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.

Combining individual participant and aggregated data in a meta‐analysis with correlational studies

“…Our small example illustrates that even with the addition of two large scale national studies, we find an important within‐study interaction that is not evident in the AD meta‐analysis. These techniques will also prove invaluable in understanding the effectiveness of interventions at the individual level, just as they have contributed to the medical literature (Cardwell et al ., ; Schmid et al ., ; Buyse, ; Angelillo and Villari, ).…”

“…Despite professional standards that urge researchers to share their data after a specified time, there are few reports of social science researchers actually analyzing data collected by other colleagues. Contrast this with medicine where many researchers using meta‐analysis have access to the individual level data gathered in a particular study (Angelillo and Villari, ; Buyse, ; Cardwell et al ., ). As medical researchers (Berlin et al ., ; Stewart and Tierney, ) have argued, the use of individual patient or participant level data in a meta‐analysis strengthens the inferences possible from a meta‐analysis, allowing the examination of relationships among variables within studies and providing more information to explore the possible confounding of variables that occurs across studies.…”

The Value of Pharmaceutical Industry-Sponsored Patient Registries in Oncology Clinical Research

“…These disease settings may necessitate the approval of drugs based on relatively small studies, single-arm trials, or meta-analyses. 5 In addition, registrational clinical trials conducted in a specific geographic region or country may have limited applicability in broader clinical practice due to differences in patient demographics, clinical practice patterns, and resource availability. 6 Clinical trials are purposely designed to include more homogeneous populations and must adhere to strict study protocols to examine the causal effects of particular clinical interventions.…”

Meta-analysis of randomized clinical trials in the era of individual patient data sharing

“…In 2009, a special issue of the International Journal of Clinical Oncology reviewed the implementation and limitations of the meta-analysis of randomized controlled trials (RCTs). This special issue included a general discussion of the role of meta-analysis [ 1 ], implementation of the tabulated-data meta-analysis [ 2 ], additional contributions of individual patient data (IPD) meta-analysis [ 3 ], and the development of the statistical methods for performing IPD meta-analysis [ 4 ]. In these articles the benefits of IPD meta-analysis were discussed, and IPD meta-analysis was presented as a gold standard for conducting a quantitative review of evidence arising from randomized clinical trials.…”