抄録
In confirmatory clinical trials, missing data may lead to a loss of power and accuracy, and represent a potential source of bias for the estimates of treatment effect (ICH-E9). Recently, the EMA guidelines (2010) and the NAS report in the U.S. (2010) have been published, and missing data has become one of the topics of high interest in biostatistics. In particular, clinical trials in the central neurological disease area (pain and psychiatric disease) have the following characteristics: missing data due to treatment withdrawal or dropout are more frequent than in other disease areas, and the reasons for discontinuation tend to differ between treatment groups. In this paper, we describe the considerations for the handling of missing data, and present actual cases in the central neurological disease area, in which the handling of missing data was discussed in new drug application (NDA) review and consultation at the Pharmaceuticals and Medical Devices Agency (PMDA).
