•Understand the rate of hospice conversion during long-term interdisciplinary, community-based palliative care, and describe the sociodemographic and clinical factors that predict this conversion.•Understand sociodemographic and clinical factors that may predict hospice referral timing after community-based, specialist-level palliative care using statistical evidence. Community-based palliative care is emerging to support seriously ill homebound patients. Timely referral to hospice is a key objective. Interdisciplinary palliative care models may provide repeated opportunities to facilitate appropriate hospice admission. We conducted a retrospective cohort analysis of a large sample to explore hospice utilization after enrollment in one such model. To evaluate the incidence and timing of hospice conversion after enrollment in a palliative care program. Patients (n=1505) were enrolled during 2010-2013 in a community-based, ‘high-touch’ model of specialist palliative care that included home visits by an MD/NP and social worker, telephonic outreach by an RN specialist, and access to 24/7 on-call care. Data entered into an electronic medical record at the time of service were extracted for this analysis. Outcomes included hospice admission (yes/no) and duration of palliative care enrollment, and covariates included sociodemographics, diagnosis, Karnofsky Performance Status (KPS), referral source, and insurance status. The sample had a mean age of 71.6 years (SD=14.5) and was 58.7% women, 32.9% White, 29.8% Black, and 28.5% Hispanic. The mean length of stay (LOS) in the palliative care program was 310 days; median was 210 (range 1-1589). A total of 334 patients (22.2%) were admitted to hospice. In multivariate analysis, predictors of hospice admission were shorter LOS in palliative care (p<0.001), cancer diagnosis (p<0.001), lower KPS (p<0.001), and lower poverty level (p=0.03). The only significant predictor of a longer LOS was higher KPS (p<0.01). During months of care by a ‘high-touch’ palliative care model, about one-fifth of patients were admitted to hospice, and relevant predictors of this outcome included cancer diagnosis, lower performance status, and higher socioeconomic status.