Variations in hospitalization and emergency department or observation (ED/OB) stays using the oncology care model (OCM) methodology in Medicare data.

Abstract

112 Background: The Medicare OCM started in 2016 to improve oncology care and reduce costs. Two metrics for care evaluation are hospitalizations and ED/OB stays. Baseline data (2012−2015) on variations in all-cause hospitalization and ED/OB stays not resulting in hospitalization in the 21 reconciliation-eligible cancer types are not well characterized. Methods: We applied OCM methodology in Medicare fee-for-service 20% sample data to estimate 6-month patient (pt) episodes triggered by chemotherapy, as well as associated cancer type and practice, from 2012 to 2015. Proportions of episodes with hospitalization and ED/OBs stay were summarized overall and by cancer for all episodes and by practice in those with ≥20 pt-episodes. Results: In total, 485,186 pt-episodes at 13,823 practices were identified (mean episodes/practice, 35.1; median, 2; IQR, 1–15). The mean number of episodes/pt was 1.90 overall (from 1.17 in anal cancer to 2.32 in multiple myeloma [MM]). The proportion of episodes with hospitalization was 25% overall (from 14% in breast cancer [BC] to 56% in acute leukemia) and with ED/OB stays was 23% overall (from 18% in BC to 36% in liver cancer). In 2,995 practices with ≥20 total episodes, at practice-level, the proportion of episodes with hospitalization ranged (20th–80th percentile) from 14% to 31% and with ED/OB stays from 17% to 29%. For select cancers, the range of proportion of episodes with hospitalization and ED/OB stays were highest for lung cancer (LC) and lowest for BC (Table). Conclusions: The proportion of 6-month OCM episodes of care with hospitalization and ED/OB stays after chemotherapy varied by cancer and was higher for certain solid and hematological malignancies. Variations between practices and cancers are important factors when evaluating OCM performance. [Table: see text]