Contributions of meta-analyses based on individual patient data to therapeutic progress in colorectal cancer

Abstract

Meta-analysis is the statistical process of combining information from several studies addressing the same question. Meta-analyses based on individual patient data are far more reliable and informative than those based on summary statistics obtained from the trialists or extracted from the published literature. Meta-analysis of randomized clinical trials may contribute to therapeutic progress through (1) establishing efficacy benefits beyond a reasonable doubt, (2) identifying sources of heterogeneity between trials, (3) studying subsets reliably, (4) confirming differences in toxicity profiles, (5) evaluating the cost-effectiveness of experimental therapies, (6) assessing surrogate endpoints, and (7) addressing ancillary questions. All of these potential contributions are illustrated with examples in early and advanced colorectal cancer.