Hospice utilization in patients with malignant gliomas

Background: Despite recommendations from professional organizations supporting early hospice enrollment for patients with cancer, little research exists regarding end-of-life (EOL) practices for patients with malignant glioma (MG). We evaluated rates and correlates of hospice enrollment and hospice length of stay (LOS) among patients with MG. Methods: Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare-linked database, we identified adult patients who were diagnosed with MG from January 1, 2002 to December 31, 2011 and who died before December 31, 2012. We extracted sociodemographic and clinical data and used univariate logistic regression analyses to compare characteristics of hospice recipients versus nonrecipients. We performed multivariable logistic regression analyses to examine predictors of hospice enrollment >3 or >7 days prior to death. Results: We identified 12437 eligible patients (46% female), of whom 7849 (63%) were enrolled in hospice before death. On multivariable regression analysis, older age, female sex, higher level of education, white race, and lower median household income predicted hospice enrollment. Of those enrolled in hospice, 6996 (89%) were enrolled for >3 days, and 6047 (77%) were enrolled for >7 days. Older age, female sex, and urban residence were predictors of longer LOS (3- or 7-day minimum) on multivariable analysis. Median LOS on hospice for all enrolled patients was 21 days (interquartile range, 8-45 days). Conclusions: We identified important disparities in hospice utilization among patients with MG, with differences by race, sex, age, level of education, and rural versus urban residence. Further investigation of these barriers to earlier and more widespread hospice utilization is needed.