Background: Symptoms are common in oncology patients, though they remain undetected and untreated by clinicians in up to 50% of cases. Integrating patient reported outcomes (PRO) within routine clinical practice has been suggested as a way to improve detection. In order to inform an effective and efficient PRO symptom screening program, we sought to determine whether outpatient symptom scores could predict emergency room use and unplanned hospitalization (ER/Hosp) in a cancer patient population. Methods: This was a population-based study of patients diagnosed with head and neck cancer who had completed at least one outpatient Edmonton Symptom Assessment System (ESAS) assessment between January 2007 and March 2018 in Ontario. Logistic regression models were used to determine the relationship between reported outpatient ESAS scores and ER/Hosp use in the 14-day period following ESAS completion. A generalized estimating equations approach was incorporated to account for possible patient-level clustering. Results: There were 11,761 unique patients identified with a total of 73,282 ESAS assessments. There were 5,203 ER/Hosp outcome events. In adjusted analysis, the odds of ER/Hosp use increased log linearly with ESAS score (1.23 per 1 unit increase in index ESAS score, [95% confidence interval (CI) 1.22 – 1.25]). This corresponds to a 9.23 (95%CI 7.22-11.33) higher odds of ER/Hosp use for the maximum index ESAS score of 10. Seven of the nine ESAS symptom scores were significantly associated with ER/Hosp use with pain, appetite and shortness of breath demonstrating the strongest association. Conclusions: ESAS scores are independently associated with 14-day ER/Hosp in head and neck cancer patients. Appropriate and timely management of symptom burden may reduce rates of ER/Hosp.

*adjusted for age, sex, rurality, comorbidity, treatment modality, subsite, diagnosis year and treatment centre