Risk factors for adverse outcomes in solid tumor interhospital transfer patients.

Abstract

191 Background: We previously reported that inter-hospital transfer patients to our institution’s oncology service experienced higher mortality and increased length of stay (LOS) compared to other origins of admission. We sought to identify risk factors for adverse outcomes in this hospital transfer population. Methods: We reviewed all inter-hospital transfers from Jan-June 2016 to the Cleveland Clinic’s solid tumor oncology service. Patient characteristics, including age, albumin, and case severity indices (mortality and severity risk scores and AP-DRG), as well as encounter characteristics, including admitting provider (house staff or hospitalist), admission time (8AM-5PM or evening), admission day (weekday or weekend), and time between transfer acceptance and admission were recorded. Adverse events examined included activation of adult medical emergency team (AMET), ICU transfer, LOS, in-hospital mortality, and 30-day readmission. Associations of patient and encounter characteristics with adverse events were assessed using Wilcoxon and Fisher’s exact tests. Results: Fifty-three transfer admissions were identified. Patients had a median age of 67 years and 58.5% were male. House staff admitted the majority of patients (81.1%) and most occurred after hours (62.3%). Age, admission time and day, and type of admitting physician were not associated with adverse events. There was a significant association between higher AP-DRG and mortality/severity risk scores with ICU transfers, AMET activations, and mortality. Patients who experienced any adverse events on average had a lower mean albumin than those who did not (2.3 vs 3.0 g/dL p=0.006). LOS and readmission were not significantly associated with any patient or encounter characteristics. Conclusions: Burden of disease as measured by mortality/severity risk and AP-DRG as well as lower albumin levels are associated with adverse events in solid tumor inter-hospital transfer patients, while encounter characteristics do not predict for poorer outcomes. This population should be targeted for improvements in communication and handoffs at the time of transfer, as well as early involvement of palliative care providers.