薬理と治療

Abstract

Meta-analysis has been one of important research methods of modern evidence-based medicine. However, exploring critical reviews in this field, many studies have pointed out its weakness of scientific validity and reproducibility. For example, LeLorier et al. （N Engl J Med 1997; 337: 536-42） systematically reviewed discordances of results among meta-analyses and their following large randomized trials, and reported that many discordant results have been found even in papers published in the leading medical journals（BMJ, JAMA, Lancet, New England Journal of Medicine）. Following these critical discussions, Feinstein（J Clin Epidemiol 1995; 48: 71-9） noted meta-analysis is “statistical alchemy for the 21st century.” In this article, we briefly review these critical discussions and how readers should interpret the evidences obtained by meta-analyses. Also, we review the developments of methodologies and systems for preventing the misleading evidences.