Mistaken inference caused by reliance on and misinterpretation of a significance test

Abstract

The high prevalence of both non-aspirin NSAID use and atrial fibrillation makes an association between these two variables of considerable clinical and public-health importance. Previously, only two studies had examined the effect of selective COX-2 inhibitors on the risk of atrial fibrillation [1,2]. Therefore we read with great interest the study by Chao et al. [3], which concluded “new NSAID use may predispose patients to atrial fibrillation but use of selective COX-2 inhibitors was not significantly related to atrial fibrillation occurrence, except in patients with chronic kidney or pulmonary disease”.