Determining patient preferences for the medical management of osteoporosis using conjoint analysis.

We used conjoint analysis-a method that assesses complex decision making-to quantify patients' choices when selecting an osteoporosis therapy. While 60% of people prioritized medication efficacy when deciding among treatments, the remaining 40% highly valued factors other than efficacy, suggesting the need for personalized shared decision-making tools. INTRODUCTION: In this study, we aimed to examine patient decision-making surrounding osteoporosis medications using conjoint analysis. METHODS: We enrolled osteoporosis patients at an academic medical center to complete an online conjoint exercise which calculated each patient's relative importance score of 6 osteoporosis medication attributes (higher = greater relative importance in decision-making). We used latent class analysis to identify distinct segments of patients with similar choice patterns and then used logistic regression to determine if demographics and osteoporosis disease features were associated with latent class assignment. RESULTS: Overall, 304 participants completed the survey. The rank order of medication attributes by importance score was the following: efficacy at preventing hip fractures (accounted for 31.0% of decision making), mode of administration (17.5%); risk of serious side effects (16.6%); dose frequency (13.9%); efficacy at preventing spine fractures (12.5%); risk of non-serious side effects (8.4%). We found that 60.9% of the cohort prioritized medication efficacy as their top factor when selecting among the therapies. Being a college graduate, having stronger beliefs on the necessity of using medications for osteoporosis, and never having used osteoporosis medicines were the only factors associated with prioritizing medication efficacy for fracture prevention over the other factors in the decision-making process. CONCLUSIONS: While about 60% of patients prioritized efficacy when selecting an osteoporosis therapy, the remaining 40% valued other factors more highly. Furthermore, individual patient characteristics and clinical factors did not reliably predict patient decision making, suggesting that development and implementation of shared decision-making tools is warranted.