Derivation and Validation of a Predictive Algorithm for Long-Term Care Admission or Death

Area of Science:

• Gerontology
• Health Services Research
• Predictive Analytics

Background:

• Older veterans prefer aging in place within their homes and communities.
• Access to home- and community-based services (HCBS) is crucial but often inconsistently organized or delayed for veterans.
• Early identification of veterans at high risk for long-term care placement or mortality is needed.

Purpose of the Study:

• To develop and validate a predictive algorithm to identify veterans at high risk for long-term institutionalization or death.
• To provide clinicians with a tool for proactive clinical decision-making and care coordination.

Main Methods:

• A retrospective observational cohort analysis was conducted using two large cohorts (Derivation: 4.6 million; Confirmation: 4.7 million) of Veterans Health Administration (VHA) users.
• 148 predictor variables (demographics, comorbidities, utilization) were selected via logistic regression.
• The algorithm predicted placement in long-term care (>90 days) or death within two years.

Main Results:

• The algorithm demonstrated good discrimination, with areas under the receiver operating characteristic curves of 0.80 in both the Derivation and Confirmation cohorts.
• Veterans in the study were predominantly male (over 92%) and older (mean age ~61.6 years) with a high prevalence of comorbid conditions.

Conclusions:

• The validated "Choose Home" algorithm effectively identifies veterans at high risk for long-term institutionalization or death.
• This predictive tool can inform clinical decision-making and care coordination, potentially enabling targeted interventions to support aging in place.
• The study lays the foundation for future research on optimizing HCBS delivery for high-risk older adults.