An exploration of the utility of mathematical modeling predicting fatigue from sleep/wake history and circadian phase applied in accident analysis and prevention: the crash of Comair Flight 5191.

On 27 August 2006 at 0606 eastern daylight time (EDT) at Bluegrass Airport in Lexington, KY (LEX), the flight crew of Comair Flight 5191 inadvertently attempted to take off from a general aviation runway too short for their aircraft. The aircraft crashed killing 49 of the 50 people on board. To better understand this accident and to aid in preventing similar accidents, we applied mathematical modeling predicting fatigue-related degradation in performance for the Air Traffic Controller on-duty at the time of the crash. To provide the necessary input to the model, we attempted to estimate circadian phase and sleep/wake histories for the Captain, First Officer, and Air Traffic Controller. We were able to estimate with confidence the circadian phase for each. We were able to estimate with confidence the sleep/wake history for the Air Traffic Controller, but unable to do this for the Captain and First Officer. Using the sleep/wake history estimates for the Air Traffic Controller as input, the mathematical modeling predicted moderate fatigue-related performance degradation at the time of the crash. This prediction was supported by the presence of what appeared to be fatigue-related behaviors in the Air Traffic Controller during the 30 min prior to and in the minutes after the crash. Our modeling results do not definitively establish fatigue in the Air Traffic Controller as a cause of the accident, rather they suggest that had he been less fatigued he might have detected Comair Flight 5191's lining up on the wrong runway. We were not able to perform a similar analysis for the Captain and First Officer because we were not able to estimate with confidence their sleep/wake histories. Our estimates of sleep/wake history and circadian rhythm phase for the Air Traffic Controller might generalize to other air traffic controllers and to flight crew operating in the early morning hours at LEX. Relative to other times of day, the modeling results suggest an elevated risk of fatigue-related error, incident, or accident in the early morning due to truncated sleep from the early start and adverse circadian phase from the time of day. This in turn suggests that fatigue mitigation targeted to early morning starts might reduce fatigue risk. In summary, this study suggests that mathematical models predicting performance from sleep/wake history and circadian phase are (1) useful in retrospective accident analysis provided reliable sleep/wake histories are available for the accident personnel and, (2) useful in prospective fatigue-risk identification, mitigation, and accident prevention.