A Model for Predicting the Resolution of Type 2 Diabetes in Severely Obese Subjects Following Roux-en Y Gastric Bypass Surgery
- 519 Downloads
Severely obese type 2 diabetics who undergo Roux-en Y gastric bypass surgery have significant improvements in glycaemic control. Little work has been undertaken to establish the independent predictors of such resolution or to develop a predictive model. The aim of this study was to develop a mathematical model and establish independent predictors for the resolution of diabetes.
A consecutive sample of 130 severely obese type 2 diabetics who underwent gastric bypass surgery for weight loss from November 1997 to May 2007 with prospective pre-operative documentation of biochemical and clinical measurements was followed up over 12 months. Logistic discrimination analysis was undertaken to identify those variables with independent predictive value and to develop a predictive model for resolution of type 2 diabetes. Consecutive samples of 130 patients with body mass index (BMI) ≥ 35 with type 2 diabetes were selected. One hundred and twenty-seven patients completed the study with a sufficient data set. Patients were deemed unresolved if (1) diabetic medication was still required after surgery; (2) if fasting plasma glucose (FPG) remained >7 mmol/L; or (3) HbA1c remained >7%.
Resolution of diabetes was seen in 84%, while diabetes remained but was improved in 16% of patients. Resolution was rapid and sustained with 74% of those on medication before surgery being able to discontinue this by the time of discharge 6 days following surgery. Five pre-operative variables were found to have independent predictive value for resolution of diabetes, including BMI, HbA1c, FPG, hypertension and requirement for insulin. Two models have been proposed for prediction of diabetes resolution, each with 86% correct classification in this cohort of patients.
Type 2 diabetes resolves in a very high percentage of patients undergoing gastric bypass surgery for severe obesity. The key predictive variables include pre-operative BMI, HbA1c, FPG, the presence of hypertension and diabetic status.
KeywordsMathematical model Diabetes resolution Gastric bypass surgery
Conflict of Interest
The authors declare that they have no conflict of interest.
- 10.Witten IH, Frank E, editors. Data Mining: Practical Machine Learning Tools and Techniques. San Francisco, Ca, Morgan Kaufmann; 2005.Google Scholar
- 11.Cohen WW. Fast effective rule induction. Proceedings of the Twelfth International Conference of Machine Learning, Tahoe City, CA, Morgan Kaufmann; 1995.Google Scholar
- 13.Barr EL, Zimmet PZ, Welborn TA, et al. Risk of cardiovascular and all-cause mortality in individuals with diabetes mellitus, impaired fasting glucose, and impaired glucose tolerance: the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Circulation. 2007;116(2):151–7.PubMedCrossRefGoogle Scholar
- 14.Standards of medical care in diabetes—2008. Diabetes Care. Jan 2008; 31 Suppl 1:S12–54.Google Scholar