(A-158) Patterns of Long-Term Bisphosphonate Use and Drug Holiday Among U.S. Medicare Beneficiaries

Background: Oral bisphosphonates (BP) are first-line therapy for osteoporosis. Clinical guidelines recommend temporary discontinuation of BP after long-term treatment (≥3 years), also called a drug holiday, to reduce serious adverse events. However, the prevalence of long-term BP treatment and drug holidays among older adults in the US has not been well-described with contemporary data.

Objectives: We aimed to describe the initial persistence to BP and the prevalence of long-term BP use and drug holidays after accounting for initial gaps in therapy in community-dwelling US Medicare beneficiaries.

Methods: We conducted this cohort study using U.S. Medicare enrollment and prescription claims data for a 20% random sample of beneficiaries. We identified Fee-for-Service beneficiaries aged ≥66 who initiated an oral BP from 01/01/2008 to 12/31/2019. We defined persistence as the length of time until a 90-day gap in BP use and we defined long-term BP use as a 3-year period with a proportion of days covered ≥80%, measured using 3-year windows. Among beneficiaries with long-term BP therapy, we identified those potentially starting a drug holiday, defined as the first 120 days without BP therapy. For all analyses, we followed patients until the outcome, death, disenrollment, entry of long-term care, or end of data (12/31/2022). Kaplan Meier estimators were used to calculate the proportion of patients with persistence, long-term BP therapy, and drug holidays over time.

Results: We identified 321,686 eligible beneficiaries who initiated oral BPs (mean age (SD) = 75.4 (6.9), 78% White, 12% male). Alendronate was the most common BP (79%), followed by ibandronate (12%), and risedronate (9%). Initial persistence to BP was 63% at 6 months, 46% at 1-year, and 21% at 3 years. However, when using a rolling-window approach, we identified that 63,231 (31%) had long-term BP use, with 10,450 (17%) having a gap between initial prescription and the start of long-term BP therapy (11% with a gap >1 year). By the end of the data, 35,731 (57%) of beneficiaries with long-term BP use started a potential drug holiday, with a median length of 421 days between the start of long-term BP treatment and the drug holiday.

Conclusions: Although initial persistence to BP was low, when accounting for initial gaps in therapy, nearly one-third of Medicare FFS beneficiaries who initiated BP had long-term BP use, and of these, more than 50% potentially had a drug holiday. Future work should assess the safety of drug holidays among generalizable patient populations.