Of the nearly 1 million patients diagnosed annually with acute heart failure (AHF) in emergency departments nationwide, more than 80% are hospitalized, even though many could safely be treated at home. The annual cost of treating AHF in U.S. hospitals exceeds $30 billion.
Six years ago, researchers at Vanderbilt University Medical Center developed a risk stratification tool to predict outcomes and avoid unnecessary hospital admissions after emergency department visits for AHF.
Now the challenge is to test the real-world implementation of the tool, called STRATIFY (Improving Heart Failure Risk Stratification in the ED), by time-strapped hospital emergency departments with varying levels of information technology support, workflows and decision-making patterns.
To that end, the researchers have received a four-year, $3 million grant from the National Heart Lung and Blood Institute of the National Institutes of Health to test a multilevel, biostatistics- and informatics-driven approach to implementing STRATIFY in the emergency care setting.
“If we could decrease the number of patients admitted to the hospital by even a small amount, it would make a huge difference,” said Alan Storrow, MD, associate professor of Emergency Medicine at VUMC who led development of the STRATIFY tool.
Storrow is one of the grant’s three principal investigators, with Sunil Kripalani, MD, MSc, professor of Medicine and director of the Center for Health Services Research and the Center for Clinical Quality and Implementation Research at VUMC, and Dandan Liu, PhD, associate professor of Biostatistics.
STRATIFY determines the risk of adverse events, including heart attack, kidney failure and death, from 13 variables measured during the first three hours of an ED visit, among them blood pressure, electrocardiogram readings, oxygen saturation, kidney function and blood chemistry tests that can detect damage to the heart muscle.
Bringing a decision tool like STRATIFY into the clinical setting, however, “is not as easy as we thought,” said Liu. “It needs to be easy to use.”
Physicians in busy EDs don’t have time to input more data into yet another clinical information system, Storrow agreed, and community hospitals may not have the necessary IT resources to implement it.
That’s why Liu’s group developed ways to calculate risk in a format that is compatible with hospital electronic health record systems. “The physician can just pull up the score,” Storrow said. “It will not require extra work.”
“This project will bring real-time predictive analytics to the bedside to support evidence-based decision-making and patient-centered care, all important aspects of learning health systems,” Kripalani said.
The implementation study will be conducted at VUMC, the Henry Ford Health System in Detroit, and Oregon Health & Science University Hospital in Portland.
To tailor implementation strategies for different emergency care environments, the researchers will conduct onsite assessments with stakeholders including physicians, nurses, other caregivers and patients. They will also conduct surveys and focus groups of patients and their caregivers.
The hope is that STRATIFY will facilitate shared decision making, “where you can talk about what you really think the risk is and make a decision together — physician, nurse and patient, and potentially home caregivers,” Storrow said.
Kripalani and colleagues will evaluate implementation of the STRATIFY clinical decision support tool in different emergency department settings to determine what factors affect how the decision support tool is actually used in practice, and how the local environment and use of the tool impact hospital admissions for heart failure.
If successful, the lessons learned here could be applied to decision tools for other complex disorders in the ED, including pulmonary embolism, stroke and chronic obstructive pulmonary disease.
“A lot of risk models … applicable to the emergency care setting are not being used because of barriers to implementation,” Storrow said. “We feel that we’ll be able to develop a process which we can put to tools already developed for other conditions and have them get used.”