Abstract Background A small percentage of high-risk patients accounts for a large proportion of Medicaid spending in the United States, which has become an urgent policy issue. Our objective was to pilot a novel patient-centered intervention for high-risk patients with frequent hospital admissions to determine its potential to improve care and reduce costs. Methods Community and hospital-based care management and coordination intervention with pre-post analysis of health care utilization. We enrolled Medicaid fee-for-service patients aged 18-64 who were admitted to an urban public hospital and identified as being at high risk for hospital readmission by a validated predictive algorithm. Enrolled patients were evaluated using qualitative and quantitative interview techniques to identify needs such as transportation to/advocacy during medical appointments, mental health/substance use treatment, and home visits. A community housing partner initiated housing applications in-hospital for homeless patients. Care managers facilitated appropriate discharge plans then worked closely with patients in the community using a harm reduction approach. Results Nineteen patients were enrolled; all were male, 18/19 were substance users, and 17/19 were homeless. Patients had a total of 64 inpatient admissions in the 12 months before the intervention, versus 40 in the following 12 months, a 37.5% reduction. Most patients (73.3%) had fewer inpatient admissions in the year after the intervention compared to the prior year. Overall ED visits also decreased after study enrollment, while outpatient clinic visits increased. Yearly study hospital Medicaid reimbursements fell an average of $16,383 per patient. Conclusions A pilot intervention for high-cost patients shows promising results for health services usage. We are currently expanding our model to serve more patients at additional hospitals to see if the pilot's success can be replicated. Trial registration Clinicaltrials.gov Identifier: NCT01292096