The trusted source for
healthcare information and
Artificial intelligence and case management can help patients stay out of the hospital. An inpatient rehabilitation hospital system uses risk stratification data from electronic health records (EHRs) to identify patients with declining health who might need to be sent to an acute care hospital.
The data also can help organizations target resources to cases where case management and care transition services provide the most efficient and cost-effect benefits.
“An exciting thing we are doing on the inpatient side is the predictive analytical model,” says Elissa Charbonneau, DO, MS, chief medical officer at Encompass Health in Birmingham, AL, which provides inpatient rehabilitation care, home health, and hospice services.
The model uses an algorithm in an EHR. It ranks patients according to their risk and need to be transferred to an acute care hospital, she says. “There is an alert on the electronic health record, and we’re now piloting a similar program for patients after they leave our hospital,” Charbonneau adds. “We can assess their risk after they are home.”
The risk stratification tool, named ReACT (Readmissions Acute Care Transfer) is used by all Encompass Health hospitals, says Dina Walker, RN, MSN, ACM, RN-BC, national director of case management, Encompass Health in Birmingham, AL. The purpose of the tool is for the hospital to react to changes in patients’ conditions as timely as possible.
“We’re trying to track and determine abnormal medical signs and symptoms, suggesting the patient is getting worse in some way,” Walker explains. “It helps us assess the patient and intervene before the patient has to go back to the acute care hospital.”
The algorithm checks for changes in various clinical areas, including:
A change for the worse in any of the metrics that ReACT monitors would suggest the hospital needs to thoroughly assess the patient and see what might be going wrong, Walker says.
Encompass Health can collect cutting-edge patient data because the health system is large and has been collecting EHR metrics for about nine years, Charbonneau says.
“We want the model to predict — not just for inpatients, but also look at after they are discharged, so we can look at what makes them a higher risk,” Charbonneau adds.
The data might point to a patient who needs help accessing medication or whose lab work indicates a higher risk. “Maybe the patient is not eating well or is missing therapy,” Charbonneau says. “Those are statistically significant predictors that [suggest they might end up] back in the hospital.”
Risk stratification tools can be embedded in health records, looking for trends continuously. “The algorithm runs in the background of the electronic health record,” Charbonneau explains. “Doctors and nurses can see the information as it happens. They can see when a patient’s risk goes from low to high or from high to very high.”
These changes will trigger an alert message to the doctor and nurse, telling them the patient’s risk has increased, she says. “It gives us the ability to stratify patients according to their risk of going back to an acute care hospital,” she adds. “They are labeled as low, high, or very high risk, using the colors green, yellow, and red.”
Encompass Health plans to expand the tool and use it in the outpatient setting, Charbonneau adds. “There will be different data points, but we’ve come up with an algorithm to look at these patients as they leave the hospital, and we’re currently piloting it,” she says.
The new algorithm is called Readmission Prediction Model. Its purpose is to determine the probability that a patient will be readmitted to an acute care hospital after he or she is discharged from the inpatient rehabilitation hospital, Walker says.
This tool examines a variety of metrics, including:
“It tells us the probability of someone being readmitted to the hospital,” Walker says. “We can foresee something happening down the road and implement an intervention to address it.”
Case managers could contribute to the risk stratification data by adding information about patients’ social determinants of health, including their access to nutritious food, transportation, community resources, and healthcare facilities near their homes, she explains.
“Case managers are learning to assess the things the predictive algorithm hasn’t learned yet,” Walker says. “We’re hoping documentation by case managers for those other types of things will inform the algorithm, so we’ll have a complete picture of the patient’s readmission risk.”
Financial Disclosure: Author Melinda Young, Editor Jill Drachenberg, Editor Jonathan Springston, Editorial Group Manager Leslie Coplin, Nurse Planner Toni Cesta, PhD, RN, FAAN, and Accreditations Manager Amy Johnson, MSN, RN, CPN, report no consultant, stockholder, speaker’s bureau, research, or other financial relationships with companies having ties to this field of study.