Professor, Kelley School of Business
Inidiana University
Friday, 3/6 @ 10:30am.
Title
Infection-Aware Nurse Staffing: Hidden-State Random Graphs and Lessons from Multy-Hospital System
Abstract
Hospitals face a painful paradox during infectious-disease surges both for cyclical outbreaks like flu season and larger-scale pandemics such as COVID-19: admissions spike while nurse absenteeism rises due to infection. While the existence of this phenomenon has been well-documented, the timing and sources of nurse infections are largely unobserved, making it difficult to craft effective staffing policies during infectious disease outbreaks. This work develops a dynamic random-graph framework with endogenous, hidden nurse health states (healthy, incubating, symptomatic) that links staffing policies to transmission dynamics and lets us infer who infected whom, where, and when from observed absenteeism patterns. Methodologically, we couple this network with an estimation procedure that maps nurse and unit characteristics to a classical disease transmission model (Susceptible-Infected-Recovered) considering patient to nurse, nurse to nurse, and community to nurse pathways.
We apply this model to data from a large 18-hospital system during the COVID-19 pandemic. The model suggests roughly 70% of nurse infections are from in-hospital sources (split nearly evenly between patient to nurse and nurse to nurse) and 30% from the community. Counterfactuals show that tuning workloads/schedules alone can cut infection-driven absenteeism by up to 25%, and a simple “on-call” coverage policy trims another 8% by avoiding spillover overload when a nurse calls in sick. We also quantify a key operational trade-off: excessive workload increase risk for those on the floor, while too low a workload exposes more nurses to infection opportunities. Finally, dedicating units only for infectious patients has limited benefit, but pairing unit dedication with a fixed nursing team (patient-nurse dedication) projects a reduction in total absenteeism of 16% and a reduction in nurse-to-nurse transmission by 26%.
These results offer a practical, data-driven framework for infection-aware staffing that strengthens health-system resilience during outbreaks.
For more information, contact: Maica Lane-Umali at opim@uconn.edu