PRM204 - There Is More to Decision Making Than Costs and Effects: Handling Practical Constraints in the Value of Information Framework

Abstract

Whether new medical technology is implemented may depend on the balance between costs and effects, but also on practical constraints. Examples are a fixed health care budget and a maximum clinically acceptable risk of adverse events. However, the impact of compliance with such constraints cannot be handled explicitly in the current value of information (VOI) framework. Our objective was to demonstrate proper handling of constraints by extending the VOI framework through separation of cost, effect, and constraint components. The proposed VOI extension was investigated in a simulation study comparing two hypothetical drugs and their side effects. The VOI extension was also applied to a clinical study concerning the cost-effectiveness of carotid intima-media thickness measurements to improve treatment guidance of patients at high risk of cardiovascular disease. Results of the standard VOI analysis, considering only costs and effects, were compared with results from the extended VOI analysis explicitly considering constraints. Standard VOI results may under- or overestimate the value of additional research compared to extended VOI results. In our clinical example, with penalties of $2 and $5 per dollar budget exceedance, standard values for the Expected Value of Perfect Information (EVPI) of $24, and $1,490 were found, with corresponding values of $239, and $565 for the extended EVPI. Ignoring the budget constraint in the standard EVPI analysis therefore resulted in a underestimation of $214 ($2 penalty) and an overestimation of $925 ($5 penalty) of the EVPI per patient. When decision-maker’s criteria go beyond costs and effects, standard VOI results may not reflect the actual value of additional research accurately and may therefore jeopardize optimal research prioritization. Determination of the extended VOI, through separation of cost, effect, and constraint components, is straightforward and can support optimal research prioritization regardless of the complexity of the decision criteria considered.