Classifying End-of-Life Patterns: Evidence From the National Health and Aging Trends Study

Abstract

Background and objectives: End-of-life (EOL) care quality encompasses multiple domains, which are typically studied in silos. This paper explores interrelationships across domains and how they simultaneously shape patterns of EOL care quality.

Research design and methods: Secondary analysis using National Health and Aging Trends Study. Latent class analysis was used to identify subgroups of decedents according to quality of symptoms and needs management, care experience, care processes, and site of death. Multinomial logistic regression was employed to investigate the association between decedent characteristics and class membership.

Results: Five groups of decedents were identified. Multiple sites of death with high care coordination (Class 4: 35%) comprised decedents with least concerns regarding symptom management and care experience. Decedents in Predominantly home deaths with no hospice use (Class 1: 26%), predominantly hospital deaths with low hospice use (Class 2: 18%), and multiple sites of death with hospice use (Class 3: 7%) presented similar symptom management and care experience patterns. Multiple sites of death with low care coordination (Class 5: 14%) comprised decedents experiencing poor quality EOL care across multiple domains. Non-Hispanic Black and Hispanic decedents were more likely to be in Class 2 and decedents with dementia diagnosis in Class 4.

Discussion and implications: There is considerable heterogeneity in EOL care patterns. Assessing quality based on discrete metrics in silos may lead to unintended consequences like stinting of necessary care or increased caregiving burden on families. There is a need for greater focus on patient-centeredness and how care processes and structures interact in the context of shifting care patterns, care preferences, and payment reforms.

Keywords: Care patterns; End-of-life; Latent class analysis; NHATS; Quality indicators.