Identification of factors predictive of outcome following Roux-en-Y gastric bypass for obesity

INTRODUCTION: Worldwide obesity levels have doubled since 1980. Bariatric surgery is the only effective long-term treatment however its results remain highly patient, procedure and surgeon dependent. This study aimed to assess the impact of a range of factors on the outcome of Roux-en-Y gastric bypass (RYGBP) surgery. METHOD: All patients eligible for RYGBP at 2 regional centres were approached for inclusion. Data pertaining to 13 factors identified as potentially associated with reduction in excess BMI (eBMI) following RYGBP were collected: genetic predisposition to obesity, ethnicity, reasons for seeking surgery, eating behaviour, physical activity level, quality of life, personality score, motivation to change, alcohol intake, smoking history, social class, working pattern and past medical history. The data were analysed using multivariate linear and logistic regression analysis. The primary outcome was the percentage eBMI loss at 12 months postoperatively. The secondary outcomes were the resolution of diabetes and/ or hypertension. RESULTS: 129 patients were recruited after written informed consent of whom just 60 were eligible for analysis. At 12 months postoperatively their percent eBMI losses ranged between 33.4% and 136.2% (mean 67.3%, SD 18.8%). Of the 13 factors investigated using linear regression none showed a significant correlation with percent eBMI loss. Logistic regression analysis showed that personality score, motivation to change score and smoking status were all significantly associated with eBMI loss of 70% or greater when combined with the other investigated factors (p = 0.013, p = 0.016 and p = 0.027 respectively) but not when analysed individually. 9 out of 12 (75%) and 5 out of 14 (35.7%) patients on medication preoperatively for type 2 diabetes mellitus and hypertension respectively were able to discontinue their medication by 12 months post-RYGBP. CONCLUSIONS: This pilot study shows that a multi-factorial approach to clinical prediction and patient selection is viable and feasible