Article inclusion

We reviewed all articles in the Journal of the American Medical Association (JAMA) and Lancet from years 2003-2017 and the New England Journal of Medicine (NEJM) from years 2011-2017 (7036 total articles for all three journals combined). After removing studies that did not report on a medical practice or a randomized controlled trial (n=4019), we were left with 3017 articles for review and coding. We coded the practices in the study as either novel or established and then we coded the results of the study as negative, positive, or inconclusive. After further removing studies that reported on a novel practice (n=1373) and studies that had inconclusive or positive results (1229), we were left with415 studies that were tentative reversals. For each of the 415 tentative reversals, we searched to see if a systematic review had been done on the same topic as the tentative reversal. If there was a systematic review done on the topic and the results of the systematic review contradicted the results of the tentative reversal study, we further excluded the study (n=19). This left us with 396 studies reporting on medical reversals, which were either confirmed by results from a systematic review (n=209) or did not have a systematic review done on the topic of the reversal (n=187).

Category