Biostatistics Journal Club: What is an Estimand? How do we Choose a Clinically Meaningful Estimand for a Comparative Clinical Study?
For a typical clinical study comparing two therapies, the investigators identify a target patient population, precisely define the treatments interventions and primary endpoint, then specify a population-level summary to quantify the treatment difference. Collectively, these choices are components of the estimand framework recently set forth by the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH). In this guideline and various related research publications, the issue of choosing the study estimand is discussed. Special attention is needed when the study’s primary outcome could foreseeably be affected by intercurrent events, such as treatment discontinuation, which occur after treatment initiation and interfere with the observation or interpretation of the outcome. When such unavoidable interruptions of the assigned study therapy are anticipated, it is important to consider at the study design stage how they will be handled when quantifying and interpreting the treatment difference. Here, the treatments of interest need to be precisely defined, and appropriate analytic procedure for handling incomplete observations pre-specified.
This journal club will discuss the more fundamental issue of quantifying the treatment difference. An appropriate population-level summary of the treatment difference preferably has the following features:
1. The summary measure for the treatment difference is clinically interpretable, ideally in layperson’s terms, and is accompanied by an appropriate summary of the endpoint in each treatment group;
2. The summary of the treatment difference does not have modeling constraints, and the corresponding inference procedures are robust and model-free.
This talk will use three recent clinical studies to illustrate the selection process for an appropriate estimand.