Hypothetical estimands in clinical trials: a unification of causal inference and missing data methods

TL;DR

This paper unifies causal inference and missing data approaches to estimate hypothetical treatment effects in clinical trials, clarifying assumptions and demonstrating methods through simulations.

Contribution

It establishes the equivalence of certain causal inference and missing data estimators for hypothetical estimands, enhancing methodological understanding.

Findings

Causal and missing data estimators can be identical for hypothetical estimands.

Potential outcome notation clarifies assumptions needed for estimation.

Using post-intercurrent event data can improve estimation accuracy.

Abstract

The ICH E9 addendum introduces the term intercurrent event to refer to events that happen after randomisation and that can either preclude observation of the outcome of interest or affect its interpretation. It proposes five strategies for handling intercurrent events to form an estimand but does not suggest statistical methods for estimation. In this paper we focus on the hypothetical strategy, where the treatment effect is defined under the hypothetical scenario in which the intercurrent event is prevented. For its estimation, we consider causal inference and missing data methods. We establish that certain 'causal inference estimators' are identical to certain 'missing data estimators'. These links may help those familiar with one set of methods but not the other. Moreover, using potential outcome notation allows us to state more clearly the assumptions on which missing data methods rely to estimate hypothetical estimands. This helps to indicate whether estimating a hypothetical estimand is reasonable, and what data should be used in the analysis. We show that hypothetical estimands can be estimated by exploiting data after intercurrent event occurrence, which is typically not used. We also present Monte Carlo simulations that illustrate the implementation and performance of the methods in different settings.