Abstract
INTRODUCTION: Timely goals of care (GOC) discussions improve end-of-life (EOL) care for patients with solid malignancies. Although timely GOC discussions might also improve EOL care for patients with blood cancers, data are sparse regarding such discussions for this population. We studied the impact of timing and location of GOC discussions on quality of EOL care for a cohort of patients who died of blood cancers.
METHODS: We assembled a retrospective cohort of blood cancer decedents from Dana-Farber Cancer Institute, Boston, MA. We included patients diagnosed with a blood cancer, who had ≥ 2 outpatient visits at the study institution, and who died between January 1 and December 31, 2014. Using the established definition from Mack et al, JCO 2012, we classified patients as having had a GOC discussion if there was any documentation in the medical record of a discussion involving resuscitation status, hospice, or preferred location of death. We abstracted content and location of the first documented GOC discussion. Next, we ascertained intensity of care near the EOL (i.e. chemotherapy ≤ 14 days before death, intensive care unit [ICU] admission ≤ 30 days before death, and hospital death) and hospice enrollment. Of note, decreased intensity of medical care and high rates of hospice enrollment are accepted indicators of quality EOL care. In univariable and multivariable models, we assessed the potential relationship between timing (> or ≤ 30 days before death) and location (outpatient or inpatient) of GOC discussions with intensity of medical care and hospice use near the EOL.
RESULTS: Of the 384 eligible deceased patients, 39.1% had leukemia/myelodysplastic syndromes, 37.2% had lymphoma, and 23.7% had myeloma. 40.6% had undergone a hematopoietic stem cell transplant (HCT). Overall, 235 (61.2%) had a documented GOC discussion. The median time between first documented discussion and death was 15 days, with 33.2% of first discussions occurring > 30 days before death, and the minority (36.2%) occurring in the outpatient setting. The most commonly discussed topic was resuscitation preferences (82.6%), followed by hospice (30.6%); preferred location of death was rarely discussed (3.4%).
Of the entire cohort, 16.9% received chemotherapy ≤ 14 days before death, 21.6% had at least one ICU admission ≤ 30 days before death, and 38.0% died in the hospital. 49.5% of the study population experienced at least one indicator of high-intensity medical care near the EOL. The rate of hospice use was 24.2%.
In univariable analyses among those who had EOL discussions (n=235), having the first GOC discussion > 30 days before death was significantly associated with a lower likelihood of ICU admission ≤ 30 days before death (14.1% vs. 40.8%, p<0.001) and a lower likelihood of hospital death (37.2% vs. 63.1%, p=0.0002; Figure). Significant differences persisted in multivariable models controlling for sex, age, diagnosis, and HCT status. There were no significant differences in chemotherapy receipt close to death or hospice use by timing of GOC discussions in univariable or multivariable models. Having the first GOC discussion in an outpatient setting was significantly associated with a lower likelihood of ICU admission ≤ 30 days before death (9.4% vs. 44.7%, p<0.001), lower likelihood of hospital death (25.9% vs. 70.7%, p<0.001) but no significant difference in chemotherapy use near death (Figure). Outpatient location of first GOC discussion was also significantly associated with higher rates of hospice use (41.2% vs. 24.0%, p=0.006). Significant associations persisted in multivariable analyses.
CONCLUSIONS: In this large cohort of patients who died of blood cancers, nearly 40% did not have a documented GOC discussion. For those who did, most GOC discussions occurred close to death and in the inpatient setting. Moreover, when patients had GOC discussions greater than a month before death and in the outpatient setting, they were more likely to experience high-quality EOL care.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
This icon denotes a clinically relevant abstract