Northeastern University

A fix for overcrowded emergency rooms

Print Friendly

Photo by mah_japan via Flickr

Last month, Atul Gawande had a popular article in the New Yorker questioning whether the hospital industry could learn a thing or two from the likes of the Cheesecake Factory. In the article, he points to qualities like management oversight and standardization common to large-scale restaurant chain operations as areas for hospitals to work on. Now, I’m not sure, but the Cheesecake Factory may also have some idea of how many customers will show up on a given night. Things like time of year, day of the week, holidays, local population, etc., could all have an impact and I bet CF knows it.

This idea of foresight is certainly something that guides production flow in large-scale manufacturing settings, according to Northeastern industrial engineering professor James Benneyan. In a recent article in Academic Emergency Medicine, Benneyan and his colleagues apply the idea to the emergency department setting, as a way to improve patient flow and minimize crowding. “Manufacturing settings improve flow by starting some production early based on predicted demand, rather than waiting for all orders to be placed,” write the authors.

The article compares three predictive measures — two computational models and expert opinion — to determine whether patients walking into the emergency department will be admitted to the inpatient unit. If they can give the clinicians in the latter department ample time to prepare, patients will more readily move out of the emergency department, scaling back a major problem that plagues hospitals across the country.

In contrast to similar studies, which typically predict bed need based on individual patient admissions, the two computational models developed here aggregate the individual patient predictions into a “summative measure” of near-future bed demand.

Of the three predictive measures, expert opinion was the least accurate. For the study, nurses performing triage were asked to categorize patients on their likelihood of admittance, based on a 6 point scale from “definitely yes” to “definitely no.” They tended to underestimate whether a person would be admitted, and by extension the need for inpatient beds.

The model that worked the best took into account patient age, primary complaint, bed type designation and mode of arrival (ie., by stretcher, wheelchair or ambulance).

The authors account for the major limitations of the study, noting that the data used to develop the models and test them were taken from only one hospital. Thus a clear direction for future work, they write, is to see if the approach works similarly across a range of hospitals with varied needs and populations. They have already begun this work.

Benneyan has long researched the strategies deployed in industrial manufacturing settings to healthcare. He’s done the same thing for ski resorts, airlines and other industry.

“Just driving slop and waste out of our healthcare system would save a trillion dollars,” he said to me back in March. “But it’s a complex problem and it necessitates a variety of approaches.”


Follow

Get every new post on this blog delivered to your Inbox.

Join other followers: