The Case for a Bayesian Approach to Benefit-Risk Assessment: Overview and Future Directions

Abstract

The benefit-risk assessment of a new medicinal product or intervention is one of the most complex tasks that sponsors, regulators, payers, physicians, and patients face. Therefore, communicating the trade-off of benefits and risks in a clear and transparent manner, using all available evidence, is critical to ensure that the best decisions are made. Several quantitative methods have been proposed in recent years that try to provide insight into this challenging problem. Bayesian inference, with its coherent approach for integrating different sources of information and uncertainty, along with its links to optimal decision theory, provides a natural framework to perform quantitative assessments of the benefit-risk trade-off. This paper describes the current state of the art in Bayesian methodologies for quantitative benefit-risk assessment, and how these may be leveraged throughout the life cycle of a medicinal product to support and augment clinical judgment and qualitative benefit-risk assessments. Gaps and potential new directions that extend the current approaches are also identified.

Bayesian joint modelling of benefit and risk in drug development

“…However, MCDA relies on the identification of appropriate weights and utility functions for each of the endpoints. Although this allows one to incorporate multiple endpoints into a single BR metric, there is a risk that the weights rather than the data drive the BR assessment and conclusions . In addition, it may be challenging to understand the impact that dependency among the multiple endpoints has on the conclusions.…”

“…These situations will usually be specific to the asset and have low dimensionality. The Bayesian framework is particularly useful for BR assessments as uncertainty can be expressed in terms of probabilities, which are simple to communicate to nonstatisticians . Decision‐making under uncertainty is described by the EMA and FDA as one of the key challenges associated with BR evaluations .…”

Bayesian Benefit Risk Analysis

“…Waddingham et al (2016) used Bayesian modelling to incorporate parameter estimation uncertainty into the final MCDA score, and demonstrated the importance of taking that uncertainty into account for drug evaluation. Costa et al (2017); Li et al (2019) showed how to account for the correlation between continuous types of data, typically the treatment efficacy outcomes, and discrete types of data, typically adverse events. Naturally, in order to account for any correlation, such models are fitted on individual level data, rather than summary data.…”

Predicting Regulatory Product Approvals Using a Proposed Quantitative Version of FDA’s Benefit–Risk Framework to Calculate Net-Benefit Score and Benefit–Risk Ratio

“…Numerous mathematical methodologies have been explored in the past in an effort to quantitatively explain the benefit–risk assessment conclusions and/or to explore a working model tool. These include probabilistic decision analysis [ 4 ], use of spatial planes [ 5 ], Bayesian approaches [ 6 ], patient preferences [ 7 , 8 ], incremental net health benefit using simulation data [ 9 ], number-needed-to-treat (NNT) and number-needed-to-harm (NNH) [ 10 ], and a variety of other quantitative approaches [ 11 ]. Some notable studies involved the benefit–risk analysis of cancer-related endpoints of pivotal studies from 20 + products for non-small cell lung cancer [ 12 ], multiple myeloma [ 13 ], and melanoma [ 14 ], and another compared the quantitative profile of new chemical entities that were initially approved but subsequently withdrawn from the market [ 15 ].…”