Predicting mid- and late-life dementia risk in primary care: A prognostic study from a national health screening cohort

Abstract

The objective of this study is to develop and validate dementia risk prediction models for mid- and late-life individuals that are based on accessible variables within a primary care setting. Using the Korean National Health Insurance Service database (2010–2019), we analyzed 6,696,073 individuals aged 40 and older who participated in a national health screening program over a mean follow-up of 8.95 years. Potential predictors were selected based on a literature review and the available data. Dementia cases were identified using claim-based codes and validated by corresponding prescription information. 5-year dementia risk prediction models for mid-life (40–59 years), and late-life (60+ years) stages were developed using the Cox proportional hazards model. Model performance was assessed through discrimination and calibration. Both models included age, sex, body mass index, smoking, alcohol consumption, physical activity, diabetes, hypertension, dyslipidemia, and chronic kidney disease. The models’ AUROCs were 0.764 for mid-life and 0.743 for late-life. The impact of predictors on dementia risk was consistently stronger in mid-life than in late-life stages. Our models showed good calibration with low-risk estimates in mid-life and overall in late-life. These findings underscore the crucial role of managing modifiable risk factors, particularly during mid-life to reduce dementia risk.