Institute of Work Health and Safety / Conditions
Fatigue is a hazard with a supply chain
The roster is a control. It is usually designed by somebody else.
Fatigue degrades judgement, reaction and risk perception in ways that are well characterised and consistently underweighted, largely because the affected person is the least reliable judge of their own state. A tired worker does not usually feel dangerously impaired. They feel like themselves, slightly slower, entirely capable of a task they have done a thousand times.
What makes it organisationally awkward is that the main controls sit outside the safety function. Roster design, shift length, overtime approval, travel time, and the scheduling pressure that produces a run of long days are operational and commercial decisions. A fatigue policy that asks individuals to self assess and declare is asking the least informed party to control a hazard created upstream.
The controls that work are structural: limits on consecutive shifts and total hours that are enforced by the scheduling system rather than by good intentions, minimum breaks between shifts that account for commuting, and a rule about who may not be asked to do certain tasks after a certain point. All of these cost money and are effective for that reason.
It is also worth separating the two problems that get discussed together. Acute fatigue from a long shift is different from chronic sleep debt accumulated over a roster cycle, and they respond to different controls. A generous break in a single shift does nothing for the second.
There is a related point about the interaction between fatigue and every other control. A tired workforce is more likely to omit a step, misread an instrument, or accept a workaround, which means fatigue degrades the effectiveness of controls that were assessed on the assumption of an alert operator. It is therefore not one hazard among many but a multiplier on several, and treating it as a line item in a register understates what it does to everything else recorded there.