Gabapentinoid use and the risk of fractures in patients with inflammatory arthritis: nested case-control study in the Clinical Practice Research Datalink Aurum.

BACKGROUND

Gabapentinoids are increasingly prescribed in inflammatory arthritis (IA), despite no trial evidence for efficacy at managing pain in this population. Observational studies in non-IA populations suggest gabapentinoids are associated with fractures but are limited by methodological heterogeneity/potential residual confounding. Patients with IA generally have an increased risk of fracture so may be particularly vulnerable. We examined the relationship between fractures and gabapentinoids in patients with IA who had all been prescribed a gabapentinoid at some point (to minimise confounding by indication).

METHODS

Our matched case-control study used linked national data from English primary care (Clinical Practice Research Datalink Aurum) and Hospital Episode Statistics. A cohort was constructed of adults with IA, contributing data 01/01/2004-31/03/2021, and ever prescribed oral gabapentinoids. Cases with an incident fracture post-cohort inclusion were ascertained and 1:5 risk set-matched (on age/gender/gabapentinoid type) with controls. Gabapentinoid prescription exposure was categorised as follows: (a) current (overlapping with fracture date); (b) recent (ending 1-60 days pre-fracture); and (c) remote (ending > 60 days pre-fracture). Conditional logistic regression models determined ORs with 95% CIs for fractures with current or recent vs. remote gabapentinoid use, adjusting for confounders.

RESULTS

A total of 2485 cases (mean age 63.0 years; 79.4% female) and 12,244 controls (mean age 62.7 years; 79.6% female) were included. Of cases: 1512 received gabapentin, 910 pregabalin, and 63 both drugs; 65.6% were remote, 5.5% recent, and 28.9% current users. In adjusted models, current gabapentinoid use had an increased risk of fracture (OR vs. remote: 1.36 [95% CI 1.22, 1.51]). Similar associations were seen with gabapentin (OR 1.38 [1.19, 1.60]) and pregabalin (OR 1.40 [1.18, 1.66]). Similar or higher levels of association were seen for all gabapentin/pregabalin doses except moderate/very high dose gabapentin. Associations were strongest in those starting gabapentinoids more recently.

CONCLUSIONS

Our study suggests a modest association between current gabapentinoid use and fractures in patients with IA, after accounting for measured and time-invariant unmeasured confounding. Whilst other unmeasured confounding remains possible, given the absence of evidence for gabapentinoid efficacy in patients with IA who are particularly vulnerable to fractures, this highlights a need for efforts to deliver safer gabapentinoid prescribing in this population.