All Times EDT
Keywords: Bayesian assessment and comparison of Benefit-Risk
Quantitatively assessing patient and/or expert opinions of a treatment’s benefit-risk profile can be challenging. The hierarchical Bayes benefit-risk (HBBR) method (Mukhopadhyay et. al., 2019) provides a way to estimate the relative magnitudes of benefits and risks of a treatment for a given indication using trade-off data collected from a specially designed survey. The HBBR approach then provided some initial proposals on how to combine the clinical data with the estimated magnitudes to assess the overall benefit-risk of a treatment. In this talk, a general framework is proposed to apply HBBR estimates to compare treatments within an indication. The framework utilizes the posterior distribution of treatment benefits and risks in combination with the relative magnitudes from HBBR to estimate the overall benefit/risk of each treatment. The methodology is model agnostic, allowing any suitable Bayesian model to be used, and aggregate level data may be utilized to obtain the posterior distributions for each of the benefits and risks. The approach is demonstrated using simulated data to compare treatments in a Bayesian meta-analysis setting.