Predictors of Hip Fracture Despite Treatment With Bisphosphonates Among Frail Older Adults

Andrew R. Zullo, PharmD, PhD; Mark N. Sorial, PharmD; Yoojin Lee, MS, MPH; Christine W. Lary, PhD; Douglas P. Kiel, MD, MPH; Sarah D. Berry, MD, MPH


J Am Geriatr Soc. 2020;68(2):256-260. 

In This Article

Abstract and Introduction


Objectives: Bisphosphonates are effective at preventing hip fractures among older adults, yet many patients still fracture while on treatment and may benefit from additional preventive interventions. Little data are specifically available to target such efforts among bisphosphonate users. We aimed to identify predictors of hip fracture unique to frail older adults initiating pharmacologic treatment for osteoporosis.

Design: Retrospective cohort using 2008–2013 linked national Minimum Data Set assessments, Medicare claims, and nursing home (NH) facility data.

Setting: NHs in the United States.

Participants: Long-stay NH residents 65 years or older who initiated treatment with a bisphosphonate (N = 17 753). Estimates for bisphosphonate initiators were contrasted with those for calcitonin initiators (control group; N = 5348).

Measurements: Hospitalized hip fracture outcomes were measured using Part A claims. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated for 36 a priori selected potential predictors.

Results: The mean (SD) age of the study population was 84 (8) years, 85% were women, and 51% had moderate to severe cognitive impairment. Predictors associated with a higher risk of hip fracture despite bisphosphonate use included age 75 years or older to 85 years (vs ≥65 to <75 y; HR = 1.25; 95% CI = 1.02–1.55), female sex (HR = 1.33; 95% CI = 1.06–1.67), white race (vs black race (HR = 1.87; 95% CI = 1.36–2.58), and body mass index = 18.5–24.9 (vs ≥30; HR = 1.93; 95% CI = 1.53–2.42). Independent ability to transfer (vs total dependence; HR = 3.11; 95% CI = 1.83–5.30) and occasional urinary incontinence (vs frequent; HR = 1.45; 95% CI = 1.18–1.78) were also important predictors. Dementia, diabetes, psychoactive drug use, and other characteristics were not associated with post-prescribing hip fracture. Predictors did not differ between bisphosphonate and calcitonin users.

Conclusion: Predictors of hip fracture among frail older adults did not differ between those who were new users of bisphosphonates vs calcitonin. Given the absence of risk factors unique to bisphosphonate users, targeting of fracture prevention efforts should extend beyond pharmacologic therapy to include existing nonpharmacologic therapies, particularly fall prevention strategies.


Bisphosphonates reduce fracture risk among frail older adults,[1] yet many treated individuals still experience fracture. As many as 10% of treated patients will fracture within 5 years of initiating therapy.[2,3] The occurrence of fractures despite bisphosphonate treatment confirms that additional fall and fracture interventions must be implemented after prescribing. However, data are critically lacking about who is most likely to fracture while on therapy, and thus the benefit from additional interventions to reduce fall and fracture risk is not known. It is possible that bisphosphonate users have unique risk factors that should guide post-prescribing fall and fracture interventions. Without an improved understanding of risk factors for fracture subsequent to the decision to prescribe osteoporosis treatment, it will be impossible to efficiently reduce the post-prescribing risk of fractures and associated morbidity,[4] mortality,[4] and worsened quality of life[5] among frail older adults. Therefore, we aimed to identify unique predictors of hip fracture despite treatment among frail older adults prescribed bisphosphonates compared with calcitonin users as a control group.