Background: Type 2 diabetes (T2D) increases dementia risk for undetermined reasons; neurodegenerative and vascular pathways are likely. Dementia risk factors might differ in people with younger (<65 years) versus older (≥65 years) age at T2D onset. We therefore determined predictors of incident dementia by T2D onset age.

Methods: The Fremantle Diabetes Study includes 1296 T2D participants recruited between 1993-96 who undertook a comprehensive baseline assessment. The cohort was linked to the WA hospital morbidity, mental health and mortality databases from 1970 onwards to identify prevalent and incident dementia using ICD diagnosis codes. Cox proportional hazards and Fine and Gray competing risk regressions with age as time-scale were used to identify independent predictors of dementia.

Results: There were 1289 participants after five with prevalent dementia and two with no T2D onset age were excluded. At baseline, they were 64.0±11.3 years old, 48.6% were male with median [IQR] diabetes duration 4.0 [1.0-9.0] years. During 12.7±5.9 years' follow-up, 179 (13.9%) developed incident dementia, 91/931 (9.8%) in those diagnosed before 65 years and 88/358 (24.6%) in those ≥65 years.  For younger onset, predictors of dementia included diabetes duration (HR (95%CI): 1.04 (1.01-1.08) /year), cerebrovascular disease (2.24 (1.17-4.32)), ApoE 4,4 genotype (7.76 (2.63-22.9)), antipsychotic medications (4.42 (1.64-11.9)), and schizophrenia (18.9 (5.27-68.1)). For older onset, dementia was predicted by exercise (0.48 (0.30-0.76)) and peripheral sensory neuropathy (1.74 (1.10-2.75)). Adjusting for the competing risk of death before dementia, older age became a risk factor for dementia (sHR (95%CI): 1.07 (1.04-1.10) /year) whereas in the conventional Cox model it was protective (0.90 (0.85-0.94) /year).

Conclusion: Competing risk may have occurred with people younger at study entry dying during follow-up before dementia diagnosis. There were differences in dementia risk factors between those with younger vs. older onset T2D. Dementia prevention in T2D needs to account for these age-related differences.