According to a release from Vanderbilt University Medical Center, nurses have long had the ability to pick up cues about patients’ health from subtle changes in behavior or appearance.
They share these clinical assessments in patients’ electronic health records — both in the content of their comments and when and how often they make them.
Unfortunately, these assessments are often buried in form fields and not analyzed as a whole. As a result, the insights are often overlooked by care teams in favor of measures seen as more objective. This can make it difficult for nurses to engage the care team when they feel patients are at risk.
Vanderbilt University Medical Center is working to change that — collaborating with nurse scientists from Columbia University Irving Medical Center and two other medical systems to implement a predictive tool that uses machine learning to extract patient behaviors documented by nurses in electronic health records and transform them into observable data. The information generated can be used to support early prediction of organ failure or other critical conditions in hospitalized adults.
In addition to VUMC, the CONCERN (Communicating Narrative Concerns Entered by RNs) implementation initiative also includes Mass General Brigham in Boston and Washington University School of Medicine/Barnes-Jewish Hospital in St. Louis. During the three-year trial, each of the health systems will test the effectiveness of an implementation toolkit, developed to support large-scale adoption of the tool. The project is one of 10 that is funded in a $14 million grant from the American Nurses Foundation (ANF) and its Reimagining Nursing Initiative.
At VUMC, the predictive tool will be implemented in the existing eStar system for patients in the medical-surgical (med-surg) and intensive care unit settings at Vanderbilt University Hospital. It will take a variety of nurse-entered information, including vital signs and documentation notes, to create an objective score for the patient in real time.
The results will be visible to nurses in the electronic medical record, such as a patient list or a nurse leader dashboard. The system will color code each patient’s record green, yellow or red based on severity. It will serve as an early warning system to detect patients with deteriorating conditions, allowing them to better prioritize care in a busy unit.
“This is a situational awareness communication tool to collate information that will assist the nurse in follow-up and communication related to patients who may be deteriorating,” said Cathy Ivory, PhD, RN, RN-BC, RNC-OB, NEA-BC, associate nurse executive and VUMC principal investigator for CONCERN. “The machine learning methods work behind the scenes in real time to collate and summarize that information in a format that’s consumable by the nurse.”
Feedback will be welcomed throughout the three-year project, from direct care nurses, nurse leaders and IT professionals. Direct care nurses who interact with CONCERN will participate in focus groups to give feedback about the usefulness of the tool, including “what format they’d like to see it in, when they’d like to see it, when it’s most useful, how the user interfaces, how they interact with it, all of those things,” Ivory said.
Ivory noted that VUMC uses other systems of early warning scores, and this is not intended to replace them. Nor will it increase documentation burden; it will summarize documentation in a way that is useful. Rather, it is another means to leverage the value of nurse-generated observations for better patient care.
“Nurses generate and document a tremendous amount of data in the context of care of their patients, and they may not perceive that what they’re documenting is valuable,” Ivory said. “This project demonstrates that what they document is indeed valuable in making decisions about patient care.”
“The CONCERN initiative is yet one more way nurses are working to better care for patients and families,” said Executive Chief Nursing Officer Marilyn Dubree, MSN, RN, NE-BC. “We are excited to be a part of this.”
