Author + information
- Received November 5, 1999
- Revision received March 15, 2000
- Accepted April 28, 2000
- Published online October 1, 2000.
- William Whang, MD, MSa,* (, )
- J.Thomas Bigger Jr., MD, FACCa,
- The CABG Patch Trial Investigators and Coordinators
- ↵*Reprint requests and correspondence: Dr. William Whang, c/o J. Thomas Bigger, MD, PH 9-103D, Columbia University Health Sciences, 630 West 168th Street, New York, New York 10032
We examined the relationship between diabetes mellitus and outcomes after coronary artery bypass graft (CABG) surgery in patients with severe left ventricular (LV) dysfunction.
Although diabetes is associated with poor outcomes after CABG surgery among unselected patients, the relationship between diabetes and mortality after CABG surgery among patients with LV dysfunction is less certain.
Using data from The CABG Patch Trial, a study of implantable cardiac defibrillator therapy, we analyzed 900 patients with ejection fraction <0.36 who underwent CABG surgery from 1990 to 1996.
Diabetics comprised 38% of the patients, and 48% of diabetics were prescribed insulin. Diabetes was associated with hypertension, peripheral vascular disease, history of stroke, clinical heart failure and rales on physical exam. Diabetics were at higher risk for postoperative superficial sternal wound infection and renal failure. With an average follow-up time of 32 ± 16 months, actuarial all-cause mortality 48 months after CABG surgery was 26% in diabetics and 24% in nondiabetics (p = 0.66, log-rank test). Diabetes was not associated with long-term mortality in Cox multiple regression analyses. Actuarial re-hospitalization rates 48 months after CABG surgery were 85% in diabetics and 69% in nondiabetics (p = 0.0001, log-rank test). Diabetics had a 44% higher risk of re-hospitalization for any cause (p = 0.0001) and a 24% higher risk of re-admission for cardiac causes (p < 0.05). Unexpectedly, fewer arrhythmic events were found in diabetics.
Diabetes was not a predictor of mortality after CABG surgery among patients with LV dysfunction despite associated comorbidities. However, diabetes was associated with increased postoperative complications and re-hospitalization.
Diabetes is a risk factor for coronary artery disease (CAD) and heart failure (1,2). Investigators have posited several possible mechanisms for the association between diabetes and heart disease, including abnormalities in lipid metabolism, nitric oxide activity, platelet function, coagulation and autonomic function (3–7).
Not only is diabetes a risk factor for CAD, but observational studies have also linked it with worse outcomes after revascularization via coronary artery bypass graft (CABG) surgery (8–13). The higher mortality among diabetics after revascularization is felt to be consistent with more severe atherosclerosis associated with diabetic heart disease, possibly in the form of more diffuse distal CAD that continues to jeopardize myocardium.
The association between diabetes and mortality after CABG surgery among patients with left ventricular (LV) dysfunction is less conclusive. Some studies that examined patients with reduced LV ejection fractions (LVEFs) have identified diabetes as a risk factor for mortality (9,14); others have not (15–17). In this study we assessed the long-term results of CABG surgery in diabetic patients with LV dysfunction using data from a multicenter trial. We tested two primary hypotheses: 1) among patients with LV dysfunction who undergo CABG, diabetes is an independent predictor of mortality, both all-cause and cardiac; 2) diabetes is a predictor of the risk of re-hospitalization, due to all causes and cardiac causes.
We performed a cohort analysis of data collected for The CABG Patch Trial, a randomized study of prophylactic implantable cardiac defibrillators (ICDs) in high risk coronary heart disease patients undergoing CABG surgery from 1990 to 1996 (18). Criteria for inclusion in the trial included the following: age less than 80 years, LVEF less than 0.36 and an abnormal signal averaged electrocardiogram (ECG).
In the main trial, 71,855 patients were screened for inclusion, and 900 patients were randomized and followed for an average of 32 ± 16 months. Patients were excluded from the trial on the basis of the following: diabetes mellitus with poor blood glucose control or recurrent infections, history of sustained ventricular tachycardia or fibrillation, previous or concomitant aortic or mitral valve surgery, concomitant cerebrovascular surgery, serum creatinine greater than 3 mg per deciliter, emergency coronary bypass surgery, noncardiovascular condition with expected survival less than two years or inability to attend follow-up visits.
Nurse coordinators at the 37 clinical centers involved in the trial ascertained whether patients had ever been diagnosed with diabetes and whether treatment included insulin. In obtaining information about early postoperative outcomes, need for mechanical support was defined as the use of an intra-aortic balloon pump or LV assist device on exit from the operating room, and renal failure was defined as renal dysfunction severe enough to require hemodialysis or peritoneal dialysis (19). After the index hospitalization for CABG surgery, follow-up visits were scheduled every three months. Data on hospitalizations included the reason for hospitalization, whether for a cardiac problem and whether a definite/probable myocardial infarction, congestive heart failure or ventricular arrhythmias occurred during the hospital stay. An independent events committee classified each death that occurred according to whether the mechanism of death was cardiac in nature, for example, due to arrhythmia, myocardial pump failure or cardiac procedure (20).
We collected information on baseline demographic, clinical, angiographic and intraoperative risk factors for postoperative complications, long-term mortality and re-hospitalization. We examined differences in baseline variables between diabetic and nondiabetic patients via chi-square tests, with a correction for continuity for dichotomous variables. From our candidate predictors for mortality/morbidity, we used chi-square tests to select independent variables for multiple logistic regression models of perioperative complications and log-rank tests to select variables for Cox multiple regression models in the case of long-term outcomes. We used backwards stepwise regression, with mandatory inclusion of diabetes status, to select models for parsimony. Diabetes status was represented via two separate models for each outcome—one model in which all diabetics were grouped together and another in which patients treated with insulin and those treated with oral hypoglycemics or diet were grouped separately. We used chi-square tests for nested models to test whether the relationship between diabetes and outcome significantly differed depending on the type of treatment for diabetes. Kaplan-Meier curves were constructed to compare unadjusted time to death or re-hospitalization between diabetics and nondiabetics. All statistical tests were performed to reject the null hypothesis at a nominal alpha level of 0.05. All analyses were performed using SAS Version 6.12.
Among the 900 patients studied, 344 (38%) of the patients were diabetic at the time of CABG surgery (Table 1); 48% of the diabetic patients were treated with insulin. Average follow-up time was 31 ± 17 months among diabetics and 32 ± 16 months among nondiabetics. Compared with patients without diabetes, diabetics were more likely to be women, obese and less likely to have a history of tobacco use. Hypertension, peripheral vascular disease, history of stroke, clinical heart failure and rales on physical exam were more prevalent among diabetics. Diabetics and nondiabetics did not significantly differ in their severity of angina, LVEF, presence of left main or triple vessel disease or use of internal mammary artery grafts as conduits. Mean LVEF was 0.27 ± 0.06 in both diabetics and nondiabetics. Before CABG surgery, diabetics were more likely to be treated with angiotensin converting-enzyme inhibitors, diuretics and digoxin and less likely to be treated with beta-adrenergic blocking agents (Table 1).
In analyses of perioperative complications, diabetes was associated with an increased risk of superficial sternal wound infection and renal failure requiring dialysis (Table 2). The odds of superficial sternal wound infection were 3.31 times greater (p < 0.05) and the odds of renal failure were 2.24 times greater (p < 0.05) for diabetic patients than for patients without diabetes. There was a trend toward higher risk for these outcomes among patients treated with insulin than there was among patients treated with oral hypoglycemic agents; however, chi-square tests for the interaction between type of treatment and both superficial sternal wound infection and renal failure were not statistically significant. Insulin therapy was associated with an increased use of mechanical support (intra-aortic balloon pump or LV assist device). Neither deep sternal wound infection nor pneumonia was associated with diabetes in multivariable logistic regression models. There was an increased risk of deep sternal wound infection and pneumonia among ICD patients; there was no significant interaction between ICD use and diabetes in regression models of either outcome.
There were 198 total deaths, 157 cardiac deaths and 43 arrhythmic deaths. The actuarial total mortality rate at 48 months after CABG surgery was 26% in diabetics and 24% in patients without diabetes (p = 0.66, log-rank test; Fig. 1). Cardiac and arrhythmic death rates were 19.4% and 6.1%, respectively, in diabetics and 20.7% and 6.9% in patients without diabetes. Diabetes was not significantly associated either with cardiac or arrhythmic mortality by log-rank test.
Neither diabetes nor treatment with insulin was significantly associated with increased long-term mortality in Cox multiple regression analyses (Table 3). Age, history of stroke, clinical heart failure, history of previous CABG surgery and body mass index >30 were independent, significant predictors of long-term postoperative mortality in diabetics and in nondiabetics. In analyses of cardiac mortality, diabetes was not a significant predictor of the risk of death due to any cardiac cause. There was a strong trend toward less arrhythmic mortality among diabetics in multiple regression analyses; however, it was not statistically significant. As has been reported previously (18), defibrillator therapy had no effect on long-term mortality, nor was there a significant interaction between defibrillator therapy and diabetes.
The actuarial rate of re-hospitalization at 12 months was 54% among diabetics and 41% among nondiabetics; the corresponding rates at 48 months were 85% and 69%, respectively (p = 0.0001, log-rank test; Fig. 2). The elapsed time before 50% of the diabetic group was hospitalized was 23.7 months; the corresponding time among nondiabetics was 44.8 months. After adjustment for other variables, diabetes was associated with a statistically significant 44% increase in the relative risk of rehospitalization (Table 4). Although there was a trend toward a higher risk of re-hospitalization for patients treated with insulin than there was for patients treated with oral agents or nonpharmacologic therapy, tests for the interaction between type of diabetes treatment and re-hospitalization were not statistically significant. Diabetes also was associated with a 24% increase in the risk of re-admission for cardiac reasons. For initial re-hospitalizations, the odds that the admission was for noncardiac reasons were 48% higher among diabetics than they were among nondiabetics (Table 5). Re-hospitalization was less likely to be due to arrhythmias among diabetics, and relatively fewer ventricular arrhythmias occurred during hospitalization of patients with diabetes (Table 5).
In a cohort study of 900 patients in The CABG Patch Trial, we tested two hypotheses of the link between diabetes and outcomes after CABG surgery: 1) among patients with LV dysfunction who undergo CABG, diabetes is an independent predictor of mortality, both all-cause and cardiac; and 2) diabetes is a predictor of the risk of rehospitalization, due to all causes and cardiac causes. Our study sample comprised 344 diabetics and 556 nondiabetics with low ejection fraction (<0.36); approximately half of the patients had clinical heart failure, and 198 died during the follow-up period. Half the patients in the study received ICD therapy, but ICD therapy had no effect on survival. Contrary to our first hypothesis, the long-term risk of death after CABG surgery was not significantly different for diabetics compared with nondiabetics, despite higher rates of comorbidities such as hypertension, heart failure, peripheral vascular disease and history of stroke among diabetics. However, diabetes was associated with a higher risk of re-hospitalization, including those for cardiac reasons.
Comparisons with other studies
Previous studies have included diabetes as an explanatory variable for mortality after CABG surgery among patients with reduced LV function with varying results. As part of a larger study of CABG surgery in diabetics, Morris and colleagues (9) included approximately 870 patients with ejection fraction ≤0.40, of whom about 27% were diabetic. Five-year mortality was 34% among diabetics and 20% among nondiabetics (p < 0.02). Trachiotis and colleagues (14) examined long-term survival after CABG surgery in 11,830 patients, of whom 156 patients had ejection fraction <0.25, and 588 patients had ejection fraction 0.25 to 0.34. In an analysis inclusive of all the patients regardless of ventricular function, diabetes was associated with a 59% increase in the relative risk of death (p < 0.0001). Gill and colleagues (15) found that among 166 patients with ejection fraction <0.31, during 4.75 years of follow-up, diabetes was not associated with reduced survival after CABG surgery. Kaul and colleagues (16) measured outcomes in 210 patients with LVEF ≤0.20 who underwent CABG surgery at University of Alabama from 1987 to 1992. During an average follow-up of 43 months, diabetes was not significantly associated with excess long-term mortality. Milano and colleagues (17) studied outcomes among 118 patients with ejection fraction ≤0.25, of whom 33% were diabetic. Diabetes was not associated with long-term mortality in single variable analyses.
Explanations for lack of mortality findings
The lack of a relationship between diabetes and mortality after CABG surgery in our analysis has multiple potential explanations. Our study had adequate power to detect an association of diabetes mellitus with long-term cumulative mortality. The mortality rate was 26% at 48 months after CABG surgery among diabetics, giving us approximately 80% power to detect a 35% relative difference or a 9% absolute difference in mortality at a two-tailed alpha level of 0.05. It is possible that diabetes does not contribute to the underlying force of mortality after CABG surgery in ischemic cardiomyopathy to as large an extent as it does in CAD with preserved ventricular function. Alternatively, diabetic patients with ventricular dysfunction may derive a larger benefit from revascularization of jeopardized myocardium and, therefore, have similar lifespans as nondiabetics after surgery.
Although the risk of death was not higher for diabetics in our study, diabetes was clearly a risk factor for morbidity after CABG surgery. There was a strong association with perioperative complications including renal failure, superficial sternal wound infection and use of mechanical circulatory support. The association with sternal wound infection is consistent with previous studies of CABG surgery in diabetic patients (21–25). The association between sternal wound infection and epicardial ICD use in our analysis has been documented in-depth by Spotnitz and colleagues (19) and may result partly from reporting bias.
The time to postoperative re-hospitalization was significantly shorter for diabetics. The one year re-hospitalization rate was 54% among diabetics compared with 41% in nondiabetics. It took almost twice as long for patients in the nondiabetic group to reach a 50% re-hospitalization rate compared with those in the diabetic group (44.8 vs. 23.7 months, respectively). Hospitalizations for noncardiac causes comprised a higher proportion of re-admissions among diabetics. However, in the context of a much higher overall re-hospitalization rate, there was a higher relative risk of cardiac hospitalization among diabetics, which is consistent with our second hypothesis regarding excess morbidity. The implications of our findings for long-term resource use and costs are significant and have yet to be fully quantified.
The CABG Patch Trial is the only study, to our knowledge, of patients after CABG surgery that has prospectively collected data on ventricular arrhythmias and arrhythmic death during long-term follow-up. The adjusted risk for arrhythmic death in diabetics was almost half that in nondiabetics, although with 43 total arrhythmic deaths, this was not a statistically significant finding. Arrhythmia was less often the reason for initial re-hospitalizations after CABG surgery in diabetics, and substantially fewer ventricular arrhythmias occurred during hospitalizations in diabetics. Given the past evidence that has linked autonomic dysfunction with diabetes, we expected more, not less, arrhythmic events among the diabetics in our study. About 40% of diabetics with and without ischemic heart disease have cardiac autonomic nerve dysfunction, that is, reduced parasympathetic and increased sympathetic modulation of RR intervals (7,26,27), and diabetics with autonomic neuropathy are at increased risk of death compared with diabetics without autonomic neuropathy (28). In addition, prolongation of the QT interval has been noted in diabetic patients with autonomic neuropathy (29,30) and ischemic heart disease (31). However, all the patients in our study had severe LV dysfunction, and the diabetics had a higher prevalence of heart failure. The UK-Heart study showed that, among heart failure patients, reduced heart rate variability was the best predictor of death due to heart failure but did not predict sudden cardiac death (32). Ischemic cardiomyopathy may interact with the effect of diabetes and autonomic neuropathy on arrhythmic events. Also, diabetics could benefit more from revascularization than nondiabetics, particularly with respect to their arrhythmic substrate. Further studies are needed to examine the incidence and significance of ventricular arrhythmias after CABG surgery in diabetics.
One potential disadvantage of this analysis is the fact that diabetic patients with poor blood glucose control or recurrent infections and patients with serum creatinine >3 mg/dl were excluded from the trial. These two groups respectively comprised about 0.6% and 0.5% of the 71,855 screened patients in the enrollment cascade (33). Thirty-eight percent of the 900 patients in the trial had diabetes, a larger percentage than most previous observational studies of CABG surgery. Another limitation of this study was that the extent of glucose control, in terms of serum glucose levels or glycosylated hemoglobin levels, was not available from our database. In addition, the retrospective nature of our analysis limited the strength of inferences from the data. However, the multicenter experience provided by our sample strengthens its generalizability.
Our analysis indicates that diabetes is not a strong predictor of mortality after CABG surgery among patients with LV dysfunction, despite its value as a predictor of postoperative complications and long-term morbidity. Further clinical investigations may clarify the long-term noncardiac risks that diabetics face after CABG surgery and the effects of CABG surgery on ventricular arrhythmias in diabetics.
We are grateful to Henry M. Spotnitz, MD, and Richard Steinman, AB, for their helpful insights and comments.
- coronary artery bypass graft
- coronary artery disease
- implantable cardiac defibrillator
- left ventricular
- left ventricular ejection fraction
- Received November 5, 1999.
- Revision received March 15, 2000.
- Accepted April 28, 2000.
- American College of Cardiology
- Kannel W.B,
- McGee D.L
- Grundy S.M,
- Benjamin I.J,
- Burke G.L,
- et al.
- Ginsberg H.N
- Ewing D.J
- Herlitz J,
- Wognses G.B,
- Emanuelsson H,
- et al.
- Bigger J.T Jr.,
- Whang W,
- Rottman J.N,
- et al.
- Fietsam R,
- Bassett J,
- Glover J.L
- Antunes P.E,
- Bernardo J.E,
- Eugenio L,
- de Oliveira J.F,
- Antunes M.J
- Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology
- Ewing D.J,
- Campell I.W,
- Clarke B.F
- Gonin J.M,
- Kadrofske M.M,
- Schmaltz S,
- Bastyr E.J,
- Vinik A.I
- Nolan J,
- Batin P.D,
- Andrews R,
- et al.