Background: Early palliative care improves quality of life and survival in patients with metastatic NSCLC. As a result, ASCO recommends early palliative care integration in patients with metastatic cancer. However, generalizing the benefits outside of a clinical trial may be difficult due to selection bias of patients that opt into palliative care due to advanced symptom burden and real-world resource constraints. Therefore, the benefits of this in real-world practice may not be realized. We aimed to evaluate early vs late palliative care referral impact on unplanned hospital visits and survival among GI and lung cancer patients at a large health system cancer center. Methods: The study design was a retrospective, single health system, review of patients treated at Orlando Health from 2021-2022.156 patients of all stage GI and Lung cancers were included. For these patients, early referral was defined as receiving a palliative referral within 8 weeks of cancer diagnosis. Any referral later than this was considered late and it was possible for patients to opt out of referral. Our main outcomes were unplanned hospital visits and survival. To compare the rates of these between palliative care referral groups, a multivariable negative binomial regression model was fitted. To assess any differences in survival, a multivariable cox hazard regression model was fitted. These were adjusted for age, sex, diagnosis, stage, ECOG, race, and insurance. Results: Patients with any palliative care referral had increased unplanned hospital visits and shorter survival than patients who did not have referrals. Specifically, patients with no palliative care referrals had a 62% lower relative rate of unplanned hospital visits compared to patients with early referral (RR = 0.38 [0.22-0.66], p<0.001). And patients with any time referral had approximately 8 months lower restricted mean survival times (RMST) when followed for 2 years for stage 1-3 cancers than patients receiving no palliative care (14 vs 22 month RMST; stage x palliative care interaction, p <0.05). In contrast, stage 4 patients had similar survival of ~14-month RMST when followed for 2 years regardless of referral. Conclusions: In real world practice palliative care referral was associated with increased unplanned hospital visits. Additionally, in this retrospective review, palliative care referral was not found to confer a survival advantage in gastrointestinal and Lung cancer patients. We postulate that this is related to selection bias in the opt in approach to palliative care referral biasing towards a sicker population with higher symptom burden and comorbidities. Our future prospective study will build upon this work through expansion of palliative care services and to better evaluate patient co-morbidities. Our goal is to increase utilization of palliative care for patients with GI and lung malignancies as part of a multidisciplinary integrative approach.