Predicting Hospital Readmissions: Reducing Healthcare Costs

Hospital readmissions, defined as a patient being readmitted to the same or another acute care facility within 30 days of an initial hospital stay, pose a significant problem in the U.S. healthcare system. Annually, there are 4.2 million adult hospital readmissions, and one in six Medicare beneficiaries is readmitted within 30 days of discharge (1)(2). For older adults with functional impairments, the risk of readmission is 40% higher than for those without impairments (3). The financial impact is substantial, with the average cost of a readmission being $14,500 and hospital readmissions costing Medicare $26 billion annually (4). In 2019, 83% of general hospitals in the HRRP were penalized by CMS, amounting to $564 million in penalties for excessive 30-day readmission rates (5). Addressing the conditions that contribute to readmissions and factors such as inadequate caregiver support and social determinants of health can significantly reduce readmission rates, with successful programs achieving reductions of up to 34% (6)(7).

• Annually, 4.2 million adult hospital readmissions occur in the U.S., with Medicare beneficiaries experiencing a 1 in 6 readmission rate within 30 days of discharge.
• Hospital readmissions cost Medicare $26 billion annually, with the average cost of a readmission being $14,500.
• In 2019, 83% of general hospitals in the HRRP were penalized by CMS, totaling $564 million in penalties for excessive 30-day readmission rates.

"ReAdmitPredict AI" is an AI predictive model designed to estimate the likelihood of 30-day hospital readmissions. It analyzes a wide range of health parameters and hospitalization details to identify patients at high risk for readmission, facilitating early interventions and potential cost savings.

The synthetic data set, created to reflect real-world patient scenarios, is based on research on hospital readmissions. The studies by Bailey et al. (1), the National Quality Forum of the Department of Health and Human Services (2), Greysen et al. (3) and the RevCycleintelligence material (4) provide an understanding of the patterns and causes of readmissions. The research of Rau (5) and the findings of Hines et al. (6) on the costs and conditions of readmissions, along with studies on patient participation in discharge processes by Kemp et al. (7), are used to shape the model parameters.