Data-Driven Physician Staffing Saves Health Systems Over $800K Annually

Study finds UPMC's dynamic staffing model reduces overtime and idle time across 11 hospitals.

Published on Feb. 13, 2026

Got story updates? Submit your updates here. ›

A new study published in Operations Research examined a data-driven physician staffing model implemented at the University of Pittsburgh Medical Center's anesthesiology department. The model combines historical data and short-term forecasts to assign physicians to specific hospitals or a shared on-call pool, adjusting assignments as demand becomes clearer. After implementation across 11 UPMC hospitals, the system reduced daily overtime by nearly 13 hours and idle time by 14 hours, leading to net estimated savings of over $800,000 annually.

Why it matters

As health systems face ongoing workforce challenges and demand variability, data-driven staffing approaches like UPMC's can generate significant cost savings while also accounting for real-world constraints like physician credentialing and fairness in on-call rotations.

The details

The staffing model developed by researchers combines historical data with updated short-term forecasts to assign anesthesiologists to specific UPMC hospitals or a shared on-call pool weeks in advance, with adjustments made closer to the day of surgery as demand becomes clearer. The model also accounts for factors like which physicians are credentialed at certain facilities, fairness rules for on-call rotations, and uncertainty in projected surgery volumes and total anesthesia hours required.

  • The study was published in the February 2026 edition of Operations Research.

The players

University of Pittsburgh Medical Center (UPMC)

A large non-profit health system based in Pittsburgh, Pennsylvania that operates 40 hospitals and more than 700 clinical locations.

Kumar Rajaram, PhD

Study author and professor at the UCLA Anderson School of Management in Los Angeles.

Got photos? Submit your photos here. ›

What they’re saying

“Combining an on-call structure with robust, data-driven planning can substantially reduce overtime and idle time. Our approach also demonstrates how fairness constraints, such as ensuring no one is placed on consecutive on-call days, can be integrated without sacrificing efficiency.”

— Kumar Rajaram, Professor (News release)

What’s next

Researchers say the data-driven staffing framework could be applied to other areas of clinical staffing facing similar demand variability and workforce constraints, including nurse scheduling.

The takeaway

As health systems grapple with staffing challenges, UPMC's experience demonstrates how a data-driven, dynamic approach to physician scheduling can generate substantial cost savings while also addressing fairness and other operational constraints.