Full text of DOI:10.1111/j.1467-842X.2003.tb00402.x

Injury
Article
A cohort study of drink-driving motor vehicle
crashes and alcohol-related diseases
Abstract
Mark Stevenson, Peter D’Alessandro, Jack Bourke,
Matthew Legge and Andy H. Lee
Objectives: To elicit whether drivers
involved in alcohol-related motor vehicle
crashes are more likely to have future
alcohol-related hospital admissions.
Method: A population-based cohort study
of 3,286 drivers involved in a motor vehicle
crash between 1988 and 1992 were
followed over an eight to 13-year period.
Results: The findings from the study
suggest a twofold increased risk associated
with an alcohol-related motor vehicle crash
and future alcohol-related hospital
Injury Research Centre, School of Population Health,
The University of Western Australia
he effect of alcohol-related disease injured drivers with a BAC >0.05 g/100 mL
on Australian health resources is has fallen from 55% in 1983 to 31% in 1997.7
T
large. Harmful use of alcohol in
The reduction in the number of people
Australia has been estimated to have a direct killed on Australian roads as a result of
health care cost of $145 million a year.1 In alcohol is due to a combination of legislative
1986 alone, alcohol-related problems were change and public education. The
the cause of 5,360 deaths and more than one introduction of random breath testing, the
million hospital bed-days in Australia.2 In implementation of lower and nationally
1998, hazardous or harmful levels of alcohol consistent BAC limits, adoption of zero BAC
consumption were responsible for an limits for certain groups of road users,
estimated 3,271 deaths and an estimated tougher judicial penalties for drink-driving,
71,422 hospitalisations in Australia.3 Falls, and the introduction of compulsory blood
alcohol dependence, assaults and road testing on crash participants who attend
injuries are the most common reasons for hospital in some States8 have contributed to
hospitalisations due to excessive alcohol the reduction.
admission. The average time between an
alcohol-related motor vehicle crash and
future alcohol-related hospital admission
was 12 years. Men and Indigenous
Australian drivers were more likely to have
a future alcohol-related hospital admission.
Conclusion: It is evident from this study
that drink-driving resulting in a motor
vehicle crash and hospitalisation could be
considered an indicator of a less overt
problem of alcohol dependency.
consumption.4 Despite the apparent decrease
The legislative changes appear to have had
in the number of alcohol-related deaths, an impact on the risk-taking behaviour of
alcohol misuse still remains one of the Australians, reflected by the decrease in
Implications: It is important that penalties
for drink-driving go beyond merely punitive
action and provide rehabilitation.
leading causes of preventable deaths.
alcohol-related mortality onAustralian roads.
In Western Australia between 1990-94, Recent figures from the 2001 National Drug
there were 1,607 alcohol-related deaths, and Strategy Household Survey show that the
between 1991-95 there were 40,881 alcohol- proportion of Australians aged 14 years and
related hospital episodes.5 Alcohol was older who, in the previous 12 months, drove
wholly attributable in an estimated 9,914 a motor vehicle while under the influence of
hospital admissions and 519 deaths over this alcohol has fallen since 1998. Between 1998
(Aust N Z J Public Health 2003; 27: 328-32)
period.6
and 2001, the reported prevalence fell from
The misuse of alcohol has been identified 18% to 13%. Among males, the proportion
for many years as the single most significant decreased from 24% to 18% and among
contributing factor to road morbidity and females from 11% to 8%.1 These results also
mortality inAustralia. In 1997, 28% of fatally highlight the fact that while there has been a
injured motorists had a blood alcohol significant reduction in the number of
concentration (BAC) >0.05 g/100 mL, mak- Australians who drive under the influence of
ing alcohol the main cause of death on Aus- alcohol, males are still more than twice as
tralian roads.7 However, over the past 20 likely to do this as females.
years the percentage of alcohol-related road
Previous studies have shown that there
fatalities has decreased substantially; in is an increased risk of dying in alcohol-
Western Australia, the percentage of fatally related motor vehicle crashes among repeat
Correspondence to:
Submitted: July 2002
Associate Professor Mark Stevenson, Injury Research Centre, School of Population Health,
The University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009.
Fax: (08) 9380 1199; e-mail: marks@dph.uwa.edu.au
Revision requested: November 2002
Accepted: March 2003
328
AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH
2003 VOL. 27 NO. 3

Injury
Drink-driving motor crashes and alcohol-related diseases
drink-drivers. Brewer et al. found that drivers between the age of
21 and 34 years who died in an alcohol-related crash were 4.3
times more likely to have been arrested previously for driving
while intoxicated compared with drivers who died in non-alcohol
related crashes.9 For drivers 35 years or older, those killed in an
alcohol-related crash were 11.7 times more likely to have had a
previous arrest for drink-driving than drivers killed in non-alco-
hol-related crashes. While Brewer et al.’s research shows that re-
peat drink-drivers are at higher risk of dying in an alcohol-related
crash, there has been a dearth of literature on whether drink-driv-
ing offenders actually have an alcohol dependency problem.
Few studies have investigated the relationship between alco-
hol-related medical conditions (defined as one or more of the
diagnoses listed in Table 2) and road crashes. A study of Western
Australian drivers investigated alcohol-related medical conditions
as a risk factor for future alcohol-related motor vehicle crashes.10
Results indicated that only 3% of patients admitted to hospital
for an alcohol-related condition would be involved in a future
alcohol-related motor vehicle crash. A limitation of that study,
however, is people with alcohol dependency are unlikely to re-
quire hospitalisation for an alcohol-related medical condition until
later in their ‘alcohol career’. Consequently, an alcohol-related
motor vehicle crash is likely to occur before, rather than after,
hospitalisation for such conditions.
external cause of injury by ethanol or methanol poisoning, and
external cause of injury by alcohol aspiration.
It was necessary to link those drivers or riders identified in our
cohort to their hospital records. This was achieved by accessing
the Health Services Linked Database.All hospital admissions and
death records for each cohort member admitted to hospital for an
alcohol-related medical condition between 1 January 1988 and 31
December 1992 (the cohort recruitment period) and 31 December
2000 were extracted from the database. This ensured there was a
follow-up period of between 8 and 13 years from the date of
hospitalisation following a police-reported motor vehicle crash.
Details (including name, gender and date of birth) of individuals
involved in a crash were linked using probabilistic techniques to
the hospital data. The specific techniques used for linking (or
matching) the records are outlined in the paper by Rosman.11 It is
important to note that each individual presenting to a hospital in
WesternAustralia is allocated a unique identifier, which minimises
the likelihood of counting an individual twice.Therefore, all drivers
or riders in the cohort were followed-up using their unique
identifier.
The survival time for this study was calculated from the date of
hospitalisation following a police-reported motor vehicle crash
through to the date of first alcohol admission (as outlined above),
loss to follow-up, or censoring.
The aim of this study is to investigate the association between
involvement in an alcohol-related motor vehicle crash requiring
hospitalisation and subsequent admission to hospital for alcohol-
related medical conditions.
A description of the demographic characteristics of the cohort
was undertaken using SPSS software. Comparison of the means
for continuous variables was undertaken using independent t-tests.
Tests of association were undertaken using the chi-square test with
continuity correction where appropriate. Survival times (alcohol
versus non alcohol-related motor vehicle crash) were calculated
and the significance of any difference was determined by the log-
rank test. To determine the factors associated with alcohol-
related hospital admission (including time to first alcohol-related
hospital admission), both logistic regression and Cox’s propor-
tional hazards analyses were undertaken. Factors included in the
models were gender, age, Indigenous status and whether the driver
was involved in an alcohol-related motor vehicle crash. All
p-values were two-sided and were considered significant at 0.05
and 95% confidence intervals were calculated using the standard
errors from the analysis.
Methods
A population-based prospective cohort study was undertaken
involving all drivers or riders of motor vehicles who were
hospitalised in Western Australia during the period 1988 to 1992.
Motor vehicle crashes were identified through the Western
Australian Road Injury Database, which links police crash reports
obtained from the Traffic Accident System of the Main Roads
Department of Western Australia with hospital discharge records
for the State ofWesternAustralia.To determine whether the driver
involved in a motor vehicle crash had a BAC exceeding 0.05gm/
100ml only police-attended motor vehicle crashes were included
in the cohort (n=3,286). A BAC exceeding 0.05gm/100ml was
defined using a calibrated breath test by a police officer. This
equipment is used to provide evidence in court.
Results
There were 3,286 drivers or riders of a motor vehicle admitted to
hospital following a police-attended motor vehicle crash between
1988 and 1992 in Western Australia. The number of observations
was equally distributed across the five-year recruitment period,
namely 20% per year (χ2=4.27, df=4, p=0.371). Seven per cent
(n=217) of the cohort was classified as an alcohol-related (had a
BAC ≥0.05 as reported by the attending police officer) motor vehi-
cle crash.The average age of the cohort was 32 years (range=15-88
years) with significantly more males (71%, n=2,345, χ2=7.94, df=1,
p=0.005). Three per cent (n=95) of the cohort were Indigenous
Australians. Descriptive statistics for alcohol and
The outcome of interest was a subsequent alcohol-related
hospital admission. An alcohol-related hospital admission was
defined as a medical diagnosis that could only have resulted from
excessive alcohol consumption. These medical diagnoses were
defined as having an aetiologic fraction of 1.0 (100% attribution)5
and were identified using the ICD-9-CM (prior to 1 July 1999)
and ICD-10-AM (from 1 July 1999 onwards) codes as alcohol
psychosis, alcohol dependence, alcohol abuse, alcohol
polyneuropathy, alcohol cardiomyopathy, alcoholic gastritis,
alcoholic liver cirrhosis, ethanol toxicity, methanol toxicity,
2003 VOL. 27 NO. 3
AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH
329

Stevenson et al.
Article
Table 1: Demographic features of the cohort.
non-alcohol-related motor vehicle crashes are presented inTable 1.
A total of 104 (3%) of the cohort members had at least one
alcohol-related hospital admission over the observational period,
with the number of alcohol-related admissions ranging from one
to 36 admissions. The percentage of various medical diagnoses
for alcohol-related hospital admission is presented inTable 2.There
was a significant association between an alcohol-related motor
vehicle crash and future alcohol-related hospital admission
(χ2=6.05, df=1, p=0.014).There was also a significant association
between Indigenous and non-Indigenous Australians and future
alcohol-related hospital admissions. Eighteen per cent (n=17) of
Indigenous Australians in the cohort had a future alcohol-related
hospital admission compared with 2% (n=87) of non-Indigenous
Australians (χ2=69.3, df=1, p<0.001).
Variables
Alcohol-related
motor vehicle
crash related
Non alcohol-related
motor vehicle
crash related
Number
217
3,069
1.06
Mean number
of admissions
(SD)
1.12
0.78
0.85
Mean age
of drivers
(SD)
30.07
32.44
14.23
15.88
Sex of the driver
Male
Female
163
54
2,183
886
71.13
(male proportion)
75.12%
Indigenous Australian
10
86
2.80%
(proportion)
4.61%
It is evident from the logistic regression results in Table 3 that
an alcohol-related motor vehicle crash is a significant predictor
of future alcohol-related hospital admissions. In fact, the likelihood
of an alcohol-related hospital admission is almost twofold greater
(OR=1.96, 95% CI 1.06-3.61) if, at the time of recruitment, the
driver was involved in an alcohol-related motor vehicle crash.This
finding applies when the age, sex and Indigenous status of the
driver or rider of the motor vehicle is taken into account.
It is also interesting to note that Indigenous Australians had an
eightfold (OR=8.07, 95% CI 4.53-14.3) increased risk of a future
alcohol-related hospital admission following an alcohol-related
motor vehicle crash resulting in hospitalisation. However, due to
the small sample size, this estimate is unstable.
Alcohol-related
hospital admissions
(proportion)
13
91
5.99%
2.99%
Table 2: Distribution of alcohol-related medical
conditions (n=104).
Alcohol-related ICD
code combinationsa
Frequency Per cent
ICD-9-CM
Other and unspecified alcohol
dependence (303.9)
25
24.04
Toxic effects of alcohol (980)
12
8
11.54
7.69
6.73
Alcohol dependence syndrome (303)
Other and unspecified alcohol
7
dependence, continuous (303.91)
There was a significant difference between alcohol and non-
alcohol-related motor vehicle crashes and time to first alcohol-
related hospital admission (Log Rank Test 7.64, df=1, p=0.005).
The average time it took for a driver involved in an alcohol-related
motor vehicle crash to be re-admitted for an alcohol-related
condition (other than a motor vehicle crash) was 12 years (95%
CI 11.7-12.2) compared with 12 years 7 months (95% CI 12.7-
12.8) for drivers involved in a non-alcohol-related motor vehicle
crash.
Other specified alcoholic psychosis (291.8),
alcohol abuse, continuous (305.01)
3
2.88
Other alcohol-related conditions
ICD-10-AM
21
20.19
Harmful alcohol use (F10.1)
Acute alcohol intoxication (F10.0)
Alcohol dependence syndrome (F10.2)
11
9
10.58
8.65
4.81
0.96
5
Acute alcohol intoxication (F10.0),
harmful alcohol use (F10.1)
1
Table 4 reports the key predictors of time to first alcohol-related
hospital admission. The Cox’s regression findings are similar to
those reported in Table 3, namely that drivers involved in an
alcohol-related motor vehicle crash are twice as likely (Hazard
Ratio 2.05, 95% CI 1.14-3.67) to be admitted, in the future, for an
alcohol-related condition compared with drivers not involved in
an alcohol-related motor vehicle crash. Figure 1 contrasts the
cumulative hazard function for the time to first alcohol-related
hospital admission between cohort members defined at recruitment
as an alcohol- or non-alcohol-related motor vehicle crash, after
adjusting for age, sex and Indigenous status of the drivers (factors
found to be important from the multivariate modelling).
Acute alcohol intoxication (F10.0),
external cause – severe alcohol
intoxication (Y91.2)
1
1
0.96
0.96
Alcoholic cirrhosis of the liver (K70.3),
alcohol dependence syndrome (F10.2)
Note:
(a) Codes include diagnosis and external cause of morbidity codes.
Table 3: Predictors of alcohol-related hospital admissions.
Variable
Odds ratio
95% CI
Alcohol-related motor vehicle crash
No
Yes
1.00
1.96
–
1.06-3.61
Sex of driver
Female
Male
1.00
2.18
–
1.28-3.73
Discussion
Indigenous Australian
No
Yes
1.00
8.07
–
The aim of this study was to investigate the association between
involvement in an alcohol-related motor vehicle crash requiring
hospitalisation and future admission(s) to hospital for alcohol-
4.53-14.3
Note:
Model adjusted for age.
330
AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH
2003 VOL. 27 NO. 3

Injury
Drink-driving motor crashes and alcohol-related diseases
Table 4: Predictors of time to first alcohol-related
hospital admission.
vehicle crashes resulting in hospitalisation may be beneficial.
VariousAustralian studies have found that age, sex, employment,
education and marital status are associated with patterns of problem
drinking.12 This study clearly highlights that being male and
Indigenous status are key determinants of future alcohol-related
hospital admissions.
Variable
Hazard ratio
95% CI
Alcohol-related motor vehicle crash
No
1.00
2.05
–
Yes
1.14-3.67
A limitation of the study is the relatively short follow-up period
(8-13 years) and, consequently, the small number of subjects
admitted to hospital for an alcohol-related disease during the
observational period. While diseases such as alcoholic psychosis,
alcohol abuse, acute alcohol gastritis, ethanol or methanol toxicity,
and injury caused by alcohol poisoning or aspiration tend to have
short aetiologic periods, diseases such as alcohol polyneuropathy,
alcohol cardiomyopathy and alcoholic liver disease usually take
many years to develop. Clearly a greater follow-up period, be-
yond the current study, is necessary to capture cases that become
serious enough to require hospitalisation. We were limited to the
study’s observational period as prior to 1988 the classification of
an alcohol-related motor vehicle crash was not reliable.
Sex of driver
Female
1.00
2.16
–
Male
1.28-3.65
Indigenous Australian
No
1.00
7.46
–
Yes
4.41-12.6
Note:
Model adjusted for age
related medical conditions. The study proposed that people in-
volved in alcohol-related motor vehicle crashes were more likely
to be admitted for an alcohol-related disease in the future. More-
over, there might be the potential, if the association was true, that
drivers involved in serious alcohol-related motor vehicle crashes
could be candidates for prevention programs focused on alcohol
dependence.
Another issue that should be considered is the fact that hospi-
talisation represents one of the most severe outcomes of alcohol-
related diseases under investigation. Without question, a number
of patients within the cohort would have developed these dis-
eases during the follow-up period, but would not have required
hospitalisation. Furthermore, a classification that the alcohol-
related disease must have an aetiologic fraction of one was very
rigorous. For these reasons, it is likely that our estimate under-
represents the true value.
The findings suggest a twofold increased risk associated with
an alcohol-related motor vehicle crash and future alcohol-related
hospital admission. In fact, the average time between an alcohol-
related motor vehicle crash and future alcohol-related hospital
admission was 12 years. This finding, however, does not
discriminate between the time from alcohol-related crash and a
specific diagnostic alcohol-related disease. Unfortunately, due to
insufficient numbers, we could not estimate this relationship but
it is possible that with a longer observational period this could be
determined.
The issue of whether drink-drivers are actually ‘driving drinkers’
has been raised numerous times over the past 30 years inAustralia.
As far back as 1973, a Law Reform Commission report on Alcohol,
Drugs and Driving13 found that a significant proportion of drivers
who had died in alcohol-related crashes were likely to be chronic
alcoholics. The report also discussed appropriate penalties and
countermeasures for drink-driving. It recommended greater fines
and licence disqualification, with the latter seen as a greater
deterrent and a powerful factor for the individual not to repeat the
offence, although this is complicated by high rates of driving under
suspension, especially for long disqualifications. Furthermore,
while it is known that more than one drink-driving offence is pre-
dictive of alcohol dependence,9 it is important to recognise that
most alcohol-related injuries involve high-risk but non-depend-
ent drinking.
This is the first population-based study to investigate the
association, prospectively, and provides preliminary evidence that
prevention programs targeting drink-drivers involved in motor
Figure 1:Time to first alcohol-related hospital admission
by alcohol- and non-alcohol-related motor vehicle crash.
.07
.06
.05
.04
.03
.02
More recently, alternative countermeasures for drink-drivers
have been considered. Several United States (US) experiments
have shown that the main thrust of countermeasures should be
rehabilitation therapy and treatment programs, especially in the
group of people with a drinking problem serious enough to cause
them to break the law. The US uniform vehicle code includes a
provision for the treatment of alcoholics and drug addicts as an
alternative to the imposition of conventional penalties and such
programs have been reported to reduce recidivism by up to 9%.14
Likewise, underTasmania’s Road Safety (Alcohol and Drugs)Act
1970, it allows a court to order an offender into treatment with
.01
0.00
0
2
4
6
8
10
Time to Admission (years)
2003 VOL. 27 NO. 3
AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH
331

Stevenson et al.
Article
2. Holman CDJ, Armstrong BK, Arias LN, Martin CA, et al. The Quantification
of Drug Caused Morbidity and Mortality in Australia 1988, part 2. Canberra:
AGPS; 1990.
3. Ridolfo B, Stevenson C. The Quantification of Drug-caused Mortality and
Morbidity in Australia, 1998. Canberra: Australian Institute of Health and
Welfare; 2001. Drug Statistics Series No.: 7, AIHW Catalogue No.: PHE 29.
4. National Alcohol Strategy: A Plan for Action 2001 to 2003-04. Canberra:
Commonwealth Department of Health and Aged Care; 2001.
respect to alcohol and drink-driving. The act also provides that a
second offender will not be reissued with a driver’s licence after
disqualification unless the court is satisfied that they are not
suffering from alcohol or drug dependency. To our knowledge,
however, there is no provision of a treatment or rehabilitation
program for drink-driving in Western Australia specifically
targeting repeat drink-drivers.
5. English DR, Holman CDJ, Milne E, Winter MG, et al. The Quantification of
Drug Caused Morbidity and Mortality inAustralia. Canberra: Commonwealth
Department of Human Services and Health; 1995.
It is evident from this study that drink-driving resulting in a
motor vehicle crash could be considered an indicator of a less
overt problem of alcohol dependency. It is important, therefore,
that penalties for drink-driving must recognise the basis of this
critical issue. There is a need (particularly in Western Australia)
to establish a flexible range of penalties using both conventional
and alternative countermeasures to successfully treat drink-driving.
6. Unwin E. Drug-caused Deaths inWesternAustralia, 1984-1994. Perth: Health
Information Centre, Health Department of Western Australia; 1996.
7. Federal Office of Road Safety. Alcohol and Road Fatalities in Australia 1997.
Canberra: Department of Transport Regional Development; 1999.
8. Federal Office of Road Safety. Alcohol and Road Fatalities in Australia 1996.
Canberra: Department of Transport Regional Development; 1997.
9. Brewer RD, Morris PD, ColeTB, Watkins S, et al.The risk of dying in alcohol-
related automobile crashes among habitual drunk-drivers. N Engl J Med
1994;331(8):513-7.
10. Cercarelli LR, Rosman DL, Kirov C, Legge M. The Relationship between
Alcohol Related Medical Conditions and Road Crashes. Perth: RoadAccident
Prevention Research Unit, The University of Western Australia; 1999. Report
No.: RR85.
Acknowledgements
11. Rosman DL. The Western Australian road injury database (1987-1996): ten
years of linked police, hospital and death records of road crashes and injuries.
Accid Anal Prev 2001;33:81-8.
The research was supported by the Road Safety Council of
Western Australia. We would like to thank the support staff in the
Injury Research Centre for assisting with the research.
12. Jonas H, Dietze P, Rumbold G, Hanlin K, et al. Associations between alcohol
related hospital admissions and alcohol consumption in Victoria: Influence
of socio-demographic factors. Aust N Z J Public Health 1999;23(3):272-9.
13. The Law Reform Commission. Alcohol, Drugs and Driving [preliminary
paper]. Canberra: Parliament of the Commonwealth of Australia; 1976.
14. Ferguson M, Sheehan M, Davey J, Watson B. Drink Driving Rehabilitation:
The Present Context. Canberra:Australian Transport Safety Board; 1999 Nov.
Report No.: CR 184.
References
1. Australian Institute of Health and Welfare. 2001 National Drug Strategy
Household Survey: First Results. Canberra: AIHW; 2002. Drug Statistics
Series No.: 9, AIHW Catalogue No.: PHE 35.
332
AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH
2003 VOL. 27 NO. 3