Stakeholder engagement is key to the successful implementation of predictive analytic tools in clinical settings, according to a report by a research team from Cornell Weill Medicine in New York and the OneFlorida Clinical Research Consortium.
The team developed a tool that uses electronic health record data to help identify patients at high risk of becoming high-need, high-cost patients in the coming months or years. This subset of patients accounts for most of the nation’s health-care spending.
High-need, high-cost patients often have complex health care needs including multiple diseases, social issues that undermine health, and functional limitations, all of which could be helped by improved access to primary care, better coordination of care, and increased social services. Identifying these patients early could lead to more timely interventions, lower health-care costs and improved health outcomes.
After developing the tool, the researchers, including OneFlorida research coordinator Katie Blackburn, interviewed health systems administrators, informatics personnel, and potential end-users, such as physicians, nurses and social workers, to determine how best to incorporate it into clinical practice.
“Predictive analytics are potentially powerful tools, but to improve health care delivery, they must be carefully integrated into healthcare organizations,” the researchers wrote in the March 6, 2020 issue of the Journal of the American Medical Informatics Association. “One critical lesson from previous health information technology implementation is that end-users, such as nurses, physicians, and support staff, must be involved to make new technology useful, usable, and actionable.”
Key takeaways from the stakeholders:
- Predictive algorithms should be developed to address specific issues identified by stakeholders, not just because the data is available.
- To establish trust with end-users, the development team should be prepared to explain or demonstrate how the algorithm works and how effective it is.
- Predictive tools should be tied to interventions that are clear and easily implemented.
“Although predictive analytics can classify patients with high accuracy, they cannot advance healthcare processes and outcomes without careful implementation that takes into account the sociotechnical system,” the researchers concluded. “Key stakeholders have strong perceptions about facilitators and challenges to shape successful implementation.”