Objective: To compare diabetes remission after bariatric surgery in subjects with body mass index (BMI) of 35 kg/m2 or more or BMI of less than 35 kg/m to determine which predictors are best.
Background: BMI is currently the only selection criterion for bariatric surgery in diabetic subjects. Many studies have challenged BMI for predicting diabetes remission.
Methods: Data sources were PubMed, Cochrane Library, and EMBASE databases from January 1980 to June 2013. The selected studies were randomized controlled trials, controlled clinical trials, or cohort studies with 10 or more patients per arm. Of 1437 screened articles, 94 studies were included with 94,579 patients undergoing surgical procedures (4944 with type 2 diabetes mellitus). Weight, BMI, glycated hemoglobin A1c, fasting glucose, and insulin were abstracted by 2 independent reviewers. The effect size was the percent diabetes remission.
Results: Meta-analysis was performed for BMI less than 35 kg/m2 (group 1) and BMI 35 kg/m2 or more (group 2). Diabetes remission was 72% [95% confidence interval (CI), 65-80] in group 1 and 71% (95% CI, 65-77) in group 2. Diabetes resolution was 89% (95% CI, 83-94) after biliopancreatic diversion, 77% (95% CI, 72-82) after Roux-en-Y bypass, 62% (95% CI, 46-79) after gastric banding, and 60% (95% CI, 51-70) after sleeve gastrectomy. The only significant predictor of glycated hemoglobin A1c reduction was waist circumference, lower baseline waist associating with higher reduction.
Conclusions: Bariatric surgery determines similar diabetes remission rates in patients with BMI of 35 kg/m2 or more or BMI of less than 35 kg/m2. Baseline BMI is unrelated to diabetes remission. The association of baseline waist circumference with glycated hemoglobin A1c reduction is likely due to selection bias. Bariatric or metabolic effects of the surgical procedures appear independent, and different indices are needed to predict them.