ICH E9 (R1) addendum on estimands and sensitivity analysis in clinical trials to the guideline on statistical principles for clinical trials

ICH E9 (R1), adopted by the ICH Assembly at Step 4 in November 2019 and implemented in Europe in 2020, supplements the original ICH E9 guideline “Statistical Principles for Clinical Trials”. The addendum responds to the need for clearer description of treatment effects in the presence of intercurrent events, so that clinical trial results properly inform decisions made by patients, prescribers and regulators. It is issued as a 19‑page harmonised guideline (EMA/CHMP/ICH/436221/2017) and should be read together with ICH E9.

The document is provided as a single PDF via the European Medicines Agency website, with sections covering introduction and scope, the estimand framework, strategies for addressing intercurrent events, trial design and conduct implications, sensitivity analysis and its role, and annexed examples. The introduction explains that the principles apply whenever a treatment effect is estimated or a hypothesis relating to a treatment effect is tested, for both efficacy and safety endpoints, and for randomised and single‑arm clinical trials. The core of the addendum defines an estimand through four attributes: the treatment condition(s) of interest, the target population, the variable (endpoint) and the summary measure, together with a description of how intercurrent events are handled.

A key contribution is the classification of strategies for handling intercurrent events, such as treatment policy, hypothetical, composite, while‑on‑treatment and principal stratum strategies, with guidance on when each may be appropriate. The addendum stresses that the choice of estimand should guide data collection, trial design (including visit schedule and handling of treatment changes) and the choice of estimator (analysis method). It also requires that clinical trial protocols and statistical analysis plans explicitly specify the main estimands, the primary estimator for each estimand and one or more pre‑planned sensitivity analyses targeting the same estimand to explore robustness to deviations from assumptions and data limitations.

Sensitivity analysis is defined and distinguished from supplementary analysis, with the addendum explaining that sensitivity analyses should investigate the impact of alternative plausible assumptions (for example about missing data mechanisms or the behaviour of patients after intercurrent events) on the estimated treatment effect. Examples and figures illustrate how to align study objectives, estimands, estimators and sensitivity analyses, and how to document this alignment in trial protocols and reports. The addendum highlights that, when estimands are clearly defined and analyses are pre‑specified to a level that allows replication, regulatory interest can focus on assessing sensitivity to assumptions rather than on reconstructing the target of inference.

For investigator‑initiated clinical trials, ICH E9 (R1) is particularly useful when designing protocols with complex intercurrent events (such as treatment switching, rescue medication or varying adherence) and when preparing statistical analysis plans and clinical study reports.