Abstract
BACKGROUND: Metformin is an antidiabetic medication that is routinely prescribed. Recent reports suggest a dose-response disease-modifying effect on osteoarthritis (OA) of the knee with patients on metformin having a lower incidence of total knee arthroplasty. However, it is unknown whether this effect translates to OA of the shoulder. Therefore, the purpose of this study is to investigate the effect of metformin on the likelihood of having shoulder arthroplasty in patients with shoulder OA. METHODS: The TriNetX database was used to identify a cohort of patients with shoulder OA. Two groups were formed. The first group was comprised of patients who were not prescribed metformin. The second group was comprised of patients who were prescribed metformin. Propensity score matching was performed to control for multiple cofounding variables including hemoglobin A1C levels. Both matched cohorts consisted of 26,327 patients each. The outcomes of interest were the incidence of shoulder arthroplasty up to 10 years after the diagnosis of shoulder OA, as well as serum inflammatory marker levels (C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and Cortisol). RESULTS: A total of 1,245 (4.729% risk) patients in the metformin group underwent shoulder arthroplasty compared to 962 (3.654% risk) patients in the nonmetformin group. The metformin group had averages of 30.703 mg/L ± 57.155 mg/L for CRP, 33.692 mm/hour ± 27.857 mm/hour for ESR and 13.224 μg/dL ± 8.450 μg/dL for cortisol. The nonmetformin group had averages of 36.577 mg/L ± 60.627 mg/L for CRP, 5.574 mm/hour ± 29.719 mm/hour for ESR and 13.326 μg/dL ± 8.571 μg/dL for cortisol. The metformin group had a significantly increased risk of arthroplasty of 1.075% [95% confidence interval (0.733%, 1.417%), P < .0001 and odds ratio = 1.309 [95% confidence interval (1.201-1.426)] compared to the nonmetformin group and had a longer follow-up period (P < .0001). Mean ESR and CRP inflammatory markers levels in the metformin group were significantly lower than the nonmetformin group (both P < .0001) However, there was a nonsignificant decrease in average cortisol levels (P = .779) in the metformin group when compared to the nonmetformin group. DISCUSSION/CONCLUSION: Patients with shoulder OA who are prescribed metformin are more likely to undergo shoulder arthroplasty than those not taking metformin, despite having significant decreases in inflammatory marker levels. These findings stand in contrast to previous data which suggest a potential dose-response disease-modifying effect on OA of the knee. Additional research is needed to better elucidate the mechanisms of metformin on OA.