Calculated risk: Weighing the options for ER visitors

// Industrial & Systems Engineering

Tags: 2019, Faculty, News

Photo of Hospital emergency department

ISyE Assistant Professor Gabriel Zayas-Caban and collaborators have created a framework for determining the impact of emergency department admission decisions on patient outcomes. Photo credit: Komu Photos/Eric Staszczak/Flickr.com.

Share this story:

When some patients with serious ailments or lengthy track records of health problems visit an emergency room, doctors can easily make a logical choice and admit them to the hospital.

Other cases aren’t quite as straightforward, though, and each option—to admit or send home—carries its own degree of risk. Hospital stays can potentially expose patients to infections or trigger declines in cognition and mobility for older patients. On the flip side, sending home a patient introduces another set of threats.

Photo of Gabriel Zayas-Caban
Gabriel Zayas-Caban

In the moment, it’s difficult for medical staff to determine the risks associated with each choice. Gabriel Zayas-Caban, an assistant professor of industrial and systems engineering at the University of Wisconsin-Madison, and collaborators have created a framework for determining the impact of admission decisions on patient outcomes.

In a paper published June 11, 2019, in the journal Statistics in Medicine, the researchers used their approach to analyze data from emergency departments in the University of Michigan Health System over a three-year span.

Their method estimated that admitting patients with higher health needs greatly reduced their 30-day risk of both returning to the emergency department and being readmitted. Conversely, admitting patients with lower health needs actually slightly increased their 30-day risks of revisits and readmission.

To cut through the obvious reality—patients who are admitted are typically in worse condition—and isolate the influence of solely the admission decision on outcomes, Zayas-Caban and his partners employed a technique called causal inference, which establishes a cause-and-effect relationship.

Moving forward, Zayas-Caban would like to use the method to home in on admissions risks for specific patient populations.

“Our hope is that our approach can provide an evidence base for supporting emergency department decisions by care providers,” he says. “For example, our approach may suggest whether admitting fewer older adults presenting with chest pain would reduce mortality rates.”

Zayas-Caban collaborated on the research with Amy Cochran, a research associate in the Department of Biostatistics and Medical Informatics at UW-Madison; Paul Rathouz, a visiting professor and former chair of the Department of Biostatistics and Medical Informatics at UW-Madison; and Keith Kocher, an assistant professor and physician in emergency medicine at the University of Michigan.

Author: Tom Ziemer