Abstract
Aims.
Sobriety checkpoints are an effective strategy to reduce alcohol-impaired driving, motor vehicle crashes, injuries, and fatalities. The aim of this study was to identify the geographic extent over which individual sobriety checkpoints affect alcohol-impaired driving.
Methods.
This spatial ecological panel analysis used geolocated breath test data from the Queensland Police Service, Australia, for January 2012 to June 2018. Data were aggregated over 338 weeks within 528 Statistical Area level 2 units (n=178,464 SA2-weeks) and 84 Statistical Area level 3 units (n=28,392 SA3-weeks). SA2 units in Queensland contain a mean population of 8,883.5 (SD=5,5018,3) and encompass 468.9 roadway kilometers (SD=1,490.0); SA3 units contain a mean population of 57,201.6 (SD=29,521.6) and encompass 2,936.0 roadway kilometers (SD=7,025.0). Independent measures were the density of sobriety checkpoints conducted per 500 roadway kilometers within local and spatially adjacent space-time units. The dependent measure was the rate of tests that detected breath alcohol concentration (a proxy for Blood Alcohol Concentration [BAC]) greater than the legal maximum value of 0.05% for fully licensed drivers in Queensland. Bayesian hierarchical spatial negative binomial models related sobriety checkpoints to the rate of breath tests with BAC≥0.05% within and between space-time units.
Results.
One additional sobriety checkpoint conducted per 500 roadway kilometers was associated with 2.5% reduction in the rate of breath tests with BAC≥0.05% within local SA2 units (IRR=0.975; 95%CrI: 0.973, 0.978), and with 5.5% reduction in the rate of breath tests with BAC≥0.05% within local SA3 units (IRR=0.945; 95%CrI: 0.937, 0.953). Associations were attenuated towards null in spatially adjacent units and in temporally lagged units (e.g., SA3-weeks; adjacent lagged 1 week: IRR=0.969; 95%CrI: 0.937, 1.003).
Conclusions.
Individual sobriety checkpoints are associated with reductions in nearby alcohol-impaired driving. Relationships decay after approximately one week and beyond local areas containing approximately 60,000 residents and 3,000 kilometers of roadway.
Keywords: alcohol, motor vehicle, crash, sobriety, checkpoint
INTRODUCTION
Motor vehicle crashes are a leading cause of death and injury around the world (1). A total of 1.3 million people died and a further 63 million were injured in motor vehicle crashes in 2017, leading to a total of approximately 70 million disability adjusted life years lost (2). Motor vehicle crashes will cost the world economy an estimated $1.8 trillion (in 2010 $US) from 2015 to 2030 (3). One of the major contributors to motor vehicle crash incidence is alcohol-impaired driving (4). In Organization for Economic Co-operation and Development (OECD) countries approximately one-fifth of motor vehicle crash deaths were attributable to alcohol consumption between 2000 and 2010 (5); this proportion varied widely between countries, from less than 5% in Israel to over 35% in Greece (6). To reduce the considerable human and economic burden, the WHO Global Road Safety Performance Targets for 2030 include the goal of halving the number of alcohol-involved motor vehicle crash injuries and fatalities (7).
Sobriety checkpoints aim to reduce alcohol-impaired driving among the full population of drivers within a geographic area (8, 9). The intervention involves law enforcement officers establishing a roadside checkpoint in which all drivers are stopped for a predetermined period and subject to the same protocolized sobriety breath alcohol tests. Drivers who are found to be driving while impaired by alcohol (or other illegal drugs) are subject to local laws and penalties. The intervention is guided by the rationalist theory that checkpoints will have general deterrent effects among the population of drivers who pass by or otherwise become aware of the activity, because these motorists will be less likely to drive while impaired during future trips due to increased perceived risks of detection and penalties (10). Concomitant specific deterrent effects will apply to people who are found to be driving while impaired at a sobriety checkpoint. However, specific deterrent effects are likely to be minimal (11) because the overall proportion of impaired drivers present on roadways who are detected at sobriety checkpoints is very low (12).
Consistent with general deterrence theory, several systematic reviews of observational studies have found sobriety checkpoints to be associated with decreased incidence of alcohol-impaired driving and alcohol-involved crashing (13-16). Two cluster randomized controlled trials that test the combined impacts of sobriety checkpoints and other universal prevention strategies provide supporting experimental evidence for their effectiveness (17, 18). Given the consistency and strength of these findings, research on sobriety checkpoints can now move from intervention science (which tests whether an intervention is effective) to implementation science (which tests how an intervention is most effective)(19, 20).
Accordingly, research is beginning to identify configurations of individual sobriety checkpoints (within the context of overall checkpoint programs) that most effectively reduce alcohol impaired driving and crashing. For example, general deterrence theory predicts that the impacts of individual sobriety checkpoints will be small and will decay over time because drivers’ perceptions regarding risks for detection and punishment for impaired driving will revert to their prior level some time after being exposed to a checkpoint (21). Studies in Los Angeles, USA (22), and Brisbane, Australia (23), found that individual checkpoints were associated with small relative reductions in alcohol-involved crash incidence for approximately one week. These results suggest checkpoints should be scheduled during the week prior to dates when impaired driving is likely to be greatest, such as public holidays and sporting events (24, 25).
An important further step for implementation science is to examine the spatial scale over which individual sobriety checkpoints are associated with risks for alcohol-involved crashes. Just as the effects of individual checkpoints decay over time, they are also likely to decay over geographic space. Prior studies have detected associations between individual checkpoints and impaired driving outcomes within areas as small as 3.2 kilometers around checkpoint sites (26) and as large as the full extent of cities such as Los Angeles and Brisbane (22, 23); but no studies have sought to identify the precise spatial scale over which these associations occur, and the optimal spatial location for checkpoints is not clear. Given these gaps in knowledge, the aim of this study was to identify the geographic extent over which individual sobriety checkpoints affect alcohol-impaired driving.
METHOD
Study Design
The study setting for this ecological (i.e., aggregate-level) space-time panel analysis was the state of Queensland, Australia. The state has a land area of 1.8 million km2 and had a population of 4.9 million on the day of the 2016 census. Data are released for the Australian census within nested geographies at the statistical area level 1 through level 4 (SA1 – SA4), the state level, and the national level. Based on our prior research (9, 22, 23, 27), we assessed that the theoretically plausible areas level over which individual checkpoints could be associated with alcohol impaired driving were the SA2 level (typically one or more suburbs with resident population of 3,000-25,000) and SA3 level (typically regional areas with resident population of 30,000-130,000). In Queensland, there are 528 SA2 areas, which have a mean land area of 3,276.8 km2 (SD = 18,559.0 km2) and a mean population of 8,883.5 (SD = 5,5018,3); and 84 SA3 areas, which have a mean land area of 20.597.3 km2 (SD = 81,496.9 km2) and a mean population of 57,201.6 (SD = 29,521.6). Following our previous findings (22, 23) we partitioned these polygons into weeks, from 5:00am Tuesday to 4:59am on the following Tuesday, because this time typically has very low activity for sobriety checkpoints and alcohol impaired driving (25). Data were available for 338 weeks beginning at 5:00am Tuesday January 3, 2012 and ending at 4:59am Tuesday June 26, 2018. Thus, the space-time analytic data sets were two balanced panels of 528 SA2 units × 338 weeks = 178,464 SA2-weeks, and 84 SA3 units × 338 weeks = 28,392 SA2-weeks.
Data
The Queensland Police Service (QPS) Road Policing Command provided data for the main dependent and independent measures for this study. Queensland Police Service has jurisdiction throughout the state of Queensland. The QPS Road Policing Command oversees a comprehensive program of roadside sobriety testing across the full extent of the state, including sobriety testing checkpoints. Operational targets during the study period included an annual 1:1 ratio of roadside sobriety tests to licensed drivers (27). Breath testing devices automatically record the date, time, and breath alcohol concentration (an approximation of Blood Alcohol Concentration [BAC]) each time a test is conducted, and these data are automatically uploaded to a central database. An embedded Global Positioning System (GPS) records a single pair of latitude and longitude coordinates each time the device is activated and the attending officer specifies that the test is part of a stationary checkpoint, otherwise a new pair of coordinates is generated for every test. The dependent measure for this ecological space-time panel analysis was the rate of breath tests per space-time unit that had BAC ≥0.05%, which is the legal maximum for fully licensed drivers in Queensland.
Checkpoints
The main independent measure was the rate of sobriety checkpoints conducted per roadway kilometer within each space-time unit. As in a previous study, we used hierarchical cluster analyses to identify geographic clusters of breath tests (23). Briefly, the latitude and longitude coordinates for each breath test were converted into X and Y values for a projected coordinate system (GDA_1994_BCSG02; WKID: 3113; Units: Meters). We then conducted complete linkage cluster analyses using data for each 24-hour period (5:00am to 4:59am) and grouped breath tests that were within 100 meters. Checkpoints were groups of breath tests that were within three standard deviations of the mean time for tests conducted within a cluster. After consulting Road Policing Command, we included checkpoints that were between 180- and 640-minutes duration, used ≥3 breath testing devices, and included ≥25 tests. We calculated counts of checkpoints per space-time unit, then denominated these values by the total length of roadway kilometers within the SA2 or SA3 units (i.e., measuring “sobriety checkpoint density”) to account for variation in spatial unit size.
The primary goal of this research was to identify the precise spatial scale over which sobriety checkpoints affect alcohol-impaired driving. To achieve this goal, we assessed associations between alcohol-impaired driving and sobriety checkpoint density both within the local space-time units of analysis and within spatially adjacent units (Figure 1). To compute the number of checkpoints in adjacent spatial areas during week t, we summed the checkpoint counts that occurred during week t in adjacent areas by queen’s contiguity (i.e., that shared a border point or segment)(28). We also calculated the total length of roadway kilometers in these neighboring areas, then divided adjacent checkpoints by the total adjacent roadway length.
Figure 1.
Adjacency was defined using queen’s contiguity.
Guiding theory (21) and our prior observations of cities (22, 23) suggest that sobriety checkpoints may affect alcohol impaired driving for approximately one week after the checkpoint date. Thus, in addition to measuring associations for spatially adjacent units, we also measured associations for temporally lagged units. To assess delayed effects, we calculated the number of checkpoints conducted in local spatial units, i, and in spatial units adjacent to i during week t-1 (i.e., in the same spatial unit one week prior: “Lag 1”) and t-2 (the same spatial unit two weeks prior: “Lag 2”), divided by the total roadway kilometers within the relevant areas.
Covariates
We calculated statistical controls for seasonality, including sine and cosine functions that captured annual rhythms in impaired driving, and linear and quadratic terms that captured additional trends over time. We also measured the total rainfall and the average maximum temperature (29, 30). To generate these measures, we extracted daily data from the Australian Government Bureau of Meteorology for all georeferenced weather stations in Queensland. Raster layers, calculated using inverse distance weighing for rainfall and temperature values, created smooth continuous surfaces that covered the extent of Queensland for each weather variable for each study week. We extracted raster values representing rainfall and temperature from the cell in which the centroid of each SA2 and SA3 unit was located, thereby creating spatially and temporally varying measures of estimated rainfall and temperature over SA2-weeks and SA3-weeks.
Statistical Analysis
Bayesian hierarchical spatial negative binomial models were used to assess the associations between the rate of breath tests with BAC ≥0.05% and the density of sobriety checkpoints conducted per roadway kilometer within SA2-weeks and SA3-weeks. Model specification is described in full in the online supplementary files. Briefly, we fitted separate models for the SA2 and SA3 geographies, and interpreted the posterior means of the regression coefficients in the model as the incidence rate ratios measuring the association between the sobriety checkpoint density and the incidence of breath tests with BAC ≥0.05%.
Sensitivity Analysis
Geographic bounding of police operations within administrative divisions may bias associations with breath test results due to the modifiable areal unit problem (33). To address this concern, we conducted a sensitivity analysis in which the units were Queensland Police Service division-weeks. There are 337 QPS divisions in Queensland, thus these polygons are slightly larger compared to the SA2 units in the main models. It is also possible that associations differ for metropolitan compared to rural areas, so we conducted subgroup analyses using SA2-weeks, SA3-weeks and QPS division-weeks that were located within the greater metropolitan regions of Brisbane. Finally, drivers who pass through a sobriety checkpoint and are randomly selected for breath testing are less likely to be alcohol-impaired compared to those who are selected for testing during other traffic stops. This observation is important for the current analysis because endogeneity could bias measures of association; therefore, we conducted additional sensitivity analyses in which we subtracted breath tests conducted within sobriety checkpoints from both the dependent measure (breath tests with BAC ≥0.05%) and the offset term (total breath tests conducted). Note the analysis was not pre-registered and that the results should be considered exploratory.
RESULTS
In the 338 weeks between January 2012 and June 2018, Queensland Police Service conducted a total of 16,891,593 breath tests, of which 159,863 (0.9%) had BAC ≥0.05%. Our algorithm identified that a total of 7,116 sobriety checkpoints were conducted in Queensland during that study period, including 1,426,492 breath tests (8.4% of all breath tests). Table 1 shows the distribution of breath tests and sobriety checkpoints per roadway kilometer within SA2-weeks and SA3-weeks.
Table 1.
Descriptive statistics for breath tests, sobriety checkpoints, and covariates, aggregated over 338 weeks within 528 SA2 polygons (SA2-weeks) and 84 SA3 polygons (SA3-weeks).
| Mean | SD | Min | Max | |
|---|---|---|---|---|
| SA2-weeks (n = 170,945) | ||||
| Tests (count) | 94.6 | 156.3 | 0.0 | 5684.0 |
| Positive tests (count) | 0.9 | 1.7 | 0.0 | 100.0 |
| Proportion of tests that were positive | 0.0 | 0.1 | 0.0 | 1.0 |
| Local checkpoints | 0.0 | 0.3 | 0.0 | 9.0 |
| Local roadway kms | 468.9 | 1489.9 | 7.7 | 17964.3 |
| Local checkpoints per 500 roadway kms | 0.2 | 1.6 | 0.0 | 128.3 |
| Adjacent checkpoints | 0.2 | 0.6 | 0.0 | 14.0 |
| Adjacent roadway kms | 2955.3 | 7182.0 | 0.0 | 63676.1 |
| Adjacent checkpoints per 500 roadway kms | 0.2 | 0.7 | 0.0 | 22.8 |
| Rainfall (mm) | 21.5 | 45.3 | 0.0 | 923.5 |
| Maximal temperature (°C) | 27.2 | 3.8 | 11.3 | 44.7 |
| SA3-weeks (n = 27,496) | ||||
| Tests (count) | 609.5 | 543.4 | 0.0 | 7244.0 |
| Positive tests (count) | 5.8 | 5.5 | 0.0 | 107.0 |
| Proportion of tests that were positive | 0.0 | 0.0 | 0.0 | 0.4 |
| Local checkpoints | 0.3 | 0.7 | 0.0 | 14.0 |
| Local roadway kms | 2935.5 | 7025.4 | 156.9 | 53327.6 |
| Local checkpoints per 500 roadway kms | 0.2 | 0.7 | 0.0 | 28.1 |
| Adjacent checkpoints | 1.2 | 1.7 | 0.0 | 19.0 |
| Adjacent roadway kms | 14968.1 | 23880.7 | 533.8 | 98815.4 |
| Adjacent checkpoints per 500 roadway kms | 0.1 | 0.3 | 0.0 | 6.1 |
| Total rainfall (mm) | 21.5 | 44.7 | 0.0 | 863.7 |
| Maximal temperature (°C) | 27.1 | 3.9 | 13.8 | 44.6 |
Sobriety checkpoint density per roadway kilometer was greatest in high population density areas along the eastern coast (particularly the greater surrounding area of the capital city, in the south-east). Figure 2 shows the sobriety checkpoint density per roadway kilometer within SA2 and SA3 units, aggregated across the full 6.5 years. Breath tests tended to have a higher proportion of readings with BAC ≥0.05% in the very rural north and north-western parts of the state.
Figure 2.
Geographic distributions of sobriety checkpoints per 100kms of roadway and the proportion of breath tests with BAC ≥0.05%; January 2012-June 2018, Queensland
Results for the Bayesian hierarchical spatial negative binomial models are presented in Table 2. At the SA2 geography, an additional sobriety checkpoint per 500 roadway kilometers conducted within local SA2-weeks was associated with 2.5% reduction in the rate of breath tests with BAC ≥0.05% (IRRlocal lag0 = 0.975; 95%CrI: 0.973, 0.978). Associations were weaker but remained negative within adjacent SA2 units at temporal lag 0 (IRRadjacent lag0 = 0.981; 95%CrI: 0.971, 0.991), were attenuated for local and adjacent SA2 units one week after the checkpoint date (IRRlocal lag1 = 0.996; 95%CrI: 0.992, 1.000; IRRadjacent lag1 = 0.994; 95%CrI: 0.983, 1.005), and were significant for local and adjacent SA2 units in subsequent week 2 (IRRlocal lag2 = 0.995; 95%CrI: 0.991, 0.999; IRRadjacent lag2 = 0.981; 95%CrI: 0.970, 0.992). At the SA3 geography, the associations were comparable, in that relationships were strongest during the week of the checkpoint (IRRlocal lag0 = 0.945; 95%CrI: 0.937, 0.953) and were mostly attenuated in subsequent weeks (e.g., IRRlocal lag1 = 0.995; 95%CrI: 0.984, 1.006). Associations were comparably strong in spatially adjacent units (e.g., IRRadjacent lag0 = 0.942; 95%CrI: 0.912, 0.973).
Table 2.
Results for Bayesian hierarchical spatial negative binomial models for SA2-weeks (n = 178,464) and SA3-weeks (n = 28,392)
| Model 1: SA2-weeks | Model 2: SA3-weeks | |||||
|---|---|---|---|---|---|---|
| IRR | (95% | CrI) | IRR | (95% | CrI) | |
| Local checkpoints per 500 roadway kms | ||||||
| Lag 0 * | 0.975 | 0.973 | 0.978 | 0.945 | 0.937 | 0.953 |
| Lag 1 * | 0.996 | 0.992 | 1.000 | 0.995 | 0.984 | 1.006 |
| Lag 2 * | 0.995 | 0.991 | 0.999 | 0.981 | 0.970 | 0.992 |
| Adjacent checkpoints per 500 roadway kms | ||||||
| Lag 0 * | 0.981 | 0.971 | 0.991 | 0.942 | 0.912 | 0.973 |
| Lag 1 * | 0.994 | 0.983 | 1.005 | 0.969 | 0.937 | 1.003 |
| Lag 2 * | 0.981 | 0.970 | 0.992 | 0.949 | 0.917 | 0.982 |
| Temporal structure | ||||||
| Timepoint * | 0.690 | 0.671 | 0.710 | 0.714 | 0.692 | 0.736 |
| Timepoint squared * | 1.458 | 1.417 | 1.501 | 1.409 | 1.365 | 1.454 |
| Cosine (annual) | 1.008 | 0.998 | 1.018 | 1.006 | 0.996 | 1.017 |
| Sine (annual) | 0.989 | 0.979 | 0.999 | 0.994 | 0.983 | 1.005 |
| Space-time varying covariates | ||||||
| Total rainfall (per 100mm increase) | 1.112 | 1.094 | 1.131 | 1.106 | 1.085 | 1.126 |
| Maximal temperature (per 10°C increase) | 1.090 | 1.068 | 1.112 | 1.100 | 1.076 | 1.124 |
| Overdispersion | 0.490 | 0.427 | 0.563 | 0.190 | 0.184 | 0.197 |
| ICAR variance | 0.576 | 0.563 | 0.590 | 0.193 | 0.143 | 0.270 |
centered and scaled before model fitting
CI: credible interval
The results of the sensitivity analyses further illuminate the associations of interest (Tables s5 to s7). Associations were slightly stronger when data were aggregated within QPS divisions (e.g., IRRlocal lag0 = 0.933; 95%CrI: 0.926, 0.939) compared SA2 and SA3 units, and after removing breath tests conducted within checkpoints from the dependent measure (e.g., SA3-weeks: IRRlocal lag0 = 0.939; 95%CrI: 0.928, 0.951). Note that QPS divisions are larger than SA2 units but smaller than SA3 units, and that the trend across many sensitivity analyses was that associations were slightly weaker for adjacent compared to local SA3 and QPS divisions. Relationships were slightly weaker for the greater metropolitan area of Brisbane compared to the full extent of Queensland (e.g., SA3-weeks: IRRlocal lag0 = 0.955; 95%CrI: 0.948, 0.961), and spatially adjacent relationships were not detectable with just the very few SA3-units in Brisbane, likely due to a lack of variation.
DISCUSSION
Observational and experimental studies provide compelling evidence that sobriety checkpoints likely cause reductions in alcohol-impaired driving and alcohol-involved crashes (9, 13-16, 27). The imperative for research is now to examine configurations of sobriety checkpoints that maximize public health benefits but minimize the operational cost burden for police (19, 20). In this spatial ecological panel analysis of 16 million breath tests conducted over 6.5 years in Queensland, Australia, we found each additional sobriety checkpoint per roadway kilometer reduced the rate of positive BAC tests over the extent of the Statistical Area level 3 (SA3) unit in which it was conducted, which is an area of around 60,000 residents encompassing a total of 3,000km of roadway.
These findings are consistent with deterrence theory (10) and the findings of prior published studies (22, 23). The guiding theory holds that rational drivers who pass by a sobriety checkpoint or otherwise learn of its presence (e.g., through social media) will have increased perceived risks of detection and punishment for impaired driving and will be less likely to drive while impaired on subsequent occasions. Previous research has detected that these associations decay temporally over approximately one week (22, 23), and that smaller checkpoints staffed by fewer law enforcement officers have similar effects compared to larger checkpoints staffed by more officers (34). Here we add that they also decay spatially within local areas that approximate the geographic extent within which individuals’ travel during routine activities. Point estimates were strongest for local SA3 units, though the negative associations for local and adjacent SA2 units suggest most of the benefit is concentrated in small areas around the checkpoint sites and that they decay over space. The association for adjacent SA3 units suggests benefits further decay beyond SA3 boundaries. The most likely mechanism to explain these collective findings is that local motorists who live within those areas alter their behavior temporarily in response to checkpoint operations. Effects over wider geographic areas diminish because the number of motorists in those areas exposed to the local checkpoints are substantially reduced.
There are important practical implications for this work. Sobriety checkpoints are an effective prevention intervention to reduce alcohol-impaired driving and crashing, but they are also costly for law enforcement (35). During this 6.5-year study period, Queensland Police Service expended 36,940.2 officer-hours across a total of 7,116 checkpoints (not including officer time related to preparing for checkpoint operations and processing positive cases). Understanding the precise spatial and temporal signatures of sobriety checkpoints on impaired driving will help law enforcement agencies such as the Queensland Police Service to deploy scarce policing resources more efficiently. The emerging work in this area (including the current study) suggest that police departments could reduce costs without reducing public health benefits by conducting checkpoints staffed by fewer police officers (34). Scheduling checkpoints in the week prior to dates with high expected incidence of impaired driving (22, 23) and within about a 60,000 resident-area of areas with high expected concentration of impaired drivers will maximize impacts. These general guidelines allow for police departments to flexibly schedule sobriety checkpoints to achieve operational imperatives (e.g., reduced costs, officer safety)(35), while using an evidence-based approach to maximize reductions in alcohol-impaired driving and minimizing law enforcement costs.
This research has important limitations. First, Queensland Police Service did not randomly allocate sobriety checkpoints to space-time units; rather Road Policing Command carefully determines where and when checkpoints should be conducted based on decades of experience and their local knowledge. If checkpoints are systematically associated with alcohol-impaired driving in local or adjacent spatial units and across temporal lags, parameter estimates may be biased. Nevertheless, one would expect checkpoints to be allocated to times and locations when impaired driving was greater, which would produce positive associations, so we are confident that the negative associations detected here were not due to confounding. Results of the sensitivity analyses further support that our results are not affected by endogeneity. Additional confounding by unknown spatially or temporally varying covariates (roadway types, school zones, permanent policing surveillance equipment) could have biased results in either direction. An essential next step is to conduct experimental studies that can control for these potential biases methodologically. Measurement error due to police conducting multiple breath tests of the same driver (e.g., to guard against false positive or false negative findings) could bias results in either direction. Large vehicle drivers and novice drivers who fall within Queensland’s graduated licensing system (i.e., learner and probationary licensees) have a maximum legal BAC of 0.00%, but driver characteristics were not available.
Society checkpoints are a highly effective prevention strategy that are an essential component of a comprehensive approach to reducing the considerable societal burden due to alcohol-impaired motor vehicle crashes. Research is beginning to explore ways to conduct sobriety checkpoints ethically, effectively, and efficiently, to maximize public benefit. Further scientific advances in this area will encourage and better enable implementation.
Supplementary Material
What this Research Adds.
Each sobriety checkpoint reduces alcohol-impaired driving in areas containing around 60,000 residents for around 1 week.
Funding:
Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award Number R21AA025749. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors wish to acknowledge the support and assistance from the Queensland Police Service in undertaking this research. The views expressed in this publication are not necessarily those of the Queensland Police Service and any errors of omission or commission are the responsibility of the authors.
Footnotes
Conflicts: None.
REFERENCES
- 1.WHO. Global status report on road safety 2018. Licence: CC BY-NC-SA 3.0 IGO. Geneva; 2018. [Google Scholar]
- 2.Global Burden of Disease Collaborators 2017. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2018;392(10159):1789–858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chen S, Kuhn M, Prettner K, Bloom DE. The global macroeconomic burden of road injuries: estimates and projections for 166 countries. The Lancet Planetary Health. 2019;3(9):e390–e8. [DOI] [PubMed] [Google Scholar]
- 4.GRSP. Drinking and Driving: a road safety manual for decision-makers and practitioners. Geneva; 2007. [Google Scholar]
- 5.NHTSA. 2018 Fatal Motor Vehicle Crashes: Overview DOT HS 812 826. In: US Department of Transportation NHTSA, editor. Washington, DC: 2019. [Google Scholar]
- 6.Vissers L, Houwing S, Wegman F. Alcohol-Related Road Casualties in Official Crash Statistics. In: Group ITSDaA, editor. Paris, France: OECD; 2017. [Google Scholar]
- 7.United Nations. Resolution adopted by the General Assembly on 15 April 2016, (2016). [Google Scholar]
- 8.NASEM. In: Negussie Y, Geller A, Teutsch SM, editors. Getting to Zero Alcohol-Impaired Driving Fatalities: A Comprehensive Approach to a Persistent Problem. Washington (DC): National Academies Press (US); 2018. [PubMed] [Google Scholar]
- 9.Ferris J, Mazerolle L, King M, Bates L, Bennett S, Devaney M. Random breath testing in Queensland and Western Australia: examination of how the random breath testing rate influences alcohol related traffic crash rates. Accid Anal Prev. 2013;60:181–8. [DOI] [PubMed] [Google Scholar]
- 10.Homel R Drivers who drink and rational choice: Random breath testing and the process of deterrence. In: Clarke RV, Felson M, editors. Routine Activity and Rational Choice: Advances in Criminological Theory; 1993. p. 59–84. [Google Scholar]
- 11.Watson B, Freeman J. Perceptions and experiences of random breath testing in Queensland and the self-reported deterrent impact on drunk driving. Traffic Inj Prev. 2007;8(1):11–9. [DOI] [PubMed] [Google Scholar]
- 12.Voas RB. A new look at NHTSA's evaluation of the 1984 Charlottesville Sobriety Checkpoint Program: implications for current checkpoint issues. Traffic Inj Prev. 2008;9(1):22–30. [DOI] [PubMed] [Google Scholar]
- 13.Bergen G, Pitan A, Qu S, Shults RA, Chattopadhyay SK, Elder RW, et al. Publicized sobriety checkpoint programs: a community guide systematic review. Am J Prev Med. 2014;46(5):529–39. [DOI] [PubMed] [Google Scholar]
- 14.Elder RW, Shults RA, Sleet DA, Nichols JL, Zaza S, Thompson RS. Effectiveness of sobriety checkpoints for reducing alcohol-involved crashes. Traffic Inj Prev. 2002;3(4):266–74. [Google Scholar]
- 15.Erke A, Goldenbeld C, Vaa T. The effects of drink-driving checkpoints on crashes--a meta-analysis. Accid Anal Prev. 2009;41(5):914–23. [DOI] [PubMed] [Google Scholar]
- 16.Shults RA, Elder RW, Sleet DA, Nichols JL, Alao MO, Carande-Kulis VG, et al. Reviews of evidence regarding interventions to reduce alcohol-impaired driving. Am J Prev Med. 2001;21(4 Suppl):66–88. [DOI] [PubMed] [Google Scholar]
- 17.Holder HD, Gruenewald PJ, Ponicki WR, Treno AJ, Grube JW, Saltz RF, et al. Effect of Community-Based Interventions on High-Risk Drinking and Alcohol-Related Injuries. JAMA. 2000;284(18):2341–7. [DOI] [PubMed] [Google Scholar]
- 18.Saltz RF, Paschall MJ, O'Hara SE. Effects of a Community-Level Intervention on Alcohol-Related Motor Vehicle Crashes in California Cities: A Randomized Trial. Am J Prev Med. 2021;60(1):38–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bauer MS, Damschroder L, Hagedorn H, Smith J, Kilbourne AM. An introduction to implementation science for the non-specialist. BMC Psychol. 2015;3(1):32-. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Community Preventive Services Task F. Publicized sobriety checkpoint programs to reduce alcohol-impaired driving: recommendation of the Community Preventive Services Task Force. Am J Prev Med. 2014;46(5):540–1. [DOI] [PubMed] [Google Scholar]
- 21.Ross HL. Deterring the Drinking Driver: Legal Policy and Social Control: Lexington Books; 1984. [Google Scholar]
- 22.Morrison CN, Ferris J, Wiebe DJ, Peek-Asa C, Branas CC. Sobriety Checkpoints and Alcohol-Involved Motor Vehicle Crashes at Different Temporal Scales. Am J Prev Med. 2019;56(6):795–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Morrison CN, Kwizera M, Chen Q, Puljevic C, Branas CC, Wiebe DJ, et al. Alcohol-Involved Motor Vehicle Crashes and the Size and Duration of Random Breath Testing Checkpoints. Alcoholism: Clinical and Experimental Research. 2021;n/a(n/a). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Redelmeier DA, Stewart CL. Do Fatal Crashes Increase Following a Super Bowl Telecast? Chance. 2005;18(1):19–24. [Google Scholar]
- 25.Lloyd B, Matthews S, Livingston M, Jayasekara H, Smith K. Alcohol intoxication in the context of major public holidays, sporting and social events: a time-series analysis in Melbourne, Australia, 2000-2009. Addiction. 2013;108(4):701–9. [DOI] [PubMed] [Google Scholar]
- 26.Nunn S, Newby W. The geography of deterrence: exploring the small area effects of sobriety checkpoints on alcohol-impaired collision rates within a city. Eval Rev. 2011;35(4):354–78. [DOI] [PubMed] [Google Scholar]
- 27.Ferris J, Devaney M, Sparkes-Carroll M, Davis G. A national examination of Random Breath Testing and alcohol-related traffic crash rates (2000-2012). Canberra, ACT: Foundation for Alcohol Research and Education; 2015. [Google Scholar]
- 28.Cromley EK , & McLafferty S (2002). GIS and public health. New York: Guilford; 2002. [Google Scholar]
- 29.Omranian E, Sharif H, Dessouky S, Weissmann J. Exploring rainfall impacts on the crash risk on Texas roadways: A crash-based matched-pairs analysis approach. Accid Anal Prev. 2018;117:10–20. [DOI] [PubMed] [Google Scholar]
- 30.Zhao S, Wang K, Liu C, Jackson E. Investigating the effects of monthly weather variations on Connecticut freeway crashes from 2011 to 2015. J Safety Res. 2019;71:153–62. [DOI] [PubMed] [Google Scholar]
- 31.Banerjee S, Carlin BP, Gelfand AE. Hierarchical Modeling and Analysis for Spatial Data, Second Edition: Taylor & Francis; 2014. [Google Scholar]
- 32.Gómez-Rubio V Bayesian Inference with INLA: CRC Press; 2020. [Google Scholar]
- 33.Openshaw S The modifiable areal unit problem. Norwich: Geo Books; 1984. [Google Scholar]
- 34.Stuster J, Blowers P. Experimental evaluation of sobriety checkpoint programs. HS 808 287. In: US Department of Transportation NHTSA, editor. Washington, DC: 1995. [Google Scholar]
- 35.Fell JC, Lacey JH, Voas RB. Sobriety checkpoints: evidence of effectiveness is strong, but use is limited. Traffic Inj Prev. 2004;5(3):220–7. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.