Association of SARS-CoV-2 infection with incident diabetes among U.S. Veterans in a prospective longitudinal cohort.

Abstract

To assess care-seeking patterns and incident diabetes risk following SARS-CoV-2 infection in a prospective longitudinal cohort. We used data from the Veterans Health Administration (VHA)-based, prospective longitudinal study Epidemiology, Immunology, and Clinical Characteristics of COVID-19 (EPIC3) and electronic health records from participants with and without a history of SARS-CoV-2 infection who were free from diabetes at baseline and enrolled between June 2020 and September 2022 (n = 1,212); participants were followed prospectively for a median of approximately 4 years. We fit Cox proportional hazard models to examine associations of prior SARS-CoV-2 infection with incident diabetes. Models were adjusted for age, sex, race, smoking status, BMI, education, and comorbidities, as well as number of laboratory test days in the year prior. Men comprised 79.4% of the cohort. Median age was 50.1 (SARS-CoV-2-positive) and 57.2 (SARS-CoV-2-negative) years. After accounting for time-varying SARS-CoV-2 infection status, compared to participants with a negative test, participants with a positive test had fewer days/year with clinic visits (18.2 vs. 25.4, p < 0.001), laboratory tests of any kind (3.1 vs. 4.3, p < 0.001), and glucose tests (1.8 vs. 2.8, p < 0.001) post-enrollment; however, the number of days/year on which they had HbA1c tests was not significantly different (0.7 vs. 0.7, p = 0.094); diabetes incidence was 16.1 and 21.2 per 1,000 person-years in SARS-CoV-2 positive and negative groups, respectively. SARS-CoV-2 was not associated with adjusted diabetes-free survival overall, in inpatients or in outpatients (adjusted HRs: 0.82 [95% CI, 0.51-1.35], 1.03 [0.27-3.96], and 0.79 [0.44-1.39], respectively). Although SARS-CoV-2-positive participants used healthcare less often than those without, HbA1c testing rates were similar. We did not replicate prior reports of higher diabetes risk after SARS-CoV-2, although small sample size may have reduced power to detect modest associations. NCT: NCT05764083.