An estimand is a precise description of the target measure that is to be statistically estimated in a clinical study. It links the clinical question with the analysis strategy and specifies what exactly is interpreted as a treatment effect, considering relevant events during the study (intercurrent events). Thus, the estimand framework creates transparency and comparability, particularly in cases of treatment discontinuation, rescue medication, or protocol deviations.
Components of an Estimand according to ICH E9(R1)
The Estimand Framework (ICH E9(R1)) defines five core elements that collectively form the estimand. First, the population (e.g., all randomized patients with a defined disease). Second, the treatment(s) being compared. Third, the variable, i.e., the specific endpoint (e.g., change in a score or time to event). Fourth, the intercurrent events, i.e., events after randomization that can influence the interpretation (e.g., treatment discontinuation, change in concomitant therapy). Fifth, the summary measure, i.e., how the effect is expressed (e.g., mean difference, hazard ratio).
The central added value: The estimand compels explicit formulation of the clinical intention and alignment of the analysis with it. This prevents analysis decisions from being made only after data review. In many projects, the estimand is closely linked to the definition of analysis populations such as intention-to-treat analysis or per-protocol analysis, but differs in content: analysis populations describe who is included in the analysis, while the estimand describes which effect is to be estimated.
Strategies for Handling Intercurrent Events
Various strategies exist for intercurrent events, which should be defined in the clinical study protocol and statistical analysis plan. Common options include:
- Treatment-policy strategy: The effect is considered irrespective of the occurrence of the intercurrent event, e.g., including rescue medication.
- Hypothetical strategy: It is estimated what the effect would have been if the event had not occurred (requires assumptions and often model-based methods).
- Composite strategy: The event is integrated into the endpoint, e.g., “treatment failure” also includes discontinuation or additional therapy.
- While-on-treatment strategy: Consideration up to the event, e.g., measurements only as long as the treatment is taken.
- Principal-stratum strategy: Effect in a subpopulation characterized by the event; methodologically demanding.
A practical pitfall is the inconsistent combination of estimand definition and missing data strategy. For example, if a hypothetical strategy is chosen, imputation or model assumptions must be consistent and tested in sensitivity analyses. If these connections are not properly documented, discussions with authorities and potentially additional requirements in the approval process are at risk.
Operationalization in SAP, Data Management, and Reporting
Implementing an estimand requires clear operational processes: intercurrent events must be defined, collected, and coded traceably. This includes, among other things, the Electronic Case Report Form, query management, data management plan, and audit trail. In many studies, close coordination between the clinical team, biostatistics, and medical writing is also necessary to ensure consistency of endpoint definition, analysis sets, and report text.
The Clinical Study Report should transparently present which estimand is primary, which sensitivity estimands are used, and how results are to be interpreted. This improves traceability for ethics committees, authorities, and subsequent meta-analyses. A clear presentation is also helpful for the Steering Committee and Data Monitoring Committee, as study decisions are often based on interim analyses and endpoint results.
Relevance for clinical trials
The estimand framework has professionalized the discussion about “What is the treatment effect?” in many therapeutic areas. It supports the planning of studies with realistic conditions where treatment discontinuations and concomitant medication occur. At the same time, it increases the demands on planning and documentation: without clear definitions, a later amendment may be necessary, which costs time and resources.
From a CRO perspective, it is advisable to define estimands early to establish a robust statistical strategy and derive data requirements. This allows sponsors to communicate study objectives more clearly, and the analysis becomes less susceptible to interpretation disputes. Especially in adaptive designs or with complex endpoints (e.g., composite endpoints), the estimand logic is an important quality feature.
Frequently Asked Questions (FAQ)
Is an estimand the same as an endpoint?
No. The endpoint is the measured variable, while the estimand additionally specifies how events after randomization are incorporated into the interpretation and how the treatment effect is summarized.
Why was the estimand framework introduced?
Because classical approaches (e.g., pure ITT analyses) did not always clearly define which clinical effect was actually being estimated, especially in cases of treatment discontinuation or rescue medication. The framework increases transparency and consistency.
Which documents must describe the estimand?
At a minimum, the clinical study protocol and the statistical analysis plan should define the primary estimand and important sensitivity estimands. The implementation and interpretation of results are documented in the Clinical Study Report.
Regulatory References
- ICH E9(R1) Statistical Principles for Clinical Trials (Addendum on Estimands and Sensitivity Analysis): Framework for defining estimands and sensitivity analyses.
- ICH E6(R3) Good Clinical Practice: Requirements for planning, documentation, and oversight, relevant for SAP and CSR consistency.
- EU Regulation (EU) No. 536/2014 (CTR): Requirements for protocol content and transparency of the analysis strategy in clinical trials.
It is also important to note that the technical infrastructure (device management, data transfer, backup, roles and rights concept) should be tested before the first patient first visit. A data management plan with clear responsibilities, a monitoring strategy (e.g., remote monitoring), and defined quality metrics helps to detect deviations early and initiate CAPA measures. Training materials and documented instructions are also essential to achieve inspection readiness.
It is also important to note that the technical infrastructure (device management, data transfer, backup, roles and rights concept) should be tested before the first patient first visit. A data management plan with clear responsibilities, a monitoring strategy (e.g., remote monitoring), and defined quality metrics helps to detect deviations early and initiate CAPA measures. Training materials and documented instructions are also essential to achieve inspection readiness.