Caroline Madrigal1, Christopher W Halladay1, Kevin McConeghy1, Natalie A Correa2, Tess E K Cersonsky2, Daniel Strauss2, Stefan Gravenstein3, Richard W Besdine4, Thomas P O'Toole3, James L Rudolph5. 1. Center of Innovation in Long Term Services and Supports, Providence VA Medical Center, Providence, RI, USA; Brown School of Public Health, Providence, RI, USA. 2. Warren Alpert Medical School of Brown University, Providence, RI, USA. 3. Center of Innovation in Long Term Services and Supports, Providence VA Medical Center, Providence, RI, USA; Brown School of Public Health, Providence, RI, USA; Warren Alpert Medical School of Brown University, Providence, RI, USA. 4. Brown School of Public Health, Providence, RI, USA; Warren Alpert Medical School of Brown University, Providence, RI, USA. 5. Center of Innovation in Long Term Services and Supports, Providence VA Medical Center, Providence, RI, USA; Brown School of Public Health, Providence, RI, USA; Warren Alpert Medical School of Brown University, Providence, RI, USA. Electronic address: james.rudolph@va.gov.
Abstract
OBJECTIVES: Older veterans prefer to remain in their homes and communities as long as possible. Although targeted delivery of home- and community-based services for veterans might delay long-term care placement, often, access to these services is inconsistently organized or delayed. To aid in early recognition of veterans at high risk for long-term care placement or death, we developed and validated a predictive algorithm, "Choose Home." DESIGN: A retrospective observational cohort analysis was used. SETTING AND PARTICIPANTS: Two cohorts of Veterans Health Administration (VHA; a large integrated health care system) users were assembled: Derivation (4.6 million) and Confirmation (4.7 million). The Derivation Cohort included Veterans Administration users from fiscal year 2013; the Confirmation Cohort included Veterans Administration users from fiscal year 2014. METHODS: A total of 148 predictor variables, including demographics, comorbidities, and utilizations were selected using logistic regression to predict placement in a long-term care facility for >90 days or death within 2 years. RESULTS: Veterans were predominantly male [92.8% (Derivation), 92.5% (Confirmation)] and older [61.7±15.5 (Derivation), 61.5±15.6 years (Confirmation)], with a high prevalence of comorbid conditions. Between the Derivation and Confirmation Cohorts, the areas under the receiver operating characteristic curves were found to be 0.80 [95% confidence interval (CI) 0.799, 0.802] and 0.80 (95% CI 0.800, 0.802), respectively, indicating good discrimination for determining at-risk veterans. CONCLUSIONS AND IMPLICATIONS: We created a predictive algorithm that identifies veterans at highest risk for long-term institutionalization or death. This algorithm provides clinicians with information that can proactively inform clinical decision making and care coordination. This study provides the groundwork for future investigations on how home- and community-based services can target older adults at highest risk to extend time in their communities. Published by Elsevier Inc.
OBJECTIVES: Older veterans prefer to remain in their homes and communities as long as possible. Although targeted delivery of home- and community-based services for veterans might delay long-term care placement, often, access to these services is inconsistently organized or delayed. To aid in early recognition of veterans at high risk for long-term care placement or death, we developed and validated a predictive algorithm, "Choose Home." DESIGN: A retrospective observational cohort analysis was used. SETTING AND PARTICIPANTS: Two cohorts of Veterans Health Administration (VHA; a large integrated health care system) users were assembled: Derivation (4.6 million) and Confirmation (4.7 million). The Derivation Cohort included Veterans Administration users from fiscal year 2013; the Confirmation Cohort included Veterans Administration users from fiscal year 2014. METHODS: A total of 148 predictor variables, including demographics, comorbidities, and utilizations were selected using logistic regression to predict placement in a long-term care facility for >90 days or death within 2 years. RESULTS: Veterans were predominantly male [92.8% (Derivation), 92.5% (Confirmation)] and older [61.7±15.5 (Derivation), 61.5±15.6 years (Confirmation)], with a high prevalence of comorbid conditions. Between the Derivation and Confirmation Cohorts, the areas under the receiver operating characteristic curves were found to be 0.80 [95% confidence interval (CI) 0.799, 0.802] and 0.80 (95% CI 0.800, 0.802), respectively, indicating good discrimination for determining at-risk veterans. CONCLUSIONS AND IMPLICATIONS: We created a predictive algorithm that identifies veterans at highest risk for long-term institutionalization or death. This algorithm provides clinicians with information that can proactively inform clinical decision making and care coordination. This study provides the groundwork for future investigations on how home- and community-based services can target older adults at highest risk to extend time in their communities. Published by Elsevier Inc.