Neonatal Physician Scheduling at the University of Tennessee Medical Center

A common issue faced by physician groups is how to schedule 24-7 coverage for hospital units such as an emergency department. The first step is to determine the shifts to be covered. The second step, assigning physicians to specific shifts, is complicated because shifts vary with respect to duration, day of week, time of day, and desirability. To ensure workload fairness, physician groups often create “equality” schedules in which they evenly divide shifts, by type, among physicians. This problem can be readily modeled and solved via optimization. This paper presents a novel approach that incorporates individual physician shift-type preference and seeks, for each physician, a schedule that is superior to his or her equality schedule. We formulate and solve the problem as a binary, mixed-integer program designed to maximize relative gains in individual physician preference. We describe the methodology and real-world implementation within a neonatal intensive care group, and present a hybrid version of the model that is capable of simultaneously accommodating physicians who prefer an equality schedule and physicians willing to deviate from the equality schedule in pursuit of a schedule that better fits their shift preferences. Increases in schedule preference for the latter group ranged from 6.3 to 8.5 percent.