Insulin Use in Diabetes Association With Cognitive Impairment and Dementia Incidence Publisher Pubmed



Hajihosseini S ; Kurd DM ; Nouroozi F ; Haghighi S ; Dini N ; Ghobadi Y ; Tehrani N ; Shahrokhi M ; Belbasi M ; Alizadeh H ; Salimi O ; Hadimaleki S ; Yarahmadi D ; Deravi N
Source: BMC Pharmacology and Toxicology Published:2025

Abstract

Background: Dementia has become an increasingly critical public health dilemma, and type 2 diabetes (T2D) is an established risk factor, yet it remains unclear whether insulin therapy alone contributes to future dementia risk. Methods: We utilized data from the Health and Retirement Study (1996–2020), an ongoing study of a representative cohort of U.S. adults aged 50 or older. Baseline was defined as the first wave of data collection, where indications of diabetes pharmacotherapy were included in study procedures. We then constructed two analytic cohorts: one with incident dementia (cognition score ≤ 6) and another with any cognitive impairment (cognition score ≤ 11). Cumulative incidence functions (CIFs) were calculated, which take death into account as a competing event, and cause-specific Cox models were implemented. To address bias, we applied stabilized inverse probability of treatment and censoring weights (IPTW × IPCW) with truncation of the 1st/99th percentiles, and uncertainty was assessed using subject-level bootstrap. Weighted linear mixed-effects models examined cognitive trajectories. Results: The analytic sample was made up of 6,671 adults with diabetes (2,359 insulin users; 4,312 non-insulin users). Following weighting, the groups were balanced for baseline age, BMI, and cognition (median cognition 15 [IQR 12–18]). CIFs for dementia separated early and remained apart; insulin users had a greater cumulative risk across horizons, with a 1.4-fold excess at 10 years. Cause-specific Cox models supported a higher dementia hazard in insulin users (HR ≈ 1.26; 95% CI 1.08–1.48; p = 0.004). Associations with cognitive impairment were generally weaker and not all significant. The cognitive decline in the mixed-effects models was evident for both groups across time, although the time×group interaction was not significant, suggesting that the average group slopes were similar. Conclusion: Although a similar mean cognitive decline was observed in both T2D groups, insulin exposure was associated with an increased risk of dementia. This calls for caution in the consideration of insulin use among the older population due to its association with dementia, and the need for future studies to consider HbA1c, duration of diabetes, and rates of hypoglycemia. Clinical trial number: Not applicable. © The Author(s) 2025.