Predicting Risk of Long-Term Institutionalization Among Community Dwelling Veterans Before the COVID-19 Pandemic

Area of Science:

• Gerontology
• Health Services Research
• Biostatistics

Background:

• Identifying Veterans at risk for long-term institutionalization (LTI) is crucial for resource allocation.
• The Veterans Health Administration (VA) seeks to optimize care delivery and support community residence for its beneficiaries.

Purpose of the Study:

• To develop and validate a predictive model for long-term institutionalization (LTI) risk among Veterans.
• To identify Veterans requiring targeted interventions to prevent nursing home placement.

Main Methods:

• The Predicted Long-term Institutionalization (PLI) risk model was developed using VA and Medicare claims and Minimum Data Set data.
• The model incorporated demographics, diagnoses, prior healthcare utilization, and risk indices for death and frailty.
• Age normalization and differential risk strata thresholds were employed to enhance predictive accuracy and minimize bias.

Main Results:

• The final PLI model achieved a c-statistic of 0.87, identifying 3.5% of Veterans as high risk for LTI.
• High-risk Veterans (PLI-high) exhibited a 13% LTI rate and accounted for 41% of all new LTIs.
• The PLI model demonstrated improved sensitivity (41%) compared to baseline models, with a specificity of 96.6%.

Conclusions:

• The PLI score effectively identifies Veterans at high risk of long-term institutionalization.
• This risk stratification enables targeted assessments and resource allocation to support Veterans' continued community residence.