Anforderung an die Daten für die Target-Trial-Emulation: Eine Diskussion unter Betrachtung von Patientenregistern

Abstract

The article first considers basic requirements (e.g. recording of relevant confounders) and frequent sources of bias (e.g. immortal time bias) when using registry and other real-world data as the basis for comparative effectiveness analyses of treatments. The target trial concept is described and estimands relevant to the benefit assessment are discussed. Furthermore, relevant requirements for the data in terms of practical feasibility, adequate emulation, and internal validity, in particular for the implementation of propensity score-based analyses, are reflected. The feasibility is considered against the background of the German register landscape. Finally, the experiences of projects that have attempted to emulate RCT effects using real-world data are presented.

Keywords

• causal inference
• target trial
• propensity scores
• patient registries
• real-world data
• routine practice data collection