Background: Benefits of hospice care for patients with advanced solid cancers have been extensively studied, but they have not been comprehensively documented in hematologic malignancies (HM). Surveys indicate that hematologists harbor concerns about adequacy of hospice services for patients with HM (Odejide et al., Cancer 2017; Hui et al., Ann Oncol 2015). A common perception is that these patients often receive chemotherapy until the very end of life (EOL) and are destined to die in the acute care setting (Sekeres & Gerds, Cancer 2015). They also experience a barrier to hospice use when transfusion dependence develops (LeBlanc et al., Blood 2018). In this context, some authors have questioned whether standard measures of EOL care quality proposed by the National Quality Forum (NQF) should apply to patients with HM, and whether the use of hospice services can improve such measures. Our objective was to describe EOL care quality measures derivable from Medicare administrative claims among patients with HM who did or did not use hospice services.

Methods: From the population-based linked SEER-Medicare registry, we selected fee-for-service beneficiaries with leukemias (acute or chronic), myeloma, myeloproliferative neoplasms (MPN), myelodysplastic syndrome (MDS), or lymphoma (any subtype), who died in 2007-2011. We identified beneficiaries who used hospice services, and ascertained claims-based indicators of aggressive EOL care corresponding to select NQF care quality measures: 1) death in acute care hospital, 2) number of days spent in acute care hospital within 30 days of death, 3) intensive care unit (ICU) admission within 30 days of death, 4) use of chemotherapy within 14 days of death, and 5) Medicare spending for care within the last 30 days of life. Binary outcomes were compared in multivariable robust Poisson models (reporting relative risk, RR), count outcomes in negative binomial models, and costs in a log-gamma model. All estimates were reported with 95% confidence intervals (CI). Models were adjusted for age, sex, race, marital status, Medicaid co-insurance (indicator of low socio-economic status), prevalent poverty in the county of residence, comorbidity index, poor performance status indicator, calendar year, and survival from diagnosis.

Results: We identified 13,556 decedents with leukemia, 7,910 with myeloma, 9,543 with MPN/MDS, and 22,990 with lymphoma, who had median age at death of 78 years (interquartile range [IQR] 72-84). Overall, 47% of patients used hospice services, enrolling at median 22 months from diagnosis (IQR, 5-59; varying from 13 in leukemia to 28 in lymphoma). Median length of stay on hospice was 9 days (IQR, 3-27), varying from 8 to 10 days between different HM. Overall, 39% of patients died in hospital settings (from 36% in myeloma to 41% in leukemia). Median number of days spent in hospital at EOL was 4 in all types of HM. ICU admissions within the last month of life occurred in 39% of patients, and chemotherapy was administered to 10% in the last 14 days (varying from 7% in MPN/MDS to 13% in leukemia). These measures significantly differed between hospice enrollees and non-enrollees (Table).

In multivariable models, use of hospice services was associated with a significant decrease in the aggressiveness of EOL care, as follows: 95% decrease in inpatient deaths (adjusted RR, 0.05; 95%CI, 0.05-0.06), 48% decrease in days spent in hospital (adjusted relative count, 0.62; 95%CI 0.60-0.63), 44% decrease in the risk of ICU stay (adjusted RR, 0.56; 95%CI, 0.54-0.57), 47% decrease in the use of chemotherapy (adjusted RR, 0.53; 95%CI, 0.50-0.57), and 38% decrease in mean Medicare spending at EOL (adjusted relative cost, 0.62; 95%CI, 0.60-0.63).

Conclusions: The proportion of Medicare beneficiaries with HM who die in inpatient setting (39%) or are admitted to ICU (39%) in the last month of life is higher than in a contemporary population of Medicare beneficiaries with cancer (22% and 27%, respectively, in 2009 per Teno et al., JAMA 2013). Across the spectrum of HM, which vary in prognosis and clinical course, use of hospice services is associated with markedly improved measures of EOL care quality, as well as lower Medicare spending in the last month of life. However, only 47% of beneficiaries with HM use hospice services (compared with 59% in Teno et al.). Chemotherapy use at EOL was uncommon, challenging the notion that pervasive chemotherapy is a barrier to hospice use in HM.


Olszewski:Genentech: Research Funding; Spectrum Pharmaceuticals: Consultancy, Research Funding; TG Therapeutics: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.