Doctor: Astrid Chevance
Title: New methods for the development of Core Outcome Sets : the example of depression
Supervisors: Philippe Ravaud and Viet-Thi Tran
Doctoral school: ED 393 Epidemiology and Biomedical Information Sciences, Université Paris Cité
Date of thesis defense: 16/12/2020
Jury: Raphaël Gaillard, Marie Tournier, Marie Herr, Pierre Thomas, Philippe Ravaud, Viet-Thi Tran.
Many treatments for depression are available such as antidepressants, psychotherapy, or neurostimulation techniques. These different treatments are evaluated in randomized controlled clinical trials (RCTs) measuring their effectiveness and safety with outcomes. However, the heterogeneity of outcomes across RCTs prevents the comparison of their results and their combination in meta-analyses. Moreover, they may be of little relevance to clinical practice. One answer to this problem is the use of a Core Outcome Set (COS), which is the minimum set of outcomes to be measured across trials of a given disease. In this work, we present new methods for developing COS by involving large samples of all relevant stakeholders, including patients, to improve their generalizability and credibility, and thus, potentially, their use in RCTs. Regarding efficacy outcomes, we propose to identify relevant domains using an international online survey involving patients, informal caregivers, and clinicians. The survey relies on a few open-ended questions about the expectations regarding the treatment. The proof of concept was performed by the PROCEED study which included 1912 patients, 464 relatives and 627 clinicians and identified 80 domains most of which, such as mental pain, are never measured in RCTs. PROCEED embodies the first step of the development of a COS for depression. Regarding harm outcomes, we defined what kind of adverse events (AEs) should be included in a COS: 1) unexpected serious AEs (as defined by the FDA) that occur in RCTs – which is a legal requirement, 2) expected serious AEs, 3) a list of AEs that are not serious but still troublesome to patients. We propose a method to identify the list of non-serious troublesome AEs by 1) identifying the non-serious AEs of a given condition through a systematic review of RCTs, 2) conducting a preference study with patients and clinicians to determine which non-serious but troublesome ARs are considered important. Using the example of RCTs of antidepressants, we have made the proof-of-concept of this method with an online survey involving 1631 patients and 282 clinicians. Participants ranked the 30 most frequently reported non-serious AEs in antidepressant RCTs. Those considered most deleterious were insomnia, sexual dysfunction, weight gain, anxiety, fatigue, and agitation. This research work has produced new methods for the development of COS allowing for a better quantitative and qualitative involvement of all relevant stakeholders of RCTs, with patients in the foreground. In addition, we have conceptualized and proposed methods for the selection of harm outcomes, whereas until now, COS research initiatives have focused on efficacy outcomes. More specifically for depression, this work has identified important areas of efficacy and safety for patients and clinicians where no research had been dedicated to this topic. The further development of the COS for depression will help establish the proof of concept for the new methods we have proposed. While research on COS has focused on their development, it is becoming increasingly clear that the next challenge is to improve their uptake in RCTs, otherwise we risk contributing to research waste rather than reducing it.