[Link] This is the Pre-Published Version.

A SIMULATION STUDY OF THE EFFECTS OF ALCOHOL ON DRIVING

PERFORMANCE IN A CHINESE POPULATION

Y. C. LIa, N. N. SZEb, S. C. WONGa, Wei YANa, K. L. TSUIc, F. L. SOc

a
Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong
b
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University,
Hung Hom, Kowloon, Hong Kong
c
Accident and Emergency Department, Tuen Mun Hospital, Tsing Chung Koon Road, Tuen Mun,
Hong Kong

ABSTRACT

Driving under the influence of alcohol (DUIA) is a significant factor contributing to road traffic
crashes, injuries, and fatalities. Although the effects of alcohol on driving performance are widely
acknowledged, studies of the effects of alcohol impairment on driving performance and
particularly on the control system of Chinese adults are rare. This study attempts to evaluate the
effects of alcohol on the driving performance of Chinese adults using a driving simulator. Method:
A double-blind experimental study was conducted to evaluate the effects of alcohol impairment
on the driving performance of 52 Chinese participants using a driving simulator. A series of
simulated driving tests covering two driving modules, including emergency braking (EB) and
following braking (FB), at 50 km/h and 80 km/h were performed. Linear mixed models were
established to evaluate driving performance in terms of braking reaction time (BRT), the standard
deviation of lateral position (SD-LANE), and the standard deviation of speed (SD-SPEED).
Results: Driving performance in terms of BRT and SD-LANE was highly correlated with the level
of alcohol consumption, with a one-unit increase in breath alcohol concentration (BrAC) degrading
BRT and SD-LANE by 0.3% and 0.2%, respectively. Frequent drinkers generally reacted faster in
their BRT than less-frequent drinkers and non-drinkers by 10.2% and 30.6%, respectively.
Moreover, alcohol impairment had varying effects on certain aspects of the human control system,
and automatic action was less likely to be affected than voluntary action from a psychological
viewpoint. Conclusion: The findings should be useful for planning and developing effective
measures to combat drink driving in Chinese communities.

Keywords: Traffic safety; drink driving; driving simulator; driving performance, Chinese
population.

© 2016. This manuscript version is made available under the CC-BY-NC-ND 4.0 license [Link]
1 INTRODUCTION

The rapid rise in both motor transport and alcohol consumption has made drink driving a major
road safety concern responsible for an increasing number of crashes and injuries in China (Li et
al., 2012). It is generally accepted that drink driving significantly increases one’s likelihood of
engaging in risk-taking behavior. Moreover, the general belief is that the drinking habit of drivers
leads to the development of physiological changes that would affect their driving performance.
Numerous studies have reported that alcohol-impaired drivers pose a higher crash risk than those
who have not consumed alcohol, particularly when the victims are killed or seriously injured (KSI)
(Mounce and Pendleton, 1992; Robertson and Drummer, 1994; Tsui et al., 2010; Li et al., 2013a
and 2013b). Despite this, as drinking alcoholic beverages has been traditionally accepted in
Chinese culture as a part of creating friendships and even successful business, getting drunk is
generally not seen as a problem in Chinese communities. Li et al. (2010) revealed that about 35%
of Chinese adults abstained from alcohol consumption (55.6% of males and 15.0% of females).
Of those who drank, 26.3% of males and 7.8% of females reported doing so frequently, i.e., 5-7
days per week (Li et al., 2010). Moreover, young Chinese adults were found to engage in more
risk-taking and risky driving behavior. In particular, 39% of the respondents reported that they had
driven after consuming alcohol, and 28% indicated that they may drive after drinking in the future
(Chan et al., 2010).

Due to rapid economic development, alcohol consumption is unfortunately increasing faster in

China than in other parts of the world (WHO, 2014; Wang et al., 2015). In light of this, it is not
surprising that drink driving has become a main cause of road crashes among the Chinese
population. In 2011, license suspensions and monetary fines were imposed to combat drink-driving
offenses and address the serious problem of drink driving in China. Nevertheless, the reduction in
drink-driving-related injuries was less significant than that of overall road injuries (Zhang et al.,
2014a). Indeed, researchers have concluded that the random breath test (RBT) is the most effective
means of enforcing drink-driving penalties (Erke et al., 2009; Li et al., 2014). Hence, the
Government of the People’s Republic of China introduced the zero tolerance law in January 2013
to coincide with the application of RBTs. As a result, traffic fatalities caused by drink driving
between 2012 and 2013 dropped substantially by 6.81% (from 25.26% to 18.45%) (Wang et al.,
2015). Notwithstanding these efforts, the risks of drink-driving fatalities remain considerable and
of great concern to Chinese communities.

Until recently, Chinese research related to the driving performance of drivers under the influence
of alcohol focused its attention mainly on drink-driving statistics in China (Wang et al., 2015), the
prevalence of drink-driving (Yuan et al., 2013; Suo, 2015), or the risks of alcohol impairment (Li
et al., 2010; Li et al., 2012; Sun et al., 2014; Zhang et al., 2014a) while ignoring the effects of
alcohol on the driving performance of Chinese adults. Indeed, a considerable number of studies
have demonstrated that the effects of alcohol degrade many of the important physiological
responses involved in driving, decrease judgment and inhibition, impair visual functions, and
decrease attention and alertness depending on the amount of alcohol consumed (Schuckit et al.,
1997a and 1997b; Fillmore and Vogel-Sprott, 1999 and 2000; Harrison and Fillmore, 2005). These
studies have clearly demonstrated that alcohol affects driving performance, which is considered
essential to the safe operation of a vehicle. Furthermore, few studies have explored the effects of
alcohol impairment on the inhibitory control of different human control systems, either

2
automatically or voluntarily from a psychological viewpoint (Abroms et al., 2006; Cho et al.,
2013). In recent years, driving simulators have commonly been used to monitor the effects of
alcohol on driving performance in road safety research (Fillmore et al., 2008; Lenne et al., 2010;
Wester et al., 2010; Ronen et al., 2010; Harrison and Fillmore, 2011). These simulation studies
have generally concluded that variability in speed and lateral position increase as alcohol
concentration increases. Simulation studies have also demonstrated that alcohol impairs driving
performance by increasing the reaction time of drivers. For example, a simulation study conducted
by Christoforou et al. (2013) demonstrated that a 10% increase in alcohol concentration could
result in a 2% increase in reaction time. Except for the simulation study conducted by Liu and Ho
(2010), who investigated the relationship between alcohol consumption and the performance of
drivers in Taiwan, all of these studies have focused on Caucasians in Western countries. The scale
of Liu and Ho’s (2010) study was relatively small, including only eight licensed drivers as
participants. As the physiological responses of Chinese people to alcohol are known to differ from
those of Caucasians (Reed et al., 1976), the findings from previous research may not be applicable
to a Chinese population.

The current simulation study attempted to evaluate the effects of alcohol on the actual driving
performance of Chinese adults using a driving simulator. This paper is organized as follows.
Section 2 details the study’s materials and methods. Section 3 explains the driving performance
data collection process and the formulation of the random-intercept model. Section 4 presents the
findings on the differing alcohol-impaired driving performance of Chines adults. Finally, Section
5 concludes the paper by discussing the effects of alcohol on driving performance.

2 MATERIALS AND METHODS

2.1 Participants

To capture the full spectrum of driver types and strive for a balanced distribution of participants
in terms of gender and age, 52 Chinese drivers (34 males and 18 females) who had held a valid,
full driving license for at least 1 year were recruited through the network of driver associations.
The mean age of the participants was 38.2 years (ranging from 21 to 61 years). Every participant
was invited to attend two or three experimental sessions, separated by two days or more. Before
the experiments began, the participants were required to pass both a health assessment and an
Alcohol Use Disorders Identification Test (Saunders et al., 1993), both of which were conducted
by a medical doctor. Those who reported previous alcohol or substance abuse or any psychiatric
disorder were excluded. Women who were pregnant or breastfeeding (determined by self-report)
were not permitted to participate. All of the participants provided informed consent before
participating and received HK$50 per hour. Table 1 presents a summary of the participants’
statistics.

[Insert Table 1 here]

2.2 Apparatus and materials

3
Simulated driving tests: Driving performance was measured using a computerized simulated
driving test in an enclosed area. A desktop-based XP-300 Driving Simulator (XPI Simulation Ltd.,
UK) was used in all of the tests. This simulator provided a fully interactive simulated driving
environment with three 19-inch LCD monitors, a Logitech G27 steering wheel, and a pedal kit.
Figure 1 illustrates the hardware setup. The available driving modules included emergency braking
(EB), following braking (FB), the two-second rule, hazard perception, and motorway driving. The
simulated driving test recorded the vehicle position, speed, acceleration, time to contact pedals
(acceleration and braking), and angle of the steering wheel in a 30-Hz sampling frame.

[Insert Figure 1 here]

Breathalyzer: To measure the BrAC, an Alcotest 9510 evidential breathalyzer (Drager Safety AG
& Co., Germany), was used. The Hong Kong Police Force currently uses this breathalyzer in
evidential alcohol tests due to its high specificity for ethyl alcohol. It quantifies the alcohol
concentration based on two separated breath samples: one measured via infrared red (IR) sensor,
the other measured by fuel cell technology (dual sensor technology). The maximum permissible
deviation between the readings of two sensors is 5%. In this study, the mean reading of the two
sensors was calculated for subsequent analysis.

2.3 Experimental session design

The participants were told that the purpose of the experiment was to study the impairing effects of
alcohol on driving performance. They were required to attend two or three experimental sessions
separated by two days or more, and given alcohol doses in random amounts (ranging from zero to
six standard drinks1). The maximum alcohol intake was limited to six standard drinks, which was
considered an acceptable and ethical level of alcohol intake (Lenne et al., 2010).

2.3.1 General procedures

Figure 2 illustrates the experimental procedures. Before the experiment, all of the participants were
instructed to abstain from food for four hours and from alcohol and sedatives for twenty-four hours.
A registered nurse conducted a clinical assessment of the physiological responses of each
participant at the start of the experiment. A standard light meal was provided to simulate a normal
social drinking occasion and allow for the effect of food on alcohol metabolism. The BrAC was
then measured to ensure that every participant was free of alcohol. The participants then completed
a 10-minute practice plus a 20-minute simulated driving test before alcohol intake (denoted as
“DT0”). Approximately 30 minutes after the meal (time 0), each participant was asked to consume
500 ml of an alcoholic drink (orange juice mixed with vodka of 40% alcohol by volume) in 15-20
minutes. To decrease the risk of possible bias, a double-blind procedure was used in which neither
the participants nor those carrying out the experiments knew the doses of alcohol taken. The BrAC
was measured at 10-minute intervals during the initial rapid absorption phase to ensure that the
peak alcohol concentration was measured. The “post-beverage” simulated driving test (denoted as
“DT1”) began about 60 minutes after the start of the alcohol intake (time 60) to coincide with the
ascending period of the peak alcohol concentration.

1
The dose of alcohol was either zero, two, four, or six standard drinks. One standard drink should contain 10 g of pure
alcohol and is equivalent to 100 ml of wine with 12% alcohol by volume.

4
Although similar experimental procedures were applied during the elimination phase, BrAC
measurements were taken at 60-minute intervals, i.e., at 120, 180, 240, and 300 minutes from the
start of the alcohol intake (time 120, 180, 240, and 300). Considering that the alcohol-impaired
effects would decrease with time after alcohol intake, two further post-beverage simulated driving
tests (denoted as “DT2” and “DT3”) were conducted at about 120 and 240 minutes after the start
of the alcohol intake (time 120 and 240), respectively.

[Insert Figure 2 here]

The participants then rested in the laboratory waiting area until the end of the session. The
experiment ended four hours after the alcohol intake or until the BrAC fell to a level of 10 μg/100
ml or below. The protocol of this study was approved by the Institutional Review Board of the
University of Hong Kong, Hospital Authority Hong Kong West Cluster.

2.3.2 Driving simulated tests

Practice test: Each session began with a 10-minute practice so that the participants could
familiarize themselves with the steering control, accelerator, and brake of the driving simulator.
The participants were trained using a free driving module in an urban area, where they could
practice speed control, turning, and other general driving techniques without encountering hazards
or vehicles on the roads. This was followed by a practical session for each driving module to be
taken during the simulated driving test.

Driving simulated test: One baseline (DT0) plus three post-beverage (DT1, DT2, and DT3)
driving simulated tests were conducted after the alcohol intake. Each test consisted of two modules,
including EB and FB, to monitor the impairment of the participants’ driving performance. In the
EB modules, the participants were required to maintain a constant speed of 50 km/h (or 80 km/h)
on a straight section of road and then engage the emergency brake in response to a virtual
instruction “STOP” given at random intervals (see Figure 3a). In the FB module, the participants
were instructed to maintain their speed and follow at the correct distance behind the car in front at
50 km/h (or 80 km/h) and to brake when the car they were following began to brake (as indicated
by its brake light) (see Figure 3b).

[Insert Figures 3a and 3b here]

3 STATISTICAL ANALYSIS

3.1 Data collection

In this simulation study, the level of driving performance was the dependent variable, measured
by the braking reaction time (BRT), the standard deviation of lateral position (SD-LANE), and the
standard deviation of speed (SD-SPEED). BRT is the time (in seconds) from the appearance of the
braking event, TEvent, to the brake onset, TBrake. Figure 4 and Equation (1) illustrate the definition
of BRT.

5
[Insert Figure 4 here]

BRT = 𝑇𝐵𝑟𝑎𝑘𝑒 − 𝑇𝐸𝑣𝑒𝑛𝑡 (1)

Measurements of the lateral position and vehicle speed within the period between TLimit and TLimit+10
were extracted from the dataset to calculate the variation in lateral position and speed (see Figure
5). For a 30-Hz sampling frame, 300 measurements were recorded in a 10-second period. Hence,
the SD-LANE and SD-SPEED of each simulated driving test are defined as

∑𝑛 ̅ 2
𝑖=1(𝑋𝑖 −𝑋)
𝑆𝐷-𝐿𝐴𝑁𝐸 = √ , for n = 1, 2, …, 300 (2)
𝑛−1

∑𝑛
𝑖=1(𝑉𝑖 −𝑉𝐿𝑖𝑚𝑖𝑡 )
2
SD-𝑆𝑃𝐸𝐸𝐷 = √ 𝑛−1
, for n = 1, 2, …, 300 (3)

where 𝑋𝑖 and 𝑉𝑖 are the ith measurements of the lateral position and speed, respectively. 𝑋̅ is the
mean of the lateral position and VLimit is the target speed limit of the testing module (either 50 or
80 km/h). The independent variables are the BrAC value; driver drinking habit (non-drinker
[never], less-frequent drinker [less than twice per month], or frequent drinker [twice or more per
month]); driver age group (young driver [18-24], adult driver [25-54], or older driver [55 or
above]); driver gender (female or male); years spent holding a driving license (less than 3 years,
3-10 years, or more than 10 years); occupation as a driver (none, part time, or full time); simulated
driving test (DT0, DT1, DT2, or DT3); speed limit (50 or 80 km/h); and driving module (EB or
FB) of the simulated driving test.

As previously mentioned, each participant was required to attend two or three experimental
sessions. Thirty-one participants attended three sessions, seventeen participants attended two
sessions, and the remaining four participants attended only one session. One hundred and thirty-
one simulated driving experiments (ninety for males and forty-one for females) were conducted.
Invalid measurements due to reported physical illness (e.g., vomiting and headaches) were
excluded, leaving 1,933 measurements taken. Table 2 summarizes the characteristics of these
measurements.

[Insert Table 2 here]

3.2 Model formulation

To account for the cross-sectional heterogeneity effects of individuals, panel data analysis
(Washington et al., 2003) was conducted to identify the possible factors affecting the driving

6
performance of Chinese adults. A linear mixed model was adopted to perform statistical analyses
using SPSS 20.0 statistical software to facilitate the model formulation.

In the proposed model, the heterogeneity effects were assumed to be constant for given fixed
components and were absorbed by the intercept term as a mean to account for individual
heterogeneity. A one-way random-intercept model with the dependent variable 𝑦𝑖𝑡 is specified as

𝑦𝑖𝑡 = 𝛽0 + 𝑿𝒊𝒕 𝜷 + 𝜇𝑖 + 𝜐𝑖𝑡 (4)

where i refers to the cross-sectional unit, i.e., the participant in the experiment (i = 1, 2, 3, …, 52),
and t refers to the time for the simulated test (t = 1, 2, and/ or 3). 𝜇𝑖 is the unobserved cross-specific
effect and 𝜐𝑖𝑡 refers to random disturbances. 𝑿𝒊𝒕 is the vector of the possible contributory fixed
variables, including the BrAC value, driver drinking habit, driver age, driver gender, number of
years spent holding a driving license, occupation as a driver, simulated driving test, speed limit,
and driving module. 𝜷 is the vector of the corresponding coefficient, estimated using the maximum
likelihood approach. For the random component, the variances of the random parameters
𝑉𝐴𝑅(𝑢𝑖𝑡 ) = 𝜎𝜇 2 + 𝜎𝑣 2 are estimated for the purpose of model interpretation.

To assess the goodness-of-fit of the proposed regression model, a likelihood ratio statistic can be
calculated with

𝐿𝑅 = −2(𝐿𝐿(𝛽𝑅 ) − 𝐿𝐿(𝛽𝑈 )) (5)

where 𝐿𝐿(𝛽𝑅 ) is the restricted log likelihood function and 𝐿𝐿(𝛽𝑈 ) is the unrestricted log likelihood
function. Under the null hypothesis that there is no association between the dependent and
independent variables, the LR-statistic is 𝜒 2 distributed with the degrees of freedom equal to the
difference between the numbers of parameters in the restricted and unrestricted models. Thus, a
significant LR-statistic indicates a good fit for the proposed model.

3.3 Levels of alcohol consumption

To determine the influence of a one-unit increase in BrAC (in μg/100 ml) on driving performance,
the odds ratio (OR) of BrAC is calculated as

𝑂𝑅 = exp⁡(𝛽𝑗 ) (6)

with 95% confidence intervals (CI) of ((𝛽𝑗 − 1.96𝑠𝛽𝑗 ) , (𝛽𝑗 + 1.96𝑠𝛽𝑗 )) , where 𝑠𝛽𝑗 is the
standard error of the coefficient 𝛽.

4 RESULTS AND INTERPRETATION

4.1 Braking reaction time (BRT)

7
Measurement records were used to establish the regression model when evaluating alcohol-
impaired driving performance in an unpredictable emergency situation. Table 3 presents the
prediction model results for driving performance in terms of BRT. Based on these results, BrAC
(coefficient = 0.0034) contributed to an increase in BRT, and a frequent driver drinking habit (non-
drinker: 0.267; less-frequent drinker: 0.097; frequent drinker: control), years spent holding a
driving license (3-10 years: 0.146; more than 10 years: control), and full-time occupation as a
driver (part-time: 0.188; full-time: control) contributed to a reduction in BRT during the simulated
driving test at a 5% level of significance. Furthermore, the type of driving module was a
contributing factor to driving performance in terms of BRT. Participants were likely to react faster
in the EB module than in the FB module (EB: -0.116; FB: control) at a 5% significance level. The
proposed prediction model generally fit well with the data (χ2 = 288.58).

[Insert Table 3 here]

4.2 Standard deviation of speed (SD-SPEED)

Table 3 also illustrates the model results for driving performance in terms of SD-SPEED. Alcohol-
related factors had no significant influence. In contrast, the demographic characteristics of the
individual participants including old (young driver: -1.220; older driver: control) and female
(female: 0.700; male: control) drivers contributed to an increase in speed variation. Moreover,
speed limit was one of the factors contributing to the driving speed variation, where SD-SPEED
decreased as the driving speed increased (50 km/h: 0.830; 80 km/h: control). Similar to the case
of BRT, participants were likely to perform better in the EB module than in the FB module (EB: -
0.975; FB: control), as shown in the SD-SPEED prediction model at a 5% significance level. The
proposed prediction model generally fit well with the data (χ2 = 371.01).

4.3 Standard deviation of lateral position (SD-LANE)

BrAC (coefficient = 0.0024) and not driver drinking habit was the major factor contributing to an
increase of SD-LANE at a 5% significance level. In addition, older drivers (young drivers: -0.142;
adult drivers: -0.122; older drivers: control) contributed to an increase in SD-LANE. However,
full-time drivers (part time: 0.169; full time: control) and a higher speed limit (50 km/h: -0.027;
80 km/h: control) contributed to a reduction in SD-LANE during the simulated driving test, both
at a 5% level of significance. No correlation was found between SD-LANE, driver gender, and
years spent holding a driving license. Unlike BRT and SD-SPEED, different driving modules were
not likely to have an influence on SD-LANE. The proposed prediction model fit well with the
experimental data (χ2 = 14.741).

To conclude, alcohol-impaired effects on BRT and SD-LANE, but not on SD-SPEED, were
observed during the simulated driving tests. Furthermore, the time the simulated driving tests took
had no significant effect on the driving performance of the participants in any of the models.

4.4 Automatic vs. voluntary action

In this simulation study, two simulated driving modules, including EB and FB, were used to
measure the alcohol-impaired driving performance of the participants. The EB module involved a

8
type of automatic control that allowed the participants to shift their attention automatically in
response to a sudden “STOP” signal. In contrast, the FB module involved voluntary control, as
from a psychological viewpoint the participants shifted their attention voluntarily. By comparing
the participants’ driving performance in terms of BRT between the EB and FB modules, it was
possible to investigate any difference in alcohol-impaired effects between the automatic and
voluntary control groups.

As shown in Table 3, alcohol-impaired driving performance in terms of BRT was highly correlated
with the simulated driving modules (EB: -0.116; FB: control) at a 5% significance level, meaning
that both the automatic and voluntary control groups were likely to be affected by alcohol.
However, the participants were likely to react faster in the EB module than in the FB module by
11% (OR = 0.890; 0.95CI = 0.878, 0.903). The effects of alcohol impairment on aspects of the
control system therefore varied, with the automatic control group less likely to be affected than the
voluntary control group.

5 DISCUSSION

Over the decades, many driving simulation studies have been conducted to examine alcohol-
impaired driving performance in Western countries (see Table 4). Researchers have generally
agreed that after alcohol consumption the human brain requires a longer time to perceive visual
stimuli and process information and the transmission of signals to the muscles is delayed
(McMillen and Well-Parker, 1987; Fillmore and Vogel-Sprott, 2000). Alcohol concentration can
thus affect various aspects of physiological driving performance, mainly in terms of driving
reaction time, lane, and speed variations.

[Insert Table 4 here]

For example, in an assessment made by Fillmore et al. (2008), the deviation of lateral position
increased by 0.6% for a one-unit (in μg/100 ml) increase in BrAC.2 Weafer et al. (2008) similarly
found that a one-unit increase in BrAC could increase the deviation of lateral position by 0.3%.
Christoforou et al. (2013) evaluated the relationship between the level of BrAC and driving
reaction time and found that a 10% increase in BrAC caused a 2% increase in reaction time. In
terms of speed variation, numerous researchers have suggested that alcohol consumption should
have no significant influence on driving speed (Weafer et al., 2008). However, some have argued
that alcohol can affect driving speed and hence the deviation of speed while driving. For example,
Lenne et al. (2010) compared the driving performance of 47 drivers under the influence of low and
high alcohol doses (19 and 28 μg/100 ml, respectively). They found that the standard deviation of
speed increased with low but not high alcohol dosages. In terms of the Chinese population, Liu
and Ho (2010) demonstrated that the variance in longitudinal speed and reaction time increased
with alcohol consumption. Nevertheless, the scale of their research was relatively small and may
be inadequate to reflect the alcohol-impaired drink-driving performance of the Chinese population.
The driving simulation experiment conducted in the current study provided more information in
this area. Indeed, consistent with many previous studies conducted in Western countries, this
simulation study found that driving performance in terms of BRT (which could be related to the

2
The value of BAC (in g/kg) given in the original research article was converted to BrAC (μg/100 ml) in a blood-to-
breath ratio of 2,300:1 in this study.

9
reaction time of drivers) and SD-LANE (i.e., the standard deviation of lane position) were highly
correlated with the level of alcohol consumption. A one-unit increase in BrAC increased BRT by
0.3% (OR = 1.003, 0.95CI = [1.002, 1.004]) and SD-LANE by 0.2% (OR = 1.002, 0.95CI = [1.001,
1.003]). In our developed model for driving performance in terms of SD-SPEED (which could be
related to speed variations), alcohol consumption had no apparent significant effect on speed
variation.

Over the years, researchers have argued that separate neural systems exist to control the automatic
and voluntary actions (Corbetta and Shulman, 2002; Gazzaniga et al., 2002). Although numerous
studies have been conducted to investigate the impairment effects of alcohol on attention (Fillmore
and Vogel-Sprott, 1999 and 2000; Harrison and Fillmore, 2005), its effects on the two control
systems are not well understood. In light of this, Abroms et al. (2006) conducted research into the
effects of alcohol impairment on the inhibitory control of the different control systems. The
findings demonstrated that vehicle control depends more on voluntary action than automatic action.
More recently, research by Cho et al. (2013) also examined the voluntary action of drivers, and
whether it is impaired and more likely to cause crashes and injuries if so. However, the alcohol-
impairment effects on automatic actions were not mentioned in their study, and we have found few
studies that review alcohol-impaired driving performance due to the difference between voluntary
and automatic actions. The results of this simulation study demonstrated that the degree of alcohol
impairment of aspects of the human control system varies, and automatic action is less likely to be
affected than voluntary action. It is commonly believed that slow response to road hazards is the
main cause of alcohol-related road crashes. In fact, the influences of alcohol on basic vehicle
control skills, such as acceleration, deceleration, and changing lanes, may be of a greater safety
concern to the driver himself.

The results of this study also support the general belief that frequent drinking leads to the
development of physiological changes that increase one’s alcohol tolerance, minimizing the effects
of alcohol consumption on the driving performance of frequent drinkers (Schechtman et al. 1999).
Table 3 indicates that frequent drinkers generally reacted faster than less-frequent drinkers and
non-drinkers in their BRT. Comparing the alcohol-influenced driving performance of frequent
drinkers and other drivers, the BRT of less-frequent drinkers was slower than that of frequent
drinkers by 10.2% (OR = 1.102, 0.95CI = [1.024, 1.182]). Alcohol-influenced driving performance
was even worse for non-drinkers, whose BRT increased by 30.6% (OR = 1.306, 0.95CI = [1.133,
1.505]). However, the drinking habits of drivers were unlikely to have different effects on vehicle
control, including driving performance in terms of SD-SPEED and SD-LANE. Although frequent
drinkers seem to exhibit better driving performance than less-frequent drinkers and non-drinkers,
this should not be an excuse to drink alcohol, as frequent or binge drinking has proved to have
adverse effects on health and risk-taking behavior (Schechtman et al., 1999; Clark et al., 2001,
LaBrie et al., 2011). Although driving performance is significantly degraded for non-drinkers,
these drivers are not likely to engage in drink driving. More efforts to combat drink driving
effectively should focus on drinking drivers and especially less-frequent drinkers, who present a
higher accident risk than frequent drinkers and other road users. For example, ignition interlock
devices should be implemented. These devices have been widely promoted in many Western
countries but are considered as a new and emerging technology for the Chinese population.

10
The findings of this study may be useful for police and policymakers to estimate the influence of
alcohol on different aspects of driving performance. For example, the Hong Kong SAR
Government imposed a three-tier sliding scale penalty system for drink-driving offenses that
applied different penalties to BrAC levels exceeding 22, 35, and 66 μg/100 ml, respectively.
Assuming a normal BRT of 2.5 seconds and based on the current results of a 0.3% increase in
braking reaction time per unit increase in BrAC, the stopping distances increase by 4, 6, and 11 m
for BrAC of 22, 35, and 66 μg/100 ml of breath, respectively, at a vehicular speed of 80 km/h. The
general practice of the two-second rule may be inadequate to provide a safe following distance for
drivers under the influence of alcohol and thereby have no effect on the frequency of rear-end
collisions. In this study, variations in lateral position were correlated with alcohol concentration
levels. Furthermore, the presence of a curve in the association between lateral stability and alcohol
level indicated the possibility of interaction, and impaired drivers tended to travel on one particular
side of the lane (Zhang et al., 2014b). Therefore, lateral stability may be an effective performance
indicator for the detection of drink-driving. This finding has implications for the planning and
implementation of enforcement against drink-driving.

6 CONCLUSION

A series of driving simulated tests was conducted to evaluate the driving performance of Chinese
adults under the influence of alcohol. Each simulated test consisted of two driving modules,
including EB and FB, to monitor the impairment of the driving performance of 52 participants.
Their driving performance was measured in terms of their BRT, SD-LANE, and SD-SPEED, and
linear mixed models were established to identify the contributing alcohol-related and other
confounding factors. Similar to the findings of many studies conducted in Western countries,
driving performance in terms of the BRT and SD-LANE of Chinese adults was highly correlated
with their level of alcohol consumption. Consistent with previous studies, the effects of alcohol
impairment on certain aspects of the human attentional control system varied, and automatic action
was less likely to be affected than voluntary action from a psychological viewpoint. It is commonly
believed that slow response to road hazards is the main cause of alcohol-related road crashes. In
fact, this study demonstrated that the influence of alcohol on basic vehicle control skills, such as
acceleration, deceleration, and lane changing, may be a greater safety concern for drivers.

ACKNOWLEDGEMENTS

The research described here was supported by a Research Postgraduate Studentship, and grants
from the University Research Committee of the University of Hong Kong (201309176024) and
the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No.
718811).

REFERENCES

11
Abroms B.D., Gottlob L.R., Fillmore M.T. (2006) Alcohol effects on inhibitory control of attention:
Distinguishing between intentional and automatic mechanisms. Psychopharmacology, 188:
324-334.
Burian S.E., Liguori A., Robinson J.H. (2002) Effects of alcohol on risk-taking during simulated
driving. Human Psychopharmacology, 17: 141-150.
Chan D.C.N., Wu A.M.S., Hung E.P.W. (2010) Invulnerability and the intention to drink and drive:
An application of the theory of planned behavior. Accident Analysis and Prevention, 42: 1549-
1555.
Charlton S.G., Starkey N.J. (2015) Driving while drinking: Performance impairments resulting
from social drinking. Accident Analysis and Prevention, 74: 210-217.
Cho S.H., Hong E.J., Kak H.B., Moon T.Y., Cho B.J. (2013) The effects of alcohol on the H-reflex
in adults. Journal of Physical Therapy Science, 25(3): 221-224.
Christoforou Z., Karlaftis M.G., Yannis G. (2013) Reaction times of young alcohol-impaired
drivers. Accident Analysis and Prevention, 61: 54-62.
Clark D.B., Lynch K.G., Donovan J.E., Block G.D. (2001) Health problems in adolescents with
alcohol use disorders: Self-report, liver injury, and physical examination findings and
correlates. Alcoholism: Clinical and Experimental Research, 25: 1350-1359.
Corbetta M. and Shulman G.L. (2002) Control of goal-directed and stimulus-driven attention in
the brain. Nature Reviews Neuroscience, 3: 201-215.
Erke A., Goldenbeld C., Vaa T. (2009) The effects of drink-driving checkpoints on crashes–A
meta-analysis. Accident Analysis and Prevention, 41(5): 914-923.
Fillmore M.T., Vogel-Sprott M. (1999) An alcohol model of inhibitory control and its treatment
in humans. Experimental and Clinical Psychopharmacology, 7(1): 49-55.
Fillmore M.T., Vogel-Sprott M. (2000) Response inhibition under alcohol: Effects of cognitive
and motivational conflict. Journal of Studies on Alcohol, 61: 239-246.
Fillmore M.T., Blackburn J.S., Harrison E.L.R. (2008) Acute disinhibiting effects of alcohol as a
factor in risky driving behavior. Drug and Alcohol Dependence, 95(1-2): 97-106.
Gazzaniga M., Ivry R., Mangun G. (2002) Cognitive Neuroscience: The Biology of the Mind. (2nd
ed.) New York: W.W. Norton & Company, Inc.: 247-252.
Harrison E.L.R., Fillmore M.T. (2005) Are bad drivers more impaired by alcohol? Sober driving
precision predicts impairment from alcohol in a simulated driving task. Accident Analysis and
Prevention, 37: 883-889.
Harrison E.L.R., Fillmore M.T. (2011) Alcohol and distraction interact to impair driving
performance. Drug and Alcohol Dependence, 117: 31-37.
Helland A., Jenssen G.D., Lervag L.E., Westin A.A., Moen T., Sakshaug K., Lydersen S.,
Morland J., Slordal L. (2013) Comparison of driving simulator performance with real driving
after alcohol intake: A randomised, single blind, placebo-controlled, cross-over trial.
Accident Analysis and Prevention, 53: 9-16.
Hong Kong Police Force (2012) RSC and TBHQ remind the public not to drive after drinking
during festive season. Hong Kong Police Force of HKSAR (cited on 24 June, 2015).
Available at [Link]
LaBrie J.W., Kenney S.R., Mirza T., Lac A. (2011) Identifying factors that increase the likelihood
of driving after drinking among college students. Accident Analysis and Prevention, 43: 1371-
1377.

12
Lenne M.G., Dietze P.M., Triggs T.J., Walmsley S., Murphy B., Redman J.R. (2010) The effects
of cannabis and alcohol on simulated arterial driving: Influences of driving experience and task
demand. Accident Analysis and Prevention, 42: 859-866.
Leung S., Starmer G. (2005) Gap acceptance and risk-taking by young and mature drivers, both
sober and alcohol-intoxicated, in a simulated driving task. Accident Analysis and Prevention,
37: 1056-1065.
Li Y., Jiang Y., Zhang M., Yin P., Wu F., Zhao W. (2010) Drinking behaviour among men and
women in China: The 2007 China Chronic Disease and Risk Factor Surveillance. Addiction,
106(11): 1946-1956.
Li Y., Xie D., Nie G., Zhang J. (2012) The drink driving situation in China. Traffic Injury
Prevention, 13(2): 101-108.
Li Y.C., Sze N.N., Wong S.C., Tsui K.L., So F.L. (2013a) Effects of drink driving on crash risk
based on random breath test data. Hong Kong Journal of Emergency Medicine, 20(3): 146-154.
Li Y.C., Sze N.N., Wong S.C. (2013b) Spatial-temporal analysis of drink-driving patterns in Hong
Kong. Accident Analysis and Prevention, 59: 415-424.
Li Y.C., Sze N.N., Wong S.C. (2014) Effective measures for combating drink-driving offences:
An attitudinal model in Hong Kong. Transportmetrica A: Transport Science, 10: 722-739.
Liu Y.C. and Ho C.H. (2010) Effects of different blood alcohol concentrations and post-alcohol
impairment on driving behavior and task performance. Traffic Injury Prevention, 11(4): 334-
341.
Lu Z-P., Engs R.C., Hanson D.J. (1997) The drinking behaviors of a sample of university students
in Nanning, Guangxi Province, People’s Republic of China. Substance Use & Misuse, 32(4):
495-506.
McMillen D.L., Well-Parker E. (1987) The effect of alcohol consumption on risk-taking while
driving. Addictive Behavior, 12(3): 241-247.
Mounce N.H., Pendleton O.J. (1992) The relationship between blood-alcohol concentration and
crash responsibility for fatally injured drivers. Accident Analysis and Prevention, 24(2): 201-
210.
Reed T.E., Kalant H., Gibbins R.J., Kapur B.M., Rankin J.G. (1979) Alcohol and acetaldehyde
metabolism in Caucasians, Chinese and Amerinds. Canadian Medical Association Journal,
115(9): 851-855.
Robertson M.D., Drummer O.H. (1994) Responsibility analysis – A methodology to study the
effects of drugs in driving. Accident Analysis and Prevention, 26(2): 243-247.
Ronen A., Chassidim H.S., Gershon P., Parmet Y., Rabinovich A., Bar-Hamburger R., Cassuto Y.,
Shinar D. (2010) The effect of alcohol, THC and their combination on perceived effects,
willingness to drive and performance of driving and non-driving tasks. Accident Analysis and
Prevention, 42: 1855-1865.
Saunders J.B., Aasland O.G., Babor T.F., de la Fuente J.R., Grant M. (1993) Development of the
alcohol use disorders identification test (AUDIT): WHO collaborative project on early
detection of persons with harmful alcohol consumption--II. Addiction, 88(6): 791-804.
Schechtman E., Shinar D., Compton R.C. (1999) The relationship between drinking habits and
safe driving behaviors. Transportation Research Part F, 2(1): 15-26.
Schuckit M.A., Smith T.L., Tipp J.E. (1997a) The self-rating of the effects of alcohol (SRE) form
as a retrospective measure of the risk for alcoholism. Addiction, 92(8): 979-988.

13
Schuckit M.A., Tipp J.E., Smith T.L., Wiesbeck G.A., Kalmijn J. (1997b) The relationship
between self-rating of the effects of alcohol and alcohol challenge results in ninety-eight young
men. Journal of Studies on Alcohol, 58(4): 397-404.
Sun Y., Huang Z., Zhao Z., Jiang Y., Ye., Y., Yu T., Rao Y. (2013) Characteristics of 1226 alcohol-
positive drivers involved in nonfatal traffic crashes in Shanghai, China. Traffic Injury
Prevention, 15(6): 532-536.
Suo Q. (2015) Investigation on deterrence effect of legal punishment measures on driving after
drinking in Chongqing, China. Traffic Injury, 16(6): 540-544.
Tsui K.L., Sze N.N., So F.L., Wong S.C. (2010) Association between drink driving and severity
of crash injuries to road users. Hong Kong Journal of Emergency Medicine, 17(1): 4-39.
Wang Z., Zhang Y., Zhou P., Shi J., Wang Y., Liu R., Jiang C. (2015) The underestimated drink
driving situation and the effects of zero tolerance laws in China. Traffic Injury Prevention,
16(5): 429-434.
Washington S.P., Karlaftis M.G., Mannering F.K. (2003) Chapter 6 - Panel Data Analysis.
Statistical and Econometric Methods for Transportation Data Analysis. Boca Raton, FL:
Chapman & Hall/CRC.
Weafer J., Camarillo D., Fillmore M.T., Milich R., Marczinski C.A. (2008) Simulated driving
performance of adults with ADHD: Comparisons with alcohol intoxication. Experimental
and Clinical Psychopharmacology, 16(3): 251-263.
Wester A.E., Verster J.C., Volkerts E.R., Böcker K.B.E., Kenemans J.L. (2010) Effects of alcohol
on attention orienting and dual-task performance during simulated driving: An event-related
potential study. Journal of Psychopharmacology, 24(9): 1333-1348.
WHO (2015) Global status report on alcohol and health 2014. Geneva, World Health
Organization (cited on 12 October, 2015). Available at
[Link]
Yuan A., Li Y., Zhang J. (2013) The result of a baseline survey on drink driving in Nanning and
Liuzhou of Guangxi Province, China. Traffic Injury Prevention, 14(3): 230-236.
Zhang G., Yau K.K.W., Gong X. (2014a) Traffic violations in Guangdong Province of China:
Speeding and drunk driving. Accident Analysis Prevention, 64: 30-40.
Zhang X., Zhao X., Du H., Ma J., Rong J. (2014b) Effect of different breath alcohol concentrations
on driving performance in horizontal curves. Accident Analysis Prevention, 72: 401-410.

14
Table 1 Summary statistics for the participants (Sample size = 52)
Factor Attribute Frequency Percent (%)
Age Young driver (18-24) 8 15.4
Adult driver (25-54) 40 76.9
Old driver (55 or above) 4 7.7
Gender Female 18 34.6
Male 34 65.4
Years spent holding a driving Less than 3 years 9 17.3
license
3-10 years 13 25.0
More than 10 years 30 57.7
Occupation as driver Full time 3 5.8
Part time 8 15.4
No 41 78.8
Drinking habit Non-drinker (never) 3 5.8
Less-frequent drinker (less 30 57.7
than twice per month)
Frequent drinker (twice or 19 36.5
more per month)

Table 2 Summary statistics for the simulated driving test (Sample size = 1,933)
Factor Mean (SD)
(Range)
Overall Simulated driving test Speed limit Module

DT0 DT1 DT2 DT3 50 km/h 80 km/h EB FB

Observation 1,933 480 487 485 481 947 986 964 969
BRT 0.75 0.70 0.79 0.77 0.74 0.76 0.75 0.69 0.81
(0.48, 2.1) (0.21) (0.16) (0.25) (0.22) (0.19) (0.29) (0.21) (0.14) (0.25)
SD-SPEED 2.41 2.45 2.41 2.41 2.38 2.83 2.01 1.93 2.89
(0.08, 17.5) (1.75) (1.70) (1.83) (1.63) (1.83) (1.63) (1.77) (1.38) (1.94)
SD-LANE 0.19 0.17 0.21 0.20 0.18 0.18 0.21 0.19 0.19
(0.003, 2.72) (0.20) (0.16) (0.21) (0.24) (0.18) (0.18) (0.22) (0.19) (0.21)
BrAC 11.30 1.50 20.70 14.57 8.27 11.40 11.30 11.30 11.30
(0, 68.75) (10.64) (0.25) (11.23) (9.19) (6.08) (10.68) (10.64) (10.65) (10.63)
Notes: BRT: braking reaction time;
SD-SPEED: standard derivation of speed;
SD-LANE: standard derivation of lane position;
BrAC: breath alcohol concentration;
SD: standard deviation.

15
Table 3 Parameter estimates of the linear mixed model for the different aspects of driving performance
1. BRT SD-SPEED SD-LANE
2. Coefficient (t-statistic) Coefficient (t-statistic) Coefficient (t-statistic)
Number of observations 1,933 1,933 1,933
3.
Fixed variables
Constant
1. 0.560 (8.414)** 3.374 (6.428)** 0.107 (1.605)
(a) Alcohol-related factors
BrAC
2. 0.0034 (6.535)** 0.0064 (1.436) 0.0024 (4.595)**
Drinking habit
3.
4.- Non-drinker 0.267 (3.792)** 0.156 (0.282) 0.122 (1.730)
5.- Less-frequent drinker 0.097 (2.777)** -0.463 (-1.684) 0.009 (0.261)
6.- Frequent drinker (Control) (Control) (Control)

(b) Confounding factors

Age
7.
8.- Young driver -0.093 (-1.322) -1.220 (-2.203)* -0.151 (-2.144)*
9.- Adult driver -0.080 (-1.952) -0.632 (-1.959) -0.143 (-3.478)**
10.- Older driver (Control) (Control) (Control)
Gender
11.
12. - Female 0.056 (1.700) 0.700 (2.673)* 0.052 (1.564)
13. - Male (Control) (Control) (Control)
Years spent holding a driving license
14.
15.- Less than 3 years 0.121 (1.909) -0.039 (-0.078) 0.057 (0.890)
16.- 3-10 years 0.146 (3.285)** 0.498 (1.424) 0.045 (1.008)
17.- More than 10 years (Control) (Control) (Control)
Occupation as a driver
18.
19. - No 0..120 (1.710) -0.550 (-0.994) 0.105 (4.481)
20. - Part time 0.188 (2.496)* -0.657 (-1.106) 0.169 (2.231)*
21. - Full time (Control) (Control) (Control)
Simulated driving test
22.
23. - DT0 -0.021 (-1.948) 0.110 (1.170) 0.008 (0.727)
24. - DT1 0.007 (0.571) -0.064 (-0.614) 0.004 (0.292)

16
 25. - DT2 -0.003 (-0.272) -0.034 (-0.363) 0.009 (0.872)
26. - DT3 (Control)
Speed limit
27.
28. - 50 0.012 (1.668) 0.830 (13.240)** -0.027 (-3.656)**
29. - 80 (Control) (Control) (Control)
Scenario
30.
31. - EB -0.116 (-15.903)** -0.975 (-15.574)** -0.002 (-0.296)
32. - FB (Control) (Control) (Control)
Random variable
Var(Intercept)#
33. 0.010 (4.219)** 0.596 (4.219)** 0.010 (4.272)**
Goodness-of-fit
Restricted log likelihood
34. 537.43 -3,615.96 686.40
Unrestricted log likelihood
35. 681.72 -3,430.95 679.03
Likelihood ratio test
36. 288.58** 371.01** 14.741**
*
p < 0.05; ** p < 0.01; BrAC: breath alcohol concentration.
#
Variance of the random variables.

17
Table 4 Simulation driving experiments on alcohol-impaired driving performance
Ref. Country Drinking Simple Descriptions
category size
Burian et al., USA 0.3, 0.5, and 0.8 13 - No effect on speed.
2002 g/kg
Harrison et USA 0.65 g/kg 28 - Increase in lane variation.
al., 2005
Leung and Australia 0.7 g/kg (male); 32 - Decrease in detection times.
Starmer, 0.6 g/kg (female) - No effect on overtaking
2005 behavior.
Fillmore et USA 0.65 g/kg 14 - Increase in speed, lateral
al., 2008 position, and lane variation.
- No effect on reaction time.
Weafer et USA 0.0, 0.45, and 24 - Increase in speed variation and
al., 2008 0.65 g/kg lane position.
Ronen et al., Israel 0.05% BAC 12 - Increase in driving speed and
2010 steering wheel deviation.
- Decrease in lane position
variation.
- No effect on speed variation.
Wester et al., The 0.00%, 0.02%, 32 - Increase in reaction time.
2010 Netherlands 0.05%, 0.08%,
and 0.10% BAC
Lenne et al., Australia 0.4 and 0.6 g/kg 47 - Increase in speed and lateral
2010 position variation.
Liu and Ho, Taiwan 0.0%, 0.05%, 8 - Increase in variance in
2010 0.08% and 0.10% longitudinal speed and
BAC reaction time.
Harrison and USA 0.65 g/kg 40 - Increase in standard deviation
Fillmore, of lane position.
2011 - No effect on driving speed.
Christoforou Greece Moderate 49 - Increase in reaction time.
et al., 2013
Helland et Norway 0.0, 0.7 and 1.05 20 - Increase in standard deviation
al., 2013 g/kg of lane position.
Charlton and New Moderate, High 44 - Significant effect on steering
Starkey, Zealand performance and mean speed.
2015
Our study Hong Kong Low, Moderate 52 - Increase in braking reaction
time and standard deviation of
lane position.
- No effect on standard
deviation of speed.
Notes: Light – an alcohol dose of less than 0.3 g/kg.
Medium – an alcohol dose of 0.6 g/kg (female) and 0.75 g/kg (male) or a target of 0.05 BAC.
High – an alcohol dose of ≥0.75 g/kg (female) and ≥1.0 g/kg (male) or a target of ≥0.08 BAC.

18
19” LCD monitors

Steering
wheel

Pedal kit

Figure 1 Hardware set up of the driving simulator

Baseline Post-beverage session

session
C&B PT DT0 A DT1 DT2 DT3
Time
-60 0 60 120 180 240 300 (min.)

Notes:
PT Practice test (10min)
DTX Simulated driving test (20min)
A Alcohol intake (20min)
C&B Clinical assessment and breakfast
BrAC measurement

Figure 2 General experimental procedures

19
Figure 3a Emergency braking module at 50 km/h in the simulated driving test

Figure 3b Following braking module at 80 km/h in the simulated driving test

Speed (km/h)
Notes:
TEvent = Time when braking event appeared
TBrake = Time when brake was applied
VLimit VLimit = Speed limit (50 or 80 km/h)

𝐵𝑅𝑇 = 𝑇𝐵𝑟𝑎𝑘𝑒 − 𝑇𝐸𝑣𝑒𝑛𝑡

0 TEvent TBrake Time (sec)

Figure 4 Definition of braking reaction time

20
 Notes:
Speed (km/h)
TLimit = Time when the target speed limit was reached
TLimit+10 = TLimit + 10 sec
VLimit = Speed limit (50 or 80 km/h)
VLimit

∆𝑇 = 10⁡𝑠𝑒𝑐

0 TLimit TLimit+10 Time (sec)

Figure 5 Measurements of lateral position and speed

21