Strategy for Combining Information from Real World Data Sources When Individual Patient Data Are Not Simultaneously Accessible

Abstract

Some research on the population in a European country, referred to as the target population, may not provide a sufficient sample size and require pooling multiple national data sources. The researcher can access multiple national data sources one at a time but cannot access them simultaneously at an individual patient level, because of regulation requirements on data sharing. We propose a strategy that enables the combination of information from multiple data sources when IPD are not simultaneously accessible to researchers. This strategy starts with a protocol for extracting information consistently from all data sources. It uses the matching adjusted indirect comparison (MAIC) approach to derive the weights to align the means of the identified covariates of each data source to those of a target study population. The final analysis will be a meta-analysis on the weight adjusted estimates of treatment effect from individual data sources.