Man , road and vehicle : risk factors associated with the severity of traffi c accidents

OBJECTIVE: To describe the main characteristics of victims, roads and vehicles involved in traffi c accidents and the risk factors involved in accidents resulting in death. METHODS: A non-concurrent cohort study of traffi c accidents in Fortaleza, CE, Northeastern Brazil, in the period from January 2004 to December 2008. Data from the Fortaleza Traffi c Accidents Information System, the Mortality Information System, the Hospital Information System and the State Traffi c Department Driving Licenses and Vehicle database. Deterministic and probabilistic relationship techniques were used to integrate the databases. First, descriptive analysis of data relating to people, roads, vehicles and weather was carried out. In the investigation of risk factors for death by traffi c accident, generalized linear models were used. The fi t of the model was verifi ed by likelihood ratio and ROC analysis. RESULTS: There were 118,830 accidents recorded in the period. The most common types of accidents were crashes/collisions (78.1%), running over pedestrians (11.9%), colliding with a fi xed obstacle (3.9%), and with motorcycles (18.1%). Deaths occurred in 1.4% of accidents. The factors that were independently associated with death by traffi c accident in the fi nal model were bicycles (OR = 21.2, 95%CI 16.1;27.8), running over pedestrians OR = 5.9 (95%CI 3.7;9.2), collision with a fi xed obstacle (OR = 5.7, 95%CI 3.1;10.5) and accidents involving motorcyclists (OR = 3.5, 95%CI 2.6;4.6). The main contributing factors were a single person being involved (OR = 6.6, 95%CI 4.1;10.73), presence of unskilled drivers (OR = 4.1, 95%CI 2.9;5.5) a single vehicle (OR = 3.9, 95%CI 2,3;6,4), male (OR = 2.5, 95%CI 1.9;3.3), traffi c on roads under federal jurisdiction (OR = 2.4, 95%CI 1.8;3.7), early morning hours (OR = 2.4, 95%CI 1.8;3.0), and Sundays (OR = 1.7, 95%CI 1.3;2.2), adjusted according to the log-binomial model. CONCLUSIONS: Activities promoting the prevention of traffi c accidents should primarily focus on accidents involving two-wheeled vehicles that most often involves a single person, unskilled, male, at nighttime, on weekends and on roads where they travel at higher speeds. D ESCRIPTORS: Accidents, Traffi c, mortality. Risk Factors. Hospital Information Systems. Mortality Registries. Urban Population. 2 Man, road and vehicles: accident risk Almeida RLF et al Traffi c accidents remain a signifi cant public health problem in Brazil and preventative activities require different approaches. The dynamic of this multi-causal phenomenon affects victims to different degrees depending on the type of accident (run over pedestrian, motorbike accident or another type of accident involving a vehicle or motorbike)1 and demographic characteristics (sex, age, skin color, marital status and schooling).12,20 Monthly distribution is differentiated by day of the week and time of day.5 Research using analysis techniques to understand these characteristics are relevant to redirecting activities aiming to reduce the severity of the accidents. The factors involved, implicitly or explicitly, which can contribute to a greater or lesser extent to casuistry are: man; vehicle; road and environment; and factors referring to legislation and complying with it.4 Separating these factors and studying their associations is necessary in order to better understand and intervene in the phenomenon of traffi c accidents. This is because these factors combined may increase the likelihood of accidents in different ways in specifi c areas.a,b A signifi cant number of studies in Brazil have looked at fatalities due to traffi c accidents. However, there are few studies that deal with survivors of these accidents, which has led some authors to emphasize the importance of research into non-fatal accidents.9,10,14,18,c In addition to being scarce, these studies make use of different methodologies different sources of data. A large part of this output refers to problems with the coverage and quality of the offi cial information systems as limiting factors to understanding this phenomenon.20 For specialists, the lack of an integrated information system based on standardized police reports of traffi c accidents makes it impossible to really know the situation of traffi c accidents in the country and, therefore, makes it impossible to put in place appropriate measures to mitigate it.7 As a way of overcoming this diffi culty, techniques to relate databases were used in order to improve the quality of information on the number of variables investigated and the number of valid records, achieving more complete information. This study aimed to describe characteristics of the victims, the roads and the vehicles involved in traffi c accidents and risk factors for fatal accidents. INTRODUCTION

Traffi c accidents remain a signifi cant public health problem in Brazil and preventative activities require different approaches.The dynamic of this multi-causal phenomenon affects victims to different degrees depending on the type of accident (run over pedestrian, motorbike accident or another type of accident involving a vehicle or motorbike) 1 and demographic characteristics (sex, age, skin color, marital status and schooling). 12,20Monthly distribution is differentiated by day of the week and time of day. 5 Research using analysis techniques to understand these characteristics are relevant to redirecting activities aiming to reduce the severity of the accidents.
The factors involved, implicitly or explicitly, which can contribute to a greater or lesser extent to casuistry are: man; vehicle; road and environment; and factors referring to legislation and complying with it. 4Separating these factors and studying their associations is necessary in order to better understand and intervene in the phenomenon of traffi c accidents.This is because these factors combined may increase the likelihood of accidents in different ways in specifi c areas.a,b A signifi cant number of studies in Brazil have looked at fatalities due to traffi c accidents.c In addition to being scarce, these studies make use of different methodologies different sources of data.A large part of this output refers to problems with the coverage and quality of the offi cial information systems as limiting factors to understanding this phenomenon. 20r specialists, the lack of an integrated information system based on standardized police reports of traffi c accidents makes it impossible to really know the situation of traffi c accidents in the country and, therefore, makes it impossible to put in place appropriate measures to mitigate it. 7As a way of overcoming this diffi culty, techniques to relate databases were used in order to improve the quality of information on the number of variables investigated and the number of valid records, achieving more complete information.
This study aimed to describe characteristics of the victims, the roads and the vehicles involved in traffi c accidents and risk factors for fatal accidents.

INTRODUCTION METHODS
A non-concurrent cohort study of traffi c accidents that occurred within the geographical limits of Fortaleza, CE, Northeastern Brazil, between January 1 st 2004 and December 31 st 2008.The incidents covered those registered in Police Incident Reports (PI) recorded in the Fortaleza Traffi c Accidents Information System (SIAT-FOR) and run by the Fortaleza Municipal Transit Authority (AMC).A traffi c accident was deemed to be any incident that occurred on a public road, including not only crashes between vehicles but also collisions with fi xed objects, collisions between pedestrians and cyclists, vehicles overturning, running over pedestrians and vehicles leaving the road. 17Criteria for inclusion were the accident being recorded in a PI, with or without injuries or fatalities among the victims and accidents that were not recorded in a PI were excluded.
Fortaleza is a metropolitan agglomeration in the Northeast of Brazil and is classed as the fi fth largest city in the country.The urban area is crisscrossed by a 3,700 km network of roads.Of this total, around 35 km are under state jurisdiction and 25 km under federal jurisdiction.d The road layout follows a radio-centric pattern and constituted the main connections between the urban area and neighboring municipalities.The roads are narrow as they originate in a layout defi ned when the city was founded and due to their use not being managed and a lack of control in occupation of the terrain, this makes it unviable to extend them and the system has become insuffi cient.e Data were collected from the SIAT-FOR system, which includes ten bodies involved in dealing with traffi c accidents in the municipality.

Data from the Habilitação e Veículos do Departamento Estadual de Trânsito (DETRAN-CE -State Traffi c
Department Driving Licenses and Vehicle) database were also used, as were data from the Informações de Mortalidade (SIM -Mortality Information System), which contains death certifi cates and data from the Sistema de Informações Hospitalares (SIH -Hospital Information System), with the aim of obtaining complementing the data.Databases on hospitalization and deaths (SIH and SIM) were accessed through the Ceará State Department of Health (SESA).The data from the SIM refer to all deaths that occurred in Ceará between a Raia Jr AA, Santos L. Acidente zero: utopia ou realidade?15º Congresso brasileiro de transporte e trânsito.2005; Goiânia, BR, Goiânia: Centro de Convenções de Goiânia; 2005.p.7. b Raia Jr AA.Identifi cação de pontos críticos de acidentes de trânsito no Município de São Carlos, SP, Brasil: análise comparativa entre um banco de dados relacional -BDR e a técnica de Agrupamentos pontuais.Anais do 2º Congresso Luso Brasileiro para o Planejamento, Urbano, Regional The databases were linked using relationship techniques to create one single database containing information on the vehicle, the road, the driver and the individuals involved.
Two methods were used in this process: deterministic and probabilistic relationship techniques. 6,13ta on the accidents collected by the SIAT-FOR system were complemented according to the two types of relationship, involving two different stages.In the first, deterministic, stage the SIAT-FOR database, provided by the AMC, regarding vehicles which had been involved in traffi c accidents, was linked to the DETRAN-CE database using vehicle license plates to obtain information on the year and type of vehicle.Data on the driver: type and year of driving license, sex, marital status, level of schooling, date of birth and mother's name were obtained from the database provided by DETRAN-CE based on the driving license number, registered in the SIAT-FOR system database of individuals involved in traffi c accidents.
Complementing the data concerning pedestrians and passengers involved in traffi c accidents was achieved by using probabilistic relationship technique, together with the SIM and SIH systems.The key variables were: name and age of victim and date of the accident when the individual died or was hospitalized in the SUS network.
In order to analyze the traffi c accidents, a set of co-variables was created based on the variables described: type and number of vehicles involved, type and number of individuals, age of the vehicles, sex of the driver, status of the drivers' licenses, length of time drivers had been driving and their age, marital status and schooling.
Descriptive data were presented according to the variables in question.Pearson's Chi-squared test, Fisher's exact test and the Student t-test were used in the comparative analysis.Estimated risk of a fatal accident was verifi ed with bivariate analysis using the odds ratio, with a 95% confi dence interval.Variables which had an association with a fatal accident of p < 0.20 according to the Chi-squared test were selected to be included in the multinomial analysis. 11e multinomial analysis was carried out using the Generalized Linear Model (GLM) with binomial distribution and the logistic link function.The choice of this distribution was justifi ed as the measurement of the outcome was dichotomous.
The modelling followed the strategy recommended by Hosmer & Lemeshow 11 and each variable was withdrawn after comparison of the models' likelihood ratio (-2logL) with and without the variable in question.Variables remained in the model according to theoretical justifi cations and statistical signifi cance.
The fi nal model was assessed for sensitivity, specifi city accuracy and based on the percentage improvement of the model in relation to the initial deviance (likelihood ratio).The value of the area above the ROC curve was 0.86, indicating high discriminatory power.

RESULTS
There were 118,830 traffic accidents in Fortaleza between January 2004 and December 2008, 1.4% were fatal and 46.6% involved serious or minor injuries.Of the incidents, 78.1% were categorized as a cash/ collision and 11.9% were pedestrians run down, 3.9% were collisions with a fi xed obstacle, 0.5% involved the vehicle overturning and 5.6% involved the vehicle falling/leaving the road and other types of accidents on public roads (Table 1).
The annual mean number of accidents in the period was 23,767 accidents/year, and the highest number of incidents was recorded in 2008 (20.8%; p < 0.001).The monthly mean for the period was 1,981 accidents/month.The quarter from October to December stands out as having the highest number of accidents (mean of 6,382 accidents), while the January to March quarter had the lowest number of accidents (mean of 5,421 accidents; p < 0.001) (Table 1).
Around 43.7% of the accidents occurred between cars and/or trucks, 26.5% involved motorcyclists and 8.5% were between motorcyclists.Accidents involving more than two vehicles were in the minority throughout the study, especially in 2005 (4.7%; p < 0.001) (Table 1).
In 75.1% of the accidents, only one driver of a motorized vehicle was involved.Accidents involving passengers in      In the multinomial analysis, the presence of drivers who do not have a license (OR = 4.1; 95%CI 2.9;5.5) or do not have a license appropriate for the vehicle (OR = 1.6; 95%CI 1.2;1.9),using roads that are under federal jurisdiction (OR = 2.4; 95%CI 1.8;3.1),early hours of the morning (OR = 2.4; 95%CI 1.8;3.0)and Sundays (OR = 1.7; 95%CI 1.3;2.2) all stand out as contributing factors to fatal accidents (Table 3).

DISCUSSION
Brazil is moving onto the world stage due to its promising economic growth, however, morbidity and mortality due to traffi c accidents is recognized as a large-scale and highly complex phenomenon.It represents the relationship between investments in road safety, economic development policies centered around the automotive industry and traffi c education.
Analyzing the factors which affect the occurrence of traffi c accidents is a complex procedure as they are numerous and they are not independent. 8The results of this study provide a broader view of the phenomenon of traffi c accidents based on its analysis of characteristics of the road, the individuals and the vehicles involved.This aspect highlights the importance of inter-sector practices to better deal with the problem, in view of its complexity and the multiplicity of factors related to diverse areas of human knowledge.
Road structure, signs and illumination, the day of the week and time of day of the occurrence are connected to the severity of traffi c accidents.Researchers have found similar results and attribute higher fatality rates to weekends and the early hours, to drinking and driving and speeding. 2On the other hand, it is important to bear in mind the role of poor quality of lighting and adequate signage as a parameter of safety. 16verity of the accident is related to the type of jurisdiction of the road, with higher risks at a federal level, followed by state, compared to municipal.The speed limit on each type of road and the fl ow of traffi c differs and lead to congestion on municipal roads, leading to less serious accidents without injuries.
Drivers who have held a license for fewer than fi ve years have the highest risk of being involved in a fatal accident, in contrast with the fi ndings of other studies, which highlight older motorists as signifi cantly more likely to be involved in serious and fatal accidents, when fi gures are adjusted for differences in exposure. 22his information places the Brazilian process of issuing driving licenses under discussion.Lack of experience on the part of new drivers is an implicit criticism of the rigidity of the traffi c code which stipulates that a learner's permit for up to a year is not suffi cient to make them fi t to drive vehicles.
Traffi c, the condition of the road and the greater fl ow of individuals commuting increase exposure to the problem.Added to this is the culture of punishment at the expense of education around this phenomenon.These obstacles impede an improvement in the indicators, despite the implementation of public policies which attempt to mitigate the phenomenon.The more vulnerable population groups (pedestrians, cyclists and motorcyclists) become victims of the conditions of the road, the vehicles and the road users.
There are differences in the seriousness of traffic accidents according to type.Running over pedestrians and accidents involving cyclists and motorcyclists are described as more serious. 1This fact is explained by the kinematics of trauma, and will continue to be a problem until equality exists in traffi c.
The individual's position in the traffi c is a determinant of the seriousness of the injury.The risk of death is higher among cyclists and pedestrians.This fact has also been presented in other studies, in which pedestrians and cyclists are the most vulnerable road users and make up the highest percentage of victims. 19Whereas cyclists lack proper cycle lanes and need to struggle with the vehicles for space amid the oppression and the fumes, pedestrians are faced with narrow poorly maintained sidewalks.This occurs despite Fortaleza being a fl at city where these methods of transport should, for diverse reasons, be prioritized: health, environment and economy.
Congestion in the large cities, ineffi cient public transport, tele-deliveries and mototaxis have led to the rapid spread of motorcycle use, representing an increase of 700% in fatalities between 1998 and 2008. 3Their low cost and facilitated fi nancing are responsible for the signifi cant increase in motorcycles.The problem of the severity of this type of accident, as evidenced in this and other studies, raises the need to think about the safety of these road users.
Accidents in which only one vehicle was involved had a higher risk of fatality compared with accidents with two vehicles.Confl icts in traffi c with more susceptible vehicles such as motorbikes and bicycles can lead to the driver falling.According to the kinematics of trauma, the injury is more severe where there is a greater transference of kinetic energy.This was proved in this study, in which the main type of accidents involving fatalities were those involving collisions with fi xed objects and running down pedestrians.Speed is the most signifi cant producer of this energy.
Both severity and incidence of accidents increase during the night and at the weekends, due to lack of congestion and drinking and driving, in view of the devastating effect of combining drink driving and speeding.
A difference was observed in the distribution of occurrence according to month, day of the week and time. 5Socio-demographic factors such as sex, age and schooling are related to the occurrence and the severity of accidents. 12,20dividuals who are single are at greater risk than married individuals of suffering a serious or fatal traffi c accident, even after adjusting for sex, age and alcohol, 21 which may be due to the fact that singles expose themselves more to risk factors, as confi rmed in this study.The times and roads on which fatal accidents occur suggest less congested driving conditions and travelling connected to leisure activities, characterized by imprudence.This indicates the need for investment in preventing accidents and promoting safe driving through educational strategies, creating a culture of peace in the traffi c.
Activities promoting the prevention of traffic accidents should prioritize and focus on accidents involving two wheeled vehicles, which often involve only one victim, on unqualifi ed drivers, males, night time, weekends and on roads where higher speeds are reached.
It is possible to unify various sources of data from different sectors in order to improve understanding of traffic accidents so as to produce inter-sector public policies which aim to reduce deaths from this problem.This study's analysis of the characteristics of risk in the accident, considering the individuals and the roads where the vehicles drive was an important contribution for increasing the number of factors related to this phenomenon.
A possible limitation of this study was using secondary data.However, we believe that the techniques used to link the databases are an innovation in dealing with factors which affect the severity of traffi c accidents.
The effort in integrating the different databases used and the results achieved go some way to substituting the need for establishing a unifi ed information system which includes the variables necessary for analyzing the traffi c situation in Brazil.
January 2004 and March 2009, as accidents occurring in December may have led to death in 2009.Data on hospitalization in the Brazilian Unifi ed Health System (SUS) network in Fortaleza are for between January 2004 and July 2009, as requests for payment can be presented up to six months after being authorized.

Table 1 .
Absolute and relative frequency of traffi c accidents according to the variables studied and the year in which they occurred.Fortaleza, CE, Northeastern Brazil, 2004 to 2008.

Table 2 .
Bivariate analysis to investigate factors related to roads, vehicles and time associated with fatal traffi c accidents.Fortaleza, CE, Northeastern Brazil, 2004 to 2008.

Table 3 .
Bivariate analysis to investigate factors related to the roads, the vehicles and the time linked to fatal traffi c accidents.Fortaleza, CE, Northeastern Brazil, 2004 to 2008.