Customize your JAMA Network experience by selecting one or more topics from the list below.
Li G, Baker SP, Smialek JE, Soderstrom CA. Use of Alcohol as a Risk Factor for Bicycling Injury. JAMA. 2001;285(7):893–896. doi:10.1001/jama.285.7.893
Author Affiliations: Department of Emergency Medicine, Johns Hopkins University School of Medicine and Center for Injury Research and Policy, Johns Hopkins University School of Hygiene and Public Health (Dr Li and Ms Baker), Office of the Chief Medical Examiner of Maryland (Dr Smialek), and Division of Trauma Surgery, the R Adams Cowley Shock Trauma Center, University of Maryland Medical Center (Dr Soderstrom), Baltimore.
Context Bicycling is one of the leading causes of recreational injuries. Elevated
blood alcohol concentrations (BACs) are found in about one third of fatally
injured bicyclists aged 15 years or older.
Objective To assess the relative risk of fatal and serious bicycling injury according
Design Matched case-control study.
Setting and Subjects Bicyclists aged 15 years or older who were fatally or seriously injured
while riding a bicycle during the day in Maryland in 1985-1997 (cases, n =
124) and bicyclists aged 15 years or older who were interviewed and given
a breath test for estimated BAC during roadside surveys that took place in
June 1996 through May 1998 at the same site, time of day, day of week, and
month of year in which a case bicyclist was injured (controls, n = 342).
Main Outcome Measure Odds ratio of bicycling injury according to estimated BAC.
Results An estimated positive BAC (≥0.02 g/dL) was detected in 12.9% of the
case bicyclists (23.5% of the 34 fatally injured and 8.9% of the 90 seriously
injured) compared with 2.9% of the control bicyclists (P<.001). Relative to an estimated BAC of less than 0.02 g/dL, the
adjusted odds ratio of bicycling injury was 5.6 (95% confidence interval [CI],
2.2-14.0) for a BAC of 0.02 g/dL or higher and was 20.2 (95% CI, 4.2-96.3)
for a BAC of 0.08 g/dL or higher. Rates of helmet use at the time of injury
or interview were 5% and 35%, respectively, for those with and without a positive
BAC (P = .007).
Conclusion Alcohol use while bicycle riding is associated with a substantially
increased risk of fatal or serious injury.
Bicycling as a form of recreation and exercise and, to a lesser extent,
as a means of personal transportation has become increasingly popular in the
United States. According to the Nationwide Personal Transportation Survey,
the annual average number of bicycle trips per capita has increased by 238%
between 1977 and 1995.1,2 Each
year, more than half a million people are treated in emergency departments
and 20 000 are admitted to hospitals for bicycling injuries.2
The prevalence of alcohol involvement in bicycling injuries to adolescents
and adults has been well documented.3-8
Elevated blood alcohol levels are found in about 8% of those treated in emergency
departments for bicycling injuries,4 in 16%
of adults admitted to hospitals,5-7
and in 32% of those who died.3,7
It has been shown that riding a bicycle requires a higher level of psychomotor
skills than driving a car and that biking performance in controlled laboratory
conditions declines progressively as blood alcohol concentration (BAC) increases.9,10 The role of alcohol in bicycling injury,
however, has not been adequately addressed based on rigorously designed epidemiologic
studies. Quantitative information about the risk of bicycling injury related
to alcohol use is valuable to health care providers in counseling their clients
and in championing public policies for improving bicycling safety. Using a
matched case-control design, this study assesses the relative risk of fatal
or serious bicycling injury associated with alcohol consumption.
Case bicyclists were those who, while riding a bicycle, were injured
fatally or so severely that they required hospitalization, as identified from
the records of the Office of the Chief Medical Examiner (OCME) of Maryland
and the trauma registry of the University of Maryland Medical Center. At OCME,
alcohol testing is routinely performed using the head-space gas chromatography
method. Medical examiner cases eligible for this study were bicyclists who
were fatally injured in Maryland during 1985 through 1997, excluding those
who were injured at night (9 PM to 5 AM), those whose BACs were not measured,
and those who survived 6 hours or longer after injury.
The University of Maryland Medical Center maintains a trauma registry
database for all patients admitted to the R Adams Cowley Shock Trauma Center.11 BAC testing using gas-liquid chromatography is performed
on admission for more than 95% of the patients.12
Eligible cases were injured bicyclists admitted to the trauma center between
1990 and 1997, exclusive of those who were younger than 15 years, injured
at night, or without BACs measured within 6 hours of injury.
For each case bicyclist, up to 4 controls were selected at random from
passing-by bicyclists who were 15 years or older at the site where the case
bicyclist was injured and at the same time of day (within 2 hours), day of
week, and month of year as the injury.
Data for case bicyclists were ascertained from medical records. A trained
research assistant reviewed the records for all cases and abstracted the required
information. Data for control bicyclists came from roadside interviews and
tests of breath alcohol using the hand-held device, Alco Sensor III (Intoximeters
Inc, St Louis, Mo).13 Trained field workers
arrived at each study site 2 hours before the index time (ie, the time of
day at which the case bicyclist was injured) and asked passing bicyclists
to voluntarily give a brief anonymous interview and, at the end of the interview,
a breath sample for alcohol test. At each site, field workers stayed for up
to 2 hours after the index time unless 4 bicyclists were successfully interviewed
and tested for alcohol. For safety concerns, fieldwork was limited to day
light hours only (from 6 AM to 8 PM, varying with seasons). Roadside data
collection was conducted between June 1996 and May 1998. Of the 354 bicyclists
who stopped at the study sites, 342 (97%) completed the interview and alcohol
Data were analyzed using SAS software.14
Statistical significance at the bivariate level was assessed based on χ2 tests or Fisher exact tests where appropriate for categorical variables,
and t tests for continuous variables. The relative
risk of bicycling injury in relation to alcohol use was estimated through
conditional logistic regression modeling with consideration of the matched
case-control design. Helmet use was not included in the multivariate analysis
because not wearing a helmet has been recognized as an intermediate factor
in the pathway of alcohol use and bicycling injury.7,15
During 1985 through 1997, the OCME recorded a total of 133 bicyclist
fatalities. Excluded from the study were 40 bicyclists who died at age 14
years or younger, 42 who were injured at night, and 12 for whom valid estimated
BAC information was unavailable. Data collection on a total of 106 control
bicyclists was completed for 34 of the 39 fatal cases eligible for fieldwork.
An additional 253 injured bicyclists were identified for the years 1990
to 1997 from the trauma registry database maintained by the University of
Maryland Medical Center. Of these, 11 were excluded from the study due to
being underage (<15 years), 36 due to the lack of estimated BAC information,
52 due to being injured at night, and 64 due to missing information on the
time and/or locality of the injury event. For the remaining 90 nonfatal cases,
a total of 236 control bicyclists were interviewed and tested for alcohol.
Overall, there were 124 case bicyclists and 342 control bicyclists available
for data analysis.
The demographic characteristics of the cases and controls were similar
(Table 1). Overall, 82% of the
study subjects were men; 69%, white; and 56%, never married with a mean age
of 33.4 years. Alcohol was detected in 23.5% of the 34 fatal cases, 8.9% of
the 90 nonfatal cases, and 2.9% of the controls (P<.001),
yielding a crude odds ratio (OR) of 5.3 (95% confidence interval [CI], 2.1-13.1;
matched analysis). For those who tested positive for alcohol, the mean estimated
BAC was 0.18 g/dL for the cases and 0.07 g/dL for the controls.
With adjustment for age, sex, and race, the OR of bicycling injury relative
to an estimated BAC of less than 0.02 g/dL was 5.6 (95% CI, 2.2-14.0) for
riders with an estimated BAC of 0.02 g/dL or higher and was 20.2 (95% CI,
4.2-96.3) for riders with an estimated BAC of 0.08 g/dL (Table 2). The risk of bicycling injury associated with a positive
BAC was much greater for fatal injury than for nonfatal injury (Table 2).
To assess the potential bias resulting from the different time periods
that cases were injured and controls were interviewed, analysis was performed
separately for cases occurring between 1985 and 1993 and cases occurring between
1994 and 1997. Odds ratios related to BACs appeared to be consistent between
the 2 time periods (Table 3).
Among the 78 cases whose helmet use status was known, 8% (1/12) of those
with an estimated BAC of 0.02 g/dL or higher were wearing a helmet at the
time of injury compared with 38% (25/66) of those with an estimated BAC of
less than 0.02 g/dL (P = .09). Sensitivity analyses
based on 2 assumptions regarding cases with unknown helmet use status (ie,
either all were wearing or all were not wearing helmets) revealed that the
rate of helmet use at the time of injury ranged from 6% to 31% for those with
an estimated BAC of 0.02 g/dL or higher and from 23% to 62% for those with
an estimated BAC of less than 0.02 g/dL. None of the 10 control bicyclists
with an estimated BAC of 0.02 g/dL or higher was wearing a helmet at the time
of interview vs 34% (114/332) of their counterparts with an estimated BAC
of less than 0.02 g/dL (P = .03). Overall, bicyclists
with an estimated BAC of less than 0.02 g/dL were 7 times as likely as those
with an estimated BAC of 0.02 g/dL or higher to be wearing a helmet (35% vs
5%, respectively; P = .007).
Although alcohol has been implicated in many types of injuries, few
controlled epidemiologic studies have examined the role of alcohol in these
injuries. Our study expands the knowledge base of alcohol as an etiologic
factor in causing trauma. The risk of bicycling injury increases significantly
with BACs. In particular, alcohol at levels of 0.08 g/dL and higher may subject
bicyclists to a 20-fold heightened risk of fatal or serious injury.
There are 2 main pathways linking alcohol to injury.15,16
First, the excess risk of injury for drinking bicyclists may be caused by
the deleterious effects of alcohol on psychomotor skills, cognitive functions,
and safety behaviors, ie, a reduction in the rider's ability to maintain balance,
negotiate traffic, and perceive and respond to situational hazards. Second,
the increased risk of injury for alcohol-impaired bicyclists may result from
a host of risk-taking behaviors that correlate with alcohol use and expose
the subjects to more dangerous circumstances, such as riding at excessive
speed, on highways, at night, or under adverse weather conditions.3,7 Our study shows that bicyclists with
a positive BAC are less likely than other bicyclists to be wearing a safety
helmet. Although this disparity in helmet use represents a plausible explanation
for the heightened injury risk for bicyclists who consume alcohol, it is unclear
to what extent the dismal rate of helmet use among BAC-positive riders is
due to alcohol-induced impairment in cognitive functions and safety behaviors.
It is possible that people who ride bicycles after consuming alcohol may simply
be less likely than other bicyclists to own a safety helmet or to opt to wear
it when available. A recent study found that 30% of injured bicyclists with
a positive BAC have a history of alcohol-impaired driving.17
It is noteworthy that this study did not take into account alcohol use
by motorists, who are involved in 90% of all fatal bicycling injuries and
30% of all bicycling injuries requiring hospitalization.2,7
Other limitations of this study include potential selection bias and reduced
external validity. Information for control bicyclists was collected only from
those who stopped in response to our verbal invitation for participation.
If bicyclists who had been drinking were less likely than those who had not
been drinking to stop in response to our invitation to be interviewed, the
ORs of bicycling injury related to alcohol use reported in this study would
be somewhat overestimated.
Individually matching cases and controls on index site and time has
probably enhanced the internal validity of our findings. However, the rigorous
design proved to be labor-intensive and detrimental to the study power and
generalizability. Out of safety concerns for field workers in collecting data
for control bicyclists, matching on site and time excluded from analysis bicyclists
who were injured at night. Previous studies3,7
indicate that about 56% of fatal bicycling injuries and 32% of serious bicycling
injuries occur between 7 PM and 6 AM and indicate that bicyclists who are
injured at night are more likely than those who are injured during the day
to have been drinking and to have become intoxicated. Given the reduced visibility
and increased demand of psychomotor skills for bicyclists at night, it is
conceivable that the risk of bicycling injury attributable to alcohol use
is actually greater than reported in this study.