The impact of parametric assumptions on the bias-variance tradeoff for causal analysis of time-to-event outcomes.

Challenge

The choice of parametric assumptions in survival analysis models for time-to-event outcomes involves a fundamental bias-variance trade-off, but the practical implications of this trade-off for causal analyses of real-world treatment effects had not been systematically examined.

Solution

Target RWE researchers presented a systematic examination of how parametric model specification choices affect the bias-variance trade-off in causal time-to-event analyses, characterizing the conditions under which flexible versus restrictive models are preferable.