Variations in hospitalization and emergency department/observation stays using the oncology care model methodology in Medicare data

Abstract

Objective: To assess variations in hospitalizations, emergency department/observational (ED/OB) stays not resulting in hospitalization, reasons for hospitalization, and hospitalization discharge destinations after chemotherapy, information not provided as part of Oncology Care Model (OCM) baseline data. Methods: OCM methodology was applied to the Medicare 20% sample data to identify 6-month patient episodes triggered by chemotherapy in 2012-2015. Proportions of episodes with hospitalization or ED/OB stays, reasons for hospitalization, and discharge destinations were summarized. Results: Of 485,186 6-month episodes for 255,229 patients in 13,823 practices, 25% of episodes led to ≥1 hospitalization (from 14% in breast cancer to 56% in acute leukemia), and 23% to ED/OB stays (from 18% in breast cancer to 36% in liver cancer). In 2,995 practices with ≥20 total episodes, practice-level proportions of episodes with hospitalization ranged from 14% to 31% (20th-80th percentile) and with ED/OB stays from 17% to 29%. For all cancers combined, the most frequent reasons for hospitalization were infection (13%), anemia (7%), dehydration (5%), and congestive heart failure (3%); the most common discharge destinations were home (71%) followed by a skilled nursing facility (13%), death (6%), and hospice (5%). Reasons for hospitalization and discharge destinations varied by cancer type; acute leukemia episodes led to the highest rates of infection and anemia, and central nervous system tumor episodes to the highest proportions of death or hospice discharge. Conclusion: The variations in frequency of and reasons for hospitalization, ED/OB stays, and hospitalization discharge destinations across cancer types should be considered when evaluating OCM practice performance.