Current Comments ” EUGENE

Mew-analyses and their proper place in medical literature are explained along with how they differ fromtraditional literature reviews. Both the qualitative and quantitative amccts of meta-analvsis are consid-ered in relation to studying the results of clinical tri This issue [1 February 1991] includes ameta-analysis of the use of steroids in treat-ing chronic obstructive pulmonary disease. 1Mets-analyses such as this one commandour atiention, both because they purport toprovide a “definitive” answer to a clinicalquestion that has eluded other reserwchersand because the work of those researchers,the grist for the meta-analytic mill, is oftensubject to embarrassingly intense scrutiny.They represent a new class of article, onethat straddles the maditionsd boundmies be-tween original research and review articles.Because meta-analysis is a relatively newand partially technical method, many physi-cians find themselves unable to appreciate the nuances or limitations of meta-analysesin the same way that they can appreciate those of a traditional review or original re-search article. Mets-anal yses therefore aresometimes distrusted by physicians who donot understand the statistical techniques andresented by researchers who do not like see-ing their years of effofi rapidly reduced to an“effect size,” with the consequent elevationof the meta-analyst to an expert in the field.These tensions have provoked contro-versyz-q as well as attention from the lay press.5The term “rneta-analysis” was coined in1976 by G1SSS6in the psychology literature,Steven N. Goochart and texts on the subject appeared in the1980s.7.8 The field is rapidly evolving, withthe annual number of such analyses in thegeneral medical literature increasing expo-nentially; the meta-meta-analysis, evaluating the quality of meta-analyses themselves,also has been develo’@.g-l IWhat are meta-analyses, and what is theirproper place in medical research? This qtsestion is best answered bv examining the wavs