Bisphosphonate drug holidays: Risk of fractures and mortality in a prospective cohort study.

Purpose: This study examined osteoporotic fractures and mortality in patients pretreated with bisphosphonates (BPs) during BP holidays and ongoing BP use.

Methods: Interview-based prospective observational study in a cohort of 1973 patients with BP treatment for at least 80% of the total time of the preceding 4 years. Patients were recruited from 146 primarily endocrinological, orthopedic and rheumatological practices and clinics across Germany between May 2013 and June 2015. Outcomes were analyzed by Cox proportional hazards regression in relation to treatment status at the time of the first interview (model 1) or using time-dependent treatment variables (model 2). Temporal changes in fracture risk during BP holidays were evaluated by comparisons among 3 incremental levels of simple moving averages of BP treatment during the preceding 12 months (BP-SMA levels 0%, >0% to <50%, and ≥50%).

Results: For an observation period of up to 25 months, the adjusted hazard ratios (HRs) in model 1 for BP holidays compared to ongoing BP use were 0.87 (95% confidence interval [CI] 0.59-1.28) for major osteoporotic fractures (MOFs), 0.95 (95% CI 0.70-1.28) for any clinical osteoporotic fracture, 0.96 (95% CI 0.55-1.68) for clinical vertebral fractures, and 0.86 (95% CI 0.50-1.48) for mortality. The risk of MOFs was higher for the BP-SMA level 0%, corresponding to a time >12 months since the start of a BP holiday, than for the BP-SMA level >0% to <50%, corresponding mainly to a time >6 to ≤12 months since the start of a BP holiday (adjusted HR 2.28, 95% CI 1.07-4.86). We found an interaction between prevalent vertebral fractures (PVFs) and BP-SMA-related time to first MOF for BP-SMA as a continuous variable (p for interaction 0.046 in the adjusted model). The adjusted HR for MOFs for the BP-SMA level 0% compared to the BP-SMA level >0% to <50% was 3.53 (95% CI 1.19-10.51) with a PVF but was 1.44 (95% CI 0.49-4.22) without a PVF.

Conclusions: Fracture risk and mortality in patients with preceding BP treatment did not significantly differ between BP holidays and ongoing BP use for an observation period up to 25 months when outcomes were analyzed in relation to treatment at the time of the first interview. However, in the presence of a PVF, the risk of MOFs was higher for a BP-SMA level corresponding to a time >12 months since the start of a BP holiday than for a BP-SMA level corresponding mainly to a time >6 to ≤12 months since the start of a BP holiday. The presence of a PVF may increase the relative risk of MOFs associated with a longer BP holiday.