BACKGROUND: Patient reported outcomes (PROs) are important cancer outcomes that can be measured electronically but are understudied in the safety-net hospital setting. Routine electronic screening to address social determinants of health (SDH) has been established in primary care clinics and the emergency department of New England’s largest safety-net hospital. The burden of SDH in safety-net oncology patients is less well-studied. This study aimed to determine the feasibility and challenges of routine administration of ePROMs in a safety-net Radiation Oncology clinic, describe treatment toxicities and quality of life (QOL) experienced by this vulnerable population during radiotherapy, and evaluate SDH and the need for SDH screening in the oncology clinic. METHODS: Patients with lung or head and neck cancer scheduled for radiation oncology consultation from 3/2019–1/2020 were deemed eligible for electronic questionnaire participation based on primary language spoken and absence of metastases. At consultation, patients were administered a set of baseline ePROMs (EQ-5D-3L, FACT, PRO-CTCAE) and a social needs screener (THRIVE) using a widely-used cloud-based, patient-centered outcomes platform. Associations between patient demographics and questionnaire completion were retrospectively evaluated. The set of ePROMs were collected at the end of treatment to characterize treatment-related toxicities and changes in self-reported QOL. RESULTS: In total, 99 eligible head and neck cancer (51.5%) and lung cancer (48.5%) patients were identified. Median age was 65. The majority of patients were male (71.7%), and English-speaking (82.8%). Whites, Blacks, and Asians/Others comprised 42.4%, 38.4%, and 6.1% of patients, respectively. Fifteen patients were Hispanic (15.2%). Patients were most likely to have private health insurance (39.4%), followed by joint Medicare-Medicaid (25.3%), Medicaid (17.2%), and Medicare (16.2%). Two patients were insured by Corrections (2.0%). Eight patients (8.1%) no-showed or cancelled, while 91 patients were seen in consultation. Forty-four patients (48.4%) completed the initial questionnaires. For the remaining 47 patients (51.6%), the most common reason for lack of ePROM completion was clinic understaffing and/or clinical decision based on the absence of indication for radiotherapy (n=27, 57.4%). Ten patients refused to complete questionnaires (21.3%), with reasons cited including length of questionnaires and low energy. Ten patients were physically unable to attempt questionnaires (21.3%), for reasons including disabilities and low-literacy. Age, language, race, ethnicity, insurance, marital status, gender, and disease site were not significantly associated with ePROM completion (P≥0.05). For patients who completed the general (QOL) questionnaire EQ-5D-3L, there was no significant difference in general QOL domains nor self-reported overall health score at baseline vs. end of treatment. For head and neck cancer patients, FACT-H&N Total scores, measuring disease-specific QOL, were significantly worse at end of treatment vs. baseline (P=0.006). For lung cancer patients, FACT-L Total scores at the end of radiation treatment were not significantly worse at end of treatment vs. baseline (P=0.953). For head and neck cancer patients who completed PRO-CTCAE, there was a significant increase in the number with moderate to very severe taste issues (P=0.008) and decrease in appetite (P=0.025) by end of treatment. For lung cancer patients, there was a trend towards an increase in the number reporting moderate to very severe nausea (P=0.083). Eighty-one of 99 patients (81.8%) were screened for at least one SDH domain using the THRIVE screener at the study hospital. Nineteen patients (19.1%) had all 8 THRIVE social determinants of health statuses documented. Only housing status was documented for 61 patients (61.6%). There was a trend for married individuals (P=0.068) and females (P=0.074) to be associated with the completion of at least one THRIVE domain. Age, race, language, and insurance status were not associated with THRIVE screening (P>0.05). Transportation to appointments (21.1%), food insecurity (20%), and affording medications (10.5%) were the most prevalent concerns among these oncology patients, with 100% of patients who reported insecurities with medication and transportation requesting resources for these needs. CONCLUSION: Routine ePROs collection in a busy safety-net oncology setting is feasible, but challenging and labor-intensive. Implementation was met with both patient and staff challenges and revealed the need for dedicated project management, staff training, and opportunities to increase patient accessibility. Preliminary PROs analyses revealed several significant detriments in quality of life and increased symptoms for this patient population during treatment, but additional data collection is required. Safety-net oncology patients report significant social needs. Routine SDH screening and resource referral should be considered in these vulnerable patients. Efforts in a specialized department such as Radiation Oncology could fill gaps in existing efforts in a large safety-net hospital. Safety-net oncology clinics can likely help vulnerable cancer patients access available community resources and reduce disparities due to SDH.