Applying different methods to the Avandia meta-analysis yields uncertainty

This paper is old news because of the media splash that resulted when it was released ahead of print, but it’s in the current print issue of Annals of Internal Medicine. Diamond et.al. point out one of the hazards of meta-analysis: conclusions may be sensitive to methodologic choices made by the authors. Based on their use of alternative methods of analysis: “We conclude that the risk for myocardial infarction and death from cardiovascular disease for diabetic patients taking rosiglitazone is uncertain: Neither increased nor decreased risk is established.”