Type 2 diabetes and age-related cognitive decline over 40 years in Danish men–A cohort study based on the Danish Aging and Cognition (DanACo) cohort

Abstract

Aim The extant literature on type 2 diabetes and cognitive decline is based on short cognitive follow-ups and assessments of baseline cognitive ability after diagnosis. The objective was to investigate the influence of type 2 diabetes on cognitive decline over a period of on average 44 years. Materials and methods This cohort study included 5,147 men from the Danish Aging and Cognition cohort consisting of a late mid-life (mean age 64.2 years) follow-up of men with intelligence test scores (IQ) available from statutory conscription board examinations in young adulthood (mean age 20.4 years). Follow-up included re-administration of the conscription board intelligence test and a comprehensive questionnaire. Exposure was self-reported but register-based type 2 diabetes and duration of disease were also calculated. Cognitive decline was defined as both IQ change (baseline-follow-up) and significant IQ decline based on the reliable change index (cut-off: 13.2 IQ-points). Associations were analyzed in linear and logistic regression models. Results Men having type 2 diabetes had a 1.81 IQ points (95%CI:1.14,2.49) larger decline compared to men without diabetes when adjusting for baseline IQ, years of education, follow-up age, retest interval, depression, and smoking status. Moreover, type 2 diabetes was associated with 1.42 times higher odds of a significant IQ decline and longer duration was associated with a larger, though not statistically significant, decline. The participation rate was 13.4%, and the participants were healthier and more well-educated than non-participants. To account for potential selection bias, inverse probability weights (IPW) were calculated based on baseline characteristics. The analyses applying these weights yielded similar estimates. Conclusion Type 2 diabetes was associated with modestly greater cognitive decline and higher odds of a statistically significant (>13.2 IQ points) and clinically relevant decline. Finally, the alignment between main and IPW results indicates the findings are robust and likely generalizable.