Estimands in equivalence trials and non-inferiority trials: a cross-sectional study of EMA scientific advice to drug developers.

Background: The Estimands framework, introduced in the Addendum to ICH E9, provides a structured method to define treatment effects in clinical trials. The main novelty of the framework is the discussion of intercurrent events as part of the treatment effect definition. It is widely believed that the application of the framework to non-inferiority and equivalence trials deserves specific consideration.

Methods: To examine the current practices of using the estimand framework in non-inferiority and equivalence trials, we reviewed the scientific advice provided by the European Medicines Agency to drug developers in 2022. This review aimed to determine how often the estimands framework is used by drug developers and/or recommended by EMA and to describe what intercurrent events and handling strategies are being proposed by drug developers and recommended by EMA.

Results: The use of the framework varied substantially by clinical development phases. While it was used for phase 3 trials in 47% (25/53) by developers, it was used in 5% (1/19) of the phase 1 trials. For 39% (11/28) of the trials where developers did not use the estimands framework in phase 3, there was no regulatory recommendation to adopt the framework in the response. The most discussed intercurrent event in our sample was 'treatment discontinuation' (n = 47), for which developers most often proposed either a treatment policy strategy (17/47, 36%) or a hypothetical strategy (11/47, 23%). In contrast, EMA most often recommended the use of two co-primary estimands with two different strategies (22/47, 47%).

Conclusions: Generally, the proposed and recommended strategies depend on the clinical setting and the respective intercurrent event. Developers almost always proposed a single primary estimand, whereas EMA often recommended two co-primary estimands differing in the strategies used to handle some or all the intercurrent events. Further interaction between academia, industry and regulators is necessary to progress the implementation process of the estimands framework for non-inferiority and equivalence trials.