Hospice length of service in patients with advanced gastrointestinal cancers treated at an NCI-designated cancer center.

Abstract

812 Background: Hospice Length of service (LOS) is a Quality Oncology Practice Initiative (QOPI) measure, and an established metric of end-of-life (EOL) care. Short hospice LOS is associated with lower-quality EOL care, poor symptom control, and patient and caregiver dissatisfaction. Oncologist- and/ or system-specific factors may contribute to shorter hospice LOS. Methods: We conducted a retrospective chart review of patients with advanced gastrointestinal (GI) cancers referred to hospice from a comprehensive cancer center in 2018, and compared data to 2015. We analyzed LOS under hospice care, looking specifically at median LOS and percent of short enrollments (%LOS ≤ 7 days), and examined the variation between individual oncologists within the GI oncology division. The coefficient of variation (Cv) represents degree of variability between oncologists, and was calculated based on median LOS for each oncologist. Mean and SD of oncologist LOS were calculated. Cv was defined as SD/mean; Cv > 100% is typically considered a high variance distribution. Results: A total of 114 patients with advanced GI cancers were discharged to hospice in 2015, and 115 patients in 2018. The median age of patients was 66 years, 47% women, and 80% white. The number of GI oncologists were 9 in both years and median patients per oncologist were 12 in 2015, and 13 in 2018. The median hospice LOS was 16 days in 2015, which decreased to 9 days in 2018. The %LOS ≤ 7 days increased from 25% in 2015 to 47% in 2018. The Cv% was < 100% in both 2015 (63%) and 2018 (44%). Conclusions: In this retrospective study of hospice LOS among patients with advanced GI cancer at a single cancer center, median hospice LOS declined, and %LOS ≤ 7 days increased from 2015 to 2018. We did not find significant variation in hospice referral patterns between oncologists within the GI oncology division. These data point to system level factors responsible for sub-optimal hospice LOS. We will present our data at division meetings, and use mixed-methods design for investigating and eventually addressing barriers to low hospice LOS.