Commentary. Oligognostic mega-analysis. Is Archie turning in his grave?

The Cochrane Collaboration has developed guidelines for the conduct of meta-analyses and systematic reviews. When applied to well-conducted short-duration randomized controlled trials, meta-analysis can be informative, but seldom in prolonged trials, because of "unblinding", non-adherence, losses to follow-up, and "crossovers". The meta-analysis of observational studies is beset by problems of bias, confounding, variable methodology, and lack of transparency, and is seldom valid. For the evaluation of partly quantitative or non-quantitative studies, systematic reviews using standardized methods have been developed. However, such reviews are inferior to well-conducted qualitative reviews tailored to the specific topics at issue.