Relationship Between Race and Mortality and Morbidity After Valve Replacement Surgery
Background— Previous studies have shown that black race is an independent predictor of increased operative mortality after coronary artery bypass surgery. Given the higher incidence of hypertension and hypertension-associated left ventricular hypertrophy in blacks, we hypothesized that black race might be associated with increased risk of mortality and morbidity after aortic valve replacement (AVR) or mitral valve replacement (MVR). We could not identify a previous study that used a multivariable model to evaluate the association between race and operative mortality after AVR or MVR.
Methods and Results— The Society of Thoracic Surgeons National Cardiac Database was used for a retrospective review of 3137 black and 46 249 white patients who underwent MVR alone or AVR alone from 1999 through 2002. Multivariate logistic regression was used to assess the association between race and mortality and 6 other adverse outcomes (stroke, renal failure, prolonged ventilation, prolonged postoperative stay, sternal infection, and bleeding) after adjustment for covariates. Unadjusted operative mortality for MVR only was 5.60% for blacks versus 6.18% for whites (OR 0.90 [95% CI 0.71 to 1.14]) and 4.60% for blacks versus 3.62% for whites for AVR only (OR 1.28 [95% CI 1.02 to 1.62]). After adjustment for other risk factors, black race was not a significant predictor of operative mortality after AVR or MVR; however, black race was associated with an increased risk of several complications: prolonged ventilation after AVR or MVR, postoperative stay >14 days after AVR or MVR, reoperation for bleeding after AVR, and postoperative renal failure after MVR. There was no significant association between race and the risk of stroke or deep sternal wound infection for either AVR or MVR.
Conclusions— In contrast to previously published results that defined race as an independent risk factor for operative mortality after coronary artery bypass surgery, race does not appear to be a significant predictor of operative mortality after isolated AVR or MVR; however, there is evidence of an association between race and certain complications.
Received October 17, 2004; revision received January 4, 2005; accepted January 11, 2005.
The first prosthetic aortic valve replacement was performed in 1951 by Charles Hufnagel.1 After the heart-lung machine was developed, other advancements followed, such as the first aortic valve homograft, completed by Donald Murray in 1956, and the first orthotopic prosthetic aortic and mitral valve replacements, in the early 1960s.1 Since that time, advances in technology and techniques have resulted in a significant decrease in operative mortality in direct proportion to the increasing prevalence of valve surgeries in the United States, particularly among elderly patients.1–4 Operative mortality for both aortic valve replacement (AVR) and mitral valve replacement (MVR) has been shown to be inversely related to hospital volume.5 An increase in operative mortality after valve replacement surgery has been observed in women, diabetics, and renal failure patients and as a function of advanced age.6,7
In previous studies, we and others have shown that black race is an independent predictor of increased operative mortality after CABG surgery.8,9 With the wealth of literature available about the predictors of outcome after MVR and AVR surgery, we have not been able to identify any studies that explicitly investigated the influence of race on the mortality and morbidity of valve replacement surgery. In view of our earlier results and given the higher incidence of hypertension and hypertension-associated left ventricular hypertrophy in blacks,10 we hypothesized that black race might be associated with increased risk of mortality and morbidity after AVR or MVR. The Society of Thoracic Surgeons National Cardiac Surgery Database (STS database) provides us with the statistical power to effectively investigate these questions.
The STS National Cardiac Database was established in 1989 to report surgical outcomes after cardiothoracic surgical procedures.11,12 The database currently captures clinical information for the majority of all US bypass procedures and includes more than half of the centers that perform adult cardiac surgery in the United States. Sites enter patient data using uniform definitions (available online at http://www.sts.org) and certified software systems. This information is sent semiannually to the STS Data Warehouse and Analysis Center at the Duke Clinical Research Institute. Although participation in the STS database is voluntary, data completeness is high, with overall preoperative risk factors missing in fewer than 5% of submitted cases.13 The accuracy of submitted cases has been confirmed in an independent comparison of hospital CABG volume and mortality rates reported to the STS versus those reported to the Centers for Medicaid and Medicare Services.14 Research performed at the Duke Clinical Research Institute on the STS database is approved by the Duke University Institutional Review Board. Principal investigators at participating institutions are responsible for reviewing their data collection efforts with their sites’ institutional review boards to ensure that patient privacy and confidentiality are protected.
We performed a retrospective review of patients entered into the STS database from 1999 through 2002. The cohort of 3137 black and 46 249 white patients who underwent MVR alone or AVR alone form the basis for this analysis. Exclusion criteria included cardiac procedures other than AVR or MVR (replacement only) and races other than black or white. A retrospective descriptive analysis of these data was done to compare operative mortality and complication rates within the black and white subgroups. Complications included permanent stroke, renal failure, prolonged ventilation (>24 hours), postoperative stay >14 days, deep sternal infection, and reoperation for bleeding. Operative mortality is defined as mortality that occurred at any time in the hospital during the same admission or within 30 days of surgery. Additional definitions are available at www.sts.org.
Descriptive statistics, outcomes, preoperative risk factors, and interventions were measured within and compared between racial subgroups. χ2 analyses were used to evaluate the categorical variables, whereas the Wilcoxon 2-sample test was used to compare differences in means for continuous variables. ORs were calculated for operative mortality rates with a 95% CI. A Mantel-Haenszel χ2 statistic was used to assess the association between race and type of valve prosthesis after adjustment for patient age.
Logistic regression analysis was used to assess the independent association between race and operative mortality and complications after adjustment for 15 known predictors of operative mortality identified in the STS isolated valve replacement model.7 All models included main effects for age; body surface area; diabetes; renal failure (no renal failure/renal failure without dialysis/renal failure with dialysis); hypertension; cerebrovascular accident; infectious endocarditis; chronic lung disease; peripheral vascular disease; previous cardiac operation; New York Heart Association class IV; pulmonary hypertension; myocardial infarction within 3 weeks; status (elective, urgent, emergent or salvage); preoperative intra-aortic balloon pump or inotropes; and the interaction between age and status. Models that predicted renal failure as a complication of surgery included baseline serum creatinine as a covariate and excluded patients with a previous history of renal failure or dialysis.
The null hypothesis of no independent race effect on mortality was tested by the addition of race to a model that included all previously identified main effects and interactions. Interaction between race and other covariates was assessed by the fitting of separate models that predicted mortality for black and white AVR and MVR patients. The fitting of separate models is equivalent to fitting a single model that includes an interaction term for race with each other predictor in the model. Interactions were judged to be statistically significant if they achieved a probability value <0.05. For presentation purposes, the nonsignificant interaction between age and status was removed from the model.
Subsequent analyses were performed to assess whether the relationship between race and mortality varied on the basis of a patient’s overall risk profile. Risk scores were computed with variables from the STS isolated valve replacement model. For both the AVR-only and MVR-only subsets, black and white patients were classified into decile groupings based on their preoperative risk score. Observed mortality rates for black and white patients in each risk group were then calculated along with black-white ORs.
In patients who underwent AVR, there was a significant difference in age as a function of race. Black patients had a mean age of 61±14 years, and white patients had a mean age of 67±14 years (P<0.05). Gender and body surface area were similar in the 2 racial subgroups (Table 1). In the patients who underwent MVR, there was a significant difference in both age and gender as a function of race. Black patients were on average 11 years younger than white patients (53±13 and 64±13 years, respectively, P<0.05). Interestingly, more women than men (≈3:2) underwent MVR in the population as a whole; however, a significantly higher percentage of black patients were female (64.3% versus 58.6% among white patients). Body surface area did not differ between the groups (Table 2).
In the patients who underwent AVR, our evaluation of preoperative risk factors showed significant differences in the majority of variables (Table 1). Given the large sample size, the power of this study allows for small differences to be identified. The subgroup of black patients was found to have an increased occurrence of the following compared with white patients: smoking history, diabetes mellitus, use of preoperative inotropic agents, cardiogenic shock, morbid obesity, renal failure, history of cerebrovascular accident, congestive heart failure, higher New York Heart Association class, elevated pulmonary artery pressure, and lower ejection fraction (50±15% versus 53±14%). A family history of coronary artery disease, hypercholesterolemia, and arrhythmias was more frequent among white patients (Table 1). There were no significant differences in the prevalence of cerebrovascular disease, chronic lung disease, peripheral vascular disease, or treated versus active infectious endocarditis as a function of race.
In the cohort who underwent MVR, black patients had an increased prevalence of smoking, diabetes mellitus, morbid obesity, renal failure, hypertension, a history of cerebrovascular accident, infectious endocarditis, congestive heart failure, and higher pulmonary artery pressures. White patients were more likely to have a significant family history of coronary artery disease, hypercholesterolemia, and arrhythmias than their black counterparts. Similar to the AVR sample, preoperative creatinine was significantly greater in black patients (2.05±2.72 mg/dL versus 1.20±0.98 mg/dL). There were no significant racial differences in the prevalence of cerebrovascular disease, chronic lung disease, peripheral vascular disease, or preoperative cardiogenic shock among the patients who underwent MVR (Table 2).
Several risk factor comparisons varied similarly as a function of race regardless of whether AVR or MVR was the surgical procedure performed. For both types of valve procedures, blacks were significantly more likely to undergo surgery under urgent or emergent conditions, whereas whites were more likely to have coronary artery disease (Tables 1 and 2⇑). Among patients who underwent AVR, whites were also more likely to have significant left main coronary artery stenosis. There were no racial differences in the frequency of a preoperative myocardial infarction or the frequency of preoperative use of an intra-aortic balloon pump for either valve surgery procedure.
Table 3 shows the distribution of the type of valve prosthesis by race for AVR and MVR. For both valve cohorts, black patients were more likely than whites to receive a mechanical valve (AVR: 40.4% of blacks versus 36.2% of whites; MVR: 67.1% of blacks versus 58.2% of whites). This result is not unexpected given that black patients were on average much younger than white patients. Nonetheless, the association between black race and increased likelihood of mechanical valve was statistically significant even after stratification by age (Mantel-Haenszel χ2, P<0.0001).
Operative Outcomes by Race
Tables 4 and 5⇓ provide the acute outcomes of AVR and MVR, respectively, for patients as a function of racial subgroup. The unadjusted mortality rate for whites who underwent AVR was 3.62%, whereas for blacks, it was 4.60% (P<0.05). There was no significant difference in the unadjusted operative mortality after MVR as a function of race. In fact, although not statistically significant, the unadjusted operative mortality for blacks (5.60%) was lower than that for whites (6.18%).
For both types of valve procedure, the total length of stay and postoperative length of stay were significantly greater in black patients (Tables 4 and 5⇑). Black patients remained ventilated longer, which correlated with increased duration of hospitalization. For patients who underwent AVR, blacks had a mean length of stay and postoperative length of stay of 12 and 9 days, respectively, which was 1 to 3 days longer than that of the white subgroup. There were no significant differences in the frequency of permanent stroke or deep sternal infection between the 2 racial groups for either type of valve surgery procedure.
For both AVR and MVR within each racial subgroup, age >65 years, diabetes, renal failure, and infectious endocarditis were associated with higher mortality (Tables 6 and 7⇓); however, several differences between the 2 racial groups became apparent. For black patients undergoing MVR, the absence of renal failure had a disproportionate statistically significant protective effect on mortality. Blacks without renal failure had an operative mortality rate of 2.86%, whereas whites without renal failure had an operative mortality of 4.99% (unadjusted black/white OR 0.56 [95% CI 0.39 to 0.80]. For white patients undergoing MVR, the presence of active infectious endocarditis had a disproportionately negative influence on operative survival. Whites with active infectious endocarditis had an operative mortality rate of 17.45%, whereas blacks with active infectious endocarditis had an operative mortality rate of 10.42% (unadjusted black/white OR 0.55 [95% CI 0.31 to 0.97]). Finally, although age <65 years was associated with lower operative mortality in both racial groups, the protective effect of younger age was far more important among white patients who underwent AVR. For whites aged <65 years, the unadjusted operative mortality rate was only 1.82%, whereas for blacks, it remained relatively high at 4.24% (unadjusted black/white OR 2.40 [95% CI 1.71 to 3.35]) after AVR. For MVR patients, in a formal test of risk-by-race interaction, only the race-by-endocarditis interaction was significant (P<0.05). For AVR patients, none of the race-by-risk interactions were significant.
After controlling for important risk factors in a multivariable model (Tables 4 and 5⇑), race was no longer a significant predictor of operative mortality after AVR (adjusted black/white OR 1.168 [95% CI 0.911 to 1.497]) or after MVR (adjusted black/white OR 0.883 [95% CI 0.671 to 1.163]). In contrast, black race was significantly associated with an increased risk of several complications: prolonged ventilation after AVR or MVR; postoperative stay >14 days after AVR or MVR; reoperation for bleeding after AVR; and postoperative renal failure after MVR. There was no significant association between race and the risk of stroke or deep sternal wound infection for either AVR or MVR.
Interaction Between Race and Established Risk Factors
To explore the interaction between race and other risk factors, 2 separate logistic regression models predicting operative mortality were fit to the data after stratification by race. The results of this analysis are presented in Tables 8 and 9⇓.
Aortic Valve Replacement
For both racial groups, the important risk factors were similar. Prior cardiac operation, New York Heart Association class IV, and chronic lung disease emerged as important preoperative predictors of mortality in both racial groups. Although the ORs differed marginally for black and white patients, these differences were not statistically significant when assessed simultaneously (P=0.66) or when tested individually. We also sought to determine whether the influence of race on mortality varied as a function of the overall patient risk profile. A formal test of race-by-risk interaction (Breslow-Day test for homogeneity of the ORs across risk deciles) was not significant (P=0.731). Thus, the effect of race does not appear to differ with respect to the risk decile.
Mitral Valve Replacement
Emergent or salvage operation, renal failure, and age emerged as significant preoperative predictors of mortality in both racial groups. Active endocarditis was associated with increased operative mortality among white patients (OR=2.344 [95% CI 1.809 to 3.036]) but not among black patients (OR=0.841 [95% CI 0.404 to 1.751]). The overall test of interaction between race and risk factors was not significant (P=0.33); however, when each interaction was tested individually, the interaction between race and active infectious endocarditis was statistically significant (P=0.010). A formal test of race-by-risk interaction (Breslow-Day test for homogeneity of the ORs across risk deciles) was not significant (P=0.227). Thus, MVR mortality rates for whites versus blacks do not appear to differ with respect to the risk decile.
The STS database provided a large sample from which to study the association between race and operative mortality after AVR and MVR procedures. In the AVR analysis, 32 748 white patients and 1762 black patients were included. The MVR analysis included 13 501 white and 1375 black patients. The unanticipated finding of this study is that there is no significant difference in risk-adjusted mortality between the 2 racial groups studied for either MVR or AVR. Black patients were, however, more likely to suffer certain complications, including renal failure and prolonged ventilation, which likely provides an explanation for the observed increase in length of stay for blacks.
There were significant differences in patient characteristics as a function of race. For both types of valve surgery considered, the black subgroup was younger and more often female. Although not significant, there was a trend toward an interaction between age and race for patients undergoing AVR. The fact that black patients were found to have a significantly higher proportion of urgent or emergent cases suggests disparate access to appropriate preventive care and diagnostic evaluation. Certainly, disparate access to cardiac catheterization, percutaneous coronary intervention, and coronary artery bypass surgery has been well documented for black patients.15,16
In both racial groups, men were more likely than women to undergo AVR. Conversely, women were more likely than men to undergo MVR. This gender disparity was even more pronounced among black patients.
The racial variation in the patients’ preoperative risk factors may reflect the higher prevalence of certain comorbidities (eg, renal disease, diabetes, and hypertension) in the black population as a whole. Congestive heart failure occurred more often in black patients as well and was reflected in higher preoperative pulmonary artery pressures and lower preoperative ejection fraction.
Although the present study used a large sample of patients, its conclusions should be interpreted in light of certain limitations. First, the study is observational in nature, and it is possible that we were unable to identify and adjust for all potential confounding variables. Black and white patients in the present study sample differ dramatically with respect to the distribution of important risk factors such as age and comorbidities. Although we adjusted for all available risk factors using logistic regression, it is possible that our assessment of the effect of race is confounded by omitted or unmeasured covariates or by imperfect model specification. Second, all of the data for the present study were drawn from the STS database, which is not necessarily representative of all cardiac surgical practices in the United States or North America. It is possible that surgery practices that volunteer to participate in the STS database may differ in important respects from practices that do not participate. For example, if STS practices have better outcomes than nonparticipants, then the present study may actually underestimate the true risk of mortality and morbidity after valve replacement surgery; however, we do not anticipate that selection of hospitals with better outcomes would bias our estimates of the effect of race.
In conclusion, the STS database provides us with a large sample of patients to begin our assessment of operative mortality in an understudied population: blacks. Given the well-documented discrepancies in access to cardiovascular care for black patients, it is important to investigate whether significant differences in outcome occur on the basis of race as an independent risk factor. In contrast to previously published results that defined race as an independent risk factor for operative mortality after CABG, race does not appear to be a significant predictor of operative mortality after isolated AVR or MVR. These data lend credence to the concept that as we endeavor to abolish racial disparities in access to cardiovascular care, patients with accepted indications should be referred for valve surgery irrespective of race.
Yacoub MH, Cohn LH. Novel approaches to cardiac valve repair: from structure to function, part I. Circulation. 2004; 109: 942–950.
Chapman JN, Mayet J, Chang CL, Foale RA, Thom SA, Poulter NR. Ethnic differences in the identification of left ventricular hypertrophy in the hypertensive patient. Am J Hypertens. 1999; 12: 437–442.
Athanasiou T, Aziz O, Skapinakis P, Perunovic B, Hart J, Crossman MC, Gorgoulis V, Glenville B, Casula R. Leg wound infection after coronary artery bypass grafting: a meta-analysis comparing minimally invasive versus conventional vein harvesting. Ann Thorac Surg. 2003; 76: 2141–2146.