Early palliative care is associated with less intense care in children dying with cancer in Alabama: A retrospective, single-site study

Abstract

Background: Regional studies show that children with cancer receive medically intense end-of-life (EOL) care, but EOL care patterns, including palliative care utilization in Alabama, remain unknown. Methods: This was a retrospective study of 233 children (0-19 years) who received cancer-directed therapy at Children's of Alabama and died from 2010 through 2019. Rates and disparities in palliative care utilization and the association between palliative care and intense EOL care, death location, and hospice were examined. Results: The median death age was 11 years; 62% were non-Hispanic White. Forty-one percent had a non–central nervous system (CNS) solid tumor. Fifty-eight percent received palliative care, and 36% received early palliative care (≥30 days before death). Children without relapsed/refractory disease were less likely to receive palliative care than those who had relapsed/refractory disease (adjusted odds ratio [aOR], 0.2; 95% confidence interval [CI], 0.1-0.7). Children with CNS tumors and hematologic malignancies were less likely to have early palliative care (aOR for CNS tumors, 0.4; 95% CI, 0.2-0.7; aOR for hematologic malignancies, 0.3; 95% CI, 0.2-0.7) than children with non-CNS solid tumors. Late palliative care (vs none) was associated with more medically intense care (aOR, 3.3; 95% CI, 1.4-7.8) and hospital death (aOR, 4.8; 95% CI, 1.9-11.6). Early palliative care (vs none) was associated with more hospice enrollment (aOR, 3.4; 95% CI, 1.5-7.6) but not medically intense care (aOR, 1.3; 95% CI, 0.6-2.9) or hospital death (aOR, 1.8; 95% CI, 0.8-3.7). Conclusions: Fifty-eight percent of children dying of cancer in Alabama receive palliative care, but EOL care varies with the receipt and timing (early vs late) of palliative care. Whether this variation reflects differences in child and family preferences or systemic factors (eg, hospice access) remains unknown.