Body Mass Index and Mortality Among Adults Undergoing Cardiac Surgery

Abstract

Background: In an apparent paradox, morbidity and mortality are lower in obese patients undergoing cardiac surgery, although the nature of this association is unclear. We sought to determine whether the obesity paradox observed in cardiac surgery is attributable to reverse epidemiology, bias, or confounding. Methods: Data from the National Adult Cardiac Surgery registry for all cardiac surgical procedures performed between April 2002 and March 2013 were extracted. A parallel systematic review and meta-analysis (MEDLINE, Embase, SCOPUS, Cochrane Library) through June 2015 were also accomplished. Exposure of interest was body mass index categorized into 6 groups according to the World Health Organization classification. Results: A total of 401 227 adult patients in the cohort study and 557 720 patients in the systematic review were included. A U-shaped association between mortality and body mass index classes was observed in both studies, with lower mortality in overweight (adjusted odds ratio, 0.79; 95% confidence interval, 0.76–0.83) and obese class I and II (odds ratio, 0.81; 95% confidence interval, 0.76–0.86; and odds ratio, 0.83; 95% confidence interval, 0.74–0.94) patients relative to normal-weight patients and increased mortality in underweight individuals (odds ratio, 1.51; 95% confidence interval, 1.41–1.62). In the cohort study, a U-shaped relationship was observed for stroke and low cardiac output syndrome but not for renal replacement therapy or deep sternal wound infection. Counter to the reverse epidemiology hypotheses, the protective effects of obesity were less in patients with severe chronic renal, lung, or cardiac disease and greater in older patients and in those with complications of obesity, including the metabolic syndrome and atherosclerosis. Adjustments for important confounders did not alter our results. Conclusions: Obesity is associated with lower risks after cardiac surgery, with consistent effects noted in multiple analyses attempting to address residual confounding and reverse causation.