Background: Patients with cancer undergoing treatment are high utilizers of emergency department (ED) services and are at increased risk of hospitalization. A “high-touch” (HT) model of identifying and increasing communication with cancer patients at elevated risk of ED visits was implemented. The goal was to assess whether patients in the HT cohort had a reduced rate of unplanned ED visits and hospitalizations when compared to patients prior to implementation of the model. Methods: In this retrospective study (1/1/2018-4/1/2021), 51 patients on primarily palliative-intent treatment for their cancer were prospectively included in the HT group based on providers’ assessment of their risk of ED visits and malignancy-related complications. This group was compared to a similar cohort (n= 71) pre-implementation of the HT model. Data on baseline characteristics, malignancy type, reason for unplanned ED visits, and patient interaction with the oncology clinic immediately prior to ED visits were obtained, along with end-of-life outcomes including palliative measures. Each ED visit was also categorized as a potentially preventable or non-preventable visit, based on established criteria from Centers for Medicare and Medicaid Services (CMS). Results: While implementation of a HT program did not significantly decrease utilization, the percent of ED visits (p < 0.001) and hospitalizations (p = 0.002) for CMS-preventable reasons was notably higher in the HT cohort. Patients in the HT cohort also reached out to clinic providers significantly more frequently before visiting the ED (p < 0.001). Over half of patients in the HT cohort were seen by a palliative care physician in the inpatient setting within the last 6 months of life. Patients in both groups had similar high rates of dying at home or inpatient hospice versus dying in a hospital setting. Conclusions: Incorporation of a HT model of care in an oncology clinic may promote closer patient interaction with oncology providers prior to emergency department visits, but may come at the cost of increased medical utilization for otherwise CMS-preventable hospitalizations.

¹T-test

²Fisher’s exact test

³Wilcoxon rank-sum SD = standard deviation.