Thursday, November 1, 2012
Noon – 1.15 pm, BYENG 210
Crisis level overcrowding conditions in Emergency Departments (ED’s) have led hospitals to seek out new patient flow designs to improve both responsiveness and safety. We present the result of several years of our collaboration with a few hospitals to improve the patient flow. We consider two proposed innovations. The first innovation that has attracted attention and experimentation in the emergency medicine community is a system in which ED beds and care teams are segregated and patients are “streamed” based on predictions of whether they will be discharged or admitted to the hospital. We use a combination of queuing analysis, Markov decision processes, and high-fidelity simulation models calibrated with hospital data to determine whether such a streaming policy can improve ED performance, where it is most likely to be effective, and how it should be implemented for maximum performance. Our results suggest that the concept of streaming can indeed improve patient flow, but only in some situations. First, ED resources must be shared across streams rather than physically separated. This leads us to propose a new “virtual-streaming” patient flow design for ED’s. Second, to take full advantage of streaming, physicians assigned to admit patients should prioritize upstream (new) patients, while physicians assigned to discharge patients should prioritize downstream (old) patients. In the second innovation, we demonstrate that the current practice of prioritizing patients solely based on urgency is less effective than a new ED triage system that adds an up-front estimate of patient medical complexity to the conventional urgency-based classification. Using a combination of analytic models and simulation analysis, we show that the proposed complexity-based triage can substantially improve both patient safety and operational efficiency. Finally, we examine different ED patient flow designs and show that streaming patients based on complexity information and prioritizing them based the urgency level is better than streaming them based on urgency and prioritizing them based on complexity.
Soroush Saghafian is an Assistant Professor of Industrial Engineering at ASU. He has a Ph.D. in Industrial and Operations Engineering and a Master’s degree in Mathematics from the University of Michigan. His research focuses on the application and development of operations research methods in modeling and control of stochastic systems with specific applications in (a) healthcare, (b) control of queuing systems, and (c) operations management. He has won numerous awards for his research including the 2012 INFORMS MSOM (Manufacturing and Service Operations Management Society) Best Student Paper Award, 2012 Richard Wilson Prize, 2011 University of Michigan College of Engineering Outstanding Ph.D. Research Award, and 2010 INFORMS Pierskalla Award for the best paper in Healthcare. He has also been a finalist for INFORMS 2011 Doing Good with Good OR Award as well as the best student paper award of the POMS College of Supply Chain in 2009. Saghafian serves as a referee for various journals including Operations Research, Operations Research Letters, Naval Research Logistics, IIE Trans., IEEE Trans. on Evolutionary Computation, and Production and Operations Management.