Risk Factors for Incident Diabetes After SARS-CoV-2 Infection Across Variant Eras: Findings From the National Clinical Cohort Collaborative (N3C) COVID-19 Enclave

OBJECTIVE SARS-CoV-2 infection has been associated with incident diabetes, but the influence of evolving clinical and population characteristics across the pandemic remains unclear. We sought to identify factors associated with incident diabetes after COVID-19 infection and assess differences across variant eras. RESEARCH DESIGN AND METHODS We conducted a retrospective cohort study, analyzing 528,710 adults with COVID-19 and without preexisting diabetes using data from the National Clinical Cohort Collaborative (NC3) COVID-19 Enclave. Multivariable logistic regression evaluated demographics, comorbidities, variant era, COVID-19 severity, and socioeconomic factors associated with incident diabetes within 1 year after acute COVID-19 infection. RESULTS Increasing age, male sex (odds ratio [OR] 1.32; 95% CI 1.27–1.38), higher BMI, cardiometabolic conditions, and hospitalization for COVID-19 were associated with incident diabetes risk. Risk was higher among Black (OR 1.49; 95% CI 1.41–1.59) and Hispanic individuals (OR 1.65; 95% CI 1.55–1.77) and those living in areas with higher poverty rates (OR 1.23; 95% CI 1.15–1.32). Compared with the ancestral period, risk was lower in Alpha (OR 0.87; 95% CI 0.82–0.92) and Delta eras (OR 0.91; 95% CI 0.85–0.98) but similar during Omicron (OR 1.01; 95% CI 0.93–1.09). The ancestral cohort included more Black and Hispanic individuals and individuals at greater socioeconomic disadvantage, whereas later cohorts were increasingly non-Hispanic White, were less disadvantaged, and had more comorbidities. CONCLUSIONS Incident diabetes risk after COVID-19 infection was associated with cardiometabolic and sociodemographic factors and remained similar in the Omicron and ancestral periods. Shifting sociodemographic characteristics across variant eras suggest that changing health care engagement may influence the populations represented in contemporary electronic health record data.