Sudden Increases in Health Utilization as a Predictor of Dementia Onset

Abstract Background Dementia remains a persistent clinical challenge, underscoring the importance of early prediction and management. In this study, we investigate whether trajectories of health utilization, such as health expenditure and adjusted patient-days, can serve as predictors of dementia incidence. Health utilization is a surrogate variable for health status and is easily collectable in digital health perspectives. We examined the relationship between changes in health utilization and dementia incidence. Methods Using Korean National Health Insurance Claims data, we identified newly diagnosed dementia incidence cases (N = 17,460) since 2018 based on ICD-10 codes(F00-F03, F05.1, G30-G31) and benefit codes(V800-V810). General population controls (N = 88,137) were matched 1:5 by sex and age in a nested case-control design. We calculated four types of momentum based on the changes in health utilization-simple and relative momentum, MACD, and RSI. A conditional logistic regression model was used to assess the association with dementia incidence. Results Approximately 9 months before diagnosis, cases showed a sharp increase in average health expenditure (from €229 to €446), while controls remained stable (€226-€271). For health expenditure, significant associations were observed for the MACD during the periods of 1-2 and 9-24 months prior (HR: 0.94, 95% CI: 0.94-0.95). Simple momentum was significantly associated with incidence when calculated at 9 and 14 months prior (HR: 1.01, 95% CI: 1.00-1.01). For adjusted patient-days, RSI showed a significant association when comparing the recent 4-month changes to the preceding 9-24 month period (HR: 1.07, 95% CI: 1.06-1.08), and the 12-14 vs. 12-24 month prior (HR: 1.05, 95% CI: 1.04-1.07). Conclusions Momentum-based health utilization indicators are significantly associated with dementia incidence. These indicators may help early prediction for dementia onset with digital health applications. Key messages • Four momentum-based indicators of health utilization can be predictors of dementia incidence. • These indicators can help to detect dementia incidence with digital health application.