Risk of fracture with gabapentinoid treatment in older adults: a multinational population-based study

BACKGROUND: Gabapentinoids are increasingly being prescribed in older adults (aged 60 years or older), but concerns have been raised that their adverse effects on the CNS can increase the risk of fractures. Previous studies have reported associations between gabapentinoid use and fracture, but many have not adequately addressed confounding by indication or examined risk across the treatment journey. Therefore, we aimed to investigate the temporal association between gabapentinoid treatment and fracture in older adults, and to assess whether concomitant opioid or benzodiazepine use further modifies this risk. METHODS: In this retrospective multinational population-based study, we used data from the UK Clinical Practice Research Datalink (CPRD) Aurum database and the South Korea National Health Insurance Service-National Health Screening Cohort (NHIS-HEALS). The analysis included individuals aged 60 years or older prescribed a gabapentinoid and who had a hospitalised fracture between Jan 1, 2010, and Dec 31, 2020, in the UK and between Jan 1, 2003, and Dec 31, 2019, in South Korea. The observation period for each included individual was divided into four mutually exclusive windows: 90 days before gabapentinoid treatment (pre-exposure window), first 60 days of treatment period (focal window 1), remaining time of the treatment period (focal window 2), and all other non-treatment periods (referent window), to capture how risk varied across the treatment course. Adjusted incidence rate ratios (aIRRs) with 95% CI of fracture during different risk windows were estimated using conditional Poisson models within each country, and the country-specific aIRRs for the same risk window were then pooled using a random-effects model. FINDINGS: We included 20 030 participants in CPRD and 2935 in NHIS-HEALS in the analysis. In the CPRD cohort, 15 366 (76·7%) were women and the mean age at event was 77·85 years. In the NHIS-HEALS cohort, 2007 (68·4%) were women and the mean age at event was 69·24 years. The pooled results showed an increased risk of fracture during the pre-exposure window (aIRR 2·92, 95% CI 1·61-5·28, p=0·0004). The aIRR was 1·31 (95% CI 1·00-1·71, p=0·051) in the first 60 days of the treatment period and did not increase for the remainder of the treatment period (0·84, 0·54-1·32, p=0·45). Concurrent prescription of opioids or benzodiazepines elevated the risk of fracture, with an aIRR of 3·15 (95% CI 2·85-3·48, p