Abstract only 213 Background: Most ambulatory oncology practices utilize an on-call service that is essential for continuity of cancer care after office hours, yet there is limited literature about the dynamics of this care delivery model and how to improve it from a quality standpoint. We hypothesize that patients with more advanced cancers are more likely to require the on-call service due to acute symptoms arising during their treatment, and that this information can be used to create a risk model to predict subsequent hospitalization. Methods: We performed a single-center, retrospective review of sequential overnight and weekend calls received by an oncology practice with 16 physicians over a 20-week period from January-May 2020. Calls were classified as being either urgent, requiring immediate attention, or non-urgent, which could be addressed during office hours. Data were summarized using descriptive statistics, continuous and categorical variables were compared using Wilcoxon rank-sum test and Fisher’s exact test, respectively. Multivariate analyses were estimated by logistic regression using the penalized maximum likelihood estimation method. Results: The data set included 236 consecutive calls among 176 patients, with 65% females and median age of 68 (range: 25-87). Of these, 185 calls (78.4%) were deemed urgent, among which 139 (75%) were symptom-related. Among the 202 calls (85%) from patients with cancer, 164 (81%) of them were urgent, mostly due to symptoms (82%). Of these urgent calls, 44 (27%) resulted in admission within 24 hours (P < 0.0001), primarily related to treatment toxicity or disease progression (81%). Patients with stage 4 cancers (42%) or hematologic malignancies (28%) were more likely to use the on-call service. There was no significant difference between call urgency and treatment regimen (P = 0.06). In a multivariable model, advanced age OR = 1.03(1.0-1.07) and urgent calls OR = 33.1(2.7-401.0) were independently associated with risk of hospitalization. Of note, there were no differences in admissions before or after the peak of the COVID-19 pandemic in New York City (P = 0.49). Conclusions: We identified an association between after-hours calls and more advanced malignancy, which was independent of treatment regimen. The majority of on-call issues were urgent and symptom-related, with advanced age and urgent calls being most likely to result in hospital admission. These results suggest that strategies can be developed to prevent hospitalizations in patients at higher-risk for adverse events based on a multivariable risk model.