Baseline HbA1c and dementia risk at SGLT2i initiation in type 2 diabetes

Abstract Aim Type 2 diabetes (T2DM) is a major modifiable risk factor for dementia, yet the optimal timing of sodium–glucose cotransporter‐2 inhibitor (SGLT2i) initiation for cognitive protection remains unclear. HbA1c is routinely measured, but its role in predicting dementia risk at the time of SGLT2i initiation is unknown. Methods We performed a retrospective cohort study using the Global Collaborative Network (>150 health systems worldwide). Adults with T2DM initiating SGLT2i between 2005 and 2025 were included, applying a 456‐day washout to reduce protopathic and immortal time bias. Patients were stratified by baseline HbA1c (<7.0%, 7.0%–8.9%, ≥9.0%) and matched 1:1 by propensity scores for demographics, comorbidities, medications and utilization. The primary outcome was incident all‐cause dementia; secondary endpoints included dementia subtypes, major adverse cardiovascular events (MACE), major adverse kidney events (MAKE) and osteoporotic fracture as a negative control. Results Among matched cohorts (median follow‐up 4.2 years), dementia risk increased with higher HbA1c: HR 1.32 (7.0%–8.9%) and HR 1.68 (≥9.0%) versus <7.0%. These associations were consistent but attenuated in propensity‐score–matched models (HR 1.08 and 1.28). Associations were strongest for vascular dementia and consistent across sensitivity analyses and cause‐specific competing‐risk Cox models. Conclusions Baseline HbA1c at SGLT2i initiation predicts dementia risk. Delaying therapy until HbA1c ≥9% may forfeit protection, whereas earlier use may safeguard brain, heart and kidney health.