The Stronger Test: Real Triage Decisions Under Repeated Load

The strongest evidence in this series comes from a focused test of healthcare professionals making real triage judgments under repeated workload. The study asks whether specialized nurses become more reliant on personal defaults or assign higher urgency when cumulative effort is plausibly greater, using comparisons designed to separate workload from ordinary time-of-day changes. Its value is not that one study can settle every question about fatigue, but that it makes a practical version of the hypothesis precise enough to support a meaningful null result.

A demanding setting with a planned comparison

The setting was a national Swedish telephone triage and medical-advice service. Specialized nurses handled patient calls, listened to symptoms, asked questions, assessed the information, and assigned an urgency level for how quickly the patient should receive in-person medical attention. The work was both repetitive and qualified: calls arrived in sequence, yet each still required a professional judgment under uncertainty.

That combination makes the setting useful for the question. If decision fatigue is a practically relevant domain-general effect, it should have a chance to appear in work that is hard enough to require sustained information processing and repetitive enough to create cumulative load. At the same time, a service with detailed records can show when calls occurred, how they were ordered, which nurse handled them, and what urgency was assigned. The data do not expose every mental state, but they make the sequence of work visible.

The target data contained 231,076 calls involving 174 nurses. The study was not a laboratory exercise in which researchers instructed participants to become tired. It was a retrospective field analysis with quasi-experimental components. Patients were assigned to nurses in pseudorandom order as calls arrived, and shifts began at staggered times. This created opportunities to compare workers who were handling similar streams of calls but had accumulated different amounts of work.

The researchers selected two comparison points in advance. The first was an overlap between morning and afternoon shifts from 13:00 to 17:00. During that window, a nurse on the morning shift had already worked for much longer than a nurse on the afternoon shift, while both groups were taking calls in a similar clock period. The second comparison used intense call sequences before and after an 8–40 minute break. Calls immediately before a break were treated as plausibly higher in cumulative effort than calls after the break.

These comparisons do not create a randomized assignment of fatigue. Scheduling still determines who is present, and the investigators had to infer work states from observed call sequences. But they improve on a simple morning-versus-evening graph. The overlap comparison narrows the clock-time difference while separating time at work. The break comparison keeps the before-and-after observations close in the workday while using a pause as a potential reset in cumulative effort.

The design was also constrained before the new target data were available.