Margolis LH, Foss RD, Tolbert WG. Alcohol and Motor Vehicle–Related Deaths of Children as Passengers, Pedestrians, and Bicyclists. JAMA. 2000;283(17):2245-2248. doi:10.1001/jama.283.17.2245
Author Affiliations: School of Public Health and the Highway Safety Research Center (Drs Margolis and Foss and Mr Tolbert) and Department of Maternal and Child Health (Dr Margolis), University of North Carolina, Chapel Hill.
Context The overall percentage of motor vehicle deaths associated with alcohol
consumption declined between 1991 and 1996, but the risk of death due to alcohol-related
crashes for children warrants analysis.
Objective To examine the association between alcohol use by drivers and mortality
of children who were passengers, pedestrians, and bicyclists.
Design and Setting Analysis of data from the Fatality Analysis Reporting System, a nationwide
US registry of motor vehicle deaths, for 1991-1996.
Subjects A total of 16,676 children younger than 16 years who were passengers,
pedestrians, or bicyclists and whose death was due to a motor vehicle crash.
Main Outcome Measure Alcohol use by drivers involved in crashes in which children died, assessed
by age and sex of the child and driver and type of crash.
Results A total of 3310 deaths (19.9%) involved alcohol-related crashes. The
percentage declined from 21.6% in 1991 to 17.8% in 1996. Considering only
crashes in which the alcohol-use status of the child's driver was relevant,
the decline was less marked, from 18.8% in 1991 to 15.1% in 1995, with an
increase to 16.4% in 1996. Among crashes involving alcohol, the child's own
driver had been drinking in 66.3% of cases, varying from 58.0% to 70.7% over
time. Drivers younger than the legal drinking age of 21 years who had been
drinking alcohol accounted for 30.3% of alcohol-related passenger deaths among
Conclusion While the overall percentage of alcohol-related motor vehicle deaths
for children declined between 1991 and 1996, experiences for passengers, pedestrians,
and bicyclists differ. Selected characteristics of children and drivers that
elevate the risk of an alcohol-related motor vehicle death point to the need
for further policy and clinical interventions.
Public concern over alcohol-related motor vehicle mortality and morbidity
reached a high point in the early 1980s, when nearly 60% of motor vehicle
fatalities were associated with alcohol use by at least 1 person involved
in the crash.1 Interventions in response to
this concern included raising the minimum drinking age in all states, enactment
of per se illegal blood alcohol–content laws, and administrative license
suspension.2 A marked decline in alcohol-related
fatalities has been associated with these policies, with the percentage of
drivers using alcohol involved in fatal crashes decreasing from 47.9% in 1991
to 41.0% in 1996.3
Although the focus of policies has been to address the high risk of
alcohol-related mortality for drivers, children as passengers, pedestrians,
and bicyclists are also at risk of death in alcohol-related crashes. Previous
analysis of North Carolina crash data indicated that between 1979 and 1982,
alcohol use was associated with 15.4% of the motor vehicle–related deaths
of children.4 The purpose of the present study
was to examine, on the national level, the association between alcohol use
by drivers and mortality for children as passengers, pedestrians, and bicyclists
for the years 1991 through 1996.
Data for this analysis come from the Fatality Analysis Reporting System
(FARS).5 During 1991-1996, 19,673 children
and adolescents younger than 16 years were killed in motor vehicle–related
crashes. For the present analysis, we excluded 1132 crashes involving more
than 2 vehicles, 1460 crashes involving unusual circumstances such as off-roadway
locations or nonmotorized conveyances other than bicycles, 313 vehicles other
than cars and light trucks, 92 cases with missing driver information, and
456 children listed as drivers rather than passengers. Any passengers riding
with the 456 underage drivers were included. The final analysis sample of
children younger than 16 years who died as passengers, pedestrians, or bicyclists
in crashes involving cars or light trucks represented 84.8% (16,676/19,673)
of the initial population.
There are 4 potential indicators of alcohol involvement in FARS, each
with strengths and weaknesses. The first is the actual measurement of a driver's
blood alcohol concentration (BAC). Although this is quantifiable and objective,
there was no BAC measurement for 60.2% of drivers in our analysis sample.
Fatally injured drivers are usually tested, but measurements are obtained
for relatively few surviving drivers. The second is to use imputed alcohol
involvement based on an algorithm developed by the National Highway Traffic
Safety Administration that estimates alcohol involvement based on characteristics
of the crash and the driver(s) involved.6 This
measure has the advantage of being available for all crashes, but most cases
involve imputation. The third indicator is to use the investigating police
officer's assessment of whether a driver had been drinking alcohol.7 Unfortunately, this information also is sometimes
missing; for various reasons, officers sometimes do not believe they have
sufficient information to make a judgment. The fourth method is to include
only those cases in which the officer is definitive in his or her assessment
that alcohol was or was not involved.7
We rejected the first and second indicators because of missing values
and assessed the validity of the third and fourth by examining their positive
and negative predictive values. For the purpose of this analysis, we report
the values for the child's driver. For the third indicator, the positive and
negative predictive values are 85.8% and 91.1% (κ=0.73), respectively.
For indicator 4, the positive and negative predictive values are 85.8% and
We have used the definitive police report as the indicator of alcohol
involvement to avoid the use of data for which 60% of the values were imputed.
The percentages of alcohol involvement obtained with this definitive report
indicator were then applied to the 16,676 deaths in our sample to generate
population-based rates of death associated with alcohol.
We assessed the relationship between the alcohol-use status of the driver
and the following variables: age and sex of the child, age and sex of the
driver, and type of crash (single vehicle, 2 vehicles, pedestrian, and bicyclist).
The relative risk (RR) of alcohol exposure, given these characteristics, was
calculated using Epi-Info, version 6 (Centers for
Disease Control and Prevention [CDC], Atlanta, Ga). Rates were determined
based on the mid-year population estimates from the US Census.8
During the 6 years studied, 16,676 children younger than 16 years were
killed in motor vehicle crashes involving cars, light trucks, and vans as
passengers, pedestrians, or bicyclists. Overall, 3310 deaths (19.9%) involved
a driver who had been drinking alcohol. Of the alcohol-related deaths, 79.5%
involved children as passengers, and the remainder involved pedestrians or
bicyclists struck by drivers who had been drinking alcohol.
Among single-vehicle crashes, 26.3% (1148) of child deaths were associated
with alcohol use. Among 2-vehicle crashes, alcohol consumption was noted for
the child's driver or for both drivers in 25.0% (1486) of the deaths. For
these alcohol-related deaths of child passengers, 66.3% involved alcohol use
by the child's own driver.
Table 1 shows that the number
of deaths per year ranged between 2686 and 2884, but the percentage of alcohol-related
fatalities has declined from 21.6% to 17.8% (1.03/100,000 to 0.80/100,000
population). Considering the alcohol-use status of only the child's driver
(excluding the deaths of pedestrians and bicyclists and the alcohol-use status
of the other driver in 2-vehicle crashes), the percentage of alcohol-related
deaths declined from 18.8% in 1991 to 15.1% in 1995 and increased to 16.4%
in 1996. Using these alcohol use–related deaths as the denominator,
the percentage of child passengers in which the child's own driver was using
alcohol showed a different pattern, declining from 67.9% in 1991 to 58.0%
in 1994 and then rising to 70.7% in 1996. For children as pedestrians and
bicyclists, the percentage of alcohol-related fatalities declined from 14.5%
in 1991 to 8.1% in 1996.
Table 2 shows the distribution
of deaths for children as passengers, pedestrians, and bicyclists, categorized
by age and sex. Overall, males accounted for 54.6% of the alcohol-related
deaths, mainly due to their substantial overrepresentation among pedestrian
and bicyclist deaths. Males accounted for 52.2% of the deaths as passengers.
The percentage of crashes involving alcohol use for male passengers
exceeds the percentage for females for most age groups until age 12 and 13
years (Table 2). Although not
shown in the table, restricting the analysis to crashes in which alcohol-use
status of the child's driver is relevant (ie, single-vehicle crashes and 2-vehicle
crashes) and focusing only on the child's driver shows similar results. The
proportion of deaths of children younger than 12 years involving alcohol use
showed a consistent decline from 16.4% in 1991 to 12.7% in 1996. In contrast,
deaths of children aged 12 through 15 years showed considerable fluctuation,
ranging from 23.0% in 1991 to a low of 17.3% in 1994 and climbing to 22.3%
Overall, the drivers of children in these crashes were slightly more
likely to be male (52.5%), but children aged 11 years and younger were more
likely to be driven by females (Table 3). Among children aged 14 and 15 years, only 30.0% of the drivers
were female. Alcohol use was associated with 23.3% of children's deaths involving
male drivers and 10.0% involving female drivers (RR, 2.33; 95% confidence
interval [CI], 2.12-2.57). Alternatively, male drivers accounted for 70.0%
of these alcohol-related deaths. The higher percentage of alcohol use by male
drivers was consistent across children's age cohorts.
The RR of alcohol involvement for drivers under the legal drinking age
compared with those aged 21 years and older was slightly less than 1 (RR,
0.93; 95% CI, 0.85-1.02). Males younger than 21 years accounted for 38.8%
of all deaths involving male drivers and 35.3% of the alcohol-related deaths.
In contrast, females younger than 21 years accounted for only 24.5% of the
deaths involving female drivers and 17.6% of the alcohol-related deaths. Taken
together, drivers under the legal drinking age accounted for 30.3% of the
alcohol-related children's passenger deaths.
To distinguish crashes in which the drivers and child passengers could
be considered social peers from those in which the driver was ostensibly a
caretaking adult, we categorized deaths according to whether the driver was
at least 6 years older than the child. For the 14- and 15-year-old victims,
alcohol use was more common in the crashes involving drivers who were more
than 6 years older than their passengers (28.9% vs 19.3%), but 62.6% of alcohol-related
deaths for this group involved drivers close in age (<6 years difference),
reflecting the greater exposure of teenagers to driving with social peers
than with older adults. For 14- and 15-year-old girls involved in crashes
with peer drivers, males accounted for 59.4% of the drivers, of whom 27.3%
had been drinking alcohol. The RR of alcohol use by male drivers of these
girls vs female drivers was 2.66 (95% CI, 1.85-3.84). For 14- and 15-year-old
boys with age-peer drivers, 83.3% had male drivers, of which 20.4% had been
drinking alcohol (RR, 2.48; 95% CI, 1.38-4.46). When riding with male drivers
who were their age peers, girls were at greater risk of alcohol involvement
than were boys (RR, 1.34; 95% CI, 1.08-1.66).
From 1991 through 1996, approximately 550 children per year died in
alcohol-related motor vehicle crashes. To put this in perspective, it is estimated
that annually 284 to 360 children younger than 18 years die from smoking-related
illnesses and fires,9 and approximately 208
children younger than 15 years die as the result of unintentional firearm
Paralleling the national experience for alcohol-related motor vehicle
deaths, the percentage of children's alcohol-related deaths as passengers
declined overall between 1991 and 1996, but patterns differed for younger
and older children. The risk for younger children (<12 years) has declined
somewhat, but there has been no consistent change for older children. The
percentage of alcohol-related traffic deaths accounted for by the child's
own driver varied from 58.0% to 70.7%. In many cases, drivers were old enough
to be considered the responsible caretaker rather than an adolescent peer
who might be caught up in the social act of drinking. However, 30.3% of the
alcohol-related child passenger deaths involved drivers younger than the legal
drinking age of 21 years.
The major limitation of this study relates to the measure of alcohol
involvement, as is nearly always the case in alcohol studies. In view of the
absence of observable indicators of low BACs, police officers are highly unlikely
to detect a BAC of less than 0.05% and hence to report that the driver was
drinking alcohol, which argues in favor of our use of definitive police reports.
Moreover, among fatally injured drivers for whom BAC values are available
in FARS, only about 2% have positive BAC values that are less than 0.05%.
The overall alcohol involvement of 19.9% is lower than the 24% reported
by the CDC for the period 1985-1996 for children younger than 15 years.11 One possible explanation for this disparity is that
the algorithm that the National Highway Traffic Safety Administration uses
to estimate alcohol involvement in crashes was not developed using the peculiar
subset in which a child was a passenger but rather using all fatal crashes.
A second possible explanation is that the CDC analysis included the late 1980s,
before the marked recent decline in alcohol involvement in motor vehicle–related
There are several implications of these findings both for clinicians
and policy makers. Although FARS data provide no information on the family
relationships among persons involved in crashes, in view of a recent estimate
that in a given year about 1 in every 6.6 children is exposed to family alcohol
abuse or dependence, many crashes likely involved parents as drivers.12 Given recent evidence on the effectiveness of patient
education on alcohol use,13 such counseling
may be even more effective when offered about risks to parents' own children.
Clinicians could also educate parents to be especially attentive to the possibility
of alcohol use by other drivers of their children. While 14- and 15-year-old
victims were disproportionately riding with drivers close to their own age,
the fact that there was evidence of alcohol use by 36.6% of the male drivers
and 15.5% of the female drivers who were at least 6 years older than the child
victim highlights the need to modify the behavior of these "responsible" adults.
For clinicians who work with adolescents the message is quite strong.
Since a disproportionate share of the deaths was associated with drivers younger
than 21 years, for whom alcohol use is per se illegal, it is important to
help adolescents and young adults understand the risks that arise from this
illegal behavior. Furthermore, effective strategies and techniques to avoid
risky behaviors, such as riding with a driver who has been drinking, should
be incorporated into health promotion programs for adolescents. Particularly
noteworthy, however, is the risk to adolescent girls of an alcohol-related
death (RR, 2.66) when male peers are driving compared with their female peers
as drivers. It is important to underscore that the health message should be
more extensive than "kids and alcohol don't mix" because nearly 40% of the
deaths of children aged 14 and 15 years involved drivers at least 6 years
older than the children.
Given the disproportionate share of deaths in which the driver was under
the legal drinking age, laws governing availability of alcohol to this age
group should be more rigorously enforced. Since adolescents and young adults
are more sensitive to price changes than are others, the tax structure on
alcohol, especially beer, which is the beverage most commonly used by drinking
drivers, could be altered to decrease its consumption among this younger group
Despite impressive declines in alcohol-related motor vehicle deaths
for adults during the past decade, the risk to children remains substantial,
warranting additional attention and carefully targeted action.