Integrated Data Analysis for Assessing Treatment Effect through Combining Information from All Sources

ABSTRACT

It is critical to use a precise estimate of treatment effect when drawing conclusions, evaluating benefit/risk, or designing a new study. Using data from all sources in an integrated data analysis/meta-analysis will help us move closer to meeting this need. Depending on the data sources and objectives, there are many approaches for integrated analyses. These include network meta-analysis, multivariate meta-analysis, model-based meta-analysis as well as methods of borrowing historical data. In this article, we discuss these methods with details for implementation and interpretation. In addition, we consider information adaptive repeated cumulative meta-analyses. We also discuss how to apply three approaches that take into account the variability of the overall treatment effect estimate obtained through integrated analysis to determine sample size for a new trial. Some computation and simulation results are provided.

KEYWORDS:

• Historical data
• Meta-analysis
• Sample size calculation
• Surrogate endpoint

Acknowledgments

The authors are grateful for helpful discussions with the other members of the EFSPI (European Federation of Statisticians in the Pharmaceutical Industry). Special Interest Group on Integrated Data Analysis: Georgina Bermann, Tim Friede, Frank Bretz, Richard Nixon, and David Ohlssen. The authors also thank an anonymous reviewer and associate editor for valuable comments.