Metformin and incident osteoarthritis: Causal insights from a co-twin control study

In a co-twin control design, we aimed to examine whether individuals who initiate metformin treatment have an altered risk of incident diagnosis of osteoarthritis (OA), for up to 11 years after starting the treatment, compared with their non-treated co-twin.

In all twin pairs different for metformin treatment (N=1 261 pairs) and all eligible twins/singletons (N=52 341) aged 35-75 years living in Sweden on December 31st 2005, without prior diagnosis of OA and no prior treatment with metformin, we studied the outcome OA in any joint as diagnosed in specialist care. We used a stratified Cox model in discordant pairs to estimate hazard ratio (HR) and a within-between Weibull shared frailty model to estimate incidence proportions and differences. Models were adjusted for age, sex, education level, body mass index (BMI) and smoking.

Among treated twins, 68 (5.4%) developed OA over the 11-year follow-up, compared to 79 (6.3%) among the non-treated twins (crude HR [95% confidence interval -CI] = 0.86 [0.62-1.19]). Treated twins had higher BMI. Adjusted HR comparing treated with non-treated was 0.60 (95% CI =0.44-0.82). After 10 years, 6.7% (95% CI=4.8-8.6) of treated twins, compared to 9.7% (95% CI=9.3-10.1) of non-treated twins had incident OA, with difference of -3.1% (95% CI=-5.1- (-1.1)).

Starting metformin treatment may be associated with a decreased risk of incident diagnosis of OA in specialist care over an 11-year follow-up period. The 10-year cumulative incidence was ∼3 percentage points lower among metformin initiators than non-initiators.