(B-095) Antidiabetic drug utilisation after COVID-19 infection amongst patients with diabetes mellitus type 2: a retrospective cohort study from the UK

Background: Coronavirus disease 2019 (COVID-19) might influence the clinical course of type 2 diabetes mellitus (T2DM) and raise the need for antidiabetic treatment. However, this association remains unclear to date.

Objectives: We aim to investigate the effect of COVID-19 infection on antidiabetic drug utilisation for up to 1-year in patients with T2DM.

Methods: This was a retrospective cohort study, including adult patients (≥18 years old) with T2DM from the Clinical Practice Research Datalink GOLD database between 01.09.2020 and 31.03.2021. T2DM patients with a COVID-19 diagnosis (i.e. index date) were matched with up to four unexposed patients to COVID-19 patients, based on age, sex and general practice. Four antidiabetic drug utilisation categories were defined based on NICE treatment guidelines in UK, as no antidiabetic treatment (category 1), one NIAD (category 2), ≥2 NIADs (category 3), and insulin ± NIAD (category 4). Multinomial logistic regression models estimated the odds ratio (OR) of various utilisation categories, compared with one NIAD, in T2DM patients with COVID-19 versus those without, during the short (0-3 months), medium (4-6 months) and long (7-12 months) terms after the index date. Also, the pattern of antidiabetic drug use (i.e., steps moving between the four categories) was assessed between the two exposure groups. The regression models were adjusted for lifestyle factors, comorbidities and baseline antidiabetic drug utilisation status.

Results: We identified 2457 patients exposed to COVID-19 (mean age 65.5 years, 55.4% men), matched to 7398 unexposed patients. Compared to treatment with one NIAD, COVID-19 infection was associated with more than 2-fold risk of receiving insulin ± NIADs in all follow-up windows after index date (OR 2.1, 95% CI 1.5-2.9 in both short- and mid-term, and OR 2.3, 95% CI 1.6-3.2 in long-term), while it was associated with 30% less risk of no antidiabetic treatment in long-term (OR 0.7, 95% CI 0.5-0.8). More exposed patients than unexposed progressed 1-step in antidiabetic drug use (i.e. initiating/adding NIAD or insulin) in short- (7.1% vs 3.6%, OR 2.1, 95% CI 1.7-2.6), mid- (9.1% vs 6.0%, OR 1.6, 95% CI 1.4-2.0), and long-term (12.7% vs 9.8%, OR 1.4, 95% CI 1.1-1.6), compared with no change in antidiabetic drug utilisation. Noteworthy, COVID-19 infection was also associated with almost all other pattern changes in the 3 follow-up time windows.

Conclusions: We found that COVID-19 infection was associated with insulin use overall, and almost all pattern changes of antidiabetic treatment during the 1-year after diagnosis, including initiating/adding NIAD or insulin. This information can help with better preparedness in the future pandemics of viral infections.