• PDF / 912,166 Bytes
• 12 Pages / 504 x 719.759 pts Page_size
• 73 Downloads / 277 Views

DOWNLOAD

REPORT

---

0092-8615/2001 Copyright 0 2001 Drug Information Association Inc.

WHY ARE SOME META-ANALYSES MORE CREDIBLE THAN OTHERS?* ARMINKOCHAND JOACHIMROHMEL Bundesinstitut fur Arzneimittel und Medizinprodukte, Berlin, Germany

Meta-analysis of randomized clinical trials plays an important role in summarizing all available evidence with respect to the comparison oJ for example, two drugs for the same indication. Presently, meta-analysis is considered to be of only minor importance in the new drug application. This fact is astonishing: in this situation, a summary evahation of all available evidence from many-although, in general, limited-independent clinical trials is necessary. The main reason, in our opinion, that meta-analysis is considered to be of minor importance is that presented meta-analyses often are not completely convincing because objectives are not appropriately chosen and study conduct orpresentation are not suflciently detailed to enable the reader to assess the evidence. This paper is intended to clarify why some meta-analyses have more credibility than others and to provide some guidance on increasing the credibility of meta-analyses. We believe that the ideas presented apply beyond the regulatory setting. Key Words: Clinical trials; Meta-analysis; International Conference on Harmonization E9; Drug regulation

INTRODUCTION

each year is still increasing. However, only in rare cases has the discussion of the approMETA-ANALYSIS HAS been defined to be priateness of biostatistical methodology in a quantitative and systematic summary of a medical research been so intensive as with collection of separate studies for the purpose meta-analysis. From the very beginning of obtaining information that cannot be demeta-analysis has split the community into rived from any of the studies alone (1). This clear proponents and those who completely definition implies that meta-analysis is also dislike this type of analysis. Feinstein named a technique that should lead to reproducible meta-analysis a synonym for “statistical alresults and that can be distinguished from the chemy for the 21st century” (2). Others exclassical review or overview, where results pressed their doubts about the credibility of from various studies are collected and qualitaresults “proven” by means of meta-analysis. tively weighted by an expert in the field. It has been repeatedly emphasized that cliniOriginally invented in the social sciences, cal trials should be designed to stand on their meta-analysis has found widespread use in own and that in consequence, meta-analyses clinical research during the last two decades should not be necessary (“If a treatment has and the number of meta-analyses published an effect so recondite and obscure as to require meta-analysis to establish it I would not be happy to have it used on me.” [3]). Reprint address: Armin Koch, Bundesinstitut fiir ArIn addition, empirical comparisons of the rezenimittel und Medizinprodukte, SeestraBe 10. 13353, sults from meta-analyses with results from Berlin, Germany. large randomiz