STRUCTURING BENEFIT-RISK MODELS IN PRESENCE OF NUMEROUS ADVERSE EVENTS: A CASE STUDY OF MULTIPLE SCLEROSIS

Purpose:

Preference-based benefit risk assessment (BRA) is increasingly commonly used to account for the patient voice in medical decision making. While it is a valuable tool to support the formalisation of patient preferences, further guidance is required on how adverse events (AEs) should be incorporated into a BRA. Challenges are posed by the limitation on the number of criteria that can be incorporated into a BRA, the diversity of AEs experienced by patients on different treatments, and limitations in the data on AEs for comparator treatments.  The purpose of this study was to review good practice guidelines and lessons from previous BRAs to generate recommendations on how to incorporate AEs into BRA, using multiple sclerosis (MS) as a case study.

Method(s):

A review of MS studies was carried out to identify risks associated with MS treatments, the different methods used to incorporate them into BRA, and the challenges and recommendations emerging from this experience. Good practice guidelines were reviewed to identify approaches that might support the incorporation of AEs into BRA, and their pros and cons.

Result(s): Six MS studies included a diversity of approaches to incorporating AEs into BRA, including: grouping the AEs into various categories based on severity levels, or shortlisting them to a manageable number. A number of challenges were identified which should inform BRA designs, including: ignoring different preferences for specific AEs when they are aggregated; splitting and availability biases when AEs are disaggregated; overlap with other criteria when discontinuation is used as a proxy for AE severity; different duration and reversibility of AEs; patients understanding of AEs if they have not experienced them before. These challenges have implications for which methods are appropriate for BRA. For instance, certain methods place less restrictions on the number of criteria; and workshop-based approaches allow for more information to be provided to patients and consistency checks to be performed.

Conclusion(s):

There currently is no consensus on the best approach to capture AEs as part of a quantitative BRA. While the choice of model will be motivated by the type of data available and the goals of the analyses, the literature provides some options and guidance on overcoming the challenges. Further testing with patients of the available methods is required.