CAN VOI ANALYSIS BE USED FOR SETTING GOVERNMENT RESEARCH AGENDAS AND PRIORITIES?: EXPERIENCE TO DATE FROM A LARGE PROVINCE-WIDE INITIATIVE USING AN ITERATIVE VOI ANALYSIS FRAMEWORK

Background: Results from Value of Information (VOI) analyses are increasingly being presented at conferences and appear increasingly in scientific journals. However, there is a scarcity of examples where this information is actually used for influencing future research questions and an even larger scarcity of examples where this information is actually used by government decision making bodies for setting research agendas and priorities.    
Methods: We developed an iterative evidence-based framework based on VOI analysis to assist the Ontario Ministry of Health and Long-Term Care (MOHLTC) in determining whether additional data collection is warranted when conducting studies of technologies based on conditional funding. PATH’s Reduction of Uncertainty through Field Evaluation (PRUFE) framework is an evidence-based framework that provides decision makers with information on clinical and cost-effectiveness by combining ‘prior’ evidence with new local/geographic-specific evidence in a Bayesian updating process. The combining of the ‘prior’ evidence with new data to form ‘posterior’ evidence is re-presented to decision makers as new evidence is generated and decisions are subsequently made about whether to continue to collect additional information and what information should be collected to further reduce decision uncertainty. To date, the results from VOI analyses have been used to make decisions to discontinue data collection for some studies while continuing to collect data in others. Case studies of Endovascular repair of Abdominal Aortic Aneurisms (AAA) and Drug Eluting Stents (DES) will be presented.
Results: VOI analysis within the PRUFE framework has been successfully used by decision makers for setting research agendas and priorities. In the first evaluation where VOI analysis was used (EVAR), the decision was made that sufficient information was available and that additional data collection was not warranted. In the second example (DES), an interim decision was made to continue data collection to reduce uncertainty around rates of revascularization and based on a subsequent update a decision was made to continue to collect data to reduce uncertainty around mortality benefits of DES. Data collection for this second stage of updating is ongoing.
Conclusion:  We believe this is the first example where VOI analysis has been used in a Bayesian updating data collection process for a government decision making body that uses results for setting research agendas and priorities for government funded research programs.