Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

Health Services Research & Development

Go to the ORD website
Go to the QUERI website
HSRD Conference Logo

2023 HSR&D/QUERI National Conference Abstract

Printable View

4011 — Predicting risk of Long-term Institutionalization: The PLI Score

Lead/Presenter: Kinosian Bruce,  geriatrics and extended care data analysis center
All Authors: Kinosian B (Philadelphia VAMC), Schmitt, S (Palo Alto) Phibbs, C ( Palo Alto) Hartronft, S (12GEC) Intrator, O (Canandaigua VAMC)

Most Veterans prefer to age at home and not end their lives in nursing homes (NH), or in long-term institutions. The VA’s Office of Geriatrics and Extended Care is running a demonstration program, the Reimagining Elder Care in America Program (RECAP), which evaluates needs of Veterans at high risk of long-term institutionalization (LTI) in order to better coordinate care that could help delay or avoid LTI. The objective of this work was to develop the statistical tool that would best identify Veterans at high risk of LTI.

Using combined VA and Medicare data, we used demographics, VA and Medicare diagnoses, health care use, and risk measures for VA users alive at the end of fiscal year (FY) 2017 to predict Veterans who will enter a NH long-term during FY2018-FY2019. Among potential measures evaluated, predictors included demographics, diagnostic clusters, prior annual cost and prior utilization of hospital and nursing home services. Risk indices of death and frailty were added. Separate logistic regression models were estimated for low-risk and elevated-risk strata. Predictors significant in either elevated or low-risk segments were included in both models. High-risk was determined by maximizing sensitivity (share of Veterans using LTI in the subsequent 2 years identified), limited by both numbers needed to screen (?8), and positive predictive value of either death or LTI (>0.5).

Using combined VA and Medicare data and separate estimation on elevated and low-risk tiers improved model performance over prior LTI prediction models. With the final thresholds of 6% LTI risk for the elevated-risk stratum and 7% LTI risk for the low-risk stratum, the High–Risk (HR) tier was composed of two thirds from the elevated-risk stratum and one third from the low-risk stratum. The Predicting Long Term Institutionalization (PLI) risk score identified 4% of VA users in the High-Risk Tier (HR) who account for 37% of new LTI and 22% of decedents in the following 2 years. Those 4% HR Veterans represented 19% of VA spending and 11% of hospitalizations over the subsequent 2 years. Model sensitivity was better in the elevated-risk stratum (.59 vs .27), while discrimination was better in the low-risk stratum (c-statistic .89 vs .77 for the ER stratum). For every eight Veterans identified in the HR tier in the next two years one became LTI and three died without becoming LTI.

PLI identified a group of VA users at high risk of LTI for clinical assessment.

The PLI statistical tool can help increase the efficiency of care management, thereby helping Veterans age in place and avoid LTI.