A Markov Decision Process to Dynamically Match Hospital Inpatient Staffing to Demand

Published in: IIE Transactions on Health Systems Engineering, v. 1, no. 2, Oct. 2011, p. 116-130

Posted on RAND.org on October 01, 2011

by James R. Broyles, Jeffery K Cochran, Douglas C Montgomery

Read More

Access further information on this document at IIE Transactions on Health Systems Engineering

This article was published outside of RAND. The full text of the article can be found at the link above.

Appropriate inpatient staffing levels minimize hospital cost and increase patient safety. Hospital inpatient units dynamically adjust premium staffing (above base staffing) levels by attempting to match their daily demand. Historically, inpatient managers subjectively adjust daily staffing from observing the morning inpatient inventory. Inpatient units strive to match staff with demand in a complex patient throughput environment where service rates and non-stationary profiles are not explicitly known. Related queue control and throughput modeling literature do not directly match staffing with demand, require explicit service process knowledge, and are not formulated for an inpatient unit. This paper presents a Markov decision process (MDP) for dynamic inpatient staffing. The MDP explicitly attempts to match staffing with demand, has a statistical discrete time Markov chain foundation that estimates the service process, predicts transient inventory, and is formulated for an inpatient unit. Lastly, the MDP application to a telemetry unit reveals a computational myopic, an approximate stationary, and a finite horizon optimal policy that is validated through hospital expert experience. The application reveals difficult-to-staff inventory levels and shows that the removal of discharge seasonality can drastically decrease required size of the premium staffing pool and the probability of full occupancy thus improving the inpatient unit's patient flow.

This report is part of the RAND Corporation external publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.