Demystifying fixed and random effects meta-analysis

Abstract

Selection between fixed or random effects should be based on the clinical relevance of the assumptions that characterise each approach. Researchers should consider the implications of the analysis model in the interpretation of the findings and use prediction intervals in the random effects meta-analysis.