Cowl CT, Jones MP, Lynch CF, Sprince NL, Zwerling C, Fuortes LJ. Factors Associated With Fatalities and Injuries From Hot-Air Balloon Crashes. JAMA. 1998;279(13):1011-1014. doi:10.1001/jama.279.13.1011
From the Mayo Clinic, Rochester, Minn (Dr Cowl), and the Injury Prevention Research Center, The University of Iowa, Iowa City (Drs Jones, Lynch, Sprince, Zwerling, and Fuortes).
Context.— Despite the increased popularity of hot-air balloon flight, data on
injuries and fatalities associated with hot-air balloon crashes are limited.
Objective.— To determine factors associated with injury and death in hot-air balloon
Design.— Retrospective review of data collected from reports and investigations
by the Civil Aeronautics Board and the National Transportation Safety Board.
Study Subjects.— Individuals involved in US hot-air balloon crashes from 1964 to 1995.
Main Outcome Measures.— Total number of crashes and factors associated with fatality or serious
Results.— From 1964 to 1995, a total of 495 hot-air balloon crashes involving
1533 persons were reported and included 92 fatalities and 384 serious injuries.
Pilot error or incapacitation was determined subjectively by crash investigators
to contribute to 85.1% of the crashes. In univariate analysis, collision with
the ground was the most significant predictor of a fatality or serious injury
(P<.001), and power-line contact was the most
significant predictor of fatality (P<.001). In
multiple logistic regression, only the type of object struck by a balloon
predicted a fatal crash or a fatality or serious injury.
Conclusions.— Although a number of factors likely contribute to increased severity
of hot-air balloon crashes, the object struck during a crash is most predictive
of fatality or serious injury. Preventive efforts are needed to decrease future
THE SPORT OF hot-air ballooning has become more popular each year since
a recreational ballooning renaissance began in the early 1960s. In 1964, only
6 balloons were registered with the Federal Aviation Administration (FAA),
but by 1995, that number had increased to 7123.1
With its increasing popularity has come an influx of new pilots and balloons,
as well as more crashes with associated property damage, severe injuries,
Consequently, the public health importance of studying hot-air balloon
crashes has increased. Hundreds of commercial hot-air balloon rides are chartered
each year in the United States, and mass ascensions have attracted thousands
of spectators and national media attention.
Despite the increased popularity of hot-air balloon flight, published
data on hot-air balloon crashes are limited.2,3
Frankenfield and Baker4 published a descriptive
study of the epidemiology of injuries due to hot-air balloon crashes using
5 years of National Transportation Safety Board (NTSB) crash data (1984-1988)
based on the NTSB's 2-page abbreviated reports.
In this study, we used a national database to determine the number and
characteristics of hot-air balloon crashes and to identify factors associated
with crash-related injuries and fatalities.
We reviewed the records of the 495 hot-air balloon crashes recorded
by the Civil Aeronautics Board (1964-1967) and the NTSB (1967-1995), including
statements recorded by passengers and pilots and medical reports filed for
injured victims. The Code of Federal Regulations mandates that pilots must
contact the NTSB for "any incident involving flight malfunction or failure,
inability of flight crew to perform normal duties as a result of injury or
illness, in-flight fires, aircraft collision in flight, damage to property
estimated to exceed $25,000 for repair or fair market value (whichever is
less), or if an aircraft is overdue and believed to have become involved in
The term fatality is defined as "any injury
that results in death within 30 days of the crash"; the term serious injury as "any injury requiring hospitalization for more than
48 hours commencing within 7 days from the date of initial injury." Serious
injuries also include a fracture of any bone (except uncomplicated fractures
of fingers, toes, or nose); severe hemorrhage; nerve, muscle, or tendon damage;
internal organ injury; second- or third-degree burns; or burns affecting more
than 5% of the body surface. Minor injuries include bruises, musculoskeletal
strain or sprains, or simple fractures noted above.6
Descriptive statistics were calculated using the SAS/STAT version 6
statistical package (SAS Institute, Cary, NC). The 457 reported crashes occurring
between 1976 and 1995 were used for risk factor identification. The unit of
measurement is a balloon crash. A fatal outcome is defined as a crash that
involves at least 1 fatality. A serious injury or fatality outcome occurs
when a crash includes at least 1 serious injury or 1 death. Logistic regression
(EGRET software, version 1, SERC Inc, Seattle, Wash) was used to analyze each
variable separately to determine if it was associated with an increased risk
of the crash having a serious injury or fatal outcome. This regression produced
estimates and 95% confidence intervals (CIs) of the odds ratios (ORs) as well
as P values of the significance of each variable's
possible association with an adverse outcome based on the likelihood ratio
test. Logistic regression also was used to create a final model containing
variables that in combination best predict the outcome variable. Categorical
variables having subcategories with no events recorded were analyzed by the
exact Fisher-Freeman-Halton test (StatXact software, Cytel Software Corp,
The distribution of crashes and injuries over the study period and the
primary mechanisms of injury are summarized in Table 1.
Collision with power lines accounted for 137 crashes (27.7%), including
41 fatalities (44.6%) (Table 2).
The most common serious injury was a fracture of the lower extremity (n=216,
56.3%). Head injuries accounted for only 37 serious injuries (9.6%). Sixty-six
adverse events were recorded without the balloon striking an object (eg, a
passenger twisted an ankle on landing).
The most crashes recorded in 1 year was 33 in 1980. Ballooning crashes
were most common in July (66 crashes in 31 years), followed by October (65),
August (57), and September (51), representing nearly half (48%) of the crashes.
This likely reflects a greater risk exposure due to the high volume of balloon
flights during the warmer seasons of the year. States with the highest frequency
of crashes during the study period were New Mexico (64), California (49),
Colorado (45), and Arizona (39). Twenty-three of New Mexico's 64 crashes occurred
during Albuquerque's annual International Balloon Fiesta, the largest single
congregation of balloonists worldwide.
Recreational flights (209 [42.2%]) and paid passenger flights (153 [30.9%])
were involved most frequently. Balloon rallies accounted for 75 recorded crashes
(15.2%). Student pilot crashes accounted for 46 crashes (9.3% of the total),
while 352 (71.1%) involved commercial pilots and 82 (16.6%) involved private
pilots. Most crashes (228 [46%]) occurred during landing attempts or while
approaching a landing spot (102 [20.6%]).
The NTSB determined subjectively that pilot error or incapacitation
contributed to 421 crashes (85.1%). The most common errors noted in the NTSB
files included inappropriate in-flight planning in 138 (27.9%), misjudgment
of wind in 94 (19.0%), selection of unsuitable landing sites in 68 (13.7%),
and incorrect positioning of passengers or the pilot inside the basket in
Equipment failure or malfunction contributed to 36 crashes (7.3%) and
13 fatalities. There were 17 documented fuel-system failures, and failure
of the fabric envelope was involved in only 9 crashes. The mean wind velocity
for all crashes was 9.3 knots. A total of 282 crashes (57%) occurred in winds
between 5 and 10 knots. Only 94 crashes (19%) occurred in winds less than
Aircraft damage was reported to be substantial in 218 crashes (44.0%),
162 (32.7%) had no significant damage, and 67 aircraft (13.5%) involved were
destroyed. Fire occurred in 92 crashes (18.6%), the gondola was severed from
the cables in 42 crashes (8.5%), and fuel tanks ruptured in 17 crashes (3.5%).
Pilot-in-command time (total hours in flight as sole operator of the
balloon controls) ranged from 1 hour to 2137 hours (median, 74.5 hours). The
median age of pilots involved in crashes was 38.0 years and ranged from 18
to 70 years. Table 3 shows risk
factors for a balloon crash to involve a fatality. The type of object struck
by a balloon during a crash was the only significant variable (P<.001). The odds of a fatality were more than 13 times greater
when contacting a power line than when hitting the ground.
For crashes involving either a fatality or serious injury (Table 3), the type of flight (P<.001), the object struck during a crash (P<.001),
and the experience of the pilot (P=.01) were statistically
significant. Paid passenger flights had an OR of 2.5 (95% CI, 1.5-4.1) relative
to recreational flights; however, no other subcategories were statistically
For the outcome of fatality and the outcome of fatality or serious injury,
multiple logistic regression demonstrated that no other factor improved on
a model containing the variable of what the balloon struck during a crash.
Hence, the final models for these 2 outcome variables contain only that 1
variable with its subcategories.
The NTSB was established in 1967 as an independent federal agency to
investigate all civil aviation accidents, maintain a national database of
crash data, and issue safety recommendations aimed at preventing future mishaps.
In April 1995, the NTSB was granted additional power to mandate detailed reporting
of aircraft crashes, making our data relatively free from differential case
ascertainment over time.
Our review of 31 years of NTSB data showed that power-line strikes contributed
to a large portion of hot-air ballooning deaths. Contact with power lines
caused the basket to overturn while airborne in some cases or, worse, created
electrical arcing that severed the load cables from the basket, causing pilots
and passengers to fall to their deaths. Of the 40 individuals involved in
crashes in which power lines arced, 24 people were electrocuted, resulting
in 8 deaths, 8 serious injuries, and 8 minor injuries.
The univariate data analysis indicates that if a balloon crash occurs,
there is a 1.4% chance (2/141) of a fatality when the ground is contacted
first compared with a 16.1% chance (92/130) of a fatality if initial contact
is made with a power line (OR, 13.4; 95% CI, 3.1-58.4) (Table 3). In the univariate analysis for fatality or serious injury,
contact with a power line appeared to be associated with a lower OR (0.11;
95% CI, 0.1-0.2) (Table 3). This
discrepancy reflects both a higher risk of sustaining a serious injury (but
not a fatality) if a balloon initially contacts the ground during a crash
and that individuals involved in crashes with power-line contacts are either
killed or sustain only minor injuries. If a balloon strikes a power line first,
there is a 53% chance that no significant injuries will occur; however, if
an injury does occur, there is a higher probability it will be a fatality.
Contact with trees is an important risk factor; of 43 collisions with
trees, 35% resulted in serious injuries or fatality. However, the OR of a
serious injury or fatality from striking a tree relative to striking the ground
is 0.06 (95% CI, 0.03-0.15), which may be attributable in part to underreporting
of crashes with the ground that resulted in nonserious injuries.
Changing weather conditions were noted by the NTSB in 35 crashes (7.1%),
and the odds of a fatality or serious injury appear to increase with greater
wind velocity (although this result is not statistically significant [P=.07]). Measurement error may have played a role in this
calculation since wind data are often collected from the pilot involved in
the crash or from nearby locations where wind speed may vary from the value
at the crash site. The major limitation of our study is the lack of exposure
or denominator data for estimating the incidence of injury and fatality. The
number of balloons registered with the FAA is relatively accurate since all
airworthy aircraft must be enrolled. Unfortunately, there are no reliable
estimates of the number of balloons actually in use, the number of active
balloon pilots, or the number of hours flown annually per balloon or per pilot.3 Given these major limitations, we did not attempt
to analyze incidence data.
In addition, it is likely that some pilots did not report crashes with
only minor structural damage to the balloon and no significant injuries to
the passengers. Current FAA regulations require pilots to report all crashes
that cause structural damage to the aircraft or injury to the occupants and
require a report to be filed within 10 days of the incident.6
Despite this, we discovered several crash reports in the NTSB database that
were filed weeks or months after the crash and only after NTSB officials demanded
that the pilot file a report or risk prosecution.
Further, the findings in this study are subject to misclassification
error. Because NTSB officials add information to their database by transferring
information from individual crash reports, individuals not familiar with balloon
flight may incorrectly label certain factors involved in balloon crashes.
For example, data collection forms used by the NTSB are designed for motorized
aircraft; thus, many variables in the standardized collection form are not
applicable to balloon flight. Consequently, some of the data entered into
the NTSB database for hot-air balloon crashes are extracted from the pilot
narrative and are subject to recall bias.
Finally, blood alcohol levels and urine drug-screening results were
not available for all pilots involved in crashes. The FAA regulations allow
for a maximum blood alcohol level of 0.09 mmol/L (0.4% by weight) to operate
a balloon, and no alcoholic beverage may be consumed within 8 hours of flying.5 However, since crash reports can be filed several
days after the occurrence, routine drug and alcohol screening has been impractical.
In conclusion, our data demonstrate that the number of fatalities from
hot-air balloon crashes has decreased, but the percentage of crashes with
either serious injury or death has increased. The type of object struck during
a crash is the best predictor of adverse outcome. Collision with the ground
is the most significant predictor of a fatality or serious injury, whereas
power-line contact is the most significant predictor of fatality.
Improper weather interpretation, incorrect management of rapid descents,
and contact with equipment inside the basket during landings are associated
with significant morbidity. The incidence of hot-air balloon serious injuries
and fatalities is difficult to estimate because of the lack of accurate denominator
Continued preventive efforts, such as mandatory attendance of educational
programs that emphasize avoidance of power lines and increased awareness of
other potential risk factors for crashes, and further safety improvements
by balloon manufacturers may help to decrease future injuries and deaths.