Assessment of maximum weight change and duration of therapeutic effect for non-surgical treatment of obesity using an exponential model.

Abstract

Efficacy of weight loss and maintenance therapies in obesity is difficult to quantify due to continuous weight changes over time. We assessed a single exponential model of weight changes during selected non-surgical therapies of non-diabetic obese subjects. We analyzed published mean weight data from 6 studies of ≥12 weeks duration, with comparable treatment groups, and ≥4 weight measurements during very low carbohydrate or fat diets, or treatment with Lorcaserin, Sibutramine or Orlistat. We fit data to a single exponential model to estimate maximum predicted weight loss or regain and duration of weight loss or regain for each therapy. A single exponential is the appropriate model as determined by Kolmogorov-Smirnov, constant variance, and Durbin-Watson tests. Validity of parameter estimates was indicated by coefficients of variation <25%. Sensitivity analysis showed that weight regain at the end of the weight loss phase affected parameter estimates in some instances, with variations of weight loss of 0.2-0.7% of basal. Estimated weight loss and regain were similar to observed weight changes in all studies. The model could also be used to assess dose-response relationships. Estimates from the model were used to compare concurrent obesity regimens using 95% confidence intervals, taking into account pre-determined minimal clinically important differences. This exponential model may provide accurate estimates of maximum achievable weight loss or regain and optimal duration of efficacy for a variety of non-surgical weight loss and maintenance regimens from published mean weight data and may be useful to more accurately evaluate weight loss and maintenance regimens.