estimand

Estimand is a precise description of the treatment effect that a clinical study aims to estimate. It defines, for a specified target population, the endpoint of interest, the treatment condition or strategy, the time frame, and how intercurrent events—events occurring after treatment initiation that can affect interpretation—are handled. The estimand specifies what is to be estimated and how the result should be interpreted, forming a direct link between study objectives, design, data collection, and analysis. The approach was formalized in the ICH E9(R1) addendum to improve transparency and comparability across trials.

Five attributes are typically described: target population, endpoint (outcome), treatment condition or strategy, intercurrent events handling strategy, and the population-level summary measure (for example, risk difference, risk ratio, or hazard ratio). Intercurrent events and their handling choices influence both design and analysis.

The estimand concept is complemented by the estimator, the statistical quantity calculated from trial data that aims to estimate the estimand. Different estimators may target different estimands, so alignment among objectives, data collection, and analysis is essential.

Examples: a treatment policy estimand for a hypertension trial would measure the effect of assigned treatment regardless of adherence or discontinuation over a defined period. A hypothetical estimand might ask what the effect would be if all participants adhered to the assigned regimen. A principal-stratum estimand could focus on patients who would comply under both treatments.

In use, the estimand framework supports clearer interpretation, regulatory communication, and cross-trial comparability.