Meta-analyses of clinical trials: are we getting lemonade from lemons?

Abstract

Meta-analyses guide planning of clinical trials and clinical care, but are subject to all the methodologic problems and potential biases present in the underlying trials. Furthermore, publication bias often contributes to overestimated benefit in meta-analyses of small trials, which are often 'corrected' by subsequent large trials. Meta-analyses are no substitute for large robust trials.

Keywords: anaesthesia; bias; clinical trial methodology; meta-analysis; outcomes; postoperative pulmonary complications; rigour.