Abstract

Evidence-based medicine is often seen as something dry, formal, and statistical, often used to justify a proscriptive approach to medicine. A more attractive approach is to use our understanding of those aspects of studies that can mislead us to identify the evidence we can trust. Evidence can, and probably should, be based on patient-centred outcomes of importance to clinical practice. The particular issues differ somewhat between clinical trials, observational studies, adverse events, diagnosis, and health economics. Here we explore some of important criteria relating to evidence from randomised trials, either alone or in meta-analyses.