AbstractBackgroundPreclinical evidence suggests that sodium‐glucose cotransporter‐2 (SGLT2) inhibitors may mitigate dementia pathology; however, current clinical evidence is limited. The primary aim is to compare dementia risk between SGLT2 inhibitors and dipeptidyl peptidase‐4 (DPP4) inhibitors, two common second‐line glucose‐lowering medications.MethodsThis population‐based retrospective cohort study utilized administrative databases housed at ICES (https://www.ices.on.ca/), and residents of Ontario, Canada with diabetes aged ≥66 years were included. Three different but overlapping new‐user cohorts were created. The primary comparison was SGLT2 inhibitors versus DPP4 inhibitors (entry period: January 1, 2016, to March 31, 2020). The secondary comparisons were SGLT2 inhibitors versus sulfonylureas (January 1, 2016, to March 31, 2020) and DPP4 inhibitors versus sulfonylureas (January 1, 2011, to March 31, 2020), because sulfonylureas are older second‐line therapies still used contemporaneously. Individuals were followed for maximum 4.5 years. The primary outcome was time to incident dementia. Exploratory outcomes were time to incident Parkinson’s disease diagnosis, stroke, and long‐term care admission, to examine outcomes related to both glucose‐lowering medications and dementia risk. Propensity‐score weighted Fine‐Gray subdistribution hazard models with all‐cause mortality as competing risk were used to obtain adjusted hazard ratios (aHR) and confidence intervals (CI).ResultsAmong 119,432 individuals, SGLT2 inhibitors compared with DPP4 inhibitors were associated with lower risks of dementia (14.5/1000 person‐years, aHR [95% CI] = 0.59 [0.52‐0.67]), stroke (10.8/1000 person‐years, aHR [95% CI] = 0.87 [0.77‐0.99]), and long‐term care admission (8.6/1000 person‐years, aHR [95% CI] = 0.43 [0.36‐0.52]). SGLT2 inhibitors compared with sulfonylureas were associated with lower risks of dementia (11.0/1000 person‐years, aHR [95% CI] = 0.57 [0.51‐0.63]), Parkinson’s disease (1.7/1000 person‐years, aHR [95% CI] = 0.73 [0.56‐0.97]), stroke (9.8/1000 person‐years, aHR [95% CI] = 0.75 [0.63‐0.84]), and long‐term care admission (5.6/1000 person‐years, aHR [95% CI] = 0.51 [0.44‐0.59]) among 89,114 individuals. DPP4 inhibitors compared with sulfonylureas were associated with lower risks of dementia (16.5/1000 person‐years, aHR [95% CI] = 0.78 [0.74‐0.82]), stroke (10.4/1000 person‐years, aHR [95% CI] = 0.85 [0.79‐0.92]), and long‐term care admission (8.7/1000 person‐years, aHR [95% CI] = 0.82 [0.76‐0.89]) among 152,559 individuals.ConclusionSGLT2 inhibitor use was associated with lower dementia risk, and with preservation of functional independence, compared with DPP4 inhibitor or sulfonylurea use. Randomized controlled trials are needed to replicate the findings.