(B-099) Comedication Patterns of Romosozumab Early Adopters Among Medicare Enrollees: An Unsupervised Clustering Approach

Background: Polypharmacy is a major concern in osteoporosis (OP) management, especially in older adults. Romosozumab (romo), approved in April 2019, is a parenteral OP medicine for postmenopausal women at high fracture risk, given monthly for 12 months. We examined comedication patterns among early romo adopters and their association with treatment persistence.

Objectives: Evaluate comedication patterns among early romo users.

Methods: This retrospective cohort study used Medicare fee-for-service data (2006-2022) to identify female romo early adopters aged 65+ who initiated romo in 2019, with>=4 years of continuous enrollment, including 3 years pre-initiation. We excluded patients with metastatic cancer or Paget’s disease.
We identified all prescription drugs covering the romo initiation day and selected 8 major drug categories for clustering: anti-HTN, lipid-lowering, anti-diabetic, antidepressants, hypnotics, NSAIDs, opioids, and steroids. We used consensus clustering analysis to identify comedication patterns.
We explored the association between comedication groups and romo discontinuation, defined as >= 60 days without injection, using a Cox proportional hazard model, adjusting for age, race, geographic region, Charlson comorbidity index (CCI), had emergency room visit, fracture history, romo billing in Part D, DXA scan, and prior OP treatment history. Patients were not considered discontinued after they completed 12 doses or 1 year of romo treatment.

Results: A total of 1,777 romo early users were included (mean age: 77.7 years; 90.9% White). Consensus clustering analysis identified 5 mutually exclusive comedication pattern groups:
1 (N=326). No patients used any pre-specified drugs of interest.
2 (N=275). All used anti-HTN only.
3 (N=281). All used lipid-lowering drugs, some (67%) also used anti-HTN.
4 (N=313). All used antidepressants, some also used anti-HTN (57%) and/or lipid-lowering (41%) drugs.
5 (N=582). Various drugs (>=1 of the 8 categories, not in other groups).
Patients in group 2 were the oldest (79.4 years, p< 0.001) and had the highest proportion from the South (47.3%, p=0.013). Group 5 had the highest CCI (4.8, p< 0.001) and the lowest percentage of White patients (88.5%, p=0.001). No significant differences in fracture or OP treatment history were observed. One-year overall romo discontinuation risk was 32.2% [29.1%, 35.7%]. Compared with group 1, all groups had higher discontinuation risk, with group 4 (HR=1.41 [1.14–1.76]) and group 5 (HR=1.39 [1.14–1.69]) significant after adjusting for covariables.

Conclusions: Distinct comedication patterns were identified among romo early adopters. Patients using antidepressants or multiple drug categories had a higher discontinuation risk and needed additional adherence support.