A Stochastic Optimization Approach for Scheduling Surgeries in Parallel Operating Rooms

Abstract

The continuous production of reliable operating room (OR) schedules greatly influences the overall performance of surgery delivery systems in most hospitals. We develop a two-stage stochastic integer programming model that schedules surgeries into multiple ORs in order to minimize surgeon and operating room inefficiency and patient waiting time. We consider elective patients and uncertainty in surgery durations. The Kolmogorov-Smirnov goodness of fit test is used to determine continuous distributions for stochastic analysis. Decisions are which surgeries to include in a daily schedule, surgery-to-OR allocations, surgery sequence within surgery listings and ORs and planned starting time for each surgery. Small instances are the basis for anecdotal support to the concept of parallel surgery processing. We also investigate the benefit of allowing patients to be treated earlier than scheduled in cases where preceding surgeries finish early. Our results indicate that a single surgeon operating in two parallel ORs may perform as much as 91.0% of the surgeries originally assigned to two surgeons. However, when surgeons operate in parallel ORs, we capture substantial risk of overtime and patient waiting time due to propagated delay (the VSS is 22.1% on average). Allowing patients to be treated 30 minutes earlier than scheduled reduces patient waiting time significantly and leads to an average cost reduction of 5.4%. We relate our research to the orthopedic department at st. Olavs Hospital in Trondheim, Norway, and recommend that they implement parallel surgery processing.