Background: Aggressive treatment near the end of life is a measure of poor quality care. Goals of Care (GoC) discussions may affect healthcare utilization among patients with advanced cancer. We coached oncologists to improve communication skills (CS) & report the effect of CS coaching on hospital, ER, hospice & ICU admission in the last month of life. Methods: We randomized solid tumor oncologists at 4 academic, community, municipal and rural hospitals to participate in a RCT of communication skills training & recruited their newly diagnosed advanced cancer patients with < 2 year prognosis. Ten CSs were assessed via checklist review of audiotaped visits. Charts were abstracted, patients or their caregivers were surveyed at 6 months to assess utilization of: chemotherapy, ICU, hospitalizations, ER visits and hospice in the last 30d of life and over the 6 months enrolled in the study. A GoC discussion included discussion of prognosis, treatment preferences, and what’s important to patients given their cancer diagnosis. We enrolled 22 of 25 eligible oncologists (88%). 263 patients were recruited & surveyed to assess whether a GoC discussion occurred. 250 (95%) had charts abstracted and were resurveyed at 6 months to assess utilization at other hospitals. Results: Patients’ mean age was 63 yrs (20-89), 60% male, 53% white, 29% black & 19% Latino. 35 patients (13%) died within 6 months of baseline survey with no difference between patients of intervention (INT) or control (CNTL) oncologists. Compared to CNTLs, INT oncologists’ skills improved with coaching (1.63 vs -0.09; p = .04). 43% INT and 46% CNTL patients reported a GoC discussion. The average rate of hospitalization/ pt = 0.32 (0-3) and ER visits = 0.44 (0-4) with no difference between INT & CNTL. Rates of aggressive treatment in the last 30d of life varied: chemotherapy = 21%; ICU stay = 7% with no difference between INT & CNTL patients. Hospice enrollment occurred 2 weeks prior to death; no difference in INT vs CNT (13.5d vs 14d). Conclusions: Despite improving oncologists’ communication skills to conduct GoC discussions, there was no impact on rates of hospitalization, receipt of aggressive treatments or enrollment in hospice at the end of life. Clinical trial information: NCT02374255