Osteoporosis treatment adherence trajectories, comedication patterns, and romosozumab persistence among early users: A data-driven analysis.

Liu, Ye; Wu, Hongke; Arora, Tarun; Burden, Andrea M; Danila, Maria I; Saag, Kenneth G; Curtis, Jeffrey R · Bone · 2026

retrospective_cohort · Level III

Where this comes from

Abstract

To evaluate whether previous osteoporosis treatment adherence trajectories and comedication patterns at romosozumab initiation are associated with romosozumab treatment persistence among Medicare beneficiaries initiating romosozumab in the early post-approval period. We used Medicare data (2016-2022) to identify women aged ≥65 initiating romosozumab in 2019. Prior osteoporosis treatment over the 36-month baseline was summarized as monthly proportion of days covered and grouped into adherence trajectories using group-based trajectory modeling (GBTM). Comedication patterns on the romosozumab initiation day were derived from eight drug categories using consensus clustering. Primary outcome was early romosozumab persistence, assessed in three successive 60-day windows (30-day dosing interval plus 30-day permissible gap) after the first, second, and third doses, defined as receipt of the subsequent dose. Secondary analysis assessed romosozumab discontinuation. Models adjusted for covariables selected via LASSO approach. Among 1746 early romosozumab users, 70.8% had prior osteoporosis therapy, with four adherence trajectories identified by GBTM. Compared with consistently high prior adherence, no prior therapy was associated with lower treatment persistence (adjusted HR[aHR] = 0.91, 95% CI[0.85, 0.97], p = 0.003), while declining prior adherence was associated with increased risk of romosozumab discontinuation before completion (aHR = 1.35[1.09, 1.68], p = 0.007). Consensus clustering analysis identified five comedication patterns. Patterns characterized by antidepressants and/or antihypertensives were associated with increased risk of discontinuation compared with those using none of the eight drug categories at romosozumab initiation. Data-driven stratification based on prior treatment behavior and comedications identifies romosozumab early users at higher risk of non-persistence and may inform individualized support to maintain treatment persistence.