Multilocation, Dynamic Staff Planning for a Healthcare System: Methodology and Application

Multilocation, Dynamic Staff Planning for a Healthcare System: Methodology and Application This study develops and implements a robust optimization model for dynamically assigning anesthesiologists across multiple hospitals in a large health system. The model addresses uncertainty in surgical demand by incorporating a three-stage decision process: presurgical location assignments, midhorizon on-call deployment, and day-of realization of overtime or idleness. The authors formulate the problem as a multistage robust mixed-integer program and solve it efficiently using a novel nested column and constraint generation algorithm. The approach accounts for location-specific constraints and fairness in on-call duties using historical data to estimate demand uncertainty with a calibrated trade-off between optimality and robustness. Implemented at the University of Pittsburgh Medical Center, the model reduced annual staffing costs by 12% or roughly $800,000 compared to existing practice. The framework is generalizable to other staffing problems in healthcare and provides operational insights on the value of forecast accuracy, location flexibility, and equitable workload distribution.