Abstract
Background
Survival patterns at admission to home-based palliative care (HB-PC) are not well characterized, particularly in low- and middle-income countries (LMICs), where patients are often referred at advanced stages.
Objectives
To identify population-level predictors of short-term survival and characterize survival patterns at admission to HB-PC in a real-world LMIC cohort.
Methods
We conducted a retrospective cohort study including 471 adult cancer patients enrolled in a public HB-PC program in Brazil between 2013 and 2023. Sociodemographic, clinical, and functional data were extracted from medical records. Overall survival was analyzed using Kaplan-Meier methods and Cox proportional hazards models. Short-term survival (30-180 days) was evaluated using logistic regression. Tumor-specific survival patterns were explored using a decision tree approach.
Results
Functional status (PPS) and cancer stage were independent predictors of survival. Patients with PPS ≤30 had markedly reduced short-term survival, particularly within the first 30-60 days. Tumor type further stratified survival patterns, with liver and lung cancers associated with poorer outcomes. A simplified model combining PPS and tumor type showed similar performance to more complex models (C-index ∼0.63). Hematologic comorbidities were associated with modestly longer survival.
Conclusion
Survival at admission to HB-PC is characterized by substantial heterogeneity, influenced by functional status, cancer stage, and tumor type. Simple, routinely collected variables can capture meaningful differences in population-level outcomes and may inform health service planning in resource-limited settings.