Fig. 1
From: Quantitative benefit-risk assessment of methylprednisolone in multiple sclerosis relapses
Overview of the applied drug benefit-risk assessment framework. 1st panel: It is identified what decision problem is considered; what the alternatives are; and which effects should be included. 2nd panel: A decision tree model is constructed. 3rd panel: For each clinical outcome, i.e. branch in the tree, distributions are specified for probability and utility variables. These are denoted by p and u, respectively. Here, one of the clinical outcomes from high-dose methylprednisolone (reduced relapse – psychosis – second unspecified consequence of psychosis) is used for illustration. Superscripts denote alternatives; subscripts denote paths as the tree branches. For each iteration of the probabilistic analysis, values are sampled for all variables, and expected utilities (denoted E) are computed for all alternatives, as seen in the 4th panel. The expected utility is the probability-weighted sum of utilities over all clinical outcomes, as indicated in the equations. In this example, the same 14 clinical outcomes are considered for all alternatives. 5th panel: Finally, the preference rate of each alternative is computed. Note that the entire scheme or selected parts thereof could be subjected to (non-probabilistic) sensitivity analysis