Estimands - what's all the fuss about?

The public consultation period for the ICH E9 (R1) addendum on estimands and sensitivity analyses to the guideline on statistical principles for confirmatory clinical trials closed at the end of February 2018, and the addendum is planned to be issued in mid-2019.

At first glance the draft addendum can appear quite confusing – with a liberal sprinkling of statistical terminology such as ‘Estimand’, ‘Estimate’ and ‘Estimator’ it can be quite a dry document to digest (even for us statisticians). Moreover, there is a real risk that it is perceived as a ‘statisticians only’ document. This couldn’t be further from the truth and statisticians and clinicians cannot design or analyse trials in blissful isolation.

Indeed, one the main aims of the addendum is “to facilitate the dialogue between disciplines involved in clinical trial planning, conduct, analysis and interpretation, as well as between sponsor and regulator, regarding the treatment effects of interest that a clinical trial should address.”

So, if you’re planning a clinical trial hopefully I’ve started to convince you that you should shouldn’t be frightened by this new concept and your friendly statistician is there to support you. Firstly, they will be there to explain to you what an Estimand is. Put very simply they can be thought of as the missing link between the objectives in the protocol and the statistical analysis proposed. In other words, it is a clear description of what analysis will be performed in order to address each objective of the study. You may already be familiar that a single objective can have a multitude of options for what is being analysed. Describing these as an Estimand, following the guidance in the draft addendum, helps to make this clearer to all involved.

An Estimand should be described according to four attributes (see Figure 1).  Attributes A, B and D should be familiar to those involved in the design, planning, conduct and analysis of clinical trials. However, specification of attribute C, how intercurrent events are reflected in the scientific question of interest, may be a new concept to many.

Estimand Description height=
Figure 1: Description of an Estimand.
Source: Estimands and Sensitivity Analyses: What’s in the new ICH E9 Addendum?
Chrissie Fletcher, 1 Day PSI Meeting on Estimands, 27th Sept 2017

Intercurrent events are those which occur after treatment initiation and either preclude observation of the clinical endpoint of interest or affect its interpretation.

Such events include: death, treatment discontinuation due to adverse events or lack of efficacy, use of other medicines affecting the outcome, whether specified or prohibited by the protocol.

In the past there has been ambiguity in the handling of intercurrent events in the analysis of clinical trials. The addendum attempts to overcome this by offering at least five strategies to take into account when considering how to deal with intercurrent events. In addition to this, the ICH E9 (R1) expert working group has worked tirelessly to share examples to show how these can be put into practice.

One key challenge of ICH E9 (R1) is how this fundamental change affects other ICH guidance documents – for example:

  • How should estimands be described in the protocol?
  • How should estimands be described in the CSR?

As mentioned earlier this requires close collaboration. 2018/2019 brings exciting times – let’s face the challenge together.