Table 1.

Overview of estimands

• Defining an estimand entails defining five attributes: treatment condition(s), population, endpoint (outcome measure), population-level summary measure, and strategy for handling ICEs • ICH E9 (R1) addendum suggests five potential strategies to address ICEs: 1. Treatment policy strategy: occurrence of the ICE is considered irrelevant, and interest lies in the treatment effect regardless of whether the ICE occurred. Note that this approach cannot be used for truncating events, e.g. where a variable cannot be measured due to death 2. Hypothetical strategy: treatment effect under a scenario where the intercurrent event(s) would not occur 3. Composite strategy: ICE is incorporated into the outcome definition, for instance, patients who experience the ICE are classed as a non-responders 4. While-on-treatment strategy: response to treatment before occurrence of intercurrent event(s) is of interest 5. Principal stratum: treatment effect in the “principal stratum” of interest (e.g. the population of patients who would not experience the ICE under either treatment assignment)

ICH International Conference on Harmonisation, ICE intercurrent event