Clinical Investigation and Reports. Effectiveness of Coronary Artery Bypass Grafting With or Without Cardiopulmonary Bypass in Overweight Patients

Clinical Investigation and Reports Effectiveness of Coronary Artery Bypass Grafting With or Without Cardiopulmonary Bypass in Overweight Patients Raimondo Ascione, MD; Barnaby C. Reeves, DPhil; Karen Rees, PhD; Gianni D. Angelini, FRCS Background Off-pump coronary artery bypass surgery has been demonstrated to reduce morbidity in elective patients. However, high-risk patients might benefit the most from this surgical procedure. Our goal was to investigate the effectiveness of on-pump and off-pump coronary artery bypass surgery on early clinical outcome in a consecutive series of overweight patients. Methods and Results From April 1996 to April 2001, data on 4321 patients undergoing coronary surgery (mortality 1.4%) were prospectively entered into the Patient Analysis and Tracking System. Data were extracted for all patients with a body mass index 25 kg/m 2. A risk-adjusted analysis was performed to assess the effect of surgical technique in the whole overweight cohort. 2844 patients were identified (2261 male, median age 63, interquartile range 56 to 68). Patients undergoing on-pump surgery (2170, 76.3%) were less likely than those undergoing off-pump surgery to have hypercholesterolemia or left main stem disease and were, on average, less obese. However, they were more likely to have unstable angina and to have had a previous myocardial infarction, and they had more extensive coronary disease and received more grafts (all P 0.05). Unadjusted analyses, taking account only of consultant team, showed significant benefits of off-pump surgery in terms of hospital deaths, arrhythmias, inotropic use, use of intra-aortic balloon pump, blood loss, transfusion requirement, postoperative hemoglobin, chest infections, neurological complications, intensive care unit and hospital stay (all P 0.05). After adjustment for confounding prognostic factors, the benefits of off-pump surgery were still significant for death in hospital, transfusion requirement, postoperative hemoglobin, neurological complications, intensive care unit and hospital stay (ORs 0.35 to 0.79, P 0.05). Conclusions These results suggest that off-pump surgery is safe and effective and is associated with a reduced in-hospital mortality and morbidity in overweight patients when compared with conventional coronary surgery with cardiopulmonary bypass and cardioplegic arrest. (Circulation. 2002;106:1764-1770.) Key Words: obesity cardiopulmonary bypass morbidity coronary disease mortality Over recent decades, improvement of socioeconomic conditions has led to an expansion of the overweight population worldwide. Obesity is well known to be a risk factor for the development of diabetes mellitus, hypertension, and coronary artery disease. 1 3 It is also thought to be a risk factor for perioperative morbidity and mortality with cardiac surgery and other major surgical procedures. 4 The latest review of information from The Society of Thoracic Surgeons National Cardiac Surgery Database indicated that morbid obesity is an independent predictor of increased operative mortality in patients undergoing coronary artery bypass grafting (CABG). 5 In recent randomized controlled trials, we demonstrated that performing coronary surgery on the beating heart without the extracorporeal circulation reduces morbidity in elective patients. 6,7 However, there is no evidence about the effectiveness of off-pump coronary artery bypass surgery in overweight patients. The use of a high-quality prospective database to document all patients undergoing coronary revascularization in our institution created the opportunity to evaluate the effectiveness of modern techniques of coronary surgery on early clinical outcome with the use of a cohort study design. Methods Patient Selection and Data Collection A standard set of perioperative data are collected prospectively for all patients undergoing CABG at our institution. The data set includes 5 different sections to be filled in consecutively by anesthetist, surgeon, intensive care unit (ICU), high-dependency unit, and ward nurses. Data are entered into the Patient Analysis and Tracking Systems, Dendrite Clinical Systems, London, UK. In the present study, the definition of overweight has been based on body mass index (BMI, defined as weight/height 2 ), because it is the body size measurement that best correlates with body fat content. 8 Patients with BMI 25 kg/m 2 were classified as overweight as specified by the American Heart Association guidelines. 3 Data were extracted from the database for consecutive patients with BMI 25 kg/m 2 who had undergone CABG between April 1996 and April 2001. Received April 30, 2002; revision received July 18, 2002; accepted July 19, 2002. From the Bristol Heart Institute, University of Bristol, Bristol, UK. Correspondence to Professor G.D. Angelini, Bristol Heart Institute, University of Bristol, Bristol Royal Infirmary, Bristol, England BS2 8HW. E-mail g.d.angelini@bristol.ac.uk 2002 American Heart Association, Inc. Circulation is available at http://www.circulationaha.org DOI: 10.1161/01.CIR.0000032259.35784.BF 1764

Ascione et al Off-Pump Coronary Surgery in Overweight Patients 1765 Allocation to on-pump and off-pump surgery was on the basis of the preference and expertise of the surgeons performing the operations, as well as (during the early part of data collection when experience was still being gained with off-pump techniques) on the basis of coronary anatomy and number of grafts required. The decision about which method to use was made after opening the chest, when the anatomy of the coronary vessels could be explored. Anesthetic and Heparin Management Anesthetic technique was standardized for all patients and has been previously reported. 6 Briefly, this consisted of intravenous anesthesia with propofol infusion at 3 mg/kg per hour combined with remifentanil infusion at 0.5 to 1 g/kg per minute. Neuromuscular blockade was achieved by 0.1 to 0.15 mg/kg pancuronium bromide or vecuronium and the lungs ventilated to normocapnia with air and oxygen (45% to 50%) without positive end expiratory pressure (PEEP). In the on-pump group, heparin was given at a dose of 300 IU/kg to achieve a target activated clotting time (ACT) of 480 seconds before commencement of cardiopulmonary bypass (CPB). In the off-pump group, heparin (100 IU/kg) was administered before the start of the first anastomosis to achieve an ACT of 250 to 350 seconds. On completion of all anastomoses, protamine was given to reverse the effect of heparin and return the ACT to preoperative levels. Surgical Technique On Pump CPB was instituted by use of ascending aortic cannulation and 2-stage venous cannulation of the right atrium. A standard circuit Bard tubing set was used, which included a 40- m filter, a Stockert roller pump, and a hollow fiber membrane oxygenator (Sorin Biomedica). Nonpulsatile flow was used and flow rates throughout bypass were 2.4 L/m 2 per minute. The systemic temperature was kept between 34 C and 36 C. Myocardial protection was achieved by use of intermittent anterograde hyperkaliemic warm blood cardioplegia. 6 Off Pump The method of exposure and stabilization to perform the anastomosis consisted of a technique previously described by our group. 9 Briefly, the target vessel was exposed and snared above the anastomotic site with the use of a 4-0 Prolene suture with a soft plastic snugger. The coronary artery was then opened, an intracoronary shunt was used in case of relative electrocardiographic or hemodynamic instability or excessive bleeding, and the anastomosis performed. Postoperative Management At the end of surgery, patients were transferred to the ICU. Forced-air warming was used until a stable nasopharyngeal temperature of 37 C had been reached. Patients were extubated as soon as they met the following criteria: hemodynamic stability, no excessive bleeding ( 80 ml/h), normothermia, and consciousness with pain control. Fluid management postoperatively consisted of 5% dextrose infused at 1 ml/kg per hour, with additional Gelofusine or blood to maintain normovolemia and hematocrit 24%. Figure 1. Percentage of overweight patients with BMI 25 distributed by year and by surgical technique (1996 [April to December]). Clinical Data Collection, Monitoring, and Definitions Data characterizing perioperative clinical outcome were entered prospectively into the Patient Analysis and Tracking System database. In-hospital mortality was defined as any death that occurred within 30 days of operation. Perioperative myocardial infarction (MI), ST-segment changes, pacing, arrhythmias, and inotropic requirements were recorded and defined as previously reported. 6 Pulmonary complications included chest infection, ventilation failure, reintubation, and tracheostomy. Postoperative blood loss was defined as total chest tube drainage. 10 Neurological complications included permanent and transient stroke. 11 Renal complication included acute renal failure as defined by the requirement of hemodialysis. Finally, infective complications included septicemia and sternal and leg wound infection, as defined by positive culture and requiring antibiotic therapy. 6 Statistical Analysis Patients with BMI 25 kg/m 2 were classified according to the surgical technique used. Conversions from off-pump to on-pump surgery were analyzed by intention-to-treat. The distributions of prognostic factors in on-pump and off-pump groups were compared with the use of 2 tests for categorical variables and Mann-Whitney rank-sum tests for continuous variables. Outcomes were initially compared between groups by use of logistic regression, adjusting only for the consultant teams performing the operation (entered in the model as fixed effects). These analyses allowed for varying performance between consultant teams but not for imbalance in prognostic factors between groups. To take account of imbalances in prognostic factors, propensity scores were calculated. 12 These scores estimated the probability of membership of on-pump or off-pump groups, as predicted by all known characteristics of the entire cohort of obese patients. Scores for each individual were derived by logistic regression and then transformed into logits. 12 Multivariable comparisons were performed by use of multiple logistic regression. The method varied depending on the frequency of the outcome of interest. For rare outcomes ( 50 events in the off-pump group), models included only consultant teams and dummy variables representing quintiles of the propensity scores. (For some outcomes, which did not occur in either group for one or more consultant teams, the effects of consultant team could not be modeled. For these outcomes, robust standard errors for the effect of off- versus on-pump surgery were estimated instead, to take account of the clustering of patients within surgical teams.) For frequent outcomes ( 50 events in the off-pump group), models included dummy variables representing quintiles of the propensity scores and Figure 2. Distribution of BMI and sex in the overweight CABG population.

1766 Circulation October 1, 2002 TABLE 1. Distribution of Prognostic Variables in On- and Off-Pump Groups at Admission On Pump (n 2170) Off Pump (n 674) Prognostic Variable n* % n* % P Age, y 63 56 68 63 56 68 0.93 Male 1717 79.1 544 80.7 0.37 Canadian Cardiovascular Society class 0.06 2 792 36.6 280 41.6 3 765 35.3 214 31.8 4 608 28.1 179 26.6 Unstable angina 922 42.6 239 35.5 0.001 Previous myocardial infarction 992 45.8 264 39.3 0.003 Diabetes 339 15.7 113 16.8 0.47 Hypercholesteremia 1566 72.5 527 78.7 0.002 Hypertension 1234 57.1 375 55.8 0.56 Current smoker 281 13.0 88 13.1 0.95 Creatinine 200 mol/l 19 0.9 11 1.6 0.09 Chronic obstructive airways disease 98 4.5 23 3.4 0.22 Cerebral vascular accident 120 5.5 46 6.9 0.21 Left main stem stenosis 50% 255 11.8 90 13.4 0.001 Peripheral vascular disease 186 8.6 65 9.7 0.38 Parsonnet score 0.24 0 10 1867 86.1 582 86.4 11 20 287 13.2 83 12.3 20 15 0.7 9 1.3 Ejection fraction 0.30 Good 1590 73.7 513 76.6 Fair 476 22.1 129 19.3 Poor 92 4.3 28 4.2 Extent of coronary heart disease 0.001 Single vessel 95 4.5 115 17.2 Double vessel 499 23.4 221 33.0 Triple vessel 1539 72.2 333 49.8 Operative priority 0.25 Elective 1283 59.2 420 62.4 Urgent 861 39.7 248 36.9 Emergency/salvage 25 1.2 5 0.7 No. of grafts 0.001 1 95 4.4 127 18.9 2 537 24.8 255 37.9 3 1044 48.1 249 37.0 4 494 22.8 42 6.2 Body mass index, kg/m 2 0.11 25 to 29.99 1573 72.5 462 68.6 30 to 34.99 501 23.1 174 25.8 35 96 4.4 38 5.6 Median body mass index, kg/m 2 28.0 26.5 28.4 26.7 0.01 30.3 30.5 *Column totals do not always sum to 2170 and 674, respectively, because there were missing data for some variables (maximum 1.5%). P values are reported for 2 tests unless otherwise indicated. For variables with more than 2 categories, a 2 test was carried out across all categories. Median and interquartile range shown for continuously distributed variables; medians compared with Mann-Whitney rank-sum tests.

Ascione et al Off-Pump Coronary Surgery in Overweight Patients 1767 TABLE 2. Distributions of Postoperative Outcomes in Patients Undergoing On-Pump and Off-Pump Surgery On Pump (n 2170) Off Pump (n 674) Outcome n* % n* % Deaths 19 0.9 2 0.3 Myocardial infarction 35 1.6 14 2.1 Inotropes support 880 40.8 219 32.6 Intra-aortic balloon pump 54 2.5 5 0.7 Arrhythmia 320 14.8 65 9.7 Postoperative VF/VT 31 1.4 1 0.2 Pulmonary complication 258 12.0 51 7.6 Chest infection 106 4.9 14 2.1 Neurological complication 59 2.7 6 0.9 Permanent stroke 15 0.7 0 0.0 Transient stroke 14 0.7 1 0.2 Infective complication 49 2.3 12 1.8 Septicemia 25 1.2 4 0.6 Renal complication 98 4.5 24 3.6 Hemodialysis 19 0.9 4 0.6 Gastrointestinal complication 23 1.1 4 0.6 Multisystem failure 11 0.5 0 0.0 Reoperation for 72 3.3 16 2.4 bleeding/tamponade Blood loss, ml 500 250 16.0 118 19.2 500 and 1000 734 47.1 334 54.3 1000 and 2000 505 32.4 151 24.6 2000 70 4.5 12 2.0 Median blood loss, ml 850 600 1180 775 550 1030 Total red blood cell transfusion, units 0 829 50.4 526 80.4 1 2 599 36.4 100 15.3 3 217 13.2 28 4.3 Total platelet transfusion, units 0 1400 85.3 612 93.9 1 2 212 12.9 37 5.7 3 29 1.8 3 0.5 Any fresh frozen plasma 210 12.8 27 4.1 transfusion Postoperative hemoglobin 10 819 50.0 202 31.1 g/dl Median postoperative 10.0 9.3 10.9 10.8 9.8 11.7 hemoglobin, g/dl No. of hours ventilated* 0 5 541 25.0 227 33.8 6 10 856 39.6 282 42.0 11 20 565 26.1 132 19.7 20 200 9.3 30 4.5 Median no. of hours ventilated* 8 6 15 7 5 10 TABLE 2. (Continud) On Pump (n 2170) Off Pump (n 674) Outcome n* % n* % ICU stay, nights 0 115 5.3 76 11.3 1 1547 71.6 494 73.4 2 273 12.6 58 8.6 3 128 5.9 28 4.2 4 99 4.6 17 2.5 Median ICU stay, d 1 1 1 1 1 1 Combined ICU and HDU stay, nights 1 125 5.9 94 14.0 2 961 44.9 340 50.8 3 498 23.3 112 16.7 4 268 12.5 58 8.7 5 287 13.4 66 9.9 Median combined ICU and HDU 2 2 4 2 2 3 stay, d Total length of stay, d 0 7 1416 65.9 529 79.0 8 10 446 20.8 95 14.2 11 15 182 8.5 32 4.8 16 104 4.8 14 2.1 Median length of stay, d 7 6 8 6 5 7 VF/VT indicates ventricular fibrillation/tachycardia; ICU, intensive care unit; and HDU, high-dependency unit. *Column totals do not always sum to 2170 and 674, respectively, because there were missing data for some variables (maximum 1.2%, except for variables describing blood loss, transfusion requirement, and postoperative hemoglobin, which were not collected during the first year). Median and interquartile range are shown for continuously distributed variables. Excluding length of stay for 21 patients who died in hospital. all prognostic factors observed to be distributed unevenly between groups (unstable angina, previous myocardial infarction, hypercholesterolemia, left main stem stenosis 50%, extent of coronary disease, number of grafts, and BMI group). Because imbalance in prognostic factors between groups was already taken into account by the propensity score, inclusion of these additional variables rarely altered the estimate of the effect of off-pump surgery. Exact P values and confidence intervals are presented throughout. The large number of prognostic variables and outcomes of interest resulted in many statistical comparisons. No correction was made for multiple comparisons, but our interpretation of the findings takes into account the consistency of the associations observed and their magnitude, as well as their statistical significance. Results A total of 2844 patients with BMI 25 underwent CABG (674 off pump, 23.7%) under the care of 4 consultant teams. (Operations performed by trainee surgeons were included in the consultant team responsible for the care of the patient.) An increase in the proportion of off-pump surgery during the study period was observed (Figure 1). There were 2 conversions to on-pump surgery among patients in whom a decision to perform off-pump surgery had been made.

1768 Circulation October 1, 2002 The distributions of prognostic characteristics in on-pump and off-pump groups are shown in Table 1 and the postoperative outcomes in Table 2. Overall distribution of BMI and sex is shown in Figure 2. Patients undergoing on-pump surgery were less likely to have hypercholesterolemia or left main stem disease (stenosis 50%), and they were, on average, less obese. However, they were more likely to have unstable angina and to have had a previous MI, and, on average, they had more extensive coronary heart disease and received more grafts than those undergoing off-pump surgery. The analyses of in-hospital outcomes showed significant benefits from off-pump surgery for transfusion requirement, postoperative arrhythmias and inotropic use, use of intraaortic balloon pump, postoperative blood loss, postoperative hemoglobin, pulmonary complications (including chest infec- TABLE 3. Effect of Off-Pump Versus On-Pump Surgery on Postoperative Outcome: Unadjusted and Adjusted Effect Sizes Unadjusted Effect Size* Adjusted Effect Size Outcome Odds Ratio 95% CI P Odds Ratio 95% CI P Deaths 0.34 0.16 to 0.71 0.004 0.37 0.18 to 0.77 0.008 Myocardial infarct 1.23 0.58 to 2.58 0.59 2.29 0.91 to 5.76 0.08 Inotropes support 0.63 0.51 to 0.78 0.001 0.81 0.63 to 1.03 0.08 Intra-aortic balloon pump 0.37 0.14 to 1.01 0.05 0.39 0.14 to 1.15 0.09 Arrhythmia 0.63 0.46 to 0.87 0.005 0.73 0.51 to 1.04 0.08 Postoperative VF/VT 0.11 0.01 to 0.88 0.04 0.14 0.01 to 1.47 0.10 Pulmonary complication 0.64 0.45 to 0.91 0.01 0.83 0.56 to 1.23 0.35 Chest infection 0.48 0.26 to 0.89 0.02 0.83 0.42 to 1.66 0.60 Neurological complication 0.29 0.11 to 0.76 0.01 0.35 0.12 to 1.00 0.05 Permanent stroke 0.00 Transient stroke 0.15 0.02 to 1.46 0.10 0.23 0.02 to 2.56 0.23 Infective complication 0.85 0.40 to 1.78 0.66 1.01 0.43 to 2.39 0.98 Septicemia 0.51 0.15 to 1.75 0.28 0.63 0.15 to 2.54 0.51 Renal complication 1.07 0.64 to 1.80 0.79 1.49 0.81 to 2.72 0.20 Dialysis required 0.68 0.32 to 1.42 0.30 0.90 0.44 to 1.85 0.78 Gastrointestinal complication 0.60 0.18 to 2.00 0.40 0.55 0.15 to 2.03 0.37 Multisystem failure 0.00 Reoperation bleeding tamponade 0.57 0.30 to 1.06 0.08 0.56 0.28 to 1.10 0.09 Blood loss 1000 ml 0.64 0.51 to 0.81 0.001 0.81 0.62 to 1.05 0.11 Any RBC transfusion (0 units vs 1 unit) 0.26 0.20 to 0.33 0.001 0.40 0.30 to 0.52 0.001 Any PLT transfusion (0 units vs 1 unit) 0.37 0.25 to 0.54 0.001 0.48 0.32 to 0.73 0.001 Any FFP transfusion (0 units vs 1 unit) 0.35 0.22 to 0.54 0.001 0.46 0.29 to 0.75 0.002 Postoperative hemoglobin 10 g/dl 0.44 0.36 to 0.55 0.001 0.55 0.43 to 0.70 0.001 Ventilated 10 h 0.67 0.54 to 0.83 0.001 0.86 0.67 to 1.10 0.22 ICU stay 1 night 0.54 0.41 to 0.70 0.001 0.69 0.51 to 0.94 0.02 ICU or HDU stay 2 nights 0.56 0.45 to 0.69 0.001 0.79 0.62 to 0.99 0.04 Postoperative length of stay 7 d# 0.49 0.38 to 0.62 0.001 0.67 0.51 to 0.88 0.004 CI indicates confidence interval; VF/VT, ventricular fibrillation/tachycardia; RBC, red blood cells; PLT, platelets; FFP, fresh frozen plasma; ICU, intensive care unit; and HDU, high-dependency unit. *Odds ratio adjusted only for consultant team carrying out the operation (except where indicated ). Unadjusted odds ratios were based on a sample size of up to 2844, and adjusted odds ratios on a sample size of up to 2743 (propensity scores could not be calculated for 101 patients who had 1 prognostic factor missing), with a maximum of 1.2% missing data for any outcome except blood loss, transfusion requirement, and post-operative haemoglobin; these latter outcomes were not documented during the first year (20% of patients). Estimates of odds ratios for off vs on-pump surgery adjusted for quintiles of propensity score (see text) and consultant team (except where indicated ). Odds ratios adjusted for quintiles of propensity score (see text), consultant team, presence of unstable angina, previous myocardial infarction, hypercholesteremia, left main stem stenosis 50%, extent of coronary disease, number of grafts, and body mass index group. Odds ratios were not adjusted for consultant team because one or more team experienced no events (see *). Confidence intervals could not be calculated, and a multivariable model could not be estimated because there were no events among off-pump patients. #Excluding length of stay for 21 patients who died in hospital.

Ascione et al Off-Pump Coronary Surgery in Overweight Patients 1769 tions), neurological complications, intubation time, ICU and hospital length of stay (all P 0.05)(Table 3). Furthermore, after adjustment for prognostic variables, odds ratios (ORs) for many of the adverse outcomes investigated still indicated significant benefit of off-pump surgery (point estimates of ORs, 0.35 to 0.79, P 0.05), ie, hospital deaths, neurological complications, transfusion requirement, postoperative hemoglobin, ICU and hospital length of stay. Discussion Obesity is often considered to be a risk factor for perioperative morbidity and mortality with cardiac surgery and other major surgical procedures despite little evidence in the literature. 13 The American Heart Association recognizes obesity as a major risk for coronary artery disease and suggests that this condition is often associated with other risk factors including diabetes, hypertension, and impaired respiratory function. 3,14,15 The results of prior studies of the relationship between obesity and the risk of adverse clinical outcomes after CABG procedures have been rather contradictory. 15 19 A recent analysis of the database of the Society of Thoracic Surgeons has shown that morbid obesity is an independent predictor of increased operative mortality in CABG patients. 5 Prasad et al, 15 in a comparison of 250 obese patients versus 250 age-matched and sex-matched controls of normal BMI, found obesity to be an independent risk factor for perioperativemorbidity. On the other hand, Christakis et al, 16 in a retrospective study on a population of 7025 patients, showed that sex, but not BMI, was a significant independent predictor of operative mortality and low-output syndrome, after adjusting for other preoperative risk variables. Similar results were obtained by Brandt et al, 17 who concluded that severe obesity does not necessarily adversely affect perioperative mortality and morbidity. Equal or even reduced operative mortality in obese versus nonobese patients has also been shown in a recent large study by Birkmeyer et al. 13 The investigation of the influence of off-pump surgery on early clinical outcome in overweight patients was the primary objective of the present study. Statistically significant benefits from off-pump surgery were identified for the majority of the outcomes investigated, and these effects remained after adjustment for confounding prognostic variables, including BMI. The magnitude of the differences between groups in terms of in-hospital morbidity is quite striking, with the odds of some outcomes being reduced by more than half and the odds of several other outcomes by one third to one half. There are several possible explanations for these findings: In a recent series of studies designed to compare off-pump with conventional CABG, we provided evidence for greater inflammatory activation and subsystem organ dysfunction in elective patients undergoing on-pump surgery. 6,11,20,21 These effects may be even more pronounced in overweight patients. Obesity has been shown to be independently associated with coronary endothelial dysfunction even in patients with normal or mildly diseased coronary arteries. 22 This association, when combined with the high CPB-related inflammatory response, 23 25 might explain the increased morbidity observed in the on-pump groups of the present study. The greater degree of hemodilution because of CPB circuit prime volume and colloid administration on weaning from CPB with the resultant greater oncotic pressure gradient may also favor fluid extravasation. Such oncotic forces, combined with the known effects of CPB on complement activation and the subsequent compromise of endothelial integrity, may lead to increased net fluid retention through capillary leak. 18,19 Increased fluid retention increases the likelihood of postoperative lung edema with deleterious effects on lung mechanisms and tissue oxygenation and increased need for mechanical ventilation. 18 It might be argued that, without blinding of the patients and their caregivers, some of the outcomes could be biased by knowledge of the type of surgery. We believe that bias is unlikely to be an explanation for the results. First, none of the healthcare staff were aware during the period of the study that the comparison was going to be made. Second, as described above (see Methods), strict local guidelines are used to make decisions about perioperative patient care management and these guidelines were applied carefully throughout the period of the study and would have minimized the opportunity for bias. Without random allocation it is not possible to exclude the possibility that the findings arise from confounding. However, we believe that confounding is unlikely to explain the benefits observed from off-pump surgery for several reasons. The method for allocating patients to on-pump or off-pump surgery (see Methods) does not appear to have created 2 groups with substantially different risks of adverse outcomes, although off-pump patients required, on average, fewer grafts. The effects of off-pump surgery were large across a wide range of outcomes. In conclusion, off-pump surgery in overweight patients is safe and is associated with reduced mortality and morbidity when compared with conventional CABG with cardiopulmonary bypass and cardioplegic arrest. Acknowledgments This work was supported by the Garfield Weston Trust and the British Heart Foundation. References 1. Barret-Connor EL. Obesity, arteriosclerosis, and coronary artery disease. Ann Intern Med. 1985;103:1010 1019. 2. Simopoulos AP, Van Nallie TB. Body weight, health and longevity. Ann Intern Med. 1984;100:285 295. 3. St Jeor ST, Brownell KD, Atkinson RL, et al. Obesity. Workshop III. AHA prevention conference III. Behavior change and compliance: keys to improving cardiovascular health. Circulation. 1993;88:1391 1396. 4. Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation. 1989;79:I3 12. 5. Edwards FH, Carey JS, Grover FL, et al. Impact of gender on coronary bypass operative mortality. Ann Thorac Surg. 1998;66:125 131. 6. Ascione R, Caputo M, Calori G, et al. Predictors of atrial fibrillation after conventional and beating heart coronary surgery: a prospective, randomized study. Circulation. 2000;102:1530 1535. 7. Angelini GD, Taylor FC, Reeves BC, et al. Early and midterm outcome after off-pump and on-pump surgery in Beating Heart Against Car-

1770 Circulation October 1, 2002 dioplegic Arrest Studies (BHACAS 1 and 2): a pooled analysis of two randomised controlled trials. Lancet. 2002;359:1194 1199. 8. Criqui MH, Klauber MR, Barrett-Conner EL, et al. Adjustment for obesity in studies of cardiovascular disease. Am J Epidemiology. 1982; 116:685 691. 9. Watters MPR, Ascione R, Ryder IG, et al. Hemodynamic changes during beating heart coronary surgery with the Bristol Technique. European Journal of Cardio-Thoracic Surgery. 2001;19:34 40. 10. Ascione R, Williams S, Lloyd CT, et al. Reduced postoperative blood loss and transfusion requirement after beating heart coronary surgery: a prospective randomised study. J Thorac Cardiovasc Surg. 2001;121: 689 696. 11. Lloyd CT, Ascione R, Underwood MJ, et al. Serum S-100 protein release and neuropsychologic outcome during coronary revascularisation on the beating heart: A prospective randomized study. J Thorac Cardiovasc Surg. 2000;119:148 154. 12. Blackstone EH. Comparing apples and oranges. J Thorac Cardiovasc Surg. 2002;123:8 15. 13. Birkmeyer NJO, Charlesworth DC, Hernandez F, et al. Obesity and risk of adverse outcome associated with coronary artery bypass surgery. Circulation. 1998;97:1689 1694. 14. Koshal A, Hendry P, Raman SV, et al. Should obese patients not undergo coronary artery surgery? Canadian J Surg 1985;28:331 341. 15. Prasad US, Walker WS, Sang CTM, et al. Influence of obesity on the early and long term results of surgery for coronary artery disease. Eur J Cardiothorac Surg 1991;5:67 73. 16. Christakis GT, Weisel RD, Buth KJ, et al. Is body size the cause for poor outcomes of coronary artery bypass operations in women? J Thorac Cardiovasc Surg 1995;110:1344 1358. 17. Brandt M, Harder K, Knut P, et al. Severe obesity does not adversely affect perioperative mortality and morbidity in coronary artery bypass surgery. Eur J Cardiothorac Surg. 2001;19:662 666. 18. Schwann TA, Habib RH, Zacharias A, et al. Effects of body size on operative, intermediate and long-term outcome after coronary artery bypass operation. Ann Thorac Surg. 2001;71:521 531. 19. Engelman DT, Adams DH, Byrne JG, et al. Impact of body mass index and albumin on morbidity and mortality after cardiac surgery. J Thorac Cardiovasc Surg 1999;118:866 873. 20. Ascione R, Lloyd CT, Underwood MJ, et al. On Pump versus off pump coronary revascularization: evaluation of renal function. Ann Thorac Surg. 1999;68:493 498. 21. Ascione R, Lloyd CT, Gomes WJ, et al. Beating versus arrested heart revascularixation: evaluation of myocardial function in a prospective randomised study. Eur J Cardio-Thorac Surg. 1999;15:685 690. 22. Al Suwaidi J, Higano ST, Holmes DR, et al. Obesity is independently associated with coronary endothelial dysfunction in patients with normal or mildly diseased coronary arteries. J Am Coll Cardiol 2001;37: 1523 1528. 23. Butler J, Rocker GM, Westaby S. Inflammatory response to cardiopulmonary bypass. Ann Thorac Surg. 1993;55:552 559. 24. Brasil LA, Gomes WJ, Salomao R, et al. Inflammatory response after myocardial revascularization with or without cardiopulmonary bypass. Ann Thorac Surg. 1998;66:56 59. 25. Ascione R, Lloyd CT, Underwood MJ, et al. Inflammatory response after coronary revascularization with or without cardiopulmonary bypass. Ann Thorac Surg. 2000;69:1198 1204.