Glucose metabolism, not obesity, predicts mortality in critically ill surgical patients

Our hypothesis was to determine if insulin resistance and hyperglycemia, rather than obesity, are predictive of mortality in the surgically critically ill. An observational study of an automated protocol in surgical and trauma intensive care units was performed. Two groups were created based on body mass index (BMI): Obese (OB) defined as BMI > or = 30 (n = 338) and nonobese defined as BMI < 30 (n = 885). Euglycemia was maintained using an automated protocol using an adapting multiplier, which we used as our marker of stress insulin resistance. The primary outcome was hospital mortality. One thousand, two hundred and twenty-three patients met criteria with 73,225 glucose values. The OB group required more insulin (4.5 U/hr vs 3.2 U/hr, P < or = 0.01) and had a higher mean multiplier (0.07 vs 0.06, P < 0.01) reflecting insulin resistance. There was no difference in mortality between OB and nonobese (11.6% vs 11.5%, P = 0.96). Logistic regression showed that insulin dose (odds ratio 0.864; 95% confidence interval 0.772-0.967, P = 0.01), and not BMI, was an independent predictor of survival in this population. Obesity is not an independent risk factor for mortality in the surgical critical care population. Insulin resistance and subsequent hyperglycemia are increased in obesity and are independent predictors of mortality.