J Clin Epidemiol Vol. 42,No. 10, pp. 1021-1024, 1989 Printed in Great Britain. All rights reserved 08954356/89S3.00+0.00 Copyright 0 1989 Pergamon Press plc Second Thoughts META-ANALYSIS IN MEDICAL RESEARCH: STRONG ENCOURAGEMENT FOR HIGHER QUALITY IN INDIVIDUAL RESEARCH EFFORTS KEITH O’ROURKE and ALLAN S. DETSKY* Clinical Epidemiology Unit of the Toronto Hospital and Departments of Health Administration, Statistics, and Medicine, University of Toronto, Toronto, Ontario, Canada M5S 1Al (Received in revised form 24 February 1989) Abstract-In this article, we counter some criticism regarding the desirability of performing meta-analysis in clinical research. These criticisms, we argue, are based mainly on current difficulties in deriving firm conclusions based on meta-analysis, resulting from poor methodology and reporting of primary studies. This is not a fault of meta-analysis. In fact, with a better understanding of meta-analysis in the context of the full scientific research process, meta-analysis is seen as a key element for improving individual research efforts and their reporting in the literature. This in turn will further enhance the role of meta-analysis in helping clinicians and policy makers answer clinical questions. Philosophy of science Overviews Combining studies Various authors have set out their opinions and offered guidelines for conducting meta-analyses in clinical research [l-3]. The general view seems to be that in attempting to answer questions of interest, all past randomized clinical trials of suitable scientific quality that addressed essen- tially the same question must be considered. Peer reviewed publication is usually taken as evidence of suitable scientific quality and it is accepted that conclusions must only follow from a suitable explicit analysis of all of these trials. On the other hand, the desirability of per- forming meta-analysis in clinical research has been disputed in the literature [4, 51.We feel that the opposition to meta-analytic techniques may simply reflect a less than optimal view of the full scientific research process. Meta-analysis is not only desirable, but in our opinion a necessary *Dr Detsky is supported by a National Health Research Scholar Award from Health and Welfare Canada (6606- 2849-48). requirement of being scientific and objective. It is desirable in that meta-analysis provides mech- anisms for improving the utilization and quality of individual research efforts by identifying cur- rent inadequacies and encouraging their resolu- tion. It is necessary in attempting to meet the scientific ideal of arriving at conclusions (and measures of uncertainty in these) only after explicit consideration and analysis of all rele- vant past and present research endeavors. From the statistical point of view, there really is no escape from performing a de facto meta- analysis. One can either judge the effectiveness of a therapy based solely on the most recent study and ignore all previous studies, a method which is equivalent to giving the most recent study weight 1 .O and all previous studies weight 0, or try to choose the weights on some scientific basis-giving 0 only to studies that are so unrelated or conducted so poorly that no one would pay heed to them. If important differ- ences in study findings exist they must be identified and explained. Thus meta-analysis 1021