OBJECTIVES : Novel treatments in development for depression offer rapid improvement in symptoms but have unique risk profiles. We sought to quantify patients’ willingness to accept multiple treatment-related adverse-event risks for improvements in depression symptoms.

METHODS : A web-based discrete-choice experiment was administered to two treatment-resistant depression study cohorts: patients enrolled in esketamine phase-3 clinical trials and a web-panel US sample. Improvements in depression symptoms corresponded to the Montgomery–Åsberg Depression Rating Scale (MADRS). Other features included time to response, drug administration characteristics, and risks of long-term cystitis (0%-5%) and cognitive impairment (0% - 5% or 10%, for split-sample scope test). Random-parameters logit models were specified to assess choices as a function of treatment features. While benefit-risk studies typically calculate maximum-acceptable risk (MAR) for individual adverse events, we calculated acceptance of more clinically-relevant simultaneous risks of cystitis and cognitive impairment.

RESULTS : From March 2017 to March 2018, 161 clinical-trial and 301 online-panel respondents completed the survey. About one-third (35%) of the clinical-trial sample and 14% of the online-panel sample always selected the treatment alternative with greater depression symptom improvement regardless of risks shown. Amongst the remaining respondents willing to accept benefit-risk tradeoffs (51%), for a MADRS-equivalent improvement from 40 to 20, an acceptable risk of cognitive impairment could be as high as 4.8% (95%CI: 3.5 – 6.6). Panel participants would accept 3.2% (2.4 – 4.1). For the same benefit, when considering the simultaneous occurrence of risks, the MAR falls below linear combinations of these thresholds, allowing assessment whether any combination of both risks would be acceptable for a given level of benefit.

CONCLUSIONS : Most participants were willing to accept relatively high risks of individual adverse events for improvements in depression. Our approach avoids overstatement of MAR for simultaneous risks that occurs with conventional approaches to risk tolerance.