# Development of a Prediction Model for Community-Dwelling Older Adults at Risk of Long-Term Care with Dementia

## TL;DR

This study created a risk-scoring tool to predict which older adults with dementia will need long-term care, using data from Japan's elderly health assessments.

## Contribution

A new predictive model was developed using administrative data to identify older adults at risk of long-term care certification with dementia.

## Key findings

- 143 participants (7.0%) were certified for long-term care with dementia within two years.
- Age, homebound status, cognitive decline, and locomotor decline were key predictors of certification.
- The model showed strong discriminatory ability with an area under the curve of 0.790.

## Abstract

Background: Early detection of modifiable risk factors for long-term care certification with dementia is essential. This study aimed to develop a risk-scoring tool using data from the Kihon Checklist and Questionnaire for the Late-Stage Elderly over a 2-year period to predict long-term care certification with dementia under Japan’s Long-Term Care Insurance system. Methods: Participants included 2041 functionally independent, community-dwelling older adults in Kure City, Japan, as of March 2021. A retrospective cohort study was conducted. Associations between KCL and LSEQ domains and certification for long-term care with dementia were examined using logistic regression. To improve practical use, a score chart was developed to predict certification for long-term care with dementia. Results: Two years after completing the Kihon Checklist and Questionnaire, 143 participants (7.0%) were certified for long-term care with dementia. Factors independently associated with certification for long-term care with to dementia included age, homebound status, cognitive decline, and locomotor decline. The prediction model, developed using these variables, showed excellent discriminatory ability, with an area under the curve of 0.790 (95% confidence interval: 0.754–0.827). Conclusions: We developed an effective predictive model for future long-term care certification with dementia using routinely collected administrative data. This tool may help healthcare providers and health planners identify older adults at increased risk of long-term care certification with dementia.