Strengths and Limitations of Systematic Reviews of Drug Trials: The Biostatistical Perspective

Introduction

The number of clinical trials that are published each year is increasing, and more than 25 000 randomized controlled trials (RCTs) were already published in 2008. Concurrently, clinicians are pinched for time to study the literature. It is important to resolve this discrepancy between the increasing knowledge and the limited time so that clinicians can keep up to date. It therefore seems to be natural to assess all existing evidence on the same medical problem so that a reliable summary of evidence is provided for answering the research question. This summary is termed systematic review. It describes the use of a structured procedure, and it should be described in a study protocol. Part of the systematic review is a meta-analysis. The meta-analysis is the statistical method for combining the results of individual studies to obtain the overall evidence.

Abstract

Some thousands of randomized controlled trials are published each year. But clinicians are pinched for time to study the literature. Hence, systematic reviews are important to combine the results for a specific medical question. Meta-analysis is only a statistical method that was used in systematic reviews to combine the results of individual studies. Our aim is to discuss the strengths and potential pitfalls of systematic reviews on the basis of five examples from the literature. We focus on aspects limiting the strength of systematic reviews, such as publication bias, imputation of missing data, and the use of inappropriate statistical methods. Although systematic reviews are key to summarizing evidence, meta-analyses of drug trials should be read with a healthy portion of scepticism.

Keywords

meta-analysis, systematic review, data interpretation, clinical trial