Managing more than the primary diagnosis to reduce hospital readmissions

Predictive Analytics
Last updated: 
Paramedics, taking patient on stretcher from ambulance to hospital Speeding ambulance,(blurred motion)
COPD, congestive heart failure and change in mental status are frequent causes of hospital readmissions from inpatient rehabilitation. Learn how predictive analytics are being used to help reduce the risk of a hospital readmission in this rehabilitation patient population.

Hospital readmissions set patients back in their recovery and increase the overall cost of care.

In an effort to reduce the risk of transfer back to acute care during an inpatient rehabilitation stay, clinical leaders at Encompass Health examined data from its Electronic Medical Records (EMR) to determine what factors were placing their patients at risk of being readmitted to the hospital. It turns out that the readmissions weren’t always related to the primary diagnoses that sent them to the hospital in the first place.

“We were identifying the reasons why we had to send patients back to the hospital while they were with us,” said Dr. Joe Stillo, Encompass Health’s vice president of medical services. “We were able to look at the data and see that one of the top reasons was respiratory distress.”

This is often caused by one of two underlying chronic conditions—congestive heart failure (CHF) and chronic obstructive pulmonary disease (COPD). The second-leading cause of a transfer from the rehabilitation hospital back to the acute was change in mental status.

Knowing this, the company sought to develop solutions that could help clinicians get ahead of issues related to these conditions in an effort to reduce the risk of a hospital readmission.

Reducing the risk of a hospital readmission with data

Using predictive analytics from its EMR database of inpatient rehabilitation admissions, Encompass Health, in partnership with Cerner, a leader in health information technologies and EMRs, created order sets or “Power Plans” for those identified conditions that commonly lead to a hospital readmission. They started with COPD and CHF and, most recently, altered mental status (AMS).

“This is just another tool we are providing our clinicians,” said Dayle Unger, clinical IT advisor with Encompass Health. “We know that a large number of our patients who do have to readmit during their stay with us or down the road is because of these chronic conditions. This gives our clinicians an extra set of tools to continue the focus on preventing hospital readmissions.”

At admission to an Encompass Health rehabilitation hospital, providers can execute the appropriate Power Plan in the patient’s EMR. This plan provides additional surveillance activities aimed at identifying worsening of these chronic conditions and, if the condition worsens, directing further management with the goal of preventing hospital readmission.

Learning from past hospital readmissions

In 2017, Encompass Health deployed ReACT, an acute care readmission prediction model, as its first proprietary predictive model.

Stillo said the CHF, COPD and altered mental status order sets are an extension of ReACT, which uses data from the company’s EMR to predict the risk of an acute care transfer (ACT) from an Encompass Health rehabilitation hospital to an acute hospital.

Leveraging data from more than 80,000 patient records, machine learning, and statistical analysis, ReACT runs throughout a patient’s stay. It categorizes the patient’s level of risk for an ACT based on 30 predictors that have been identified as statistically significant. These risk factors include a variety of clinical signs such as loss of appetite, vital signs, Braden score, missed therapy and medications. There are three levels of ReACT risk— low, high and very high. Because the ReACT model runs in near-real time, as new results are available in the EMR, the model will update. A change in patient risk level from low to high or from high to very high risk will trigger an alert to clinicians, leading to further evaluation for the need for enhanced monitoring and treatment.

The ReACT model and surveillance plans are key strategies used by all Encompass Health hospitals to reduce risk of a hospital readmission.

Empowering patients and clinicians

While the order sets present a guide as to how to best monitor a patient’s chronic condition while in the hospital, patient and caregiver education is also a key part of managing the condition.

“We are doing the surveillance management of the condition when they are at the hospital,” Unger said. “At the same time, our clinicians are working with the family and the patient to educate them on self-management for when they go home. It all works together.”

Ongoing efforts to reduce hospital readmissions

The order sets are part of Encompass Health’s ongoing efforts to reduce hospital readmissions. In addition to ReACT and these clinical decision support tools, the company deployed its Readmission Prevention Program in 2020. Also using predictive analytics, this algorithm determines the risk of a hospital readmission after the patient is discharged from an Encompass Health rehabilitation hospital.

Using its inpatient rehabilitation EMR, the company is also developing a fall prevention model that is specific to the rehabilitation hospital setting. This will be designed to provide a tool superior to existing risk tools in predicting a patient’s risk for fall.

The content of this site is for informational purposes only and should not be taken as professional medical advice. Always seek the advice of your physician or other qualified healthcare provider with any questions you may have regarding any medical conditions or treatments.