Motor vehicle traffi c and cardiovascular mortality in male adults

MÉTODOS: Foram analisadas informações sobre vias e volume de tráfego no ano de 2007 fornecidas pela companhia de engenharia de tráfego local. Mortalidade por doenças do aparelho circulatório no ano de 2005 entre homens ≥ 40 anos foram obtidas do registro de mortalidade do Programa de Aprimoramento de Informações de Mortalidade do Município de São Paulo, SP. Dados socioeconômicos do Censo 2000 e informações sobre a localização dos serviços de saúde também foram coletados. A exposição foi avaliada pela densidade de vias e volume de tráfego para cada distrito administrativo. Foi calculada regressão (α = 5%) entre esses indicadores de exposição e as taxas de mortalidade padronizadas, ajustando os modelos para variáveis socioeconômicas, número de serviços de saúde nos distritos e autocorrelação espacial.


INTRODUCTION
Air pollution is one of the top environmental risks with greater impact on health due to its ubiquitous and pervasive nature.A recent study estimated the impact on overall mortality of exposure to ozone (O 3 ) and particulate matter ≤2.5 μg/ m³ (PM 2,5 ) from human-related (anthropogenic) sources of pollution. 1Exposure to O 3 was associated with estimated 0.7 (standard deviation [SD] = 0.3) million deaths from respiratory causes and 6.3 (SD = 3.0) million years of life life lost annually, while exposure to PM 2,5 was associated with 3.5 (SD = 0.9) million cardiorespiratory deaths and 220,000 deaths (SD = 80,000) from lung cancer (30 [SD = 7.6] million years of life lost) annually.
Children, elderly and individuals with cardiopulmonary conditions are the most vulnerable to air pollution.Every year approximately 5% of deaths from respiratory causes among children (≤5 years of age) and elderly (≥65 years of age) can be attributed to particulate matter (PM 10 ) pollution in seven Brazilian cities: Belo Horizonte (Minas Gerais State), Curitiba (Paraná State), Fortaleza (Ceará State), Porto Alegre (Rio Grande do Sul State), Rio de Janeiro (Rio de Janeiro State), São Paulo (São Paulo State) and Vitória (Espírito Santo State). 810][11][12] though the adverse effects of different pollutants on the cardiovascular system are not yet clear.
Air pollution may cause oxidative stress in the respiratory epithelium when pollutants (PM or O 3 ) have direct contact with alveolar epithelial cells that are not covered by the epithelial lining fl uid or with other cells such as macrophages.Oxidative stress induces cell apoptosis and triggers inflammation.This inflammation have cardiovascular effects including blood hypercoagulation, atherosclerosis progression and increased platelet breakdown 13 leading to acute atherosclerotic and ischemic cardiovascular complications. 2,3Air pollution also affects the autonomic system causing cardiac arrhythmias. 13tor vehicle traffi c is a major contributing factor to air pollution in the city of São Paulo with a fl eet of 7,012,795 motor vehicles reported as of March 2011.a These vehicles account for 97% of emissions of carbon monoxide (CO), 97% of hydrocarbons, 96% of nitrogen oxides, 40% of PM and 32% of sulfur oxides, as well as resuspension of soil dust and formation of secondary aerosols in the particulate matter.b Motor vehicles are the main cause of air pollution and higher concentrations of pollutants are seen near busy road with high traffi c and they gradually decline with distance. 14Thus, many studies have used indirect methods to assess exposure based on road and traffi c data, which allows to measure the impacts of vehicle emissions on health of the exposed population and can provide input for developing emission control measures.
The current study aimed to assess the association between indicators of exposure to motor vehiclerelated air pollution and cardiovascular mortality in male adults.

METHODS
The study was conducted in São Paulo (Southeastern Brazil), a city with a population of 11,324,102 inhabitants, high population density (≈7,500 inhabitants per km 2 ), an area of 1,509 km² and 98.9% urbanization.The city is divided into 96 administrative districts grouped into west, north, east, south and central areas.
Mortality information was obtained from the Program for Improvement of Mortality Data of the City of São Paulo (PROAIM) database.This database includes information from death certifi cates such as age, gender, and area of residence of individuals at the time of their death, in addition to their primary cause of death coded according to International Classifi cation of Diseases and Related Health Problems, Tenth Revision (ICD-10).Deaths in men ≥40 years of age that occurred in 2005 where the underlying cause was diseases of the circulatory system (ICD-10 Chapter I00 to I99) were selected.Mortality rates from cardiovascular diseases for each administrative district were estimated (per 1,000 inhabitants).The mortality rate was standardized by age group (standardized mortality rate, SMR) using direct standardization with mortality rates in the city as reference.
Population and socioeconomic data obtained from the 2000 Population Census conducted by the Brazilian Institute of Geography and Statistics including total population by gender and age group and average monthly income of the head of the household were grouped by administrative districts.
It was assumed that increased access to health services could have an impact on mortality rates from cardiovascular diseases in the population.Data on health services was collected from the Digital Map of the City of São Paulo database for the construction of the variable inhabitants per health service (HS).HS were public or private facilities including primary care units, outpatient care units, general emergency rooms, health centers, hospitals, general hospitals, single or and multispecialty medical centers.Data on road system including traffi c count and simulation of the main roads (collector and arterial roads and rapid transit corridors) for 2007 were provided by the Traffi c Engineering Company of São Paulo.Charts with road networks, average daily number of vehicles, and traffi c type (light/heavy vehicle) were also provided.Since this information was grouped by administrative districts any spatial autocorrelation between districts could affect the model's explanatory power.Thus, the estimates included the spatial structure in the potential association between SMR and other independent variables.Spatial regression was performed (α = 5%).
Spatial regression analysis takes into account spatial dependence by adding a new term to the regression model in the form of spatial relationship for the dependent variable, c expressed as: where Y is the dependent variable, ρ is the spatial autoregressive coeffi cient, W is the neighborhood matrix and the WY product expresses the spatial dependence in Y.
The null hypothesis for non-spatial autocorrelation is ρ = 0.
The neighborhood matrix W (n x n), a component of the model, estimated spatial variability of data from a set of n districts {D 1 ,..,D n }, where each element w ij represents a measure of proximity between D i e D j .This measure of proximity was calculated using the following criteria: Regression analyses between the dependent variable (SMR) and independent variables (per capita monthly income of people living in the districts and inhabitants per HS) were performed in the univariate models.The variables with p≤0.2 were retained in the multivariate models including road density and traffic volume (overall, light and heavy traffi c).The analyses were performed in GEO Data Analysis.

RESULTS
The correlation between traffi c volume and road density was modest but statistically signifi cant (p<0.001),mainly for light vehicle traffi c.The correlation between road density and heavy traffi c volume was weak and not statistically signifi cant (Figure 1).The highest road density was found in central districts (República, Bela Vista, Sé, Santa Cecília, Bom Retiro and Consolação).
The southernmost districts of the city showed the lowest road density (Marsilac, Parelheiros, Grajaú) and the lowest traffi c volume as well.The heaviest traffi c volumes were not seen in central districts but rather in those near major highways of the city (Figure 2).
A total of 31,476 deaths were reported in men ≥40 years in 2005, 35.2% from cardiovascular diseases.There were excluded 11.5% of the records due to missing information on district of residence.The SMR ranged from 1.37 deaths per 1,000 in the district of Marsilac (south of the city) to 8.64 deaths per 1,000 in Pari (central area).Per capita income decreased with distance from the city center (Figure 2).
The single variable that was signifi cantly associated with SMR in the univariate spatial regression was average per capita income.Road density and traffi c volume were not statistically signifi cant (Table 1).
Monthly per capita income and inhabitants per HS were included in the multiple spatial regression analysis.
Only monthly per capita income remained statistically signifi cant and was retained in the fi nal model including road density and traffi c volume (overall, light and heavy).After adjustment for monthly per capita income the strength of association between road density and mortality from cardiovascular diseases increased (p=0.017),showing that a 10-km increase of road/km² was associated with an increase in SMR of the district by about 1 death/1,000 inhabitants.The multivariate models including overall, light and heavy traffi c volumes were not statistically signifi cant even after adjustment (Table 2).

DISCUSSION
The current study showed an association between road density in the district of residence and SMR for cardiovascular diseases in male adults in the city of São Paulo.This result is in agreement with the literature which suggests a relationship between motor vehicle-related air pollution and increased morbidity and mortality rates from cardiovascular disease.highest quartile of exposure to road traffi c up to 300 m from the residence was 32% higher (95%CI: 6;65) than that found in the lowest quartile.
Maheswaran and Elliott 7 investigated mortality from stroke in the UK and found an relative risk (RR) of 1.7 (95%CI: 1.4;1.9) of death from stroke in men living within 200 m of a major road compared with those living at a distance of more than 1,000 m.
Medina-Ramon et al 10  In Germany there was found an association between coronary heart disease and exposure to vehicle emissions in people living up to 150 m from a main road compared to those living >150 m (OR = 1.85 [95%CI: 1.21;2.84]). 5But no association was found with PM 2,5 measured by monitoring stations, which stresses the importance of using indirect measures to assess motor vehicle-related air pollution.
In a case-control study conducted in Boston, U.S., 12 the interquartile increase in traffi c near the residence was associated with OR of 1.4 (95%CI: 1.02;1.07)for acute myocardial infarction.For this outcome, living near highways showed an OR of 1.05 (95%CI: 1.03; 1.06).
One of the limitations of the current study was not analyzing data about behaviors that might affect the development of cardiovascular diseases such as smoking.However, in their study, Kan et al 6 adjusted for risk factors (smoking, obesity, cholesterol, hypertension) but it did not signifi cantly change their results.
A potential confounder controlled for in the analysis was socioeconomic condition.Lower socioeconomic condition has been associated with higher prevalence of smoking, 11 obesity and lower schooling and, consequently, less prevention and lower access to health care. 7his may explain the statistically signifi cant association between average monthly per capita income in the districts and mortality rates from cardiovascular diseases.
The use of place of residence as an indicator of exposure to motor vehicle-related air pollution is another limitation of the study since it may not be the place where people spent most of their time and therefore does not refl ect the actual exposure scenario, 4 and may result in exposure misclassifi cation.are conducted by the Traffi c Engineering Company every 10 years and the most current and complete information at the time of the study was from 2007.However, we believe that there have been no signifi cant changes in traffi c distribution in the last two years, and a possible increase in vehicle fl ow during this period was likely consistent in the city.
No association between traffi c volume and SMR was seen, even after stratifying by traffi c type (light/heavy).This can be partly attributed to limited accuracy and timeliness of traffi c count and simulation, which are estimated as annual averages disregarding potential daily and annual variations in traffi c fl ow.Another explanation could be low speed of vehicles in more congested, narrower central roads with emission of high volume of pollutants despite low vehicle fl ow.
Traffi c estimates were stratifi ed by vehicle type (light and heavy) and thus allowed to identify the type of fuel used, as heavy vehicles use diesel while light vehicles generally use gasoline or ethanol in Brazil.This was a limitation pointed out by Kan et al 6 as their traffi c data did not have this level of detail.
The association between living near major roads and mortality from cardiovascular diseases can be attributed to traffi c noise as both exposures (traffi c noise and pollution) are concurrent. 9It is likely that they may have an effect at the same time, or one is confounding the other in cardiovascular disease-related outcomess. 4he measures used do not refl ect weather conditions that may infl uence the interaction between air pollutants and their entrainment.
Although no association between traffi c volume and SMR for cardiovascular diseases was found, the study results obtained from road density data in the districts support the use of this information to construct indirect measures of exposure.These measures can be especially useful in sites where there is no air quality monitoring while investigating potential negative impacts of motor vehicle traffi c on people's health.
Two indirect indicators of exposure to motor vehiclerelated pollutants were constructed for each administrative district using MapInfo Professional geographic information system (GIS) software (version 8.5, MapInfo Corporation, New York, NY, USA): a Departamento Estadual de Trânsito.Frota de Veículos em SP, por tipo de veículo.DETRAN-SP.São Paulo; 2011[cited 2011 Apr 07].Available from: http://www.detran.sp.gov.br/b Companhia de Tecnologia de Saneamento Ambiental.Relatório de qualidade do ar no estado de São Paulo 2009.São Paulo; 2010[ 2011 Jan 01].Available from: http://www.cetesb.sp.gov.br/ar/qualidade-do-ar/31-publicacoes-e-relatorios• Road density: sum (km) of arterial and collector roads and rapid transit corridors divided by the area (km²) of an administrative district.• Traffi c volume (subdivided by light and heavy vehicles): sum of traffi c fl ow divided by the length of arterial and collector roads and rapid transit corridors (in km) of an administrative district.

Table 1 .
Means and standard deviations of independent variables, r² and coeffi cients obtained in the spatial regression analysis.São Paulo, Southeastern Brazil, 2007.