If all the beds in a hospital are full, how does the hospital accept new patients?
No simple riddle, that question motivates much of the research of Jing Dong, an assistant professor in the division of Decision, Risk, and Operations. In a series of new and forthcoming studies, Dong and her colleagues have developed data-driven models that might help hospitals increase the number of admitted patients, decrease lengths of stay, and speed up discharge processes — all by focusing on the number of beds in the hospital.
“Given the high cost of operating hospitals, administrators need to pay more attention to the efficiency of their operations,” says Dong, a specialist in service operations management. “I think I’m in a unique position to help them look at whether there’s room for improving that efficiency.”
Dong’s insights have the potential to put a dent in the high cost of US hospital care. About 18 percent of gross domestic product, or $3.3 trillion, was spent on healthcare in 2016, with one-third going to hospital care. Maintaining a single hospital bedroom costs about $1 million a year — so if patients can be safely discharged faster and more people can use the same bed over the course of a year, overall costs have the potential to come down.
Get Out of Bed
How does a hospital get the most of its beds? Hospital physicians typically conduct rounds once a day to examine whether patients are ready for discharge. In a study with Linda Green, the Cain Brothers & Company Professor of Healthcare Management, and Associate Professor Carri Chan, Dong found that hospitals can cut more than six hours off patients’ average length of stay by conducting two rounds at specific times of the day.
The optimal time for a physician to conduct rounds is six hours before the peak patient influx, the professors found; if they’re allowed to add a second inspection, it should be scheduled 12 hours later, according to their 2017 paper published in Operations Research. (Conducting any more inspections makes a marginal difference.)
Another way to reduce hospital congestion is by posting ER wait-time, as some hospitals are now doing. But this can backfire if hospitals are not constantly updating their information, as Dong found in a study to be published in Management Science this year with Galit Yom-Tov of Technion and Elad Yom-Tov from Microsoft Research.
In their analysis of more than 200 hospitals that provide wait-time information, Dong found that patients may act strategically to help balance the workload between hospitals. However, when the information is not up-to-date, strategic behavior of patients may worsen hospital congestion due to the delayed feedback effect.
“We proposed that, instead of hospitals using a four-hour moving average to estimate their delays, which is their current practice, they should use more up-to-date information,” says Dong.
Make Your Bed
A hospital can also add new beds to reduce congestion, although this introduces new costs. To find the optimal number of beds, Dong teamed up with Ohad Perry of Northwestern University to develop a mathematical model that quantify the relationship between the number of beds and the quality of service for each inpatient unit.
Their paper, “Queueing Models of Patient-Flow Dynamics in Inpatient Wards,” is to be published in Operations Research this year. The aim, says Dong, is that “each hospital can decide on the desirable level of service it wants to offer, and then find the corresponding number of beds that is needed.”
In a related working paper, with Fanyin Zheng of Columbia Business School, Pengyi Shi of Purdue, and Xin Jin of the Singapore National University Health System, Dong zoomed in on what kind of hospital beds are most needed to increase overall operational efficiency. The researchers found that when patients are assigned a bed outside their primary specialty, there can be an associated 19 percent increase in their length of stay. They also found that by reallocating 6 percent of bed capacity, the hospital can reduce patient waiting time by more than 10 percent and save more than 2,000 patient days a year.
“A lot of the improvements are achieved by taking a network view of the hospital units,” says Dong. “These units are all interconnected, helping one unit doing better will have a spillover effect to the others.”
“It’s not so much about treatment of patients or clinical decisions, but how hospitals operate,” she adds. “By working with hospitals, I see a lot of potential for efficiency improvement. It’s generating interesting research questions for me, but I also see how it could help hospitals save money and offer better quality of service to patients.”
Read the research
About the researcher
Jing Dong is an Assistant Professor in the Decision, Risk, and Operations division at the Graduate School of Business, Columbia University. Her primary research...Read more.