Use of the analytic hierarchy process for medication decision-making in type 2 diabetes

Aim: To investigate the feasibility and utility of the Analytic Hierarchy Process (AHP) for medication decision-making in type 2 diabetes. Methods: We conducted an AHP with nine diabetes experts using structured interviews to rank add-on therapies (to metformin) for type 2 diabetes. During the AHP, participants compared treatment alternatives relative to eight outcomes (hemoglobin A1c-lowering and seven potential harms) and the relative importance of the different outcomes. The AHP model and instrument were pre-tested and pilot-tested prior to use. Results were discussed and an evaluation of the AHP was conducted during a group session.We conducted the quantitative analysis using Expert Choice software with the ideal mode to determine the priority of treatment alternatives. Results: Participants judged exenatide to be the best add-on therapy followed by sitagliptin, sulfonylureas, and then pioglitazone. Maximizing benefit was judged 21% more important than minimizing harm. Minimizing severe hypoglycemia was judged to be the most important harm to avoid. Exenatide was the best overall alternative if the importance of minimizing harms was prioritized completely over maximizing benefits. Participants reported that the AHP improved transparency, consistency, and an understanding of others' perspectives and agreed that the results reflected the views of the group. Conclusions: The AHP is feasible and useful to make decisions about diabetes medications. Future studies which incorporate stakeholder preferences should evaluate other decision contexts, objectives, and treatments.