Bulletin of the World Health Organization

Effect on road traffic injuries of criminalizing road traffic offences: a time–series study

Ana M Novoa a, Katherine Pérez a, Elena Santamariña-Rubio a & Carme Borrell a

a. Agència de Salut Pública de Barcelona, Pl Lesseps 1, 08023 Barcelona, Spain.

Correspondence to Ana M Novoa (e-mail: anovoa@aspb.cat).

(Submitted: 19 August 2010 – Revised version received: 22 March 2011 – Accepted: 23 March 2011 – Published online: 31 March 2011.)

Bulletin of the World Health Organization 2011;89:422-431. doi: 10.2471/BLT.10.082180


Although road traffic injuries in Spain are decreasing, they still constitute a major public health problem. During 2008, 134 047 people suffered road traffic injuries and 3100 lost their lives in collisions.1 In recent years, the Spanish government has implemented several measures to reduce the burden of traffic injuries. In 2004 it established road safety as a political priority and created the Special Road Safety Measures 2004–20052 and the Strategic Road Safety Programme 2005–20083, which comprise several interventions focused primarily on the enforcement of traffic regulations. These regulations and their enforcement have reduced traffic injuries in Spain by 9% among men and 11% among women.4 In addition, the introduction of a penalty points system in July 2006 was followed by reductions of 11% and 12% in the number of men and women, respectively, who were seriously injured in traffic collisions across Spain.5

Legislation alone is unlikely to deter road users from engaging in risky behaviours. To be effective, it must be rigorously enforced and must strongly deter unlawful behaviour by generating awareness and fear of the consequences of breaking the law. This, in turn, depends on the degree of traffic surveillance, the severity of the penalty issued and the swiftness with which it is imposed. Public awareness campaigns can make legislation more effective.68

Despite existing laws, in Spain the number of injuries and deaths attributable to speeding and drunk driving is still extremely high. One year after the penalty points system was introduced, the main traffic offences punishable by penalty points were occurring at the following rates: speeding, 39.3%; non-compliance with wearing passive restraint devices, 15.5%; drunk driving, 11.6%.9 To further reduce road injuries linked to these behaviours, the penal code was modified on 1 December 2007. Several traffic offences were criminalized, with a change in the legal process for trying offenders from a civil to a criminal procedure. The main criminalized offences were driving over the speed limit, drunk driving, reckless driving and driving without a licence. The penalties for these violations depend on the severity of the offence but include imprisonment, a fine, compulsory community service or licence suspension (Table 1). Prior to this reform, speeding and drunk driving were also considered crimes but the penalties were much more lenient and there was no officially established speed or blood alcohol level marking the threshold for criminality, which left it up to the judge to decide. The penal code reform eliminates subjectivity and incorporates stricter penalties, including compulsory imprisonment in certain cases and a possible criminal record. An important publicity campaign was launched in all news media and intense public debate ensued.

So far, studies have only assessed the effectiveness of criminalizing driving under the influence of alcohol1015 but not other behaviours. The present study fills that gap by examining the effect of criminalizing several road behaviours on the numbers of drivers involved in injury collisions and of people injured in traffic collisions in Spain, broken down by gender, age, injury severity, type of road user, road type and time of collision. The working hypothesis is that criminalizing risky behaviours has reduced traffic injuries in Spain in the context of a previous downward trend resulting from road safety interventions implemented during previous years.


Study design and population

We used an interrupted time-series design to conduct an evaluation study in two study populations: (i) the number of drivers (injured or unharmed) involved in traffic collisions resulting in injury to self or to others (i.e. injury collisions) and (ii) the number of people injured in traffic collisions in Spain in 2000–2009.

Sources of information

We obtained traffic injury data from the Road Traffic Crashes Database of Spain’s General Traffic Directorate (Dirección General de Tráfico, DGT), which records injury collisions, the characteristics of the collision, and the vehicle and people involved. In Spain this information is collected and submitted to the DGT by the national and local police forces, who attend to non-urban and urban roads, respectively. Data on national fuel consumption, used as a proxy for exposure to traffic, was obtained from the Spanish Ministry of Public Works.


The dependent variables were the number of drivers involved in injury collisions and the number of people injured in traffic collisions. Analyses were stratified by the following: gender;16 age; injury severity as classified by the police (no harm [drivers only], slight, serious non-fatal [hospitalized > 24 hours], fatal); type of road user (car, motorcycle or moped, or pedestrian [only for the injured]); road type (urban, non-urban), and time of collision (weekday daytime, weekday night-time, weekend daytime, weekend night-time).

The main explanatory variable was the reform of the penal code. A dummy variable was created to compare rates before (1 January 2000 to 30 November 2007) and after (1 December 2007 to 31December 2009) the intervention. To adjust for the effect of road safety prioritization in 2004 and the introduction of the penalty points system in July 2006, we included two additional dummy variables in the models for the periods before and after these interventions were introduced. To adjust for changes in traffic exposure over the study period, we also included in the analyses a variable representing national fuel consumption as a proxy for motorised mobility across the population.

Statistical analysis

We performed time-series analyses using Poisson regression models adjusted for over-dispersion (quasi-Poisson).17 The number of drivers (and of people injured) per month was compared throughout the time series with adjustments for time trend and seasonal patterns using linear trend and sine and cosine functions.18 The model for each outcome can be summarized as follows:

where t is the time period (t = 1 for the first month of the series, t = 2 for the second, etc.); Xt identifies the pre- and post-intervention periods (Xt = 1 for the post-intervention period); k takes values between 1 and 6 (k = 1 for annual seasonality; k = 2 for six-monthly seasonality, etc.); T is the number of periods described by each sinusoidal function (e.g. t = 12 months); Pt is the dummy variable for road safety prioritization, multiplied by the time trend (t) (i.e. an interaction term) to take into account the differences in the time trend before and after the year 2004;19 Zjt represents other co-variables introduced (penalty points system and national fuel consumption); j is the number of co-variables introduced, and ε is the error term. We derived relative risks (RRs) and their 95% confidence intervals (CIs) from the adjusted models. They indicate the difference between the number of drivers involved (or people injured) in injury collisions before and after the intervention, after adjustments for time trend and seasonality. From the RRs we computed the percentage change in the number of drivers (or people injured) between the two periods.

The number of people prevented from being injured attributable to the reform of the penal code was calculated as the difference between the number of people injured in the post-intervention period and the number predicted by the statistical models.

We conducted the statistical analyses using Stata statistical software, release 10 (StataCorp. LP, College Station, United States of America).

Ethical approval

The present study was approved by the ethics committee (Comitè Ètic d’Investigació Clínica) of the Institut Municipal d’Assistència Sanitària.


In 2000–2009, 1 668 889 drivers were involved in injury collisions in Spain (annual median: 170 879). Most of them (78.7%) were male drivers, 70.7% of whom were between 18 and 44 years of age. An additional 1 454 971 people were injured in traffic collisions (annual median: 146 949). Again, most (63.4%) of them were male, and 65.1% of these males were between the ages of 18 and 44 years. Table 2 shows the distribution of these subjects by sex, age, injury severity, type of road user, road type and time of collision.

Drivers involved in injury collisions

Fig. 1 depicts the observed and expected numbers of male and female drivers (overall and seriously or fatally injured) involved in injury collisions throughout the study period. The graphs show a clear reduction in the number of male drivers involved in injury collisions compared with the expected numbers; notably, no such reduction was observed among female drivers.

Fig. 1. Observed and expected numbers of drivers involved in injury collisions, by sex and injury severity, Spain, 2000–2009
Fig. 1. Observed and expected numbers of drivers involved in injury collisions, by sex and injury severity, Spain, 2000–2009

Male drivers

For male drivers the overall risk of being involved in an injury collision in the post-intervention period was reduced by 7% (RR: 0.93; 95% CI: 0.89–0.97). The largest reductions in risk were observed for seriously injured and fatally injured drivers (14% and 11% reductions, respectively). No reduction in the risk of being an unharmed driver was observed (Table 3). The risk of being involved in an injury collision was reduced among all drivers under 65 years of age, but especially among those under 30.

A greater reduction in the risk of being involved in an injury collision was observed on non-urban roads than on urban roads (17% and 6% reduction, respectively). In addition, the effect varied by type of driver and road. Among motorcycle and moped riders, a reduction in the risk of having an injury collision was observed on both urban and non-urban roads, although the effect was larger on non-urban roads. However, for car drivers this risk was reduced only on non-urban roads. Finally, the risk was reduced during both daytime and night-time among all drivers.

Female drivers

Among female drivers, the overall risk of having an injury collision showed no significant change (RR: 0.99; 95% CI: 0.95–1.03) in the post-intervention period. Although a protective effect was observed in most subgroups analysed, especially against the risk of serious or fatal collisions, significant risk reductions were only observed among drivers aged 45 to 64 years old, car drivers on non-urban roads and motorcycle and moped riders on urban roads, and during daytime on weekdays (Table 3).

Injuries prevented

After the reform of the penal code, the number of people prevented from being injured in traffic collisions was consistent with the reduced risk of being involved in injury collisions observed among drivers. However, no protective effect was observed among pedestrians – except for a reduction in the women seriously injured or killed – or among children under 14 years of age.

Over the 25 months that followed the intervention, the number of men injured was 7.2% lower than the number expected to be injured had the penal code not been reformed (P < 0.01). Among women, the number injured was 3.0% lower (P = 0.241) (Table 4).

The greatest number of people prevented from being injured was observed among men, individuals aged 14 to 30 years and motorcycle or moped users.


Regulation of traffic behaviour is an essential component of road safety policy and imposing strict penalties for traffic offences can increase the deterrent effect of the law. The present study shows a reduction in both the number of drivers involved in injury collisions and the number of people injured in traffic collisions following the reform of the penal code in Spain. Greater reductions were observed among young male drivers, especially those riding motorcycles and mopeds. This may be, at least to some extent, because men and young drivers tend to engage in the riskier behaviours that were criminalized by the reformed penal code.20,21 Since females and older drivers are generally more compliant road users, the stricter penalties imposed for violating traffic laws are less likely to reduce their risk of being involved in traffic collisions than the risk among less compliant drivers.

The greater risk reductions observed among motorcycle and moped riders could be due in part to the generally younger age of these road users and to the fact that riders of powered two-wheel vehicles, especially mopeds, tend to be less compliant with road safety legislation.22 In addition, greater risk reductions were observed on non-urban roads perhaps because compliance with the speed limit and with laws against driving while intoxicated, both included among the criminalized behaviours, is lower on these roads than on urban roads.

Comparison with previous studies

No previous studies have assessed the effect of criminalizing multiple traffic behaviours on the rates of traffic injuries. However, several authors have evaluated the effect of criminalizing drunk driving, one of the behaviours included in the reform of the penal code in Spain. The findings vary greatly; they range from no effect to a 73% reduction in the number of alcohol-related collisions attributable to this measure. In Canada, where the legal blood alcohol concentration (BAC) limit is 0.05 g/l, an 18% decrease was observed in the number of fatally injured drunk drivers after the criminal law was passed.14 In Taiwan (China) where the BAC limit is 0.05 g/l, a 72.6% reduction in the number of collisions in which drivers had a positive alcohol breath test was observed.10 Studies in the United States of America (BAC limit 0.08 g/l) have shown lower reductions – 6%,13 5%11 and none12 – in the number of alcohol-related fatalities after the criminalization of drunk driving, perhaps because the use of several variables for criminal law in the study models could have confounded the real effect of the measure due to correlation. Finally, in Norway, where the BAC limit is 0.02 g/l, and in Sweden, where the BAC limit is 0.02 g/l, the number of traffic fatalities did not increase after criminal laws were attenuated.15 However, the authors did not analyse alcohol-related injuries or adjust for the effect of coexisting laws aimed to reduce drunk driving.

The results of these studies suggest that criminalizing drunk driving, as was done in Spain (BAC limit 0.05 g/l), can reduce alcohol-related crashes. This is consistent with the results of the present study. The smaller effect observed in Spain compared with other countries may be explained by the fact that much of the reduction in traffic injuries observed in recent years as a result of the prioritization of road safety and the penalty points system was already adjusted for in the models. Thus, the burden of traffic injuries, especially serious ones, has followed a downward trend in Spain since 2004, and the criminalization of certain traffic behaviours has prompted a further reduction. The effect is particularly evident among motorcycle riders, whose risk was the least affected by previous road safety interventions.4,5

Limitations and strengths

Since the intervention was nationwide, we had no comparison group. However, such a group is not essential in time–series analyses, although it can strengthen the evidence, because per cent changes are compared between time points in the same series.19 We controlled for time trend, seasonality, fuel consumption and previous road safety interventions. We included time trends in the analyses to account for changes in potential confounders, such as improvements in vehicle safety or in road behaviour, throughout the study period. To control for changes in exposure, we adjusted for national fuel consumption in the models, but we assumed that mobility changed uniformly during the study across age, sex, and user-type subgroups. We did not use the number of kilometres travelled by all vehicles because it was only available for non-urban roads. The number of vehicles registered would not have accurately captured changes in mobility; whereas this number increased steadily, fuel consumption suddenly dropped during 2008–2009. This is not likely to have resulted from greater vehicle efficiency, which would have involved a gradual change, but the economic crisis could explain it. Moreover, the number of kilometres travelled on non-urban roads showed a similar reduction.

Although Spain is currently experiencing a serious economic crisis, we believe that its influence on the results of the present study is small, as an analysis of the data up to 2008 (data not shown), when the crisis had barely begun to exert its negative effects, yielded similar findings.

The methods we used did not allow us to determine what fraction of the observed effectiveness was attributable to the reformed penal code or to the stricter enforcement of traffic laws that accompanied the reform. Both probably had an impact, since a law is only effective to the degree to which it is enforced.23 In Spain, the enforcement of traffic laws has been prioritized since 2004. After the reform of the penal code, the number of prosecutions, most of them for driving without a licence and or driving while intoxicated, increased from 43 296 in 2007 to 87 755 in 2008.

Although we were unable to separately analyse the effect of each of the road behaviours penalized, we assessed the overall effect of the reform of the penal code both on drivers, who represent the law’s target population, and on those injured in traffic collisions, who represent the Spanish population at large.

We used Poisson regression models, which yield estimates similar to those obtained with autoregressive integrated moving average (ARIMA) models and with similar goodness of fit. This allowed us to calculate RRs, which permit a straightforward interpretation of the intervention’s effectiveness.24,25

Moreover, the large sample size allowed us to stratify the analysis by relevant variables such as age, sex, type of road user, road type and time of collision. Finally, the long pre-intervention period provided analytical stability.


The results of the present study suggest that criminalizing certain road traffic behaviours can effectively improve road safety by reducing both the number of drivers involved in injury collisions and the number of people injured in traffic collisions. These findings can probably be generalized to other countries that have efficient road traffic administration and that prioritize traffic law enforcement.


We would like to thank Pilar Zori, Fernando Ruiz Cuevas and Marc Marí for their valuable contributions. This paper is part of Ana M Novoa’s doctoral dissertation at Pompeu Fabra University in Barcelona, Spain.

All authors are affiliated with the Institute of Biomedical Research (IIB Sant Pau); Ana M Novoa also works with the PhD Programme in Biomedicine in the Universitat Pompeu Fabra; Carme Borrell, Catherine Pérez and Elena Santamariña also work with the CIBER Epidemiología y Salud Pública (CIBERESP) and Carme Borrell also works with the Universitat Pompeu Fabra.


This work was partially supported by the Agencia Española de Tecnologías Sanitarias (Plan Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica and the Instituto de Salud Carlos III-Subdirección General de Evaluación y Fomento de la Investigación) [PI07/90157] and by the Intensificación de la Actividad Investigadora (Carme Borrell) programme, funded by the Instituto de Salud Carlos III and the Departament de Salut, Generalitat de Catalunya. The views expressed in the publication are those of the authors and not necessarily those of the Agencia Española de Tecnologías Sanitarias, the Instituto Carlos III or the Departament de Salut.

Competing interests:

None declared.