Incidence and Predictors of Hospice Utilization at the End of Life in Elderly Patients with Brain Metastases

Abstract

Brain metastases cause significant morbidity and mortality. Integration of hospice services toward the end of life improves symptom control and quality of life for patients. Dedicated studies relating to the incidence and predictive factors of hospice enrollment in patients with brain metastases are lacking. We used the Surveillance, Epidemiology and End Results (SEER)-Medicare database to study patterns of hospice enrollment at the end of life in elderly patients with brain metastases. We hypothesize the existence of sociodemographic predictors of hospice utilization in this population. We obtained SEER-Medicare datasets for primary cancer sites that most commonly metastasize to the brain, namely lung cancer, melanoma, breast cancer, renal cancer, colorectal cancer, esophageal cancer, ovarian cancer and testicular cancer. We identified patients with brain metastases aged 66 years and older, who were diagnosed between 2005-2016 and passed away prior to December 31, 2016. The primary outcome was hospice enrollment (yes versus no). We tested the association between hospice enrollment and sociodemographic and hospital-based covariates using univariable and multivariable logistic regression. As a subset analysis, among patients who enrolled on hospice, we identified predictors of late (<7 days before death) enrollment given the importance of earlier initiation of hospice services. We identified 50,696 eligible patients with brain metastases who died between 2005-2016. The incidence of hospice enrollment was 71.5% (95% CI: 71.1-71.9). Using multivariable regression analysis, we found that female gender (OR 1.285, 95% CI: 1.233-1.332) and older age (OR 1.009, 95% CI: 1.006-1.012) predicted enrollment in hospice. The odds of enrollment for black patients (OR 0.76, CI 0.71-0.82), Hispanic patients (OR 0.82, CI: 0.75-0.90) and Asian patients (OR 0.52, CI: 0.47-0.56) were significantly lower compared to white patients. Admission to a teaching hospital (OR 1.08, CI: 1.04-1.11) or to a government-owned hospital (OR 1.06, CI:1.03-1.10) within six months of death were associated with higher odds of hospice enrollment. Among patients enrolled in hospice, 67.7% (95% CI: 67.2-68.1) were enrolled ≥7 days prior to death. The median length of stay among all hospice enrollees was 13 days with IQR 5.0-33.0 days. Using multivariable regression analysis, we found that female gender (OR 1.352, CI: 1.291-1.416) predicted earlier hospice referral while urban residence (OR 0.877, CI: 0.812-0.948) predicted late hospice enrollment. Sociodemographic disparities exist in hospice utilization at the end of life among elderly patients with brain metastases. Further investigation to facilitate targeted interventions addressing such disparities may be warranted.