One and two-year persistence with different anti-osteoporosis medications: a retrospective cohort study

Adherence to anti-osteoporosis medications is poor. We carried out a cohort study using a real-world population database to estimate the persistence of anti-osteoporosis drugs. Unadjusted 2-year persistence ranged from 10.3 to 45.4%. Denosumab users had a 40% lower risk of discontinuation at 2 years compared to alendronate users.

The purpose of this study was to estimate real-world persistence amongst incident users of anti-osteoporosis medications.

This is a retrospective cohort using data from anonymised records and dispensation data (www.​sidiap.​org). Eligibility comprised the following: women aged ≥50, incident users of anti-osteoporosis medication (2012), with data available for at least 12 months prior to therapy initiation. Exclusions are other bone diseases/treatments and uncommon anti-osteoporosis drugs (N < 100). Follow-up was from first pharmacy dispensation until cessation, end of study, censoring or switching. Outcomes are 2- and 1-year persistence with a permissible gap of up to 90 days. Persistence with alendronate was compared to other bisphosphonates, strontium ranelate, selective oestrogen receptor modulators, teriparatide and denosumab. Cox models were used to estimate hazard ratios of therapy cessation according to drug used after adjustment for age, sex, BMI, smoking, alcohol drinking, Charlson co-morbidity index, previous fractures, use of anti-osteoporosis medication/s, oral corticosteroids and socio-economic status.

A total of 19,253 women were included. Unadjusted 2-year persistence [95% CI] ranged from 10.3% [9.1–11.6%] (strontium ranelate) to 45.4% [43.1–47.8%] (denosumab). One-year persistence went from 35.8% [33.9%–37.7%] (strontium ranelate) to 65.8% [63.6%–68.0%] (denosumab). At the end of the first year and compared to alendronate users, both teriparatide and denosumab users had reduced cessation risk (adjusted HR 0.76, 95% CI 0.67–0.86 and 0.54, 95% CI 0.50–0.59 respectively) while at the end of the second year, only denosumab had a lower risk of discontinuation (adjusted HR 0.60, 95% CI 0.56–0.64).

Unadjusted 2-year persistence is suboptimal. However, both teriparatide and denosumab users had better 1-year persistence and only denosumab had 2-year better persistence compared to alendronate users. Unmeasured confounding by indication might partially explain our findings.