Oral bisphosphonates and risk of subtrochanteric or diaphyseal femur fractures in a population-based cohort

Abstract

Bisphosphonates are the primary therapy for postmenopausal and glucocorticoid-induced osteoporosis. Case series suggest a potential link between prolonged use of bisphosphonates and low-energy fracture of subtrochanteric or diaphyseal femur as a consequence of oversuppression of bone resorption. Using health care utilization data, we conducted a propensity score-matched cohort study to examine the incidence rates (IRs) and risk of subtrochanteric or diaphyseal femur fractures among oral bisphosphonate users compared with raloxifene or calcitonin users. A Cox proportional hazards model evaluated the risk of these fractures associated with duration of osteoporosis treatment. A total of 104 subtrochanteric or diaphyseal femur fractures were observed among 33,815 patients. The estimated IR of subtrochanteric or diaphyseal femur fractures per 1000 person-years was 1.46 [95% confidence interval (CI) 1.11-1.88] among the bisphosphonate users and 1.43 (95% CI 1.06-1.89) among raloxifene/calcitonin users. No significant association between bisphosphonate use and subtrochanteric or diaphyseal femur fractures was found [hazard ratio (HR) = 1.03, 95% CI 0.70-1.52] compared with raloxifene/calcitonin. Even with this large study size, we had little precision in estimating the risk of subtrochanteric or diaphyseal femur fractures in patients treated with bisphosphonates for longer than 5 years (HR = 2.02, 95% CI 0.41-10.00). The occurrence of subtrochanteric or diaphyseal femur fracture was rare. There was no evidence of an increased risk of subtrochanteric or diaphyseal femur fractures in bisphosphonate users compared with raloxifene/calcitonin users. However, this study cannot exclude the possibility that long-term bisphosphonate use may increase the risk of these fractures.

Fig. 1

Study design. Subjects were required to have at least one claim each during the prior three 6-month intervals. For both the primary (“as treated”) and secondary (“first exposure carried forward”) analyses, follow-up began on the 91st day after filling the first prescription of either exposure of interest. The second prescription fill for the same exposure drug group was required during the 90-day lag period. For the primary analysis (A), we continued the follow-up until 90 days after the last drug available date. Last drug available date was calculated with a number of days of supply after the last prescription fill date. For the secondary analysis (B), the follow-up continued until 365 days after the index date. Patients were considered “always exposed” for the first exposure drug group during the follow-up period. In a sensitivity analysis (C), follow-up began at the first prescription fill and ended 90 days after the last drug available date.

Fig. 2

Flow diagram of cohort selection. PACE = Pennsylvania Pharmaceutical Assistance Contract for the Elderly; PAAD = New Jersey Pharmaceutical Assistance to the Aged and Disabled.

Fig. 3

Kaplan-Meier curves for fracture-free survival in oral bisphosphonates versus raloxifene/calcitonin nasal spray.

Fig. 4

Hazard ratios (HRs) for subtrochanteric or diaphyseal femur fractures according to osteoporosis treatment duration.