(105) Impact of the COVID-19 pandemic on initiation of anti-diabetic drug consumption pattern in Sweden - An interrupted time series study

Publication Number: 442

Background: Despite the availability of effective anti-diabetic drugs with different blood-glucose-lowering mechanisms, treatment of patients with diabetes mellitus (DM) became a challenge during the coronavirus disease 2019 (COVID-19) pandemic. There are very few studies examining the association between COVID-19 and antidiabetic drug consumption.

Objectives: To examine the pattern of antidiabetic drug initiation two years before and two years after the COVID-19 pandemic in Sweden. We examine the trend of some particular antidiabetic subclasses as well.

Methods: Interrupted time series analyses were carried out on individual-level data for the whole population of Sweden, based on the SCIFI-PEARL (Swedish Covid-19 Investigation for Future Insights – a Population Epidemiology Approach using Register Linkage) database. Data on the first dispensing of all non-insulin antidiabetic drug classes (ATC A10B) from March 2018 to December 2022 were analyzed, using a one-year washout period. An ARIMA model was used to check the trends of antidiabetics before and after the pandemic and possible immediate or late trend changes. Equation y= α + β1T + β2X+ β3XT+ε was used in the model where y = outcome variable, α = intercept, β = coefficients (β1= pre-intervention trend, β2=level change following the intervention, β3= post-intervention trend, β1+ β3=post intervention slope), T= time, X = study phase, XT= time after the interruption, and ε = error or residual. We calculated cumulative incidence based on a monthly number of patients dispensed.

Results: We included 167,889 people initiated on non-insulin anti-diabetic drugs during the period. A statistically significant immediate decrease in the initiation of drug consumption at the start of the pandemic was found for any type of anti-diabetics (β2=0.087, P=0.01). Immediate decreases for sodium-glucose co-transporter 2 (SGLT2) (β2=0.033, P=0.003), and dipeptidyl peptidase 4 (DPP4) inhibitors (β2=0.087, P=0.01), but not biguanides, were observed. For SGLT2, the immediate decrease was significant for both sexes and age subgroups. In patients ≥65 years, the change in trend slope after versus before the pandemic was significantly positive for biguanides (β3=0.025, P=0.021) and SGLT2 (β3=0.017, P=0.0009) but not for the DPP4 inhibitors group (β3=0.002, P=0.213).

Conclusions: The pandemic appears to have had an immediate negative effect on the initiation of anti-diabetic drugs in Sweden, with some evidence of a reactive positive trend in initiation later during the pandemic, particularly in anti-diabetic drug users at age 65 years or older.