Your Easy Guide to Clinical Studies

An Estimand defines the treatment effect we want to measure and specifies how Intercurrent Events (ICE) are handled.

 

Example: a study assesses the effect of talking therapy on stress. Participants may decide to take additional medication, such as anxiolytics. Anxiolytics intake is therefore an (ICE) affecting study outcome and interpretation.

 

Researchers may decide to use a:

• Treatment policy Estimand: analyse participants regardless of whether they took additional medication or not. This strategy corresponds to the intention-to-treat (ITT) principle
• Hypothetical Estimand: the treatment effect is estimated under the hypothetical scenario that the ICE did not occur (e.g. in the example above it addresses the question “What would the treatment effect have been if the participant had not taken anxiolytics?”)
• While on treatment Estimand: the treatment effect is estimated before ICE occurrence (e.g. assesses the effect of the therapy up to the initiation of anxiolytics treatment)
• Principal stratum Estimand: estimates the treatment effect in a subgroup of participants who do not experience ICEs (e.g. in participants who would not start anxiolytics)

As a SP-INV, decide which question is most useful and therefore which Estimand strategy should be used in the primary analysis.

Video

• What are Estimands