Causal treatment effect estimation in randomised clinical trials using endpoint measurements post-intercurrent events

Specialeforsvar: Ida Frandsen

Titel: Causal treatment effect estimation in randomised clinical trials using endpoint measurements post-intercurrent events

Abstract: In many clinical trials with continuous longitudinal data, a Mixed Model for Repeated Measurements (MMRM) has been routinely used to estimate an estimand with one or more intercurrent events handled by a hypothetical strategy. Data, after the intercurrent event has occurred, has been disregarded for the estimation of the estimand and is therefore not included in the estimation. The underlying assumptions of this method have been questioned and viewed as problematic in a regulatory decisionmaking context.
Research is ongoing on how to reliably estimate an estimand that uses a hypothetical strategy to handle intercurrent events. Therefore, the main objective of this thesis is to implement, test and discuss recently described methods using causal inference to estimate a treatment effect based on a hypothetical strategy in the context of a randomised clinical trial, namely Pioneer 1. In contrast to the standard implementation of the MMRM, these methods make active use of post-intercurrent event data. The resulting implementation will be used to re-analyse the data from the Pioneer 1. Results from the analyses will be discussed alongside the underlying assumptions of the various methods and compared to those of the MMRM. Furthermore, a simulation study mimicking the Pioneer 1 study will be conducted to explore how violations of assumptions may affect the results.

Vejledere: Niels Richard Hansen
                  Helle Lyngaard and Christian Bressen Pipper
Censor: Søren Wengel Mogensen,  Lunds Universitet