Estimands: A More Strategic Approach to Study Design and Analysis
An RCT is a scientific experiment that is conducted to allow a fair comparison between different treatment arms. We assign patients randomly to receive the different treatments under investigation. This random assignment serves to create balanced groups in terms of measured and unmeasured background characteristics, such that any differences observed after randomization are expected to be due to assigning patients to the randomized treatments irrespective of compliance and adherence.
In order to conclude that the differences between the treatment arms observed after randomization are due to taking the assigned treatments, we require, in addition, that patients take their treatment throughout the course of the study. However, in clinical studies and practice, events may occur that prevent patients from taking their treatment for the intended duration (e.g., adverse events leading to treatment discontinuation). These events, which are also referred to as intercurrent2 events, occur after randomization, and may themselves be treatment‐related. The more such intercurrent events occur, the less certain we can be that differences observed after randomization are due to taking the randomized treatments.