Author + information
- Received August 2, 2001
- Revision received December 4, 2001
- Accepted December 14, 2001
- Published online March 6, 2002.
- ↵*Reprint requests and correspondence:
Dr. Kevin E. Kip, University of South Florida, FMHI, MHC 2605, 13301 Bruce B. Downs Boulevard, Tampa, Florida 33612-3807, USA.
Objectives We sought to investigate the impact of body mass index (BMI) on short- and long-term outcomes after initial revascularization with percutaneous transluminal coronary angioplasty (PTCA) or coronary artery bypass graft surgery (CABG).
Background Equivocal results exist on the impact of BMI on the risk of in-hospital complications after PTCA or CABG, and no long-term mortality data exist from a large series of revascularized patients.
Methods From the randomized series and observational registry of the Bypass Angioplasty Revascularization Investigation (BARI), 2,108 patients who had PTCA and 1,526 patients who had CABG were evaluated by taking their BMI at study entry. They were classified as follows: low (<20 kg/m2), normal (20 to 24.9 kg/m2), overweight (25 to 29.9 kg/m2), class I obese (30 to 34.9 kg/m2) and class II/III obese (≥35 kg/m2). In-hospital complications and short- and long-term mortalities were compared between levels of BMI within each mode of initial revascularization.
Results Among patients who had PTCA, each unit increase in BMI was associated with a 5.5% lower adjusted risk of a major in-hospital event (death, myocardial infarction, stroke, coma); among patients who had CABG, no difference in the in-hospital outcome was observed according to BMI. In contrast, BMI was not associated with five-year mortality in the PTCA group; among the CABG group, adjusted relative risks of five-year cardiac mortality according to levels of BMI were 0.0 (low), 1.0 (normal), 2.02 (overweight), 3.16 (class I obese) and 4.85 (class II/III obese) (linear p < 0.001).
Conclusions Body mass index appears to have a differential impact on short- and long-term outcomes after coronary revascularization. These results underscore the need for further research to identify factors responsible for the apparent short-term protective effect of a higher BMI in patients undergoing PTCA and to study the impact of weight reduction on the long-term survival of obese patients undergoing CABG.
Obesity is a common and growing problem; almost one-third of American adults are obese (1). Annual health care costs attributable to obesity have been estimated to be approximately $68 billion, with an additional $30 billion being spent on weight-reduction programs and special diets (1). The relationship of obesity to long-term survival is complex—most studies have found a J- or U-shaped curve with increasing mortality in the very lean or very obese (2,3). However, when the data have been adjusted for smoking and concurrent illness, the relationship has been more linear—the risk of death rises as body mass index (BMI) increases (4,5). Body mass index is a validated measure of adiposity (6)and has been consistently used in previous analyses of obesity and mortality.
Cardiovascular disease is the leading cause of mortality in the U.S. (7), and obese adults are at an increased risk of cardiovascular mortality (8). In the past two decades, there has been a great increase in the number of patients who undergo coronary revascularization, with an estimated 607,000 coronary artery bypass graft surgeries (CABG) and 407,000 percutaneous coronary interventions (PCI) performed in 1997 (7). Data from PCI registries suggest that the very lean and the very obese have an increased risk of in-hospital complications and short-term mortality (9). In contrast, among patients undergoing CABG, most studies have failed to identify an increased risk of mortality or serious complications in obese patients (10–12). Collectively, these studies have been valuable, but have been limited by a general failure to account for important confounding factors, including smoking and concurrent illness. Furthermore, no long-term data are available on the impact of BMI on survival in a large series of patients who have undergone PCI or CABG.
Herein, we evaluate the impact of BMI on short- and long-term outcomes in a large series of patients undergoing coronary revascularization enrolled in the Bypass Angioplasty Revascularization Investigation (BARI).
The details and design of BARI have been published elsewhere (13,14). Briefly, BARI is a multicenter trial comparing CABG with percutaneous transluminal coronary angioplasty (PTCA). Patients with clinically severe angina or objective evidence of ischemia and angiographically documented multivessel disease suitable for both CABG and PTCA were eligible for enrollment. Patients with a history of coronary revascularization, single-vessel coronary artery disease or primary congenital, valvular or myocardial disease and those <17 years or >80 years were excluded. After obtaining written, informed patient consent, 1,829 patients were randomly assigned to undergo either initial CABG or PTCA between August 1988 and August 1991. Another 2,010 eligible patients who declined randomization were also followed in the BARI observational registry. The study was approved by the Institutional Review Board of the University of Pittsburgh. The present report is based on patients in both the randomized trial and the observational registry who underwent initial revascularization (as defined subsequently).
The initial revascularization was carried out within two weeks of randomization for patients in the randomized trial and within three months of enrollment for patients in the observational registry. Patients in the observational registry who did not undergo revascularization within three months were considered to be “medically treated.” Staged PTCA procedures were counted as a single procedure. The Core Angiographic Laboratory at Stanford University Medical Center (Stanford, California) interpreted the angiographic data from patients in the randomized trial, but not from those in the observational registry. Therefore, clinical site angiographic readings were used for all patients. Balloon angioplasty was the only percutaneous technique utilized in BARI, although at a few sites, stenting was available as a bailout device for threatened closure under a Food and Drug Administration investigational protocol. Follow-up clinic visits were conducted at weeks 4 through 14 and at 1, 3 and 5 years. Telephone contact was made at 6 months and at 2 and 4 years. Complete follow-up over 5 years was obtained in 99% of all patients.
Definitions and outcomes
Vital data were obtained for all patients. A Q-wave myocardial infarction (MI) was defined as a new pathologic Q-wave in two or more contiguous leads, or as new left bundle branch block with twice the normal level of total creatine kinase and an abnormal MB isoenzyme level. A major in-hospital event was defined as death, MI, stroke or coma. A cardiopulmonary event was defined as congestive heart failure, pulmonary edema, hypotension requiring treatment, nonfatal cardiac arrest (requiring direct current shock or cardiopulmonary resuscitation), cardiogenic shock or respiratory failure requiring re-intubation or intubation for more than 72 h. Emergency revascularization was defined as emergency PTCA, emergency PTCA with the use of another device or emergent CABG. A local complication was defined as a peripheral arterial embolus or the loss of a peripheral pulse requiring treatment, re-operation for bleeding, wound dehiscence or infection. Long-term mortality (three and five years) included all-cause mortality and cardiac mortality, as determined by an independent Morbidity and Mortality Classification Committee.
Body mass index was calculated as weight in kilograms divided by the square of height in meters (kg/m2). For the purpose of our analysis, we divided the BMI into low (<20 kg/m2), normal (20 to 24.9 kg/m2) overweight (25 to 29.9 kg/m2), obese (30 to 34.9 kg/m2) and severely obese (≥35 kg/m2).
Chi-square analysis and analysis of variance were used to compare differences in baseline and initial procedural characteristics between patients in the five study-defined BMI groups. With the exception of baseline clinical and angiographic characteristics, all analyses were performed separately for PTCA and CABG patients. Crude in-hospital event rates between the BMI groups were compared by chi-square tests; long-term mortality rates (all-cause and cardiac-only) were estimated by the Kaplan-Meier method (15). Deaths due to noncardiac causes were censored in determining cardiac mortality. Unadjusted and adjusted odds ratios of in-hospital events according to the level of BMI were estimated by use of logistic regression; relative risks for long-term mortality were estimated by Cox regression analysis (16). Statistical adjustment included demographic and clinical factors that differed in prevalence between the study-defined BMI groups and were associated with at least one of the outcomes of interest. Both linear and curvilinear effects were evaluated across the five BMI groups. In the analysis of long-term mortality, initial in-hospital deaths were excluded to assess the effect of BMI on late outcomes. We chose to exclude in-hospital deaths, because they are more likely to represent the outcome in the more seriously ill patients and probably differ from late mortality in terms of etiology. All analyses were performed with the SAS System for Windows, version 8.01 (SAS Institute, Cary, North Carolina).
A total of 3,634 patients underwent initial revascularization and were available for analysis. Angioplasty patients included 919 from the randomized series and 1,189 from the observational registry; surgically treated patients included 901 from the randomized series and 625 from the observational registry. According to the study-defined BMI groups, in 64 patients (2%) the BMI was <20 kg/m2, in 867 (24%) it was 20 to 24.9 kg/m2, in 1,677 (46%) it was 25 to 29.9 kg/m2, in 777 (21%) it was 30 to 34.9 kg/m2and in 249 (7%) it was ≥35 kg/m2.
The characteristics of the patients at study entry differed markedly in terms of initial BMI. Body mass index was inversely associated with age, being a current smoker and a history of cancer and was positively associated with a history of treated diabetes, hypertension and previous evidence of major Q-waves (Table 1). In addition, patients at the extremes of BMI were more likely to be female and to have a history of congestive heart failure and unstable angina. Patients with a low BMI (<20 kg/m2) were more likely to have a history of renal dysfunction, and those at the highest level of BMI (≥35 kg/m2) were least likely to be involved in a regular exercise program at study entry (Table 1).
Initial procedural characteristics
Among the 2,108 patients who underwent initial PTCA, the number of intended lesions and attempted lesions and procedural success did not appreciably differ according to BMI (Table 2). Similarly, among the 1,526 patients who had CABG, the presence of calcified aorta, number of grafts placed and ability to graft all intended vessels did not differ according to BMI. However, there was a modest indication of less frequent use of an internal mammary artery graft among patients with BMI <25 kg/m2.
Among patients who had PTCA, the incidence of a major in-hospital event (death, MI, stroke or coma) was highest (5.6%) among the very lean (BMI <20 kg/m2), lowest (2.1%) among the severely obese (BMI ≥35 kg/m2) and intermediate among those not at extremes of BMI (linear p = 0.02) (Table 3). After statistical adjustment, the inverse association between BMI and major in-hospital events was marginally attenuated (linear p = 0.07) and suggested a protective effect at higher levels of BMI. The incidence of an in-hospital cardiopulmonary event was exceptionally high (27.8%) among the very lean (BMI <20 kg/m2) and suggested a strong inverse linear relationship with BMI before (linear p < 0.001) and after (linear p = 0.003) statistical adjustment (Table 3). Local complications were infrequent, but again suggested a lower adjusted risk at higher levels of BMI. Finally, there was a strong inverse linear relationship between BMI and the risk of any in-hospital event before (linear p < 0.001) and after (linear p < 0.001) statistical adjustment.
Among patients who had CABG, baseline BMI was unrelated to the incidence or adjusted risk of in-hospital major events, cardiopulmonary events, local complications or any event or complication (Table 4). Moreover, unlike patients who had PTCA, there was a modest suggestion that very lean (BMI <20 kg/m2), surgically treated patients experienced a relatively more favorable in-hospital outcome.
Long-term mortality after initial hospital discharge
Compared with patients with a normal BMI (20 to 24.9 kg/m2) who had PTCA, the adjusted risk of five-year mortality after initial hospital discharge was lower (p = NS) among PTCA-treated patients classified as overweight (BMI 25 to 29.9 kg/m2) or class I obese (BMI 30 to 34.9 kg/m2) (Table 5). When classified by cause of death, the adjusted risk of five-year cardiac mortality appeared to be curvilinear with a higher risk of mortality at BMI extremes; however, this did not attain statistical significance, perhaps in part due to low statistical power.
Among CABG-treated patients, BMI was positively associated with the risk of five-year mortality, especially cardiac mortality, after the initial in-hospital discharge (Table 6). Indeed, compared with patients with a normal BMI (20 to 24.9 kg/m2), those with the highest BMI (≥35 kg/m2) were estimated to have a nearly fivefold higher adjusted risk of cardiac mortality. Moreover, the adjusted risk of cardiac mortality increased monotonically at higher levels of BMI (linear p < 0.001).
In this study, we observed a surprising dichotomy between PTCA- and CABG-treated patients in short- and long-term outcomes across levels of BMI. Among patients who had PTCA, each unit increase in BMI was associated with a 5.5% lower adjusted risk of a major in-hospital event, whereas among patients who had CABG, no difference in the in-hospital outcome was observed. In contrast, BMI was not associated with five-year mortality or cardiac mortality in PTCA-treated patients, whereas among CABG-treated patients, each unit increase in BMI was associated with an 11% higher adjusted risk of five-year cardiac mortality.
Possible explanations for the findings
Our findings in the PTCA group are similar to those in a previous report (8). Patients classified as overweight or obese had the lowest risk of an in-hospital major event, cardiopulmonary complications or local complications after PTCA. We postulate that this improved procedural outcome may be a function of a larger vessel size and a relatively favorable artery/device ratio (17). The outcome after PCI is worse in patients with smaller vessels, and it may be speculated that a smaller BMI is associated with smaller coronary vessels (18). Furthermore, as the local risk of complications increases with an increasing sheath size (19), it may well be that the relative sheath/artery mismatch may contribute to an increased risk of local complications in lean patients or may have a protective effect in those with a larger body mass.
In agreement with previous studies (10–12), BMI was not associated with a short-term outcome among CABG-treated patients. However, long-term mortality in CABG-treated patients appears to be strongly influenced by BMI. The increased risk of mortality at higher levels of BMI is most likely due, at least in part, to the increased likelihood of obese patients having multiple coronary risk factors. Furthermore, obesity is associated with a high rate of cardiovascular and noncardiovascular morbidity and mortality, and it is unlikely that coronary revascularization completely attenuates this increased risk. Finally, the adverse effect of BMI in CABG-treated patients appears to be independent of the concomitant risk factors (confounding variables) associated with BMI.
The variable long-term outcome in the PTCA group, and its inconsistency with the long-term outcome in the CABG group, needs to be interpreted with caution. The apparently comparable (if not higher) long-term survival in the overweight and class I obese patients after PCI (as compared with lean and normal-weight patients) may speculatively be partly due to delayed protective effects of a favorable device/artery ratio. Whether this finding reflects a chance variation rather than a valid result should be explored in other large series of PCI-treated patients.
The number of patients at the extreme levels of BMI was small; thus, the apparent worse short-term outcome in the lean, PCI-treated patients needs confirmation from other studies. Similarly, statistical power was limited for groups at the extremes of BMI. In addition, it is possible that BMI may play a role in the choice of modality of revascularization. To verify or refute such a possibility, we analyzed the patients followed in the registry and did not find an interaction between BMI and the choice of revascularization modality. Finally, the tools and techniques of PCI have undergone significant change since the time of BARI, and the increased complication rate observed in lean patients may not persist today with current interventional devices and adjunctive platelet glycoprotein IIb/IIIa inhibitor usage. Hence, our data underscore the need to study the impact of BMI on the outcome of coronary revascularization in the current era. If similar results are observed, then BMI would need to be one of the factors taken into consideration while choosing the modality of revascularization (i.e., a case could be made for preferring CABG over PCI in lean patients, and vice versa for obese patients).
Body mass index appears to have a differential impact on short- and long-term outcomes after coronary revascularization. Although a higher BMI seems to be associated with a better short-term outcome after PCI, BMI does not significantly impact the early outcome after CABG. In contrast, an increased BMI is associated with a worse long-term outcome after CABG, but not after PCI. Our findings underscore the need for further research to determine the factors that may be responsible for the apparent short-term protective effect of a higher BMI in patients undergoing PCI and to study the impact of weight reduction on the long-term survival of obese patients undergoing coronary revascularization.
☆ This study was supported by the following grants from the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland: HL38493, HL38504, HL38509, HL38512, HL38514-6, HL38518, HL38524-5, HL38529, HL38532, HL38556, HL38610, HL38642 and HL42145.
- Bypass Angioplasty Revascularization Investigation
- body mass index
- coronary artery bypass graft surgery
- myocardial infarction
- percutaneous coronary intervention
- percutaneous transluminal coronary angioplasty
- Received August 2, 2001.
- Revision received December 4, 2001.
- Accepted December 14, 2001.
- American College of Cardiology Foundation
- American Heart Association
- Jousilahti P,
- Tuomilehto J,
- Vartiainen E,
- Pekkanen J,
- Puska P
- Northern New England Cardiovascular Disease Study Group,
- Birkmeyer N.J,
- Charlesworth D.C,
- Hernandez F,
- et al.
- Cox D.R
- Schunkert H,
- Harrell L,
- Palacios I.F
- Foley D.P,
- Melkert R,
- Serruys P.W