Evaluating Risk Tolerance from a Systematic Review of Preferences: The Case of Patients with Psoriasis

Background: Stated-preference methods have been widely used to evaluate patient-relative preferences for the benefits and potential harms of psoriasis treatments. However, risk tolerance measures for treatment-related harms, a corollary of preferences, are rare despite their critical role in shared decision making and regulatory benefit-risk evaluations. This article presents a method to enhance information on patient risk tolerance through previously published preference results.

Objective: The objective of this article was to conduct the first meta-analysis of preferences to characterize the distribution of patients' maximum acceptable risk of harms associated with psoriasis treatments.

Data Sources: Maximum acceptable risks for treatment-related adverse events were extracted or derived from preference results published between 2011 and 2017.

Synthesis Methods: Four different analyses were conducted to evaluate maximum acceptable risk information across studies: (1) listing of maximum acceptable risk values, (2) naïve aggregation of maximum acceptable risks, (3) estimation of maximum acceptable risk mother distribution, and (4) random-effect regression analysis of maximum acceptable risks.

Results: Nine publications with maximum acceptable risk results, or with enough information to derive maximum acceptable risks, were identified from the search and screening of preference studies. The most commonly evaluated treatment benefits were duration of benefits, percentage and probability of improvement, and reductions in the coverage of lesions. The adverse-event risks most often included in the publications were those commonly associated with biologics, such as serious infections and malignancies. As expected, maximum acceptable risks changed with treatment benefits and treatment-related adverse events.

Conclusions: The results confirm the feasibility of using previously published preference information to characterize patient risk tolerance. The estimated distributions of maximum acceptable risk provide a benchmark against which future results can be compared, and signal gaps in our understanding of risk tolerance for specific health outcomes.