Time-varying association of acute and post-acute COVID-19 with new-onset diabetes mellitus among hospitalized and non-hospitalized patients

Introduction Previous studies have shown disruption of glycometabolic control and new diabetes mellitus (DM) diagnosis among patients with COVID-19. It is still unclear how the association of COVID-19 and new-onset DM may be modified by disease severity or vary over time, during acute and post-acute phases. Research design and methods In this retrospective matched cohort study, 157 936 patients with COVID-19 (aged ≥25 years, diagnosis date between March 01, 2020 and August 31, 2021) were compared with individuals without COVID-19, separately for non-hospitalized, hospitalized, and severe hospitalized patients. Stratified Cox proportional hazards models, with changing baseline time (starting at the date of COVID-19 diagnosis, and at 1, 2, 3, and 4 months afterwards), were used to evaluate the occurrence of new DM in relation to COVID-19 infection in different time frames - from each landmark date until end of study. Results During mean follow-up time of 10.9 months, there were 1145 (0.72%) new diagnoses of DM compared with 1013 (0.64%) in the individuals without COVID-19 (p=0.004). Non-hospitalized patients with COVID-19 were not at higher risk of new DM neither during the acute phase nor afterward. Hospitalized patients with COVID-19 had a higher risk of developing DM, with the highest risk among severe hospitalized patients. This risk among hospitalized patients was highest in the acute phase (HR 2.47 (95% CI 1.86 to 3.29)), attenuated over time, but remained significant at 4-month landmark analysis (HR 1.60 (95% CI 1.12 to 2.29)). Conclusions Acute and post-acute COVID-19 were associated with new DM only among hospitalized patients, with the highest risk among those hospitalized with severe disease. Those patients should be followed and monitored post-discharge for new DM. Patients who were not hospitalized did not have higher risk of new-onset DM.