Supported Scales

The SAFE predictive model suite includes seven scales: one for fatigue, two for sleepiness, and four dedicated to performance metrics.

The Samn Perelli scale (Fatigue)

In their original paper on linking scores to fatigue, (Samn, SW and Perelli, LP. Estimating Aircrew Fatigue: A Technique with Application to Airlift operations. Report SAM-TR – 82-21. December 1981. USAF School of Aerospace Medicine.) authors Samn and Perelli classified their Samn-Perelli (SP) scores between 5 and 6 as being associated with moderate to severe fatigue with some performance impairment possibly occurring. They considered flying duties permissible but not recommended.

Initially designed for military pilots, the scale was based on data gathered during computer-simulated airline operations. The scale is shown below.

Samn Perelli Subjective Fatigue Scale

1. Fully Alert, wide awake
2. Very lively, responsive, but not at peak
3. Okay, somewhat fresh
4. A little tired, less than fresh
5. Moderately tired, let down
6. Extremely tired, very difficult to concentrate
7. Completely exhausted, unable to function

The advantage of using this scale is that the scale is validated with studies of tired aircrew in simulators. Moreover, the descriptions are easy to understand.

The Karolinska Sleepiness Scale (Sleepiness)

Designed by the Karolinska Institute in Stockholm, the Karolinska Sleepiness scale (KSS) started its life as a 9-point scale but with definition covering only alternative levels as follows:

1. Extremely alert
2.
3. Alert
4.
5. Neither sleepy nor alert
6.
7. Sleepy - but no difficulty remaining awake
8.
9. Extremely sleepy - fighting sleep

In time, the definitions of each level were revised and published with the missing levels completed.

Karolinska Sleepiness Scale Definitions

1.  Very alert
2.  Very lively, responsive, but not at peak
3.  Alert, normal level
4.  Okay, somewhat fresh
5.  Neither alert nor sleepy
6.  Moderately tired, let down
7.  Sleepy, but no effort to keep awake
8.  Extremely tired, very difficult to concentrate
9.  Very sleepy, great effort to keep awake

More recently, a 10th level was included to match level 7 on the Samn Perelli scale.

10. Extremely sleepy, cannot keep awake

The advantage of using the KSS scale is its validation in studies using an electroencephalogram (EEG) to measure brain activity.

As the SP scale is a 7-point scale and the KSS is a 9/10-point scale, some users believe the two to be interchangeable when measuring either fatigue or sleepiness in a fatigue management programme. However, SP scale is a fatigue scale and the Karolinska Scale is a sleepiness scale.

The 100-point alertness scale – developed by FRMSc

FRMSc team developed this scale in the 1970s to identify and measure the fatigue experienced by aircrew of (initially) military aircraft. This is a linear 0-100 scale where 0 indicates low alertness, and 100 indicates high alertness.

As no fatigue scales were available when our original research programme started, a scale was required before anything could be measures. Two points were created – one at 0 and the other at 100, that defined the range of likely alertness change experienced by commercial aircraft pilots at that time as they went about their occupation.

Since then, the range of alertness challenges has changed. Even with the emergence of other scales such as the SP scale, the 100-point Alertness Scale (AS) is still relevant. However, the Samn Perelli scale is easier to use. We suggest clients to use SP fatigue scale.

Karolinska Probability Scale

Some biomathematical models have only one score for each duty. In these circumstances it is useful to know how the fatigue increases across the duty, specifically whether the fatigue or sleepiness at the end of the duty is at the higher levels. The Karolinska probability scale has been created to indicate whether the stated duty sleepiness metric is likely greater or equal to 8 at the end of the duty.

There is no equivalent in the SP scale. A typical output from the SAFE model would be to declare “At this mean level of alertness, the probability of a value of at least 8 on the Karolinska Sleepiness Scale is 0.15”

Vigilance Degradation scale

Vigilance Degradation scale measures visual vigilance degradation. It compares visual vigilance during a standard laboratory test to that of the personnel to identify if fatigue caused a reduction in vigilance. The SAFE model provides an indication of how much visual vigilance is degraded by fatigue when using that test.

Complex Reaction Degradation scale

The Complex Reaction Degradation scale compares the speed of reaction during a standard laboratory test to that of the personnel when fatigued. The SAFE model indicates degradation when using that test by declaring, “At this level of alertness, the response time on a warning light on a complex task is degraded by 36.1% compared with a typical rested value”.

Percentage Missed Response scale

The Percentage Missed Response scale does not suggest that the aircrew may miss an ATC call, as aircrew are sensitive to these and, unless extremely fatigued, will recognise these calls. Like the Vigilance Degradation and Reaction Degradation scales, the Percentage Missed Response scale is based on a standard laboratory test and indicates the likely general degradation of human performance.

The SAFE model provides an indication of likely degradation when using that test, “At this level of alertness, the percentage of missed responses in a sustained attention task is 17.64%. This compares with a typical rested value of 6%”

There is also an 8th scale for comparing human performance when fatigued with when having ingested alcohol. This scale was removed from the SAFE models after consideration of its overall value given that the performance degradation due to alcohol intoxication or fatigue are not adequately coherent.

There are some similarities with the degradation of vigilance when fatigued and after drinking alcohol, but in other ways, performance is not the same. When fatigued, humans speak little, but after drinking alcohol, humans become garrulous.

Overall, we suggest there are not enough similarities for the blood-alcohol scale to be helpful. Consequently, we have removed it from our models.

Risk Scales

Duty Risk Metric

FRMSc has developed a Duty Risk metric to resolve the difficulties in identifying the likelihood of a reportable incident occuring due to fatigue levels in aircrew.

This metric identifies duties that present as higher risk but are not easily identifiable by fatigue alone, e.g. consecutive early morning starts.

It has been designed to serve as a single composite metric that considers the physiological response of the aircrew to fatigue levels they experience when performing their duties. It also considers the likely risk to operations at any given point from the nature of the undertaken task and the corresponding fatigue score.

Like the fatigue scales, each operation will determine its specific categorisation of the Duty Risk Metric. Our recommendation is for pilots, to consider 1 to 13 being acceptable risk, 13-18 being heightened but still acceptable risk with caution, and 18+ being high risk for pilots. For cabin crew, we would suggest 1-15 being acceptable risk, 15 to 20 being heightened but still acceptable risk, with caution.

The metric will read zero if the aircrew is stationary or involved with duties that contain no flights, but it will read high, for example, if the pilots are landing when the fatigue levels are high.

Similarly, during the cruise phase, the risks of a reportable incident for pilots will be reduced, reducing the risk score. It means that even if the fatigue levels are relatively high, the overall risk is relatively low during the cruise phase compared with the landing phase. But, if a pilot has flown multiple sectors and landed numerous times, they will also have a higher risk score than an equivalent pilot flying a single-sector duty.

The reportable incidents are different for pilots and cabin crew. For cabin crew, the cruise phase is difficult as unruly passengers and maleficent actors may choose that phase of the flight to create a nuisance, increasing cabin crew's risk score. However, the risk score is usually lower for them during take-off and landing unless circumstances cause a difficult take-off or landing.

Pilots have fewer reportable incidents, but those may have a high impact. Cabin crew may have more reportable incidents, but from a safety perspective, the impacts of these incidents on the risk score are likely to be lower than those experienced by pilots. The cabin crew incidents might lower the passenger satisfaction, which is crucial for the marketing department that aims to maintain both passenger safety and loyalty.