.
Molenberghs G, Kenward MG. Missing Data in Clinical Studies. 1st ed. New York, NY: John Wiley and Sons; 2007: 105–117.