In the past 25 years, demand for emergency healthcare in United States has exploded, pushing capacity-strained emergency rooms across the country to the limit and prompting organizations from the National Academy of Medicine to the General Accounting Office to label the American emergency care system “in crisis.”
Between 1992 and 2012, emergency room utilization, defined as the number of visits per 1,000 people, grew by approximately 18 percent in the United States, according to a 2012 survey by the American Hospital Association. And its not just in America To combat over-crowding, hospitals increasingly have been forced to “go on diversion” — routing ambulances and new patients to other hospitals.
For many of those patients — the majority of whom require immediate care — even just a few minutes can make a significant difference in their health, but by the time most hospitals announce a diversion, long wait times have become inevitable. Our recent work, however, shows that it doesn’t have to be this way. By making use of future information to develop proactive diversion policies, we estimate that hospitals could decrease wait times for treated patients by up to 15 percent, saving time — and lives.
The use of predictive analytics in healthcare has grown rapidly in just a few years. Much effort has been spent developing models to predict patient risk of bad outcomes, such as admissions (and/or readmissions). These models have largely been developed with the intent to guide operational decision-making and allow hospital administrators and clinicians to better utilize limited healthcare resources.
While the development of these models has been met with substantial hype, as yet, too little has been done to demonstrate whether — and how — they can actually be utilized to improve healthcare at the system level.
For patients, even small differences in wait times could have profound impacts. One study from 2013 found that patients who arrived during times of high ER crowding had a 5 percent higher chance of in-patient death. They also faced longer stays and higher costs than other patients.
Hospital diversions are intended to help patients get care faster by directing them away from overcrowded ERs and towards facilities that can care for them. In current practice, diversion decisions are typically made only based on information about current congestion — once a maximum threshold is reached, patients will be diverted. But by utilizing predictions of patient arrivals to determine when congestion is likely to build in the future, hospitals could strategically divert patients — announcing a diversionand minimize delays.
Intuitively, our proposed policy diverts patients at the beginning of an episode of high patient arrivals — known to operations researchers and network engineers as a “bursty” period. Because these patients, if admitted, would delay all subsequent arrivals during the bursty episode, diverting them can help reduce the overall delay faced by the ER.
By contrast, an ER relying exclusively on present information can start diverting only after a bursty period has materialized, at which point long wait times have already become inevitable. By switching to proactive diversion, hospitals can treat the same number of patients, while incurring much lower wait times.
One of the challenges with using any predictive model to guide operational decisions is that they can only provide noisy estimates of actual demand realizations. When it comes to healthcare, that could mean diverting a critically ill patient to a more distant hospital on account of expected patients who never actually arrive — with very real consequences.
Under our proposed policy, however, diversion decisions based on noisy predictions will never increase delays for admitted patients — in most cases delays will even be reduced. These gains though may be achieved at the expense of over-diversion.
To minimize that risk, each emergency room can be associated with a quantifiable noise tolerance — a range within which reduced delays can be guaranteed. Our models, however, are surprisingly robust. Even when using a predictive model with very high noise and a predictive capacity only minimally better than simple average arrival rates, wait times can be reduced by up to 8 percent compared to current best practice.
Truly eliminating the over-crowding and long wait times that plague emergency departments — not just in this country, but around the world — is likely to require an end-to-end transformation of the way care is delivered across the entire the hospital. As our work demonstrates, however, predictive analytics offer opportunities to make meaningful improvements within the existing healthcare system, ensuring patients receive the care they need when they need it.
A version of this article originally appeared in Manufacturing & Service Operations Management Review under the title “Reducing Waiting Times in the Emergency Department via ‘Diversion’.”
About the researcher
Professor Chan teaches the core MBA class, Operations Management. Her primary research interests are in data-driven modeling of complex stochastic systems, dynamic optimization, and...Read more.