Effects of air pollution on the risk of acute myocardial infarction in Charleroi, Belgium ()
1. INTRODUCTION
Analysis of the relation between atmospheric pollutants and health problems is complex both as regards methodological approaches and the difficulties in approaching data often collected with very different purposes. Major pollutants in the atmosphere are fine particles, CO2, hydrocarbons, NOx, CO and SO2 [1]. One difficulty is; therefore, the consideration of the multiple and varied exposures which are dependent on the time spent inside and outside, active or passive smoking, professional exposure, nutrition, sensitivity and individual sociodemographic characteristics among others. Furthermore, meteorological conditions can have a considerable effect on atmospheric pollution (example: Meuse Valley in 1931 [2]) and have significant consequences on health. However, since a few years ago, researchers who adopt rigorous methodological approaches have analysed the effects of pollutants on mortality and respiratory and cardiovascular morbidity. Thus some European and North American studies confirm that an increase in exposure to an average of PM10 particles of 10 µg/m³ over one year translates into an increase of 16% to 20% of the global mortality rate and an increase of more than 40% in the mortality rate by infarction (aggravated by the other major cardiovascular risk factors) [3-5]. Exposure to shortterm peaks of ozone (O3) also increases the risk of cardiovascular pathologies in adults from 35 to 64 years of age [6].
Currently, air quality is regulated by European directives that aim in particular at setting ambient air quality targets and assessing air quality on a European scale [7]. In Wallonia, air quality is watched by a network of 22 telemetry stations. According to the site, the pollutants measured are SO2, NOx, O3, CO and suspended particles (PM10 and PM2.5) [7].
1.1. Atmospheric Pollution/Health Studies in the Walloon Region
A network of cities was set up on a European level to experiment and apply an epidemiological surveillance tool regarding air pollution (APHEIS). This network pursued and refined its work in several successive projects to assess the contribution of atmospheric pollution to the prevalence of respiratory disease. In Belgium, the APHEIS/EN APHEIS/EN HIS-1 version of the tool was tested within the framework of the National Environmental Health Action Plan (NEHAP) for Belgium at the request of the Ministerial Conference on the Environment and Health in 3 urban areas in Belgium (Liège, Brussels-capital and Antwerp). The study analyses environmental (concentration of PM10 and of ozone) and health data for 2004. The work concluded the feasibility of this type of study for urban areas in Belgium [8]. The last project (APHEKOM) concerned a wider range of health effects such as effects on cardio-respiratory disorders [9].
1.2. The Charleroi Ischaemic Cardiopathies Registry
This health information tool has worked in a standardized way for 27 years. All suspected cases of infarction are detected by a system involving all the hospitals of the entity capable of hospitalizing patients living in Charleroi that have undergone an acute coronary event [10-12]. This research aims at using this registry as a source of data validated according to a standardized protocol MONICA (Monitoring of Trends and Determinants in Cardiovascular Diseases) of coronary accidents and to quantify the short-term link between atmospheric pollution and the risk of myocardial infarction over a period of 10 years.
2. METHODOLOGY
2.1. Air Pollution Data
Concentrations of pollutants (PM2.5, PM10, O3, SO2, NO2, CO) were supplied by the Walloon Air and the Climate Agency (AWAC). The data of 5 telemetric measuring stations were analysed to calculate average concentrations. The normality of variables was visually assessed by means of a normal probability plot. As most daily averages of pollutants are not normally distributed, the nonparametric statistics (medians, percentiles and interquartile range) are presented. However, for comparison with European standards, the averages were also calculated. The comparisons of medians were performed using a Kruskal-Wallis test.
2.2. Myocardial Infarction Data
The methodology is based on the precise criteria defined within the framework of the MONICA protocol for the detection of cases (fatal and non-fatal). Classification of the events (myocardial infarction) is carried out using an algorithm based on three diagnostic criteria: symptoms (the pain felt by patients), cardiac enzymes (the dosage of enzymes, troponins in particular), and electrocardiogram (the ECG reading). The sources of data for the hospitalized patients came from 7 hospitals of the study region.
Since 1998, minimum clinical data (MCD) have allowed additional validation of every suspected case by using the codes of the International Statistical Classification of Diseases and Related Health Problems ICD: 410 - 414 (ischemic heart disorder). For fatal cases, the tenth revision of the ICD is used at present. Death certificates are collected by the Wallonia Brussels federation. All death certificates are validated and verified against the hospital databases and the certificates of the emergency medical aid service (SMUR).
The age variable was grouped into 4 categories (25 - 44, 45 - 54, 55 - 64 and 65 - 69 years). Analyses were carried out for a period of 10 years from 1999 until 2008 and are presented in five periods of two years (1999- 2000, 2001-2002, 2003-2004, 2005-2006, and 2007-2008). Certain analyses were spread to the 70 - 74 years category for the 2004-2008 period.
The evolution of attack rates was taken from community data (hospitalized cases and non-hospitalized cases). The descriptive statistics (proportions, medians and interquartile range) were used to describe the sample. Pearson’s chi2 test and the Kruskall-Wallis test were used to compare percentages and age respectively.
Fatal attack rates correspond to all the cases of infarcttion in which death occurred within 28 days of the coronary event. Attack rates were standardized (direct standardization) in comparison with the population of 2004. These analyses were performed on all hospitalized and non-hospitalized cases.
2.3. Measures of Association between Atmospheric Pollution and Myocardial Infarction
To measure the association between atmospheric pollution and the number of cases of myocardial infarction, Poisson regression was carried out. Non-dispersal of the data was verified (mean, variance). In the case of suspected overdispersion, the results were confirmed using robust Poisson regression or by negative binomial regression. A chi2 test was performed to verify the goodness of fit of the data. Graphs of the number of cases of infarction according to pollutant concentration categorized in 25-quantiles were also plotted. The excess risk of myocardial infarction was analysed according to the month of admission. The cold months are December, January and February; the warm months are June, July and August. The tests were considered statistically significant when the p-value was lower than 0.05. The statistical analyses were performed using Stata 10.1 software (StataCorp LP, TX, United States) or R version 2.15.0 (R Foundation for Statistical Computing).
3. RESULTS
3.1. Analyses of Atmospheric Pollution and the Environmental Data
Except for PM2.5 and ozone, the number of missing data concerning the average of the various stations is lower than 10 a year and lower than 20 for all 11 years (on 1999 in 2009). For ozone, measured only at one station, the number of missing data reached 56 in 2003 and the total is 239 for all 11 years. PM2.5 was only analysed in 2008 and 2009.
During the period of the study, the annual limit value of 40 µg/m3 for PM10 is surpassed in 2003, 2006 and 2007. For this pollutant, the authorized number of exceedances of the daily limit value was surpassed from 1999 to 2007. The years 2008 and 2009 seem markedly less polluted as regards PM10 with average concentrations of about 25 µg/m3 and less than 35 exceedances. This improvement is in accordance with the closure of a polluting enterprise and a review of environment licences (obligation for certain companies to install filters in chimneys). The European standards are observed for the other pollutants.
The concentration of ozone seems stable during the study period whereas a slight decrease is noticed for nitrogen dioxide and carbon monoxide. A net decrease in the concentration of sulphur dioxide is observed. This concentration fell from about 10 µg/m3 for 1999-2001 to approximately 3 µg/m3 for 2005-2009 (Table 1). Having only been analyzed since 2008, it is difficult to draw conclusions on the evolution of the concentration of PM2.5.
The long-term trends of the association between myocardial infarction and the concentration by polluting are confirmed by the graph and by linear regression. The standardized coefficients of regression are 0.09, 0.02, 0.12, 0.23 and 0.43 respectively for PM10, ozone, nitrogen dioxide, carbon monoxide and sulphur dioxide. These coefficients thus confirm the strongest decrease in the concentration of sulphur dioxide.
The warmest months are June, July and August with median temperatures of 16.3˚C, 17.7˚C and 17.1˚C respectively. The coldest months are December, January and February with median temperatures of 3.7˚C, 4.4˚C and 4.4˚C respectively. With the exception of ozone, the concentrations of pollutants (PM10, NO2, CO and SO2) are higher during the cold months than during the warm months. These differences are statistically significant for all the analysed pollutants (Kruskall-Wallis test, p < 0.001) as well as for temperature (Table 2).
Most of the pollutants vary according to season. With the exception of O3, concentrations are generally higher in winter. This seasonal effect is a consequence of less favourable weather conditions for the dispersal of pollutants (example: temperature inversion phenomenon) but also greater emissions in winter than in summer [4].
Table 1. Average values in pollutants in Charleroi (Walloon Region) according to the year.
Table 2. Variation of the concentrations in pollutants and meteorological parameters according to the season.
3.2. Analyses of Morbidity/Mortality by Myocardial Infarction in Charleroi
The data of 3139 community myocardial infarctions were analysed in five periods of two years to cover the period ranging from 1999 until 2008 (Table 3).
The median age of the cases is 57 years (IQR: 16 years). Women represent 25.5% [CI 95%: 24.0; 27.1] of all myocardial infarctions. The percentage of women as well as their median age remained stable over the ten years. At the time of the coronary event, women are older, on average, than men [60 against 56 years, p < 0.001]. With the exception of the first period of analysis, a higher proportion of cases is observed for the 55 - 64 years age group. The analysis of attack rates (fatal and non-fatal) shows an increase according to age category with significant disparity between men and women. Men present very high rates in the 55 - 64 years age range and even more so in the 70 - 74 years range. Women also present high rates in the 65 - 69 years and 70 - 74 years age categories (Table 4).
Table 4 presents the evolutions of standardized attack rates for men and women from 1999 to 2008. Myocardial infarction attack rates decrease over 10 years both for men and women. Although the attack rate in women remains systematically lower by more than 50% in comparison with men, we observe a slower decrease in this category over time (from 14.2 to 10.4 per 10,000 inhabitants from 1999 to 2008). In men, the rates of attack remain high (between 27.6 and 22.4 per 10,000 inhabitants from 1999 to 2008) in spite of a progressive year-on-year decrease.