Exploratory 6-year prediction of LTCI certification using KCL data: A decision tree analysis in a retrospective...

Area of Science:

• Gerontology
• Public Health
• Biostatistics

Background:

• Aging societies require accurate prediction of Long-Term Care Insurance (LTCI) needs.
• Existing predictive models have limitations in examining complex risk factor interactions and prediction span.
• There is a need for a robust model to stratify risk for LTCI services among community-dwelling older adults.

Purpose of the Study:

• To develop and evaluate a predictive model for Long-Term Care Insurance (LTCI) certification using Kihon Checklist data.
• To assess the model's ability to predict LTCI needs over a 6-year follow-up period.
• To identify key predictive factors for LTCI certification in older adults.

Main Methods:

• Retrospective cohort, longitudinal, and observational study design.
• Utilized data from 3263 community-dwelling individuals aged 65 and older.
• Employed decision tree analysis with Kihon Checklist domains, age, and gender as predictors, and LTCI certification status as the outcome.

Main Results:

• The decision tree model identified age, gender, and depression risk as significant predictors.
• The model achieved high negative predictive value (97.1%) and moderate specificity (79.4%).
• Sensitivity (52.2%) and positive predictive value (11.1%) were limited, indicating better performance in excluding future LTCI users.

Conclusions:

• The developed model shows potential for preliminary, population-level risk stratification for LTCI over a 6-year horizon.
• The model is more effective at identifying individuals unlikely to require LTCI services than predicting future users.
• Further refinement and validation are necessary to improve predictive performance for clinical and policy decision-making.