Qualitative assessment of studies included in a meta-analysis: DES and the risk of pregnancy loss.

Meta-analysis, the qualitative and quantitative integration of available research information, is increasingly used in clinical research. This report examines the qualitative elements of meta-analysis through critical review of two published examples that assessed the effect of diethylstilboestrol (DES) on miscarriage and neonatal mortality, respectively. Our subjective review showed the internal validity of three of five trials aggregated in the former meta-analysis to be severely compromised. The remaining two trials (data from which were aggregated in the latter meta-analysis), while internally valid, studied different clinical populations. Therefore, pooling of their results may have been clinically inappropriate. In conclusion, meta-analysis involves substantial elements of subjective judgment. Qualitative assessment of studies is of pivotal importance, because the validity of any summary estimate of effect hinges on both the methodologic quality and the combinability of the aggregated studies.