Meta-analyses of molecular association studies: methodologic lessons for genetic epidemiology.

Meta-analyses of population-based molecular association studies have become increasingly common over the last 10 years, but little attention has been paid to methodology. In addition to the traditional considerations pertinent to any meta-analysis, there are genetic issues particular to molecular association studies: checking Hardy-Weinberg equilibrium, handling data from more than two groups while avoiding multiple comparisons, and pooling data in a way that is sensitive to genetic models. We systematically reviewed all meta-analyses of molecular association studies identified via MEDLINE. Of a total of 37 studies, eight (22%) described the search terms. Nineteen (51%) did not state inclusion or exclusion criteria. Heterogeneity was assessed in 28 (76%), but only 7 of 37 (19%) studies checked for publication bias. Nine (24%) studies assessed the goodness-of-fit of Hardy-Weinberg equilibrium, and eight (22%) gave any biological rationale to justify the choice of genetic model used for pooling. There is a need for greater communication between epidemiologists and geneticists to develop methods appropriate to this area.