Neurology
Estimating early-onset dementia population attributable fractions in the U.S. population using a counterfactual approach Katherine Giorgio* Katherine Giorgio Giorgio Giorgio Giorgio Giorgio University of Minnesota Twin Cities, School of Public Health, Division of Epidemiology and Community Health, Minneapolis, MN
Introduction
Early-onset dementia (EOD; onset before age 65) affects 3.9 million people globally. While genetic predisposition contributes to EOD risk, there is growing evidence that modifiable risk factors also play an important role. We estimated the proportion of EOD cases in the U.S. which can be attributed to modifiable risk factors.
Methods
Depression, obesity, high cholesterol, diabetes, hypertension, low education, physical inactivity, smoking, alcohol over-consumption, and mild to severe head injury were examined as risk factors. Individual-level data was used to estimate population attributable fractions (PAFs). Data for effect estimates on EOD came from the UK Biobank study. Risk factor prevalence data for the U.S. population aged 45-65 years came from the National Health and Nutrition Examination Survey (NHANES). A Cox regression model for EOD, with age as the timescale and all risk factors, sex, and race as independent variables, was fit in the UK Biobank data. Model estimates were used to estimate PAFs comparing EOD incidence under observed NHANES risk factor distributions to the counterfactual outcome if no one was exposed. CIs were calculated by bootstrapping. A sensitivity analysis calculated PAFs using summary-level data.
Results
Risk factor prevalence is high, with 97% of the population having at least one. Large individual PAFs include head injury (34%), diabetes (20%), and physical inactivity (20%), and results were similar in sensitivity analysis. 74% (95% CI: 65 – 82) of EOD cases are attributable to these ten risk factors combined (43% in sensitivity analysis).
Discussion
Several risk factors may have a larger impact on the proportion of EOD cases than they do on dementia at any age, including head injury, diabetes, and physical inactivity. The combined estimates are similar, however, when using the same methods. These estimates are sensitive to the combination method. EOD incidence may be reduced by interventions for common health issues.