Observational Study of Metformin and Risk of Mortality in Patients Hospitalized with Covid-19

Abstract

Background Type 2 diabetes (T2DM) and obesity are significant risks for mortality in Covid19. Metformin has been hypothesized as a treatment for COVID19. Metformin has sex specific immunomodulatory effects which may elucidate treatment mechanisms in COVID-19. In this study we sought to identify whether metformin reduced mortality from Covid19 and if sex specific interactions exist. Methods De-identified claims data from UnitedHealth were used to identify persons with at least 6 months continuous coverage who were hospitalized with Covid-19. Persons in the metformin group had at least 90 days of metformin claims in the 12 months before hospitalization. Unadjusted and multivariate models were conducted to assess risk of mortality based on metformin as a home medication in individuals with T2DM and obesity, controlling for pre-morbid conditions, medications, demographics, and state. Heterogeneity of effect was assessed by sex. Results 6,256 persons were included; 52.8% female; mean age 75 years. Metformin was associated with decreased mortality in women by logistic regression, OR 0.792 (0.640, 0.979); mixed effects OR 0.780 (0.631, 0.965); Cox proportional-hazards: HR 0.785 (0.650, 0.951); and propensity matching, OR of 0.759 (0.601, 0.960). TNF-alpha inhibitors were associated with decreased mortality in the 38 persons taking them, by propensity matching, OR 0.19 (0.0378, 0.983). Conclusions Metformin was significantly associated with reduced mortality in women with obesity or T2DM in observational analyses of claims data from individuals hospitalized with Covid-19. This sex-specific finding is consistent with metformin reducing TNF-alpha in females over males, and suggests that metformin conveys protection in Covid-19 through TNF-alpha effects. Prospective studies are needed to understand mechanism and causality.

Figure 1:. Study diagram detailing selection of patients in United Healthcare Covid-19 Database

This is a flow diagram representing how patients were selected into the analysis.

Figure 2:. Kaplan Meier curve by metformin use, with propensity matching in persons with type 2 diabetes and obesity hospitalized for Covid-19.*

This is a Kaplan-Meier survival curve. The top panel compares metformin use to no metformin use in males only. The bottom panel compares metformin use to no metformin use in females onluy. In both panels, the blue line represents no metformin use and the red line represent metformin use. The Y axis is survival, the X axis is hospital days. *With variables selected by LASSO with AIC, matching caliper 0.2 *With variables selected by LASSO with AIC, matching caliper 0.2

Figure 3.. Subgroup analysis of mortality associated with metformin use vs no metformin use in hospitalizations for Covid-19, among women and men with type 2 diabetes or obesity Bars represent 95% CI’s.

Bar graph comparing survival among women and among men, comparing those without metformin to those with metformin. The left two columns are women, without and with metformin, analyzed by cox proportional hazards. The 3^rd and 4^th columns (blue) are men, without and with metformin, analyzed by Cox proportional-hazards. The 5^th and 6^th columns (light orange), are women, without and with metformin, analyzed by propensity matching. The 7^th and 8^th columns (blue) are men, without and with metformin, analyzed by propensity matching.