Risk Prediction Model for Late Life Depression: Development and Validation on Three Large European Datasets

Area of Science:

• Gerontology
• Psychiatry
• Epidemiology

Background:

• Risk prediction models (RPMs) are crucial for disease prevention but underutilized in psychiatry.
• Late-life depression is a leading cause of disability and functional decline in older adults.
• A significant gap exists in validated RPMs for predicting late-life depression.

Purpose of the Study:

• To introduce DRAT-up, the first prospective risk prediction model for identifying late-life depression in community-dwelling individuals aged 60-75.
• To validate the DRAT-up model's performance across diverse European cohorts.

Main Methods:

• DRAT-up was developed by appraising literature, extracting risk estimates, and integrating them into model parameters.
• A validation study was conducted on three European cohorts: ELSA, Invecchiare nel Chianti, and TILDA.
• The model was assessed for accuracy, Brier scores, and area under the curve (AUC), including sensitivity analyses for missing values.

Main Results:

• DRAT-up demonstrated accurate risk estimation across all three validation cohorts.
• Brier scores ranged from 0.041 to 0.133, and AUC values ranged from 0.736 to 0.768.
• Sensitivity analyses confirmed the model's robustness to missing data, with minimal impact on Brier scores and AUC.

Conclusions:

• DRAT-up is a validated, prospective risk prediction model for late-life depression.
• The tool is accurate, robust to missing values, and suitable for clinical application and policy development.
• DRAT-up addresses a critical need for depression risk assessment in older adults, supporting mental health initiatives.