We study patient flow management in an inpatient department of a Singaporean hospital. We focus on understanding the effect of an "early discharge" policy, implemented in late 2009, on fraction of patients who have to wait six hours or longer in the emergency department (ED) to get a bed. We propose a novel stochastic network model that has the following critical features: (1) A patient's service time is endogenous, depending on her admission and discharge times, and the number of nights she stays. As a consequence, the service times are not independent, identically distributed. (2) There are pre- and post-allocation delays for each bed-request even if a bed is available at the time of request; these allocation delays allow one to model secondary bottlenecks such as temporary nurse shortage at certain hour of a day. (3) When a patient is waiting for a bed, she can be overflowed to a "wrong ward" when her overflow trigger time reaches a certain threshold. We show that our model is able to capture the hourly waiting time statistics of the inpatient operation. The model allows one to evaluate the impact of operational policies on waiting times and overflow rates. In particular, our model predicts that a hypothetical, Period 3 policy can eliminate the excessively long waiting time for morning bed-requests at this hospital.
This is joint work with Pengyi Shi (Georgia Tech), Ding Ding (University of International Business & Economics, Beijing), James Ang and Mabel Chou (NUS), and Jin Xin and Joe Sim (NUH).