Palliative care consultation reduces hospital costs and length of stay among decedents with hematologic malignancies: A propensity score matched analysis from the national inpatient sample, 2016-2022

Abstract Introduction: Patients with hematologic malignancies face unique end-of-life challenges including rapid clinical deterioration, ongoing cytopenias requiring transfusion support, and infection risks that often preclude hospice discharge. Unlike solid tumors where hospice utilization is common, the economic and clinical impact of inpatient palliative care (PC) consultation in this population remains poorly characterized. We examined whether PC consultation reduces resource utilization for thsese patients. Methods: We conducted a retrospective cohort study using the Healthcare Cost and Utilization Project National Inpatient Sample (2016-2022). Adults (≥18 years) with primary diagnoses of acute myeloid/lymphoblastic leukemia (ICD-10: C92. 0, C92. 4-6, C92. A, C91. 0), diffuse large B-cell lymphoma (C83. 3), or multiple myeloma (C90. 0) who died during hospitalization (DIED=1) were identified. PC consultation was captured using ICD-10 code Z51. 5. Analyses were performed using STATA 19. 5. Propensity scores were estimated via logistic regression incorporating: age categories (18-49, 50-64, 65-79, ≥80), sex, race (White, Black, Hispanic, Asian/Pacific Islander, Native American, Other), primary payer (Medicare, Medicaid, Private, Self-pay, No charge, Other), All Patient Refined DRG Severity of Illness, median household income quartile, and year. We performed nearest-neighbor matching without replacement using caliper width 0. 2 (psmatch2 command). Balance was assessed using pstest examining standardized differences. Primary outcomes were length of stay and total charges. Temporal trends were analyzed using linear regression. Analyses were unweighted. Results: From 430, 232 hospitalizations screened, 23, 060 (5. 4%) met criteria: 9, 786 (42. 44%) leukemia, 4, 932 (21. 39%) lymphoma, 8, 342 (36. 18%) myeloma. Overall, 12, 388 (53. 72%) received PC, increasing from 49. 95% (1, 525/3, 055) in 2016 to 55. 86% (2, 049/3, 668) in 2022 (annual increase 0. 72%, p=0. 061). PC recipients had higher baseline severity: for leukemia, APRDRGSeverity=4 comprised 83. 2% (4, 355/5, 235) of PC versus 89. 6% (4, 079/4, 551) of non-PC patients (p0. 001). PC utilization also varied by demographics: among leukemia patients, PC rates were 60. 9% for age ≥80 versus 42. 0% for ages 18-49 (p0. 001) ; 41. 9% for Black patients versus 57. 0% for White patients (p0. 001). Propensity score distributions showed substantial overlap enabling matching. Common support exceeded 94% across all cancers: 9, 289/9, 786 (94. 9%) leukemia, 4, 691/4, 932 (95. 1%) lymphoma, 7, 964/8, 342 (95. 5%) myeloma patients successfully matched. Balance diagnostics confirmed all covariates achieved standardized differences 10% post-matching. After matching, PC consultation significantly reduced resource utilization across all malignancies. For leukemia (4, 323 matched pairs), PC reduced LOS by 1. 02 days (11. 78 vs 12. 80, ATT: -1. 02, SE: 0. 60, p=0. 09) and charges by 49, 685 (267, 741 vs 317, 426, ATT: -49, 685, SE: 17, 113, p0. 01). For lymphoma (2, 052 pairs), PC reduced LOS by 1. 57 days (10. 43 vs 12. 00, ATT: -1. 57, SE: 0. 68, p=0. 02) and charges by 52, 326 (209, 352 vs 261, 678, ATT: -52, 326, SE: 19, 108, p=0. 01). For myeloma (3, 807 pairs), PC reduced LOS by 0. 85 days (8. 15 vs 9. 00, ATT: -0. 85, SE: 0. 42, p=0. 04) and charges by 34, 300 (136, 405 vs 170, 705, ATT: -34, 300, SE: 8, 605, p0. 01). Conclusions: In this large analysis we demonstrate two critical findings. First, the complete absence of hospice discharges confirms these patients' unique inability to utilize traditional end-of-life care models due to ongoing transfusion needs and acute complications. Second, despite this constraint, inpatient PC consultation achieved substantial cost savings (34, 300-52, 326 per patient) and significant LOS reductions (0. 85-1. 57 days). These benefits likely reflect PC's role in clarifying goals, avoiding non-beneficial interventions, and optimizing symptom management within necessary hospital-level care. With 44% of dying patients still not receiving PC consultation in 2022, expanding access could yield significant healthcare savings while improving end-of-life care quality. These findings support policy efforts to integrate automatic palliative care triggers into electronic health records (EHRs) for high-risk hematologic oncology inpatients and to prioritize PC availability in resource-limited hospitals. Incorporating inpatient PC consultation as a quality metric for hospitalized cancer decedents may further incentivize uptake.