Graduate assistant University of Florida Gainesville, United States
Background: Cumulative evidence has shown promising effects of sodium-glucose cotransporter 2 (SGLT2) inhibitors on decreasing the risk of dementia, especially in people with type 2 diabetes (T2D); nevertheless, it is unclear which population subgroups will benefit the most.
Objectives: To compare the risk of dementia with initiation of an SGLT2 inhibitor to initiation of a sulfonylurea and explore its potential heterogeneous treatment effects among people with T2D.
Methods: Using electronic health records data from the OneFlorida+ Clinical Research Network, we identified individuals with T2D (aged ≥ 50 years) who initiated SGLT2 inhibitors or sulfonylureas from Jan 1, 2016 - Jan 31, 2022. The study outcome was incident dementia. We applied a causal machine learning approach – doubly robust learning – to estimate risk difference (RD) and 95% CI in the overall cohort and subgroups identified using a single decision tree model.
Results: Among 35,458 individuals with T2D, 1.8% in the SGLT2 inhibitor group and 4.7% in the sulfonylureas group developed dementia over a 3.2-year follow-up. In the overall cohort, SGLT2 inhibitors were associated with a lower risk of dementia (RD, -2.5%; 95%CI, -3.0% to -2.1%) compared to sulfonylureas. A single decision tree model identified two important variables (Hispanic ethnicity and chronic kidney disease) to define four subgroups in which RD of dementia ranged from -4.3%(-5.4 to -3.1) to -0.9% (-1.9 to 0.1). The findings were consistent across a range of sensitivity analyses.
Conclusions: SGLT2 inhibitors showed a lower risk of dementia in people with T2D, with significant variability across subgroups. Data-driven subgroup analyses may guide personalized treatment for diabetes care.
