Concerns with the statistical analysis in

To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Naude et al. [1] evaluated whether there were additive effects of prescribing bupropion/naltrexone on top of a glucagon-like peptide 1 (GLP-1) analogue as weight loss therapy. GLP-1 analogues are known to be an effective treatment for weight loss, and the authors stated based on their analysis that the addition of bupropion/naltrexone was associated with greater weight loss, including in patients who were initially non-responsive to GLP-1 analogues. On our initial review of the article, the authors found the description and results of the analysis to be hard to follow. Specifically, our concerns focus on the following: The main conclusions: The article states that both the responder and non-responder groups had greater total body weight loss (TBWL) than the monotherapy group and based this conclusion on the linear regression models presented in Table 3 [1]. However, the regression coefficient for responders was positive (0.038) and the regression coefficient for non-responders was negative (−0.07). This would seem to indicate that the non-responders experienced less weight loss than the monotherapy group. If we have interpreted these data correctly, the conclusions as stated in the paper are invalid.