Background To assess the effect of the Demerit Point System (DPS), introduced in Spain on 1 July 2006, on the number of fatalities due to road traffic accidents, using a methodology that controls for the seasonal variation and trend in the data series.
Methods Time-series analysis by ARIMA models of 29 113 fatalities in road traffic accidents (at the accident scene or within 24 h thereafter), between January 2000 and December 2007. The model permitted estimation of an intervention parameter, together with its 95% CI, to calculate the number of fatalities that would have occurred if the DPS had not been implemented, after controlling for the effect of other measures introduced in 2004.
Results It was estimated that 618 persons (95% CI 259 to 977) would have died in traffic accidents in the 18 months after implementation of the DPS had it not been in effect, which represents a reduction of 14.5% (95% CI 6.1% to 23.0%) from a total of 4252 deaths.
Conclusion Implementation of the DPS in Spain has led to a significant reduction in the number of traffic accident deaths in the context of a downward trend after the implementation of the 2004 measures.
- Demerit points system
- road traffic accidents
- trend series analysis
- road safety
Statistics from Altmetric.com
- Demerit points system
- road traffic accidents
- trend series analysis
- road safety
Demerit Point Systems (DPS) have been incorporated in the road safety laws of most developed countries. These systems, aimed primarily at those road users who consistently violate traffic laws,1 involve a loss of points according to the severity of the driver's infraction, with suspension of the drivers' licence when all points have been lost. However, despite their widespread implementation, practically no studies have adequately evaluated the effects of DPS upon traffic accident mortality. The usual practice is to quantify the percentage reduction in the number of fatalities after DPS implementation in comparison with a period before the system was introduced, without considering previous trends and seasonality in the series of deaths.2 3 Therefore, the true effectiveness of DPS remains unclear.
The DPS came into effect in Spain on 1 July 2006,4 accompanied by an extensive and ongoing public awareness campaign in the communications media. Up to January 2008, 4670 drivers had lost their licence, amounting to 8.6/day.
The decline in traffic accident mortality detected in the following months led those responsible for its introduction to evaluate its impact in the way previously described, without taking into account that a downward trend was in fact clearly present in Spain some years before, especially marked from 2004 thereafter, when a strong package of measures (training, education, fines, judicial measures) came into effect that strengthened sanctions on recidivist drivers, including prison sentences and loss of licence for driving under the influence of psychotropic drugs.
The objective of this work is to evaluate the impact of the DPS in reducing the number of traffic accident fatalities in Spain, using an analytical methodology that identifies the seasonal component and time trend in the death data series before the system was introduced.
Material and methods
The data source was the Spanish register of road traffic crashes compiled by the government's General Traffic Directorate. The data set refers to accidents in which at least one person dies at the accident scene or during the 24 h after the accident on any kind of roads except for urban routes, as data in the latter case were not available at the time of the study. The Spanish register of road traffic defines ‘urban routes’ as streets or roads within the city limits. We used fatalities grouped by month from January 2000 to December 2007.
The Box–Jenkins Autoregression Integrated Moving Average (ARIMA) approach was used to describe the time series data, taking into account time trend and seasonality.5 The ARIMA method calculates the probability structure of this series by means of moving average and auto regressive parameters; it permits introduction in the model of a dichotomous intervention variable that takes the value ‘0’ before implementation of the DPS (1 July 2006), and ‘1’ thereafter. A similar method was used to add another dichotomous variable to control for the effect of the measures introduced before January 2004.
The model thus made it possible to estimate an intervention parameter, with its 95% CI, to calculate the number of fatalities that would have occurred if the DPS had not been implemented, while controlling for the effect of the 2004 measures.
The ARIMA models use the notation ‘(p,d,q) (P,D,Q),’ where the coefficients in the first group provide information about the overall evolution of the series, and those in the second group provide information about the seasonal behaviour. The resulting model was validated by the Akaike information criterion and the portmanteau test, accepting a significance of p<0.05.
As can be seen in figure 1, the monthly number of deaths from January 2000 to December 2007 reveals both a downward trend, especially beginning in January 2004 and a marked seasonal component throughout the series, with peaks of deaths in summer and troughs in February.
The model best fitted to the data was ‘ARIMA (0,1,1) (2,0,0)12’ (Akaike information criterion=914.97; p=0.840). The intervention parameter was −34.339. With this value, it was estimated that 618 persons (95% CI 259 to 977) would have died in road traffic accidents in the 18 months after introduction of the DPS (July 2006 to December 2007) if it had not been implemented. As a proportion of the 4252 deaths observed in the same period, this represents a 14.5% reduction in fatalities.
After adjusting for secular and seasonal components and the effect of the January 2004 measures using ARIMA models, mortality from traffic accidents in the 18-month period after introduction of the DPS dropped by 14.5%, which means that 618 persons would have died had the DPS not been implemented.
As stated in the Introduction, evaluation of the effectiveness of the DPS has been the object of little analysis, which limits the comparability of our results with those of previous studies. We have found only one study that applies ARIMA models to estimate the expected number of deaths.6 In this study, performed in Italy, a reduction of 18% in the 18-month period after the introduction of DPS was observed. However, unlike the situation in Spain, the annual number of fatalities recorded in the 4-year period before DPS was stable. In the Italian region of Lazio, Farchi et al observed,7 using Poisson models, a non-significant 4% reduction in the number of road crash related deaths in the 12-month period after DPS.
Our study has some drawbacks, which should be noted. We recorded only deaths occurring within 24 h after the crash, and only on non-urban routes, because data on 30-day mortality and deaths on urban routes were not available from December 2006 on. However, we repeated the analysis for deaths occurring in the 30 days after the crash (ending the temporal series in December 2006), and the results (not shown) were quite similar. Furthermore, we could not disaggregate deaths according to some other influential variables such as age, sex or road user type. Finally, we did not take into account the larger number of vehicles on the roads and increasing fuel consumption in Spain during the study period.8 9
In conclusion, after taking into account the overall trend in road crash mortality in Spain in the years before introduction of the DPS, this measure seems to be related to a reduction in the number of deaths, a fact that adds further evidence to the effectiveness of this type of measure in different countries and settings.
What is already known on this subject
Despite the widespread implementation of the DPS, practically no studies have evaluated its effects on mortality from traffic accidents after controlling for the seasonal component and trend in the data series.
What this study adds
The implementation of the DPS in Spain seems to be related to a reduction in the number of deaths, although this effect has been produced in the context of a strong downward trend brought about by other measures introduced previous.
Our thanks to the Statistical Service of the National Observatory on Traffic Safety for providing the older data on fatalities. We also thank KM Fitch for translation and suggestions.
Funding Supported mainly by CIBER Epidemiología y Salud Pública (CIBERESP AE08_021) and Plan Nacional Sobre Drogas (PNSD/2003).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.