Rostykus PS, Cummings P, Mueller BA. Risk Factors for Pilot Fatalities in General Aviation Airplane Crash Landings. JAMA. 1998;280(11):997–999. doi:10.1001/jama.280.11.997
From the Department of Epidemiology, University of Washington School of Public Health and Community Medicine (Drs Rostykus, Cummings, and Mueller); and the Harborview Injury Prevention and Research Center, University of Washington (Drs Cummings and Mueller), Seattle. Dr Rostykus is now with the Emergency Department, Ashland Community Hospital, Ashland, Ore.
Context.— Most pilots survive airplane crash landings in small airplanes. Factors
associated with pilot death have not been well studied.
Objective.— To identify factors associated with fatalities in general aviation airplane
Design.— Case-control study.
Setting.— The United States.
Subjects.— All pilots in general aviation crash landings of airplanes with 10 seats
or fewer, from 1983 through 1992.
Main Outcome Measure.— Pilot death.
Results.— Pilots died in 437 (5.2%) of 8411 crash landings. A fire or explosion
on the ground was strongly associated with pilot death (relative risk [RR],
20.4; 95% confidence interval [CI], 15.5-26.9), adjusted for pilot age, pilot
flight hours, type of landing gear, and the filing of an instrument flight
plan. Pilots who failed to use both lap belt and shoulder harness were more
likely to die (adjusted RR, 6.8; 95% CI, 1.8-25.5), as were those who used
only the lap belt (adjusted RR, 1.7; 95% CI, 1.3-2.2), compared with pilots
who used both restraints.
Conclusion.— Pilots may be able to reduce their risk of death in a crash landing
by using lap and shoulder restraints.
CIVIL (nonmilitary) flights are classified as either general aviation
or air carrier operations. Air carrier operations include passenger or cargo
transports for hire. General aviation comprises recreational flying, flight
instruction, agricultural operations, sightseeing, and business travel flown
in a variety of aircraft of all sizes and types, including airplanes, helicopters,
balloons, and gliders. The majority of civil aviation crashes, deaths, and
injuries are attributed to general aviation operations.
Although general aviation airplane crashes typically generate much media
coverage, they are infrequent occurrences. In 1992 there were 39.6 million
flight departures and 2075 crashes.1 In 78%
of the crashes there were no deaths and in 68% there were no injuries.2 Airplanes are designed with features that can dissipate
the kinetic energy of the occupants and minimize injury in the event of a
crash landing. If a crash landing is necessary, pilots are taught to keep
the plane under control, to land in an upright position at the slowest possible
speed, and to avoid obstacles as much as possible.
We examined data from a large number of general aviation airplane crash
landings to identify pilot, airplane, environmental, and crash factors associated
with pilot fatalities and to estimate the relative risk (RR) of a pilot fatality
associated with these factors after adjusting for potential confounders.
We analyzed data from computer tapes containing the standardized findings
of National Transportation Safety Board investigations of airplane accidents
and incidents. These tapes included data on crashes only if they had occurred
during landing or takeoff and involved loss of engine power, which we defined
as crash landings. Such events often involve an attempt
at a controlled landing. Thus, we attempted to eliminate crashes where the
airplane hit a mountain, was torn apart in a thunderstorm, or where aerobatics
or low-altitude cruise flight may have occurred.
Crash landings in this study occurred during 1983 through 1992 and were
severe enough to result in serious injury or death within 30 days of the crash
landing to a pilot or passenger or caused substantial damage to the airplane.
Damage to the wings, fuselage, rudder, elevators, or cockpit that adversely
affected the ability of a crashed airplane to fly and would usually require
repair or replacement is defined as substantial by the National Transportation
Safety Board regulations.3 In contrast, damage
to an airplane's propeller, landing gear, or wing tips, dents or small holes
in the plane's skin, or engine failure are considered minor.
A serious injury is defined as one that results
in a hospitalization of more than 48 hours' duration; fractures (except simple
ones of the fingers, toes, or nose); internal organ damage; severe nerve,
tendon, or muscle damage or bleeding; or a significant second- or third-degree
burn. Only airplanes with 10 seats or fewer (98.8% of all the airplane crashes)
were selected because they represent most general aviation activity and have
different characteristics from other types of aircraft, ie, helicopters, gliders,
balloons, or larger airplanes.
Cases were defined as crash landings in which
the pilot died, while controls were defined as crash
landings in which the pilot survived. Pilot, airplane, environment, and crash
variables related to the event were categorized according to Federal Aviation
Administration regulations,4 National Transportation
Safety Board regulations,5 or to the standardized
reporting categories of National Transportation Safety Board accident and
Exposure to flying was defined as pilot flight
hours, ie, the total time a pilot had flown any aircraft. The biennial flight review is a Federal Aviation Administration requirement
that all pilots receive theoretical and practical flight training with an
instructor at least every 2 years.7
Whether or not an instrument flight rules flight plan was filed was
used to indicate a potentially high-risk flight condition. An instrument flight
rules flight plan is required whenever the flight is under instrument control,
takes place 5400 m (18000 ft) or more above mean sea level, or when instrument
flight rules are to be followed, which may be due to weather conditions or
Logistic regression was used to adjust for the effects of potential
confounders.8 Only variables for which the
data were at least 90% complete and might be logically related to pilot death
were examined. Adjustments were made only for variables that altered the risk
estimates by at least 10%. Odds ratios, estimated by the maximum-likelihood
method, were used to approximate RRs and were adjusted for pilot age, pilot
flight time, airplane landing gear category, and the filing of an instrument
flight plan. Other variables examined, but not adjusted for, included pilot
characteristics of sex, principal profession, certification, airplane rating,
and medical certificate; airplane characteristics of certificated maximum
gross weight, engine horsepower greater than 200, airplane hours of use, and
whether or not the airplane was owned by the pilot; and environmental and
crash characteristics of instrument meteorological conditions, night time,
number of occupants, and flight purpose.
There were 8411 eligible crash landings during 1983 through 1992; the
number of crash landings decreased from 1235 in 1983 to 681 in 1992. One third
of the crash landings occurred in the summer, one quarter in the spring, one
fifth in the fall, and one fifth in the winter. Most crash landings occurred
in daylight (85.1%), with no restriction of visibility (88.7%) and no weather
precipitation (94.2%). A pilot fatality occurred in 437 (5.2%) of the crash
landings. In 69.1% of the crash landings there was no pilot injury reported,
a minor injury in 17.2%, and a serious injury in 8.3%. The airplane was destroyed
in 15.2% of the crash landings, suffered substantial damage in 84.6%, minor
damage in 0.1%, and no damage in 0.1%.
Relative risks were adjusted for pilot age, pilot flight hours, airplane
landing gear type, and the filing of an instrument flight plan, except that
the estimates for each of these 4 variables were adjusted only for the other
3 variables. Relative risk estimates showed little change with further adjustment
for other variables.
A fatal pilot injury was most strongly associated with destruction of
the airplane (RR, 42.6) or an airplane fire or explosion on the ground (RR,
20.4) (Table 1). Failure to use
both a lap belt and shoulder harness and failure to use a shoulder harness
were associated with an increased RR of death compared with use of both a
lap belt and shoulder harness (RR, 6.8 and 1.7, respectively). The RR of pilot
death was also increased if the crash site was not on an airport or airstrip
The RR of pilot death in an airplane with retractable tricycle landing
gear was about twice that in one with fixed tricycle landing gear (RR, 2.2).
Multiengine planes were associated with an elevated RR of pilot death (RR,
Older pilots were more likely to die; for the 11% of pilots aged 60
years or older, the RR of death was 2.9 (95% confidence interval [CI], 1.7-4.9);
for the 19% of pilots aged 50 to 59 years, the RR for death was 1.9 (95% CI,
1.1-3.2), compared with the 6.8% of pilots younger than 25 years. The RR of
death was twice as great among pilots without a current biennial flight review,
a requirement established to attempt to ensure periodic pilot flight training.
We found that general aviation airplane crash landings were usually
survivable, with pilot fatalities occurring in only 5% of the crashes we examined.
The increased risk of fatal injury associated with failure to use restraining
belts is consistent with other studies of both motor vehicle and aircraft
crashes.9,10 Federal Aviation
Administration regulations require that restraints be used during takeoff
and landing,11 and it is possible that some
pilots who survived may falsely claim that they used restraints; restraint
use by fatally injured pilots might be less likely to suffer from this bias.
If this occurred, it would exaggerate the strength of the association that
we found when we compared pilots who used no restraints with those who used
both lap belts and shoulder harnesses. However, this should not explain the
difference in survival when we compared pilots who used lap belts only with
those who used both lap and shoulder belts, since both types of restraints
meet the legal requirements.
Our finding that crashes that occurred off an airport or airstrip were
more deadly seems plausible, due to the obstructions that would be encountered
in many such landings. The increased RR of death associated with destruction
of the airplane was expected because of the large amount of kinetic energy
that is involved in such a crash. We found that the type of landing gear best
controlled for airplane size characteristics of certificated maximum gross
weight, engine horsepower, number of engines, and type of landing gear. The
presence of multiple engines may increase the risk of death because planes
with more than 1 engine have to land at a higher velocity, resulting in the
potential for transferring more energy to the occupants during a crash landing.
Some of the increase in mortality associated with fire or explosion may reflect
more forceful crashes that might have been lethal even without thermal or
The reasons for the association of a pilot death with some of the other
factors examined are less apparent. The increased risk of pilot death associated
with airplanes that had more than 1 engine or had retractable tricycle landing
gear may be related to flight complexity. We adjusted for the filing of an
instrument flight plan as a surrogate measure of flight complexity. The filing
of an instrument flight rules flight plan, which may be done because of flight
conditions or for pilot convenience, does not necessarily describe flight
complexity. Other measures of flight complexity, such as total duration of
flight or weather conditions during flight, were not available in the data.
These measures might be related to pilot fatigue and therefore might affect
the ability of a pilot to cope with a crash landing. The lack of a current
biennial flight review, associated with a 2-fold risk of pilot death, may
indicate pilots lacking the skills needed for a crash landing or pilots prone
to risk-taking behavior.
Previous analyses of aviation crashes have examined a number of risk
factors for aircraft crashes.12 A recent investigation
of commuter and air taxi aircraft crashes found that the characteristics of
multiengine aircraft, off-airport crash site, night flights, instrument meteorological
conditions, nonuse of shoulder restraints, and fire or explosion after the
crash were associated with a greater risk of pilot fatality, although pilot
age, sex, and flight experience were not.10
Some of the risk factors identified are more amenable to change than
others. Certain characteristics associated with fatal pilot injury during
crash landings, such as travel in airplanes that land at a high velocity,
have multiple engines or retractable tricycle landing gear, or travel in complex
flight conditions may be difficult to alter. Stricter enforcement of the Federal
Aviation Administration's requirement for a biennial flight review might decrease
the likelihood of pilot death in some crash landings. Federal Aviation Administration
regulations require that starting in 1978, new aircraft must have shoulder
harnesses for pilots as well as seat belts13;
however, a large proportion of general aviation aircraft were manufactured
before that year, as the average date of aircraft manufacture is 1969.14 Retrofitting restraints is technically a fairly easy
process but costs $300 to $800 per seat. Pilots who wish to reduce their risk
of death in a crash landing should ensure that their airplanes are equipped
with lap and shoulder restraints.