Does clinical data capture modifiable midlife risk factors for Alzheimer’s disease?

Background The 2020 Lancet Dementia Commission Report highlighted 5 modifiable midlife risk factors for dementia across multiple datasets: hearing loss, traumatic brain injury (TBI), hypertension, alcohol use, and obesity. We tested associations of these risk factors with Alzheimer’s disease (AD) 10 or more years later within clinical care data from a large health system. Method We used University of Pittsburgh Medical Center electronic health record data including diagnoses, demographics, social history, and medications from inpatient, outpatient, and emergency visits. Cases were aged 50‐75 at the time of a first ICD‐9 or 10‐based diagnosis of AD between 2016‐2020. Controls were age‐matched to cases 4:1, selected with replacement; they were free from neurodevelopmental, neurodegenerative, or demyelinating disorders, and had a visit from 2016‐2020. Cases and controls had at least one visit 10+ years prior (“early visit”). Risk factors from the early visit were defined as follows: hearing loss and TBI were based on diagnoses; obesity and hypertension were based on diagnoses or 2 measures of body mass index ≥ 30 or systolic blood pressure (BP) ≥ 140 and diastolic BP ≥ 90, respectively; alcohol use was based on diagnoses or self‐reported use ≥ 21 drinks/week. We tested associations of risk factors with AD using logistic regression models; results are presented as unadjusted odds ratios [95% confidence intervals]. Result Participants had a mean age of 67 in 2016‐2020; N=2382 cases (56% female, 80% white, 19% Black) and N=9528 controls (58% female, 88% white, 10% Black). All midlife risk factors were associated with significantly increased odds of AD. In order of largest to smallest effect size: TBI, 4.5 [4.1, 5.0]; alcohol use, 3.5 [3.0, 4.0]; hypertension 2.6 [2.4, 2.9]. hearing loss, 1.9 [1.7, 2.1]; and obesity, 1.2 [1.1, 1.4]. Narrower definitions of alcohol use and TBI only slightly altered their respective effect sizes. Conclusion In agreement with the Lancet Report, we find that these modifiable factors in late‐early to midlife are associated with AD. Information about these risk factors is already captured in clinical electronic health records and should be used to guide dementia risk reduction beginning early in midlife.