Comparison of in-flight measures with predictions of a bio-mathematical fatigue model.

INTRODUCTION: Bio-mathematical models are increasingly used for predicting fatigue in airline operations, and have been proposed as a possible component of fatigue risk management systems (FRMS). There is a need to continue to evaluate fatigue models against data collected from crews conducting commercial flight operations. METHODS: A comparison was made between several in-flight studies of pilot fatigue, conducted over a 10-yr period on a variety of operations, and the predictions of a widely used bio-mathematical model, the System for Aircrew Fatigue Evaluation (SAFE). The in-flight studies collected a variety of subjective ratings as well as reaction time on a performance task. RESULTS: Overall correlation between observed and predicted fatigue was stronger for subjective fatigue than reaction time. More detailed analysis on selected studies shows discrepancies between predicted and observed fatigue, which may be explained by a variety of confounders. Closer analysis of the duty time, time of day, and schedule length show discrepancies of up to 15% between observed and predicted fatigue. DISCUSSION: This study provides comparison between the predictions of one bio-mathematical model, SAFE, and observed fatigue measures across a number of operations. Possible causes of discrepancies are discussed. There is potential for more comparison studies of this type with the various available models.

Automated collection of fatigue ratings at the top of descent: a practical commercial airline tool.

INTRODUCTION: There is a need to develop an efficient and accurate way of assessing pilot fatigue in commercial airline operations. We investigated the validity of an automated system to collect pilot ratings of alertness at the top of descent, comparing the data obtained with existing results from previous studies and those predicted by the validated SAFE fatigue model. METHODS: Boeing 777 pilots were prompted to enter a Samn-Perelli fatigue scale rating directly into the flight management system of the aircraft shortly prior to descent on a variety of short- and long-haul commercial flights. These data were examined to evaluate whether the patterns were in line with predicted effects of duty length, crew number, and circadian factors. We also compared the results with data from previous studies as well as SAFE model predictions for equivalent routes. RESULTS: The effects of duty length, time of day, and crew complement were in line with expected trends and with data from previous studies; the correlation with predictions from the SAFE model was high (r = 0.88). Fatigue ratings were greater on longer trips (except where mitigated by adding an extra pilot) and on overnight sectors (4.68 vs 3.77). DISCUSSION: The results suggest that the automated collection of subjective ratings is a valid way to collect data on fatigue in an airline setting. This method has potential benefits for the crew in assessing fatigue risk prior to approach, as part of a fatigue risk management system, with the possibility of wider safety benefits.

Fatigue in airline pilots after an additional day's layover period.

INTRODUCTION: We investigated the effect of an additional day's layover on reducing fatigue in two different duties: a two-pilot crew flying between Auckland and Brisbane, and a three-pilot crew flying between Auckland and Los Angeles. METHODS: Pilots completed a reaction time task, the Samn-Perelli fatigue scale, and the Karolinska Sleepiness Scale on both outward and return flights. The flights were conducted with and without a 1-d layover (Brisbane) and with a 1- or 2-d layover (Los Angeles). RESULTS: On the Brisbane route, the addition of a layover resulted in a significant reduction of fatigue, sleepiness, and reaction time. At top of descent, Samn-Perelli fatigue was reduced from over 5.0 to under 4.5. In contrast, the addition of an extra day layover in Los Angeles had no significant effect on the same measures during the return flight; on both flights Samn-Perelli fatigue was over 5.0 at top of descent. DISCUSSION: The results suggest that the addition of an extra night's layover has different effects depending on the type of operation. Layover periods need to ensure adequate opportunity to recover from any sleep deficit arising from the outbound journey, but the benefit of increased layover time may be limited if time-zone shifts cause a mismatch between local time and the circadian rhythm of sleep.

Fatigue in two-pilot operations: implications for flight and duty time limitations.

OBJECTIVES: Two-pilot operations make up the majority of commercial flights. Fatigue is an important consideration in these operations as there is little opportunity for in-flight rest. We investigated the role of duty length, time of day, and whether one or two sectors were flown on reported fatigue at the top of descent in two-pilot regional operations. METHODS: Pilots flying two-pilot operations ranging from 3-12 h completed Samn-Perelli fatigue ratings prior to descent at the end of each rostered duty over a 12-wk period. We collected 3023 usable ratings (72% of rostered duties) comprising 26% single and 74% double sector duties. RESULTS: We found that time of day has a marked effect on the pattern of fatigue at the start of the duty and on the rate at which fatigue levels increased, with the highest levels in the window of circadian low (0200-0600). Fatigue also increased with the length of duty and was 0.56 higher at the end of a two-sector compared with a single-sector duty. DISCUSSION: The results imply authorities should consider increasing existing limits for daytime duties and reducing those for nighttime two-pilot operations.

Pilot fatigue in short-haul operations: effects of number of sectors, duty length, and time of day.

INTRODUCTION: There is little research on what factors are associated with fatigue in short-haul pilots. The aim was to investigate how length of duty, number of sectors, time of day, and departure airport affect fatigue levels in short-haul operations. METHODS: Pilots completed Samn-Perelli fatigue ratings prior to descent at the end of each rostered short-haul duty over a 12-wk period. Overall, 1370 usable responses were collected (67% of rostered duties) and fatigue scores were examined in relation to the departure airport, the number of sectors flown, time, and the length of duty period. RESULTS: The most important influences on fatigue were the number of sectors and duty length. These were associated with fatigue in a linear fashion. Time of day had a weaker influence, with lower levels at midday and increased fatigue later in the day. Fatigue was also higher during duties originating from an airport where pilots needed to position the night before and spend the night in a hotel. DISCUSSION: Data from the study enabled the quantification of fatigue at this critical phase of flight in duties lasting between 2 and 12 h and finishing between 08:00 to 24:00. The data obtained may be useful for identifying factors associated with fatigue, evaluating existing fatigue models, and identifying or predicting problem duties within an airline operation. The methodology used in the study may be successfully applied to gather fatigue data in other airline operations.

Estimating the circadian rhythm in the risk of occupational injuries and accidents.

The authors recently published a prototypic Risk Index (RI) to estimate the risk of critical errors associated with shift systems. This RI was based on published trends in the relative risk of injuries and accidents, and a simple additive model was proposed to estimate the risk for a given shift system. However, extending the RI to irregular work schedules requires an estimation of the phase and amplitude of the circadian rhythm in risk. This paper integrates the published evidence on three independent sources of data that allow such estimations to be made: the trend in risk over a 24 h day, over the course of the night shift, and across the three different (8 h) shifts. Despite potential confounders, maximum risk (i.e., acrophase = peak time) estimates across these three trends showed a remarkable consistency, with all three estimates occurring at about midnight, although the amplitude estimates varied considerably. The best estimate of the amplitude of the circadian rhythm in risk would appear to be that based on trend over the three (8 h) shifts, as this trend is the least confounded. The estimated acrophase (peak time) in risk appeared earlier than would be predicted from consideration of the circadian rhythm in alertness, fatigue, or performance on simple interpolated tasks, such as reaction time or performance on the Psychomotor Vigilance Test.