Knowella

Shift Fatigue Screening Template

Biomathematical fatigue models take scheduled hours as their input and produce a number about an average person. Neither of those is the thing you need to know. The roster does not record that someone commuted ninety minutes each way and slept four hours, and the average person is not the individual who is about to drive home.

KnowErgoAssessmentERG-024Pinned in navigationFull guide
Models predict
The average person
Input is
The roster, not the sleep

Summary

In short

  • Model outputs describe an average person and do not account for individual differences, task type or work context. Aviation regulators state this explicitly.
  • Guidance from CASA and IATA is unusually direct: model predictions should never be the sole basis for operational fatigue decisions, and cannot provide a green light for safety.
  • Most models take scheduled hours as input rather than sleep actually obtained. A roster that models well conceals a long commute, a second job, a young child or insomnia.
  • The two physiological drivers are sleep history and time of day. A comparative study concluded it is not clear current models provide more information than an expert with a firm grasp of both.
  • Different models take different inputs and produce different scales, so a score from one cannot be compared against a threshold set for another.
  • The model belongs inside a fatigue risk management system with reporting, monitoring and controls, providing a supporting role rather than the answer.

What it is

What it is

What is shift fatigue screening?

Assessment of a shift pattern and the work performed on it to identify where fatigue risk is concentrated, using shift design factors, task demands, observed indicators and existing controls, and where warranted a biomathematical model.

What is a biomathematical fatigue model?

A computational model estimating fatigue from work and sleep schedules, based on sleep history and circadian effects. Examples include the HSE Fatigue Risk Index and SAFTE. They take different inputs and produce different scales, so results from different models are not directly comparable.

When to use it

When to use it, and when not to

This screens a shift pattern and the work on it. Individual fitness for duty is separate.

Use it for

  • Assessing a shift pattern before implementation or after a change
  • Identifying which shifts and which points within them carry concentrated fatigue risk
  • Reviewing patterns following incidents, errors or near misses clustered by time
  • Screening overtime, callout and extended-hours arrangements
  • Establishing where controls are needed and what kind

Not for

  • Individual fitness for duty assessment, which addresses a person on a day
  • Working time compliance, which is a legal limit rather than a fatigue assessment
  • Occupational health referral for sleep disorders or shift intolerance
  • The fatigue risk management system itself, which this feeds
  • Journey risk and driving hours assessments, which have their own regimes

Standards

What it is built against

Fatigue is regulated prescriptively in transport and through general duties elsewhere.

ClauseRequirementWhere it lands
ISO 45001 cl.6.1.2Hazard identification including work organisation, hours, shift patterns and human factorsShift design
ISO 45003Psychosocial risk management including workload, work scheduling and shift workTask demands
HSE Fatigue Risk IndexScreening tool assessing shift schedules for fatigue and risk, based on sleep history and circadian factorsShift design
49 CFR 395Hours of service for commercial drivers, a legal limit distinct from a fatigue assessmentHeader
Working Time RegulationsLimits on working hours, night work and rest periods (GB and EU)Header
ISO 45001 cl.8.1.2Hierarchy of controls, applied to schedule design before reaching alertness measuresControls
ISO 45001 cl.5.4Worker consultation on arrangements affecting them, including shift designIndicators
ISO 45001 cl.10.2Incident investigation, including whether fatigue contributedOutcome

What it does not cover

  • Individual fitness for duty assessment, addressing a person on a day.
  • Working time compliance, a legal limit rather than a fatigue assessment.
  • Occupational health referral for sleep disorders or shift intolerance.
  • The fatigue risk management system, which this feeds.
  • Driving hours and journey risk assessments, with their own regimes.

Filling it in

Filling it in well

Screen the design, ask about sleep rather than hours, and treat the model as one input.

Assess the shift design factors directly

Shift length, start and finish times, direction and speed of rotation, consecutive shifts, rest between shifts, night shift frequency, overtime and callout, and predictability. These are the levers that change fatigue, and most of them can be assessed without any model at all because the physiology is well established.

Ask about sleep obtained, not hours rostered

The model's input is the schedule. The relevant quantity is sleep actually achieved, which is affected by commute, caring responsibilities, second jobs, noise, daylight and shift-related insomnia. A pattern that models acceptably can produce four hours of sleep in practice, and only asking will reveal it.

Match the task to the time

Fatigue affects tasks unequally. Monotonous monitoring, complex decision-making and driving degrade fastest, while physically active varied work degrades more slowly. Scheduling the most fatigue-sensitive tasks away from the circadian low and the end of long shifts is a control that costs nothing but planning.

Work the hierarchy on the schedule

Change the pattern, change the rotation, change task allocation, before reaching for napping policies, caffeine guidance and alertness monitoring. Schedule design is the engineering control here, and programmes that skip to individual countermeasures have moved the problem onto the person who is already tired.

Audit findings

Common audit findings

Fatigue findings concentrate on model misuse and on what the schedule actually produces.

FindingClauseWhat fixes it
Model output treated as clearance for a pattern.CASA guidanceModels cannot give a green light; they are one control among several.
Scores from different models compared against a common threshold.HSE FRIModels take different inputs and produce different scales; they are not interchangeable.
Sleep obtained never asked about, only hours rostered.ISO 45003Commute, caring, second jobs and insomnia all sit outside the roster.
Fatigue-sensitive tasks scheduled into the circadian low.ISO 45001 cl.8.1.2Task allocation by time of day is a control costing only planning effort.
Controls limited to individual countermeasures.ISO 45001 cl.8.1.2Schedule design is the higher-order control; countermeasures come after.
Hours compliance treated as evidence of adequate fatigue management.49 CFR 395A legal limit is not a fatigue assessment; compliant patterns can be dangerous.
No fatigue reporting route, so the model has no field data to check against.ISO 45001 cl.5.4A system needs reported fatigue to validate its predictions.
Incidents not analysed by time of day and position in shift.ISO 45001 cl.10.2Clustering by hour and shift position is the most direct fatigue evidence available.
Commuting after night shifts not addressed.ISO 45001 cl.6.1.2The drive home follows the shift and is frequently the highest-risk period.
Pattern changed without consultation with those working it.ISO 45001 cl.5.4They hold the information about sleep achieved on each variant.

Worked case

Case in point: what the roster could not see

A site modelled a proposed twelve-hour rotating pattern and obtained results within acceptable limits. The pattern was implemented and the modelling was cited as evidence that fatigue had been assessed.

A subsequent survey of the crew found the average sleep obtained before a night shift was substantially below what the model had assumed. Several people commuted over an hour each way, two had caring responsibilities in the afternoon before nights, and daytime sleep in summer was poor for most of the crew.

None of that information was available to the model, because its input is the schedule. The pattern was, on paper, acceptable. What people were actually sleeping was not.

Definitions

Definitions and key terms

Biomathematical model
A computational estimate of fatigue from work and sleep schedules, based on sleep history and circadian rhythm.
Sleep history
Time awake and accumulated sleep debt, one of the two core physiological drivers of fatigue.
Circadian low
The period of the internal daily cycle when alertness is lowest, typically in the early hours.
Sleep opportunity
Time available for sleep according to the schedule, which is not the same as sleep obtained.
Fatigue risk management system
A multi-layered system of controls, reporting and monitoring within which a model provides a supporting role.
Rotation direction
Whether shifts advance forwards or backwards through the day, which affects adaptation.
Sleep inertia
Impaired performance immediately after waking, relevant to napping arrangements and callouts.
Fatigue reporting
A route for people to report being too tired to work safely, without which a system has no field data.

FAQ

Frequently asked questions

Can a fatigue model clear a shift pattern?+

No, and authoritative guidance says so directly. Civil aviation guidance states that biomathematical models cannot provide a green light for operational safety and should be one of several controls, complemented by monitoring and practices ensuring adequate rest. Industry guidance adds that predictions should never be the sole basis for operational fatigue decisions.

Why not?+

Because models predict fatigue for the average person, without accounting for individual differences, task type or work context. They also take the schedule as input rather than sleep actually obtained, so a pattern that models acceptably can still be producing severe fatigue in people whose sleep is constrained by commute, caring responsibilities or shift-related insomnia.

Can we compare scores between models?+

No. Different models take different inputs and produce outputs on different scales, so a threshold set for one tool means nothing applied to another. This matters when an organisation changes tool or compares itself to a published figure, because the numbers look comparable and are not.

What should we assess without a model?+

Shift length, start and finish times, rotation direction and speed, consecutive shifts, rest between shifts, night frequency, overtime, callout and predictability. The physiology behind these is well established, and a comparative study concluded it is not clear current models add more than an expert with a firm grasp of sleep history and time-of-day effects.

What is the strongest control?+

Schedule design, followed by task allocation by time of day. Moving fatigue-sensitive work away from the circadian low and the end of long shifts costs planning effort only. Napping policies, caffeine guidance and alertness monitoring come after, and a programme that starts there has placed the problem on the person who is already tired.

The agents

What the agents do with it

The screen assesses a pattern. What fails is a model output treated as clearance and sleep nobody asked about.

KnowErgo

Screens shift design factors directly, records sleep obtained alongside hours rostered, and keeps model output as one input among several.

Ella

Analyses incidents and errors by hour and position in shift, which is the most direct fatigue evidence an operation holds.

KnowSafe

Provides a fatigue reporting route so predictions can be checked against what people actually experience.

KnowLogistics

Allocates fatigue-sensitive tasks away from the circadian low and the end of long shifts, which is a scheduling control.

This template lives in KnowErgoergonomics. Task assessment, video posture analysis, rotation and workstation redesign.

Meet KnowErgo

Sources

Sources

Start in Minutes, Not Weeks

Launch a Ready-Made Template and Customize It Your Way

Every template is fully editable. Adjust fields, workflows, and branding to match your processes, then deploy to your team instantly.