What’s Really Wrong with Medical Research and How to Fix It.

Abstract

Something is fundamentally wrong with the medical literature. It’s not just that authors misunderstand p values. It’s not that studies are under-powered. Misuse of statistics remains common enough but a general uncertainty and mistrust appear to have more global roots. Richard Horton, editor-in-chief of The Lancet points to “studies with small sample sizes, tiny effects, invalid exploratory analyses, and flagrant conflicts of interest.” [1] Horton’s critique remains only one of many [1-4], some of which make the case that half of published material may be in error [1-2]. Traditional criteria, both from common sense and established principles such as Bradford Hill’s criteria [5] are known but disregarded. In my view, blame falls on editors whose failure to recognize controversial subjects allows de facto (or intentional) bias. Alternative points of view to published papers appear in PubPeer, in Letters-to-editors and blogs and social media from people with experience and credentials. The