Background: With the aging of the Veteran population, the demand for end-of-life (EOL) care, including hospice services, is increasing. VA commits to ensuring that Veterans have access to high-quality hospice in their setting of choice. In addition to home hospice services, VA offers to provide hospice care within VA Community Living Centers (CLCs; i.e., VA nursing homes) or in VA-contracted community nursing homes (CNHs). Until recently, little was known about the quality of EOL care provided to Veterans receiving CNH-based hospice care. Our preliminary work found that an alarmingly low percentage (58%) of family members of Veterans receiving hospice in CNHs gave an “excellent” rating of the overall care received in the last 30 days of life (compared to 83% of families of Veterans who received hospice in CLCs). We have also observed significant variation across CNHs in quality of EOL care ratings. Significance: CNHs are playing an increasingly critical role in VA’s commitment to ensuring Veterans’ access to hospice care. A comprehensive understanding of the structures and processes that underlie variation in quality of EOL care provided to Veterans receiving CNH- based hospice is needed to illuminate levers for intervention to improve quality. Our study aligns with HSR&D’s priority areas related to Community Care and Long-Term Care and Aging and VA’s Strategic Plan that includes a key focus on “Aging, Frail, and End-of-Life Veterans.” Innovation and Impact: Our application of positive deviance and machine learning methods to identify factors underlying the quality of CNH-based hospice care are innovative contributions to the field of EOL care research. Our findings have high potential for impact to guide VA contracting decisions and the delivery of CNH-based hospice care to Veterans. Specific Aims: 1) Examine the associations between CNH and hospice agency structural characteristics, EOL care processes and bereaved family ratings of overall EOL care quality for Veterans receiving hospice in CNHs; 2) Identify high- and low-performing CNHs based on the BFS overall rating for Veterans who receive hospice; and determine the structures and EOL care processes that are associated with high and low performance; 3) Explore bereaved family member perceptions of the quality of EOL care for Veterans who received hospice in high- and low-performing CNHs; and 4) Understand the delivery of CNH-based hospice care for Veterans in high- and low-performing CNHs from the staff perspective. Methodology: The project will employ a sequential explanatory mixed methods design. The quantitative phase will use multi-level logistic regression models to understand the structural characteristics and EOL care processes associated with BFS overall ratings (Aim 1) and a machine learning approach to identify and study high- and low-performing CNHs based on BFS ratings (Aim 2). The qualitative phase will deepen our understanding of the delivery of hospice care in high- and low-pe...