Trafﬁc Injury Prevention, 6:230–234, 2005
2005 Taylor & Francis Inc.
ISSN: 1538-9588 print / 1538-957X online
Unlicensed Drivers and Car Crash Injury
STEPHANIE BLOWS and REBECCA Q. IVERS
The George Institute for International Health, University of Sydney, Sydney, Australia
JENNIE CONNOR and SHANTHI AMERATUNGA
School of Population Health, University of Auckland, Auckland, New Zealand
MARK WOODWARD and ROBYN NORTON
The George Institute for International Health, University of Sydney, Sydney, Australia
Objective.Previous studies have indicated that unlicensed drivers are more likely to engage in risky driving behaviors,
and are more likely than licensed drivers to be at fault and more seriously injured when involved in a crash. However, the
prevalence of unlicensed drivers in the general driving population has not been measured, and the risk of an unlicensed
driver being involved in an injury crash has not been quantiﬁed. We examined the association between unlicensed driving
and car crash injury using data from a population-based case control study.
Methods. The study population was the drivers of all cars on public roads in the Auckland region. Cases were 571 vehicles
involved in a crash resulting in any occupant being hospitalised or killed, from the study base, during the recruitment period.
Controls were 588 vehicles selected from the driving population using a random cluster sampling method. The drivers of all
vehicles completed a structured interview covering multiple potentially crash-related factors.
Results. Driving unlicensed was reported by 12% of case and 1% of controldrivers. Unlicensed drivers were at signiﬁcantly
higher risk of car crash injury than those holding a valid licence (odds ratio 11.1, 95% conﬁdence interval 4.2 to 29.7) after
adjustment for age and sex. After further adjustment for education level, ethnicity, driving exposure, time of day, sleepiness
score, year of vehicle manufacture, passenger carriage, seatbelt use, blood alcohol concentration, and travelling speed at
time of crash, the increased risk was still present but no longer signiﬁcant (OR 3.9, 95% CI 0.7–22.4).
Conclusions. Unlicensed drivers are a high risk group for car crash injury after taking other crash-related risk factors
into account. Strategies to reduce unlicensed driving may therefore facilitate reductions in road crashes, although further
work is needed in this area.
Keywords Transportation; Motor Vehicle Injury; Unlicensed Drivers; Case-Control Study
Many countries require drivers to hold a licence to legally
operate a motor vehicle, and reserve the right to suspend or dis-
qualify this licence if road rules are not obeyed. Despite such
laws, a number of people drive unlicensed, either having never
obtained a licence or continuing to drive after their licence has
been suspended or disqualiﬁed (Grifﬁn & DeLaZerda, 2000;
Knox et al., 2003). These unlicensed drivers are thought to be at
higher risk of motor vehicle injury than their licensed counter-
parts, but this has been difﬁcult to prove (Federal Ofﬁce of road
Safety, 1997a; Grifﬁn & DeLaZerda, 2000; Knox et al., 2003).
Several studies of unlicensed drivers have suggested that
they are more likely to engage in risky driving behaviors such
as speeding, drink driving, red light running, and non-use of
Received 22 September 2004; accepted 14 March 2005.
Address correspondence to Stephanie Blows, The George Institute for Inter-
national Health, University of Sydney, Level 10, King GeorgeV Building, Royal
Prince Alfred Hospital, Missenden Road, Camperdown, NSW 2050, Australia.
seatbelts, than those with a valid licence (Federal Ofﬁce of
Road Safety, 1997; Retting et al., 1999; Grifﬁn & DeLaZerda
2000; Kim & Kim, 2003). It has also been suggested that unli-
censed drivers may be over-represented in crash statistics (Fed-
eral Ofﬁce of Road Safety, 1997; Grifﬁn & DeLaZerda, 2000),
but this has not been conclusively demonstrated because of a
lack of comparative data on the prevalence of unlicensed driv-
ing in the general driving population (DeYoung et al., 1997;
Knox, 2003). Studies of crashed drivers have found that com-
pared to licensed drivers, unlicensed drivers are more likely to
be at fault (Perneger & Smith, 1991; Land Transport Safety Au-
thority, 2003) and more seriously injured (Harrison, 1997) when
involved in a crash. These studies used control drivers who were
involved in a crash, rather than population-based controls, so
were unable to determine the excess risk of car crash injury for
Only one previous study has estimated the driving expo-
sure of suspended/revoked and unlicensed drivers, using a
“quasi-induced” exposure estimation method (DeYoung et al.,
UNLICENSED DRIVER CRASHES 231
1997). This study found that both groups were at higher risk of
crash involvement compared to validly licensed drivers, but cau-
tion that their exposure estimates are likely to be subject to sig-
niﬁcant bias. There remains a lack of estimates of the prevalence
of unlicensed driving amongst the general driving population,
and because of this the excess risk of an unlicensed driver’s in-
volvement in a crash that leads to injury has not previously been
In New Zealand, a driver’s licence is obtained by completing
a series of applications and driving tests, and payment of a num-
ber of fees. Unlicensed drivers may have not undertaken this
licensing process, or they may have had their licence suspended
or disqualiﬁed for a variable period of time as a result of offences
such as drink driving, speeding, or other serious infringements of
the road rules (Land Transport Safety Authority, 2004). We used
data from a population-based case-control study conducted in
the Auckland region of New Zealand to examine the prevalence
of unlicensed driving in the regional driving population, and the
relationship between unlicensed driving and car crash injury.
A complete description of the methodology of the study has
been published previously (Connor et al., 2002). Recruitment
took place from 1998 to 1999 in the Auckland region, which
contains the largest city in New Zealand and a mixture of other
urban, suburban and rural areas. The regional population is about
1.1 million people (Statistics New Zealand, 2001). The study
base was deﬁned as all light vehicles driving on non-local pub-
lic roads in the region. Case vehicles were identiﬁed when any
occupant of a vehicle from the study base was hospitalized or
killed in a crash during the recruitment period. Case identiﬁca-
tion took place through surveillance at the four hospitals serving
the Auckland region, and through the Auckland Coroner.
During the study period 615 eligible case drivers were iden-
tiﬁed and interviews were completed for 571 (93%) of these.
Control selection aimed to achieve a representative sample of
all driving time for the study base during the recruitment period.
Control vehicles were identiﬁed during the same time interval
and at approximately the same rate as cases. To select controls, a
list of roads in the Auckland region was obtained and 69 roadside
sites were randomly selected from this list. A day of the week,
time of day, and direction of travel were randomly assigned to
each site. Study staff then visited the site at the selected time
and vehicles that passed the site during a deﬁned period were
randomly selected as control vehicles. The number of vehicles
selected from each site was proportional to the volume of trafﬁc
at the site. These vehicles were stopped at the roadside and a suit-
able time for a telephone interview was arranged. During these
roadside surveys 746 control cars were identiﬁed, and of these,
interviews were completed with 588 drivers (79% response rate).
Interviews for the drivers of case and control vehicles were
conducted by telephone for 204 (36%) case drivers and 576
(98%) control drivers; the remaining interviews were conducted
in person. Proxy respondents were interviewed for 57 case
drivers and two control drivers who were fatally injured or un-
able to be interviewed for other reasons. All interviews were
based on a structured questionnaire that included characteristics
of the driver, circumstances of the crash, and vehicle character-
istics. For control drivers, the interview was referenced to the
time of being sampled in the roadside survey. Licence status was
determined by asking drivers what type of car licence they held
at the time of the crash or survey.
For these analyses, “unlicensed” drivers were those that had
never held a car licence or whose licence was disqualiﬁed or sus-
pended at the time of the crash/survey; other types of licence,
including full licences, learner licences, and overseas licences,
were considered valid. Blood alcohol level wasdetermined using
a breathalyser for controls and from hospital and police records
for cases. Missing data for blood alcohol level was imputed ac-
cording to self reported alcohol consumption prior to the crash
and the suspicions of ambulance and hospital staff. Details of al-
cohol imputation have been previously published (Connor et al.,
2004). Environmental surveys of crash and control recruitment
sites were conducted to measure environmental factors poten-
tially related to crashes.
Odds ratios (OR’s) and 95% conﬁdence intervals (CIs) were
calculated from linear logistic regression models using SU-
DAAN software, which accounts for intra-cluster correlation of
control data sampled from the same site (Shah et al., 1997). For
the multivariable analyses, we identiﬁed potential confounders
from the epidemiological literature and adjusted for these if they
were signiﬁcantly associated with car crash injury in our data
after controlling for driver’s age and sex. Because unlicensed
driving may inﬂuence crash risk indirectly through its associa-
tions with other risky driving, we examined the association be-
tween unlicensed driving and car crash injury by ﬁrst adjusting
only for age and sex.
We then adjusted for ethnicity, education level, and driving
exposure (average hours spent driving per week), plus acute
driving-related exposures at the time of the crash/survey (pas-
senger carriage, time of day, Stanford sleepiness score, year of
vehicle manufacture, blood alcohol concentration, seatbelt use,
and travelling speed). We also examined the contribution that
each of these confounders made to the age and sex adjusted odds
ratio for the association between unlicensed driving and car crash
injury. This was done by adding each variable to this model in-
dividually and estimating the percentage by which this changed
the odds ratio, using the formula 100 ([ORU-ORA]/[ORU– 1])%
where ORUand ORAare, respectively, the odds ratios for unli-
censed driving and car crash injury unadjusted (except by age
and sex) and after further adjustment for each risk factor alone.
The mean age of case drivers was 36.6 years, and control
drivers 40.8 years. The case group was 65% male and the control
group was 59% male. There were no signiﬁcant differences in
age group, sex, or driving conditions, between drivers who were
interviewed and all eligible drivers, for both case and control
232 S. BLOWS ET AL.
Figure 1 Unadjusted, age and sex adjusted, and multivariable adjusted odds
ratios (95% conﬁdence intervals) for the association between unlicensed driving
and car crash injury, Auckland Car Crash Injury Study. The multivariable odds
ratio is adjusted for age, sex, ethnicity, education, driving exposure, passenger
carriage, time of day, sleepiness score, year of car manufacture, blood alcohol
concentration, seatbelt use, and traveling speed at the time of crash.
Table I shows the frequency distributions of licence status
and confounding variables by case/control status. The preva-
lence of unlicensed driving at the time of the crash/survey was
11.6% (n =66) amongst cases and 1.1% (n =7) amongst con-
trols. Missing data for both cases and controls was less than 1%
for licence status, and less than 10% for all variables used in
these analyses after imputation for blood alcohol level.
Figure 1 shows the unadjusted, age and sex adjusted, and mul-
tivariable adjusted odds ratios and 95% conﬁdence intervals for
the association between unlicensed driving and car crash injury.
Unlicensed drivers were at signiﬁcantly higher risk of car crash
injury than those holding a valid licence in the unadjusted model
(OR 12.1, 95% CI 3.8 to 38.4), and after adjustment for age and
sex (OR 11.1, 95% CI 4.2 to 29.7) However, after adding edu-
cation level, ethnicity, driving exposure, time of day, sleepiness
score, year of vehicle manufacture, passenger carriage, seatbelt
use, blood alcohol concentration, and travelling speed at time
of crash, the increased risk was no longer signiﬁcant (OR 3.9,
95% conﬁdence interval 0.7 to 22.4). These adjustments had a
similar effect when restricted to the subset of participants with
Figure 2 shows the proportion of the age and sex adjusted odds
ratio for unlicensed driving and car crash injury explained by the
confounders we examined. Ethnicity and education level were
the two largest contributors, accounting for 47% and 34% of
the odds ratio respectively, followed by Stanford sleepiness score
(32%), blood alcohol level (25%) and driving exposure (22%).
This population based case control study allowed us to exam-
ine the prevalence of unlicensed driving the Auckland regional
driving population and the excess risk of car crash injury for
unlicensed drivers. Driving unlicensed was reported by 12%
of cases in this study, of whom 10% had never held a licence
and 2% held a licence that was currently suspended or disquali-
ﬁed. Unlicensed drivers had about 11 times higher risk of being
Table I Frequency distributions of licence status and confounding variables
by case/control status, Auckland Car Crash Injury Study
Cases (n =571) Controls1(n =588)
No. (%) No. (%)
Unlicensed (total) 66 (11.6) 7 (1.1)
Never held 55 (9.6) 6 (1.1)
Disqualiﬁed/suspended 11 (1.9) 1 (0.02)
Valid 502 (87.9) 580 (98.8)
Don’t know/missing 3 (0.5) 1 (0.1)
Age of driver (years)
<25 195 (34.2) 91 (13.7)
25–34 133 (23.3) 125 (22.3)
35–44 85 (14.9) 154 (24.5)
45–54 61 (10.7) 107 (19.6)
55–64 39 (6.8) 80 (14.2)
65+58 (10.2) 31 (5.6)
Female 198 (34.7) 226 (41.3)
Male 373 (65.3) 362 (58.7)
Post secondary 178 (31.5) 276 (49.3)
Secondary school, >3 years 137 (24.2) 154 (25.1)
Secondary school, ≤3 years 252 (44.4) 157 (25.6)
White/European 313 (54.8) 444 (74.7)
Maori 117 (20.5) 61 (9.2)
Paciﬁc Islander 86 (15.1) 36 (6.1)
Other 55 (9.6) 47 (10.0)
(average hours per week)
≤5 219 (42.1) 171 (30.5)
6–10 205 (39.4) 216 (39.3)
11–20 63 (12.1) 135 (22.3)
21–30 11 (2.1) 32 (3.8)
>30 22 (4.2) 27 (4.1)
Time of day
Not between 2–5 am 525 (91.9) 571 (99.6)
Between 2–5 am 46 (8.1) 17 (0.4)
Stanford Sleepiness Score
1–3 (sleepy) 447 (87.7) 578 (99.0)
4–7 (not sleepy) 63 (12.3) 8 (1.0)
Year of vehicle manufacture
<1984 118 (23.9) 58 (8.9)
1984 to 1988 157 (31.8) 164 (31.0)
1989 to 1993 169 (34.2) 207 (37.8)
≥1994 50 (10.1) 133 (22.4)
Number of passengers
0 285 (50.3) 355 (62.9)
1 140 (24.7) 144 (23.4)
2ormore 142 (25.0) 88 (13.7)
Yes 469 (82.1) 568 (97.4)
No 81 (14.2) 4 (0.8)
Blood alcohol concentration (mg %)2
<3 397 (69.7) 565 (96.6)
3–50 41 (7.2) 16 (2.6)
>50 132 (23.2) 6 (0.8)
0–30 kph 87 (16.9) 78 (13.9)
31–50 kph 113 (21.9) 229 (41.5)
51–80 kph 196 (38.1) 220 (33.7)
>80 kph 119 (23.1) 55 (11.0)
1Proportions of controls are adjusted for the clustered sampling design.
2Missing data imputed (Connor et al., 2004).
UNLICENSED DRIVER CRASHES 233
Figure 2 Proportion of the age and sex adjusted odds ratio for unlicensed
driving and car crash injury explained by other risk factors, Auckland Car Crash
involved in a serious injury crash compared to drivers holding a
valid licence after adjustment for age and sex. This positive as-
sociation was still present after adjusting for other crash-related
factors, although it was no longer signiﬁcant probably due to
lack of power.
Our prevalence estimates are consistent with several previous
studies. Crash statistics from New Zealand indicate that in 1998,
11% of fatal crashes involved a disqualiﬁed or unlicensed driver
(Land Transport Safety Authority, 2000). Grifﬁn and DeLaZerda
(2000) examined 278,078 drivers involved in fatal crashes us-
ing the Fatal Accident Reporting System in the United States.
Of these drivers, 11% of these held an invalid licence or had
no known licence. Harrison (1997) found that of fatal crashes
in Victoria, Australia, 2% had a disqualiﬁed licence. However,
there are few previous estimates of the frequency of unlicensed
driving amongst the general driving population. In the control
population of our study, conceptually representing the Auckland
regional driving population, 1% reported being unlicensed or
disqualiﬁed/suspended at the time of the roadside survey. Us-
ing a “quasi-induced” exposure method, also using data from
the Fatal Accident Reporting System, DeYoung et al. (1997)
estimated that 9% of the driving population in California had a
suspended or revoked licence. This is higher than our estimate,
but the authors of this study note several serious limitations of
their estimation methods. Further representative surveys are re-
quired to measure the prevalence of unlicensed driving more
We also found a strong association between unlicensed driv-
ing and car crash injury after adjustment for age and sex. Even
after adjustment for other crash-related risk factors, the point
estimate for the odds ratio was about four, although this was no
longer signiﬁcant. This lack of signiﬁcance is likely to be due to
small numbers producing wide conﬁdence intervals. Our ﬁnd-
ing that unlicensed drivers are a high-risk driving population is
consistent with previous research. Two studies have examined
the effect of driving unlicensed on crash severity. Shibata &
Fukuda (1994) used data from police-reported trafﬁc crashes
of both cars and motorcycles in Fukuoka Prefecture, Japan,
comparing characteristics of crash-involved drivers who died
to those who were uninjured (Shibata & Fukuda, 1994). Af-
ter adjustment for age, unlicensed drivers of cars were three
times more likely, and drivers of motorcycles nine times more
likely, to be killed when involved in a car crash compared to
licensed drivers. The association was signiﬁcant only for motor-
Harrison (1997) conducted a similar study of crash-involved
drivers in Victoria, Australia, and found a signiﬁcant difference
between the frequency of disqualiﬁed licences amongst fatally
injured drivers (4.6%) compared to uninjured drivers (0.7%)
in unadjusted analyses (Harrison, 1997). Another study exam-
ined the effect of various driver characteristics, including licence
status, on involvement in an at-fault crash (Perneger & Smith,
1991). Using paired crash data from the Fatal Accident Report-
ing System in the United States, this study found that drivers
with an invalid licence were about twice as likely to have ini-
tiated the crash compared to those holding a valid licence. Our
results conﬁrm unlicensed drivers to be a population at high
risk of serious car crashes and suggest that unlicensed drivers
have three times excess risk of involvement in an injury crash
compared to licensed drivers.
Driving unlicensed is unlikely to directly increase the risk
of car crash injury and the mechanism by which this popu-
lation is at risk is probably through associations with other
crash-related factors. Of the acute risky driving behaviors we
examined, sleepiness and blood alcohol level accounted for the
largest proportion of the age and sex adjusted odds ratio. Other
studies have also found evidence that unlicensed drivers may be
more likely to display risky driving behaviors, including speed-
ing, drink driving, red-light running, and non-use of seatbelts
(Federal Ofﬁce of Road Safety, 1997b; Retting et al., 1999; Grif-
ﬁn & DeLaZerda, 2000; Kim & Kim, 2003). In our data, educa-
tion level, ethnicity, and driving exposure were also important
contributors to the relationship between unlicensed driving and
car crash injury.
There were signiﬁcant associations between unlicensed driv-
ing and ethnicity (p =0.01), with more Maori and Paciﬁc Is-
landers being both never licensed and holding a disqualiﬁed or
suspended licence. There were no signiﬁcant associations be-
tween licence status and other demographic variables, including
age (p =0.2). Because of small numbers we were not able to
fully investigate the relationships between licence status and
other associated variables; this will be an interesting area for
future research. However, although a variety of crash-related
variables may account for the relationship between unlicensed
driving and car crash injury, explicit knowledge of these need
not be a prerequisite for the implementation of countermeasures
aimed at unlicensed drivers.
Our study has several potential limitations. Licence status
and many of the other variables were self reported. Because
driving unlicensed is illegal in New Zealand, this may be un-
derreported and is therefore a potential source of measurement
234 S. BLOWS ET AL.
bias. However, questions on illegal behaviors were embedded in
a large questionnaire containing multiple items relating to driv-
ing; interviewers were highly trained and assured participants of
complete conﬁdentiality. Our data on excess alcohol consump-
tion prior to driving (also illegal in New Zealand) suggest that
self report is a valid measure, as the correlation between self re-
ported and objective measures of alcohol consumption was high
(Spearman correlation coefﬁcient =0.77).
The differential response rate between cases and controls may
have introduced selection bias, particularly if non-responders
amongst the control population tended to be unlicensed. If this
is the case, we are likely to have underestimated the prevalence
of unlicensed driving in controls, which would result in an over-
estimate of the risk of injury crashes. Confounding variables,
particularly those that relate to risky and illegal driving, may
have also been inaccurately measured. Case control studies using
self-reported data may also be subject to recall bias (Woodward,
2005), although it is difﬁcult to predict what effect this would
have on our estimate of effect.
The increased risk of injury to vehicle occupants when the
driver is unlicensed supports the need for interventions aimed at
this population. Other authors have suggested a variety of mea-
sures that may be effective, including increasing police resources
to enable enforcement, reviewing and broadening the penalties
applied, and increasing public awareness of the dangers of un-
licensed driving and the penalties involved (Knox et al., 2003).
Others have proposed applying barriers to driving including ve-
hicle impoundment, electronic driver licences, and ignition in-
terlocks when licences are suspended or disqualiﬁed (Grifﬁn &
For most jurisdictions, interventions in the short term are
likely to focus on enforcement and education strategies. Unli-
censed driving is unlikely to be a randomly distributed charac-
teristic and the identiﬁcation of high-risk groups and risk fac-
tors for unlicensed driving, for example, age, gender, ethnicity,
socioeconomic status, or lifestyle factors such as hazardous al-
cohol use, would aid in targeting health promotion strategies.
There may also be value in reviewing the suitability of current
licensing processes and driver training for these speciﬁc groups.
Although the primary aim of such interventions should be to re-
duce unlicensed driving, a harm minimisation approach aiming
to decrease risky driving amongst unlicensed drivers may also
achieve a reduction in injuries.
The Auckland Car Crash Injury Study was funded by the Health Research
Council of New Zealand.
Connor J, Norton R, Ameratunga S, Jackson R. (2004) The contribution
of alcohol to serious car crash injuries. Epidemiology,Vol. 15, No. 3,
Connor J, Norton R, Ameratunga S, Robinson E, Civil I, Dunn R, Bailey
J, Jackson R. (2002) Driver sleepiness and risk of serious injury to
car occupants: Population based case control study. British Medical
Journal,Vol. 324, pp. 1125–1128.
DeYoung DJ, Peck RC, Helander CJ. (1997) Estimating the exposure
and fatal crash rates of suspended/revoked and unlicensed drivers in
California. Accident Analysis & Prevention,Vol. 29, No. 1, pp. 17–
Federal Ofﬁce of Road Safety (1997a) Proﬁle of unlicensed motorists
in fatal crashes. Federal Ofﬁce of Road Safety, Canberra.
Federal Ofﬁce of Road Safety (1997b) Road behaviour of unlicensed
motorists involved in fatal crashes. Federal Ofﬁce of Road Safety,
Grifﬁn LI, DeLaZerda S. (2000) Unlicensed to kill. AAA Foundation
for Trafﬁc Safety, Washington, DC.
Harrison WA. (1997) An exploratory investigation of the crash involve-
ment of disqualiﬁed drivers and motorcyclists. Journal of Safety Re-
search.,Vol. 28, No. 3, pp. 213–219.
Kim S, Kim K. (2003) Personal, temporal, and spatial characteristics
of seriously injured crash-involved seatbelt non-users in Hawaii. Ac-
cident Analysis and Prevention,Vol. 35, pp. 121–130.
Knox D. (2003) Research into unlicensed driving—Literature review.
Department for Transport, London.
Knox D, Turner B, Silcock D, Beuret K, Metha J. (2003) Research into
unlicensed driving: Final report. Department for Transport, London.
Land Transport Safety Authority. (2000) Stats: Road crash data from
the LTSA. Land Transport Safety Authority, Auckland.
Land Transport Safety Authority. (2003) Disqualiﬁed and unlicensed
drivers and road crashes: What’s the connection? LTSA, Auckland.
Land Transport Safety Authority. (2004) Getting your car driver licence.
Perneger T, Smith GS. (1991) The driver’s role in fatal two-car crashes:
A paired ‘case-control’ study. American Journal of Epidemiology,
Vol. 134, No. 10, pp. 1138–1144.
Retting RA, Ulmer RG, Williams AF. (1999) Prevalence and charac-
teristics of red light running crashes in the United States. Accident
Analysis and Prevention.,Vol. 31, No. 6, pp. 687–694.
Shah BV, Barnwell BG, Bieler GS. (1997) Software for the statistical
analysis of correlated data: SUDAAN User’s Manual, Release 8.0.0.
Research Triangle Institute, Research Triangle Park, NC.
Shibata A, Fukuda K. (1994) Risk factors of fatality in motor vehicle
trafﬁc accidents. Accident Analysis and Prevention,Vol. 26, No. 3,
Statistics New Zealand. (2001) Resident population of the Auckland
Region, March 2001. Statistics New Zealand, Auckland.
Woodward M. (2005) Epidemiology: Study design and data analysis.
Chapman & Hall/CRC Press, Boca Raton.