EMA Releases draft ICH E9 Addendum on Estimands and Sensitivity Analyses for Public Consultation

In August 2017, the EMA issued an addendum to ICH E9 (Statistical Principles for clinical trials) on estimands and sensitivity analyses to address the issue. 

Confirmatory clinical trials are conducted to quantify the effects of a treatment and to provide evidence of efficacy and safety to support regulatory decision making, however many clinical trial protocols are written in such a way that the trial objectives lack clarity and/or details. Vague objective statements can result in a statistical analysis that addresses a different scientific question to the one intended. Furthermore, the assumptions particularly around missing data are often not clearly stated and often clinically unrealistic. Subsequently clinical trials can report results that are not helpful to clinicians, patients or regulators.

Structured framework and tools for estimation and testing

The addendum presents a structured framework to link trial objectives to a suitable trial design and the tools for estimation and hypothesis testing. The framework primarily focuses on the concept of estimands which translates trial objectives into a precise definition of the treatment effect that is to be estimated. Sensitivity analyses for an estimand are also covered.

As outlined in the addendum, an estimand defines the target of estimation for a particular trial objective (i.e. “what is to be estimated”) through specification of

  • Population
  • Variable
  • Handling of intercurrent events
  • Population-level summary for the variable

Once defined a suitable method of estimation (i.e. the analytic approach, referred to as the main estimator) can then be selected.  The main estimator is typically underpinned by certain assumptions which need to be stated.  To explore the robustness of inferences from the main estimator to deviations from its underlying assumptions, sensitivity analyses are conducted, targeting the same estimand (see figure below).

Figure 1: Aligning target of estimation, method of estimation, and sensitivity analysis, for a given trial objective

How will this impact the way trials are designed?

Randomised trials are free from baseline confounding but certain events that occur after randomisation complicate the description and estimation of treatment effects. These events are termed “intercurrent events” in the addendum and include among others, use of rescue medication, discontinuation of treatment, treatment switching and death.

The assumptions made regarding intercurrent events are a key component when defining an estimand. However, the four attributes as discussed above (ie population, variable, intercurrent events, population summary for the variable) are not independent and subsequently should be consciously and explicitly considered in relationship to each other.

The ICH E9 addendum outlines five strategies for addressing intercurrent events, 1) treatment policy, 2) composite, 3) hypothetical, 4) principal stratum and 5) while on treatment strategy. The addendum expects a protocol to identify and document key intercurrent events for a trial and apply, with scientific reason, the proposed strategies. More than one strategy can be applied per study to accommodate different assumptions for different intercurrent events.

If you would like to know more about the potential systematic effect of intercurrent events on the estimate of treatment effect and how the ICH E9 addendum may impact your business, please contact Dr. Alan Phillips, Vice President ICON Clinical Research at alan.phillips@iconplc.com.


Draft ICH E9 (R1) addendum on Estimands and sensitivity analysis in clinical trials to the guideline on statistical principles for clinical trials, step 2b - Revision 1