[International collaborative trial: from the viewpoints of statistics and data management].

ICHE 5 released in 1998 opened the possibility of utilizing foreign clinical trials data for getting approval in Japan and the concept of "bridging" studies have been discussed intensively with less emphasis on statistical aspects. In this presentation, statistical methodology for assessing similarity between a bridging study and a bridged study is reviewed including our new proposals: concept of overlap coefficient for assessing overlap of two distributions and application of propensity-score matching for selecting a similar patient subpopulation from a global database. For accelerating development, international collaborative studies become more preferred approach than bridging as illustrated in the sentence "Best bridging is no bridging" by R. Puzdur; an approach for determining the sample size from Japan is proposed, which keeps the probability of producing an apparent heterogeneity (a reverse direction of treatment effect) low (say, 10-20%). Finally, requirements for success of international collaborative studies including incentives of researchers and data management are discussed.