A novel methodological approach which allows structured benefit risk assessments to incorporate both uncertainty and correlations between endpoints and weights
*Ian Hirsch, AstraZeneca 

Keywords: Structured, Benefit, Risk, Quantitative

Structured Benefit Risk assessments are being implemented across most internal AstraZeneca projects through a core Benefit-Risk Implementation Team via the BRAT framework. This has facilitated a systematic evaluation and allowed for a more transparent and standardized way of presenting a single Benefit-Risk assessment for each compound facilitated using an internally developed BRAT tool. Most assessments so far have been qualitative in nature using discrete comparisons between key benefits, risks and tolerability endpoints. Externally methods allowing for quantitative assessments have been developed which allow uncertainty around observed effects to be incorporated however there are still some key limitations. This work presents 2 methods which were developed and piloted internally to incorporate not only uncertainty in the observed data but also the correlations between endpoints, the ranges of weights and also target values for each endpoint. This then allows for probabilistic statements about Benefit-Risk profile to be made and the key components, together with their uncertainty, to be pictorially presented.