Hospital Bed Refresh Has Entered Formal Review
A county hospital dashboard showed bed availability updated every four hours. The decision being made required current information. The information was four hours old.
Incident Report: Hospital Bed Refresh Latency Classification: Operational
Incident Summary
A regional hospital capacity dashboard, used to allocate transfer patients across facilities, displayed bed availability data at a refresh interval of four hours. During a period of elevated volume, decisions were made based on this data. Several transfers were directed to facilities that were at or near capacity at the time of transfer.
The data was accurate as of four hours prior.
System Description
The dashboard aggregates bed availability data from participating facilities. Data is pulled from each facility's internal system on a scheduled basis. The schedule was set at four hours at the time of implementation.
Four hours was the standard interval for administrative reporting purposes at the time the system was built.
Incident Analysis
A bed that was available four hours ago may or may not be available now. The dashboard cannot say. The dashboard says it anyway.
The number displayed carries no visual indicator of age. A bed availability figure of twelve looks identical whether it reflects current conditions or conditions from four hours ago. A person using the dashboard to make a time-sensitive decision has no mechanism for knowing which of these they are seeing.
Contributing Factors
The four-hour interval was not reviewed when volume increased. Review would have required identifying that the interval was a design parameter, locating the technical contact responsible for the system, submitting a change request, and waiting for the request to be processed.
This sequence was not completed before the incident. It has not been completed since.
The Governing Contradiction
The dashboard was built to provide visibility. It provides visibility into conditions as they existed four hours ago, which is a different thing from visibility into current conditions. The two are confused because they look identical on the screen.
A system designed to improve decision-making has introduced a new kind of error: confident decisions made on stale data.
Recommendation
Update the refresh interval to fifteen minutes or less. Add a timestamp indicating when data was last updated and a visual indicator when data exceeds a defined age threshold.
Status of Recommendation
Under review. A working group has been formed. The working group will meet monthly.
The dashboard continues to refresh every four hours.