Cancer-related emergency department (ED) visits and hospitalizations can be detrimental to both patients and health care systems, especially when they could have been appropriately managed in the outpatient setting.
These avoidable hospitalizations can result in substandard care and add significant costs to the healthcare system. This is particularly true for acute care use (ACU), which accounts for almost half of oncologic costs in the United States. As such, there is a growing need to identify interventions that can reduce avoidable ACU.
This quality improvement (QI) project aimed to leverage patient risk-based prescriptive analytics at a community oncology practice to reduce avoidable ACU. The project utilized the Jvion Care Optimization and Recommendation Enhancement augmented intelligence (AI) tool, which applies continuous machine learning (ML) to predict the risk of preventable harm and generate patient-specific recommendations. The tool allowed nurse case managers to identify and resolve critical clinical issues, reducing avoidable ACU and improving long-term care and outcomes.
According to Journal of Clinical Oncology, a Quality Improvement Project at an Oncology Care Model (OCM) practice has successfully reduced the number of avoidable cancer-related emergency department (ED) visits and hospitalizations by leveraging patient risk-based prescriptive analytics.
According to research, about 50% of ED visits and hospitalizations due to cancer-related symptoms can be managed in a clinic, urgent care, or physician office setting, which can result in substandard care for patients and add significant costs to the healthcare system. Acute care accounts for almost half of oncologic costs in the United States, making it crucial to identify interventions that can reduce avoidable ACU.
The QI project utilized the Jvion Care Optimization, and Recommendation Enhancement augmented intelligence (AI) tool, which applies continuous machine learning (ML) to predict the risk of preventable harm and generate patient-specific recommendations. Using the Plan-Do-Study-Act (PDSA) methodology, nurses contacted at-risk patients and provided interventions to avert the ACU. The AI tool allowed nurse case managers to identify and resolve critical clinical issues, reducing avoidable ACU and improving long-term care and outcomes.
The study notes that the AI tool’s success highlights the importance of patient risk-based descriptive analytics using ML in reducing ACU. The tool’s flexibility and inherently complex nature necessitate proper reporting to realize its potential and attenuate misuse. As new practice reimbursement schemes such as the Enhancing Oncology Model (EOM) replace OCM, predictive modeling of patient risk, prescriptive analytics, and nurse outreach will become increasingly prominent in oncology care.
The OCM practice plans to expand the implementation of the AI tool to further reduce avoidable ACU. In addition, the practice will assess other important metrics in oncology, such as pain, deterioration, depression, and timing of palliative care/hospice referral.
In conclusion, the integration of the AI tool with nurse outreach resulted in timely, patient-centric risk identification and intervention, demonstrating a feasible and effective workflow that can lead to clinical practice and outcome improvements in oncology care.
