Preoperative Hemoglobin Level as a Predictor of Survival After Coronary Artery Bypass Grafting
A Comparison With the Matched General Population
Background— The predictive value of the preoperative hemoglobin value after coronary artery bypass grafting (CABG) has not been well established. We studied how the preoperative hemoglobin level affects the survival of patients after CABG. Late mortality was compared with that of a general population.
Methods and Results— Early and late mortality of all consecutive patients undergoing isolated CABG between January 1998 and December 2007 were determined. Patients were classified into 4 groups stratified by preoperative hemoglobin level. The cutoff point for anemia was 13 g/dL for men and 12 g/dL for women. Expected survival of a matched general Dutch population cohort was obtained from the database of the Dutch Central Bureau for Statistics. After the exclusion of 122 patients who were lost to follow-up and 481 patients with missing preoperative hemoglobin levels, complete data were obtained in 10 025 patients. Multivariate logistic regression analyses revealed anemia to be an independent risk factor for higher early mortality. Cox regression analyses revealed low hemoglobin level, both as a continuous variable and as a dichotomous variable (anemia), to be a predictor of higher late mortality. Compared with expected survival, patients with the lowest preoperative hemoglobin levels had a worse outcome, whereas patients with the highest hemoglobin levels had a better outcome.
Conclusions— A lower preoperative hemoglobin level is an independent predictor of late mortality in patients undergoing CABG, whereas anemia is a risk factor for early and late mortality. Compared with the general population, anemic patients had worse survival than expected, whereas nonanemic patients had better survival than expected.
Received January 27, 2009; accepted May 15, 2009.
The role of preoperative hemoglobin as a predictor of short- and long-term outcomes after coronary artery bypass grafting (CABG) has not been well established. Using univariate and multivariate logistic models, Zindrou and coworkers1 found that a low hemoglobin level was a predictor of in-hospital death after CABG. In the risk-prediction model of Magovern and associates,2 anemia was an independent predictor for mortality or morbidity. Kulier and colleagues3 reported that a low preoperative hemoglobin level was an independent risk factor for renal and neurological complications, whereas an increase in cardiac events was due to other factors associated with preoperative anemia, such as renal dysfunction. In a study by Cladellas and colleagues,4 a preoperative hemoglobin level of <12 g/dL was a risk factor for in-hospital mortality and serious adverse outcomes in patients undergoing valve surgery, whereas Karkouti and colleagues5 found that preoperative anemia (hemoglobin <12.5 g/dL) was independently associated with adverse outcomes in patients after cardiac surgery, including valve surgery. In patients undergoing elective vascular surgery, Dunkelgrun and colleagues6 found anemia to be an independent predictor of poor perioperative and long-term cardiovascular outcome.
Clinical Perspective on p 125
According to the World Health Organization, anemia is defined as hemoglobin <13 g/dL for men and <12 g/dL for women. Several cutoff points have been described for preoperative hemoglobin as a risk factor for poor early outcomes, including 10 g/dL,7 11 g/dL,3 and 12 g/dL.4 It is not clear whether poor outcomes are caused by the low hemoglobin level itself or by other risk factors that are frequently present in anemic patients. To the best of our knowledge, no data are available on the association between low preoperative hemoglobin levels and long-term outcomes in patients undergoing CABG. We therefore analyzed the data of patients undergoing CABG in a single center to determine the predictive value of low hemoglobin level in combination with comorbidities for early and late mortality.
Data for all patients undergoing isolated CABG in a single center between January 1998 and December 2007 were analyzed. Starting in January 1998, clinical data, including demographic data, risk factors, and complications, were collected prospectively in a database. The present study was performed after permission was obtained from the local ethics medical committee.
Patients were classified into 4 groups stratified by the preoperative hemoglobin level. In the first group (referred to as “very low”), the hemoglobin level in men was <12 g/dL, and in women, it was <11 g/dL. In the second group (referred to as “low”), the hemoglobin level in men was ≥12 and <13 g/dL, and in women, it was ≥11 and <12 g/dL. According to the World Health Organization definition, patients in the very-low and low groups had anemia. In the third group (referred to as “normal”), the hemoglobin level in men was ≥13 and <14.5 g/dL, and in women, it was ≥12 and <13.5 g/dL. In the fourth group (referred to as “high normal”), the hemoglobin level in men was ≥14.5 g/dL, whereas in women, it was ≥13.5 g/dL.
To calculate survival of a general population cohort, data were obtained from the Dutch Central Bureau for Statistics. This is the database that registers information about all citizens living in The Netherlands. Every year, a report about mortality within the normal population, stratified by age and sex, is available online from the Dutch Central Bureau for Statistics. We have matched each group in the present study with the general population according to age and sex. Because the incidence of mortality within the general population varies per year, matching was also done to compare the survival of each group with the survival of the general population for the same year the studied patients underwent surgery. We considered the survival of the matched general population cohort to represent the expected survival of the patient group.
All patients received short-acting anesthetic drugs to facilitate early extubation. Normothermic extracorporeal circulation was performed with nonpulsatile flow. Cold crystalloid cardioplegia (St Thomas solution) or warm blood cardioplegia was used to induce and maintain cardioplegic arrest, according to the surgeon’s preference. Patients undergoing CABG with the use of extracorporeal circulation received low-dose aprotinin (2 million kallikrein inactivating units) during extracorporeal circulation, administered in the prime solution of the heart-lung machine. Patients undergoing off-pump surgery did not receive aprotinin.
Follow-up data on patient mortality were collected from the databases of health insurance companies. The data of 9% of the total patient group initially could not be retrieved from these databases. We therefore contacted the patients’ general practitioners to obtain mortality data for those patients or, if necessary, we contacted the city authorities of the cities in which the patients lived at the time of the operation. Early mortality was defined as death due to any cause within the first 30 postoperative days, whereas late mortality was defined as all-cause mortality beyond 30 days.
Discrete variables were compared with the χ2 test and are presented as numbers and percentages. Continuous variables were compared with the t test and ANOVA and are presented as mean±SD. Univariate and multivariate logistic and Cox proportional hazard regression analyses were performed to investigate the impact of preoperative hemoglobin on early and late mortality. Univariate analyses were used to test for the potentially confounding effect of biomedical and demographic factors on mortality. If significant at P<0.05, confounders were included in the multivariable regression models. Cumulative long-term survival was estimated according to the Kaplan–Meier method, with differences between groups compared with the log-rank test. Timetable analyses were performed to compare survival of patient groups with expected survival by use of the Wilcoxon test to determine statistical significance. The zero time point indicates the time of CABG. A P value <0.05 was used for all tests to indicate statistical significance. Hazard ratios with 95% confidence intervals (CIs) are reported. All statistical analyses were performed with SPSS statistical software (Statistical Product and Services Solutions, version 15.0, SPSS Inc, Chicago, Ill).
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
During a 10-year period (January 1998 through December 2007), 10 626 patients underwent CABG. After the exclusion of 122 patients who were lost to follow-up and 481 patients with missing preoperative hemoglobin levels, data from 10 025 patients were analyzed (the preoperative hemoglobin level was missing in 2 patients who were lost to follow-up). The patients were divided into 4 categories: 635 patients (6.3%) in the very-low-hemoglobin group, 973 patients (9.7%) in the low-hemoglobin group, 4392 patients (43.8%) in the normal-hemoglobin group, and 4025 patients (40.2%) in the high-normal hemoglobin group. The distribution of preoperative hemoglobin levels for men and women is shown in Figure 1. The mean preoperative hemoglobin level for men was 14.1±1.2 g/dL, and for women, it was 12.9±1.2 g/dL. The minimum follow-up for surviving patients was 60 days, with a mean follow-up of 1693±1052 days and a median of 1622 days (range 0 to 3708 days; 0 days for patients who died on the first postoperative day). Baseline characteristics stratified by preoperative hemoglobin levels are shown in Table 1. Patients with higher preoperative hemoglobin were more often younger men. Furthermore, lower preoperative hemoglobin was associated with diabetes mellitus, chronic obstructive pulmonary disease, peripheral vascular disease, low left ventricular ejection fraction, low creatinine clearance, emergency operation, history of previous cardiac surgery, the use of donor blood transfusion(s), and the need for intra-aortic balloon pump support.
Early and late deaths occurred more frequently in the groups with lower preoperative hemoglobin levels, as shown in Table 2. The annual incidence of death was 7.8 per 100 patient-years in the very-low group, 5.3 in the low group, 2.6 in the normal group, and 1.8 in the high-normal group. During the study period, the overall early mortality rate fell from 2.9% in 1998 to 1% in 2007.
Risk factors for early mortality identified by univariate and multivariate logistic regression analyses are shown in Table 3. Univariate logistic regression analysis revealed low preoperative hemoglobin level, as both a continuous variable and a dichotomous variable (anemia), to be a risk factor for early mortality. Other risk factors for early mortality by univariate analysis were age, chronic obstructive pulmonary disease, diabetes mellitus, low creatinine clearance, left ventricular ejection fraction <35%, number of units of donor blood, previous cardiac surgery, peripheral vascular disease, emergency situations, and the year of operation. Complications such as perioperative myocardial infarction, perioperative use of intra-aortic balloon pump support, and reexploration for any reason were also identified as risk factors for early mortality. Sex, hypertension, and the use of extracorporeal circulation were not risk factors for early mortality.
Preoperative risk factors for early mortality identified with univariate logistic regression analysis were entered into the multivariate logistic regression model. Separate analyses were performed for hemoglobin as a continuous variable and as a dichotomous variable (anemia). Anemia was a predictor of early mortality (hazard ratio 1.48, 95% CI 1.08 to 2.02, P=0.013), whereas the preoperative hemoglobin level as a continuous variable was not. When perioperative and postoperative factors (number of received red blood cell units, perioperative myocardial infarction, reexploration, and use of intra-aortic balloon pump) were entered into the multivariate model, neither anemia nor preoperative hemoglobin level was a significant risk factor for early mortality. Other independent preoperative risk factors for early mortality by multivariate analysis were age, chronic obstructive pulmonary disease, diabetes, low creatinine clearance, ejection fraction <35%, and previous cardiac surgery. Emergency operation, the year of operation, and peripheral vascular disease were not risk factors for early mortality.
Results of Cox regression analyses for risk factors for late mortality are shown in Table 4. Univariate analyses revealed preoperative hemoglobin level, both as a continuous variable and as a dichotomous variable (anemia), to be a risk factor for late mortality. Other risk factors for late mortality by univariate analysis were age, chronic obstructive pulmonary disease, hypertension, diabetes mellitus, low creatinine clearance, ejection fraction <35%, peripheral vascular disease, previous cardiac surgery, the year of operation, the number of grafts, use of extracorporeal circulation, and the number of units of red blood cells. Complications such as perioperative myocardial infarction, perioperative use of intra-aortic balloon pump, and reexploration were also risk factors for late mortality.
Preoperative risk factors for late mortality identified with univariate analyses were entered into the multivariate Cox regression model. The preoperative hemoglobin level, entered as a continuous variable and as a dichotomous variable (anemia), was an independent risk factor for late mortality. Calculations of the hazard ratio for preoperative hemoglobin as a continuous variable and as a dichotomous variable were performed with separate analyses. Other preoperative risk factors for late mortality by multivariate analysis were age, male sex, chronic obstructive pulmonary disease, diabetes mellitus, low creatinine clearance, ejection fraction <35%, peripheral vascular disease, the year of operation, and previous cardiac surgery; however, hypertension was not an independent risk factor for late mortality.
When perioperative and postoperative factors (number of grafts, use of extracorporeal circulation, number of received red blood cell units, perioperative myocardial infarction, reexploration, and use of intra-aortic balloon pump) were entered into the multivariate model, hemoglobin level (hazard ratio 0.844, 95% CI 0.804 to 0.885, P<0.0001) and anemia (hazard ratio 1.614, 95% CI 1.407 to 1.852, P<0.0001) were still significant as independent risk factors.
In Figure 2, the long-term survival rates of the 4 groups and the expected survival rates are shown. The log-rank test of the survival curves (Figure 2) showed a value of P<0.001 for differences between the 4 patient groups. In Figures 3 and 4⇓, the survival curves of the patient groups are compared with the survival curves of the matched cohorts of the normal citizens of the general Dutch population, which we considered to be the expected survival. Figure 3 shows that the survival of patients in groups 1 and 2 was worse than that of their matched cohorts in the general population (P<0.001). Figure 4 shows that the survival of patients in groups 3 and 4 was better than that of their matched cohorts of the general population (P<0.005). Survival at 1, 5, and 9 years, corresponding with Figures 3 and 4⇓, is shown in Table 5.
In the present retrospective analysis of prospectively collected data from >10 000 consecutive patients who underwent CABG in a single center, we found both anemia, defined as hemoglobin level <13 g/dL for men and <12 g/dL for women, and the preoperative hemoglobin level as a continuous variable to be independent risk factors for late mortality. Anemia was a predictor of early mortality, whereas the hemoglobin level as a continuous variable was not. Compared with a matched general population cohort, survival of anemic patients was worse, whereas nonanemic patients had better than expected long-term survival.
Anemia has been identified as a predictor of poor short- and long-term outcomes in a nonoperative setting in the general and elderly population,8,9 in patients with coronary artery disease,10,11 and in patients with congestive heart failure.12 Recently, several studies addressed preoperative anemia as a predictor of poor short-term outcome after cardiac surgery.1–5 Most of these studies used cutoff points for preoperative hemoglobin that ranged from 10 to 12.5 g/dL. Only 2 studies found the preoperative hemoglobin level to be a predictor of poor short-term outcome specifically in patients after CABG.3,7 Multivariate analyses performed by Bell and colleagues7 revealed that a preoperative hemoglobin level <10 g/dL was not significant as an independent risk factor for early mortality and that it was only mildly significant for morbidity. Multivariate analyses described by Kulier and colleagues3 showed that preoperative anemia and red blood cell transfusion were both independent but additive risk factors for adverse early outcomes. At the same hemoglobin level, the risk for adverse events increased significantly with transfusion of red blood cells. In the present study, we also found an increased use of transfusion in patients with low hemoglobin levels, but multivariate analysis showed that the preoperative hemoglobin level used as a dichotomous variable with a cutoff point of 13 g/dL for men and 12 g/dL for women was an independent predictor of early mortality; however, preoperative hemoglobin level used as a continuous variable was not an independent risk factor of early mortality. It is difficult to establish whether or not the correlation between hemoglobin level and early mortality is linear because of confounders that occur more frequently in patients with low hemoglobin levels, such as advanced age, diabetes mellitus, peripheral vascular disease, renal function impairment, and low left ventricular ejection fraction. This can also explain the discrepancy in findings between different studies. Each study included different variables for statistical adjustment, and some studies differed in their availability or treatment of preoperative transfusion information.7 Discrepancy may be due in part to methodological differences. Zindrou and coworkers1 defined low hemoglobin as 10g/dL or less; Magovern and associates2 defined anemia as 12.5 g/dL or less for men, 11.0 g/dL or less for women, or need for preoperative transfusion.
To the best of our knowledge, no data are available on preoperative hemoglobin level as a predictor of long-term outcome of patients undergoing CABG. Recently, Dunkelgrun and colleagues6 identified anemia as a predictor of major adverse cardiac events in patients undergoing elective vascular surgery. In the present study, preoperative hemoglobin level as a continuous variable was a predictor of late mortality. Also, when used as a dichotomous variable with the World Health Organization criteria for anemia, anemia was a predictor of late mortality. To describe a cutoff point of a variable that is also significant as a continuous variable is debatable. As stated earlier, the cutoff points used in the literature vary. The risk of late mortality increases with lower preoperative hemoglobin levels; however, the quantitative nature of this correlation is difficult to establish because of confounders that occur more frequently in patients with low hemoglobin, such as advanced age, diabetes mellitus, peripheral vascular disease, renal function impairment, and low left ventricular ejection fraction, as mentioned above.
It is not clear why patients with low preoperative hemoglobin have lower long-term survival after CABG. These patients may be particularly sensitive to anemia, having limited ability to compensate through increased heart rate and stroke volume.13 Failure of CABG patients to compensate adequately for the effects of anemia may result in tissue hypoxia, cellular failure, and eventual organ dysfunction and failure.7 Such death would occur over the postoperative course.
Clearly, advanced age is associated with worse late survival. Although appropriate statistical tests can be used to distinguish between the effects caused by age and those caused by hemoglobin level, the interpretation of survival curves remains difficult. Life expectancy also varies with sex and race. Over the years, variation in life expectancy and mortality rates of the Dutch population has been well documented by the Dutch Central Bureau for Statistics. We used data from the Central Bureau for Statistics to compare survival of patients in the present study with the survival of general population cohorts matched for age and sex (expected survival). Cumulative survival of patients in the high-normal group as well as the expected survival were better than cumulative survival in the normal group of patients, whereas survival of patients in the very-low group was worse than in the low group. In the general population, a certain percentage of people have anemia and coronary artery disease. In nonanemic patients undergoing CABG, the coronary problem is treated. Most patients after CABG are given aspirin and hypocholesterolemic drugs and are treated for underlying diseases such as hypertension. This treatment, as well as the protection provided by revascularization, may contribute to improved survival and may explain why nonanemic patients have improved survival compared with a matched general population cohort. Furthermore, before undergoing CABG, patients are screened for other severe underlying disease. If severe underlying disease is present, alternative treatment is considered instead of CABG, thus enriching the CABG group with patients of above-average health (except for the coronary artery disease).
A low hemoglobin level should be considered not only as a marker for comorbidities but also as an independent predictor of higher short- and long-term mortality. Whether treatment of anemia before surgery can improve clinical outcomes remains to be established, despite the promising results of some studies.14,15
This is a retrospective study with some missing data. The study was based on the retrospective evaluation of patient charts. We could not retrieve the cause of death on long-term follow-up. Exact values of ejection fraction were not available in some patients, which is why we used ejection fraction as a dichotomous variable (ejection fraction <35%). We recommend caution in interpreting the results of the comparison with the general population. The Central Bureau for Statistics database includes the total Dutch population. Therefore, data for the patients described in the present study and those treated at other Dutch cardiac surgery centers are included. Because of this, the magnitude of differences between groups tends to be lessened. The annual number of patients undergoing CABG in the Netherlands is small (10 000 patients) compared with the total number of the general population, which limits the effect of this inaccuracy. During the study of long-term follow-up data, it is inevitable that operative results from the past will be included. Because early mortality dropped from 2.9% in 1998 to 1% in 2007, we must be cautious in using these results to predict the outcome of future patients.
In the present study, preoperative hemoglobin level, used as both a continuous variable and as a dichotomous variable (anemia) with a cutoff point of 13 g/dL for men and 12 g/dL for women, was an independent risk factor for late mortality of patients undergoing CABG, whereas anemia was a predictor of early mortality. Compared with expected survival of a general population cohort, patients with the lowest hemoglobin levels had a worse outcome, whereas patients with the highest hemoglobin levels had a better outcome.
Kulier A, Levin J, Moser R, Rumpold-Seitlinger G, Tudor IC, Snyder-Ramos SA, Moehnle P, Mangano DT; Investigators of the Multicenter Study of Perioperative Ischemia Research Group. Impact of preoperative anemia on outcome in patients undergoing coronary artery bypass graft surgery. Circulation. 2007; 116: 471–479.
Cladellas M, Bruguera J, Comín J, Vila J, de Jaime E, Martí J, Gomez M. Is pre-operative anaemia a risk marker for in-hospital mortality and morbidity after valve replacement? Eur Heart J. 2006; 27: 1093–1099.
Karkouti K, Wijeysundera DN, Beattie WS; Reducing Bleeding in Cardiac Surgery (RBC) Investigators. Risk associated with preoperative anemia in cardiac surgery: a multicenter cohort study. Circulation. 2008; 117: 478–484.
Dunkelgrun M, Hoeks SE, Welten GM, Vidakovic R, Winkel TA, Schouten O, van Domburg RT, Bax JJ, Kuijper R, Chonchol M, Verhagen HJ, Poldermans D. Anemia as an independent predictor of perioperative and long-term cardiovascular outcome in patients scheduled for elective vascular surgery. Am J Cardiol. 2008; 101: 1196–1200.
Sabatine MS, Morrow DA, Giugliano RP, Burton PB, Murphy SA, McCabe CH, Gibson CM, Braunwald E. Association of hemoglobin levels with clinical outcomes in acute coronary syndromes. Circulation. 2005; 111: 2042–2049.
Reinecke H, Trey T, Wellmann J, Heidrich J, Fobker M, Wichter T, Walter M, Breithardt G, Schaefer RM. Haemoglobin-related mortality in patients undergoing percutaneous coronary interventions. Eur Heart J. 2003; 24: 2142–2150.
Tang YD, Katz SD. Anemia in chronic heart failure: prevalence, etiology, clinical correlates, and treatment options. Circulation. 2006; 113: 2454–2461.
Mancini DM, Katz SD, Lang CC, LaManca J, Hudaihed A, Androne AS. Effect of erythropoietin on exercise capacity in patients with moderate to severe chronic heart failure. Circulation. 2003; 107: 294–299.
Silverberg DS, Wexler D, Blum M, Keren G, Sheps D, Leibovitch E, Brosh D, Laniado S, Schwartz D, Yachnin T, Shapira I, Gavish D, Baruch R, Koifman B, Kaplan C, Steinbruch S, Iaina A. The use of subcutaneous erythropoietin and intravenous iron for the treatment of the anemia of severe, resistant congestive heart failure improves cardiac and renal function and functional cardiac class, and markedly reduces hospitalizations. J Am Coll Cardiol. 2000; 35: 1737–1744.
This single-center observational study investigated the association between preoperative hemoglobin level and both early and late mortality in >10 000 coronary surgery patients. To the best of our knowledge, no data are available on preoperative hemoglobin level as a predictor of long-term outcome of patients undergoing coronary artery bypass grafting. The predictive value of preoperative hemoglobin level indicates the importance of better investigation and management of preoperative anemia in coronary surgery patients. An important finding of this study is that patients with a normal preoperative hemoglobin level had a better long-term survival than age- and sex-matched groups of the general Dutch population. Although this information has no significance in preoperative assessment and decision making, it gives patients a rough idea of their long-term prognosis.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.