2DkcTXceO

34Meta-analyses: Heterogeneity can be agood thingNan M. LairdDepartment of BiostatisticsHarvard School of Public Health, Boston, MAMeta-analysis seeks to summarize the results of a number of different studieson a common topic. It is widely used to address important and dispirit problemsin public health and medicine. Heterogeneity in the results of differentstudies is common. Sometimes perceived heterogeneity is a motivation for theuse of meta-analysis in order to understand and reconcile differences. In othercases the presence of heterogeneity is regarded as a reason not to summarizeresults. An important role for meta-analysis is the determination of designand analysis factors that influence the outcome of studies. Here I review someof the controversies surrounding the use of meta-analysis in public health andmy own experience in the field.34.1 IntroductionMeta-analysis has become a household word in many scientific disciplines.The uses of meta-analysis vary considerably. It can be used to increase power,especially for secondary endpoints or when dealing with small effects, to reconciledifferences in multiple studies, to make inferences about a very particulartreatment or intervention, to address more general issues, such as what isthe magnitude of the placebo effect or to ask what design factors influencethe outcome of research? In some cases, a meta-analysis indicates substantialheterogeneity in the outcomes of different studies.With my colleague, Rebecca DerSimonian, I wrote several articles on metaanalysisin the early 1980s, presenting a method for dealing with heterogeneity.In this paper, I provide the motivation for this work, advantages and difficultieswith the method, and discuss current trends in handling heterogeneity.381