Characteristics of Medicare beneficiary decedents with hematologic malignancies included in the study. (A) Cohort selection for analysis; the main analysis was conducted in the population of beneficiaries whose death was ascribed to their hematologic malignancy on the death certificate, but a sensitivity analysis confirmed findings in the entire population, regardless of reported cause of death. (B) Diagram showing claims-based covariates (indicators of palliative care needs ascertained within 30 days before hospice enrollment for hospice enrollees or before death for nonenrollees) and EOL outcomes (indicators of aggressive EOL care), including 3 NQF EOL care quality indicators. (C) Trends in the proportions of decedents with hematologic malignancies using hospice, dying in the acute care hospital, having an ICU admission, or receiving chemotherapy at EOL. (D) Linearized trends in median hospice LOS and in median number of days spent in the inpatient setting (within the last 30 days of life). P values for trends were derived from univariate robust Poisson models. MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm.