(071) Discontinuation of bisphosphonates in older people with complex health needs: self-controlled case series to assess fracture risk

Background: Substantial effectiveness of bisphosphonates has been shown in the prevention of fractures in older people. However, it is unclear whether and when bisphosphonate therapy can be safely discontinued in this particular population.

Objectives: To assess the effect of stopping oral bisphosphonate treatment on the risk of fractures in older patients with complex health needs.

Methods:

Design: Self-controlled case series (SCCS).
Setting: UK primary care electronic health records.
Participants: Patients aged >65 years registered in CPRD GOLD for ≥1 year before study start (01/01/2010), who in 2009 had a non-elective hospitalisation, were frail or received poly-pharmaceutical treatment were selected. People who used bisphosphonates in 2009 were subsequently excluded. Of those, we included all people with a new bisphosphonate prescription and a record of a fracture recorded in primary care data during the study period.
Exposure: Discontinuation of oral bisphosphonate, defined as a treatment gap of >90days.
Main outcome measures and analyses: Age-adjusted incidence rate ratios (IRR) were calculated for comparing fracture rates during bisphosphonate treatment vs. the post-discontinuation period. The main model included all people irrespective of the duration of continuous bisphosphonate use, while subgroup analyses required >1 year of bisphosphonate treatment prior to discontinuation.
Assumptions of the SCCS model were tested before progressing with the analyses. As no association between statins and fracture risk is known, we repeated our analyses using statins as negative control exposure to assess residual confounding.

Results: Our study included 171,048 bisphosphonate-naïve people with complex health needs. Of those 6,245 individuals started bisphosphonate treatment during the study period, 1,262 fractures were observed during 9,796 person-years of treatment (event rate 128.8/1,000 person-years [95%CI 121.7 to 136.0]. This compared to 942 fractures during 6,428 person-years of follow-up post-discontinuation (event rate 146.6/1,000 person-years [137.2 to 156.1]). No increased fracture risk was seen after discontinuing bisphosphonates, with IRR 0.94 [0.81 to 1.1].
However, discontinuation among people with >1 year history of continued bisphosphonate therapy pointed towards increased fracture risk, with IRR of 1.26 [1.0 to 1.58].
No association between statins and fracture risk was seen in the negative control exposure analyses (IRR 0.98 [0.79 to 1.21]).

Conclusions: Our study showed risks associated with discontinuing oral bisphosphonate in people using them for at least 1 year, likely explained by their continued efficacy. Further research focussing on the risk-benefit of these treatments for most vulnerable older adults is needed.