Beyond estimand specification: Considerations for estimand-aligned estimation

Abstract

The ICH E9(R1) addendum published in 2020 provides a framework for clearly constructing an estimand for use in clinical trials with the objective of increasing transparency, comparability, and “dialogue between disciplines.” In addition to aligning the trial objective with the target of estimation (i.e., estimand), the framework requires direct communication as to how intercurrent events (ICE) and other data complications will be handled. Researchers have since contributed substantially to the discussion around best practices for defining meaningful and specific targets for estimation, yet pragmatic guidance around estimand-aligned estimation strategies is still needed. Such guidance is particularly critical for driving the discovery of novel therapeutics for conditions such as tuberculosis (TB) where the burden of the disease far outweighs the funding available, and ICEs are a common complication during treatment and follow-up. We present the results from a simulation study performed to evaluate the feasibility and robustness of popular estimand-aligned estimation strategies under varying ICE settings, using TB as a case study.