Type 2 Diabetes Remission After Gastric Bypass: What Is the Best Prediction Tool for Clinicians?

Obes Surg. 2015 Jul;25(7):1128-32. doi: 10.1007/s11695-014-1511-8.


Background: Statistical models and scores have been recently suggested to predict remission of type 2 diabetes after bypass surgery, but their relevance in routine clinical practice still needs evaluation. Our objective was to assess these methods on a French cohort and to compare them with other easy-to-use models.

Methods: We investigated a cohort of 84 diabetic obese subjects who underwent Roux-en-Y gastric bypass surgery. Diabetes remission 1 year after surgery was defined based on the American Diabetes Association criteria. We tested six methods from the literature and four other models to predict remission of diabetes after bypass surgery using pre-operative bioclinical parameters. Predictive methods for diabetes remission were assessed using cross-validation error rates when appropriate.

Results: Sixty percent of the subjects had diabetes remission. Models from the literature had high error rates in our cohort (from 22.6 to 40.5 %), while published simple scoring systems behaved much better (15.9 and 16.7 %). Using other apprehensible models learned on our cohort did not improve the prediction error (from 17.2 to 19.9 %).

Conclusions: We showed that the scoring system DiaRem is easy to use and provides the best prediction error (15.9 %) compared to other methods. We additionally propose a DiaRem score threshold of ≤6 for likely remission of a subject 1 year after surgery, which may be considered in clinical decision-making.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Cohort Studies
  • Diabetes Mellitus, Type 2 / diagnosis
  • Diabetes Mellitus, Type 2 / surgery*
  • Female
  • Gastric Bypass* / methods
  • Humans
  • Male
  • Middle Aged
  • Obesity, Morbid / diagnosis*
  • Obesity, Morbid / surgery*
  • Predictive Value of Tests
  • Prognosis
  • Remission Induction
  • Research Design