Data on short-term effect of nitrogen dioxide on cardiovascular health in Wallonia, Belgium

Data presented in this article are related to the research paper entitled “Short-term effects of nitrogen dioxide on hospital admissions for cardiovascular disease in Wallonia, Belgium.” (Collart et al., in press) [1]. Nitrogen dioxide concentrations showed a strong seasonal pattern with higher levels in the cold period than in the warm period. A minimum of 13.1 µg/m3 in July and a maximum of 26.9 µg/m3 in January were observed. The coldest months are December, January and February and the hottest months are June, July and August. Temperature and nitrogen dioxide were negatively correlated in the cold period and positively correlated in the warm period. For the period 2008–2011 there were 113 147 hospital admissions for cardiovascular disease. Forty-five percent of patients were women and 66.5% were 65 and older. Heart rhythm disorders account for the majority of hospital admissions for cardiovascular disease. Our data confirms the existence of an association between NO2 and cardiovascular disease. Apart from haemorrhagic stroke, the strongest association between NO2 concentrations and number of hospital admissions is observed at lag 0. For haemorrhagic stroke, the association is strongest with a delay of 2 days. All associations calculated without stratification are statistically significant and range from an excess relative risk of 2.8% for myocardial infarction to 4.9% for haemorrhagic strokes.


a b s t r a c t
Data presented in this article are related to the research paper entitled "Short-term effects of nitrogen dioxide on hospital admissions for cardiovascular disease in Wallonia, Belgium." (Collart et al., in press) [1].
Nitrogen dioxide concentrations showed a strong seasonal pattern with higher levels in the cold period than in the warm period. A minimum of 13.1 µg/m 3 in July and a maximum of 26.9 µg/m 3 in January were observed. The coldest months are December, January and February and the hottest months are June, July and August. Temperature and nitrogen dioxide were negatively correlated in the cold period and positively correlated in the warm period.
For the period 2008-2011 there were 113 147 hospital admissions for cardiovascular disease. Forty-five percent of patients were women and 66.5% were 65 and older. Heart rhythm disorders account for the majority of hospital admissions for cardiovascular disease. Our data confirms the existence of an association between NO 2 and cardiovascular disease. Apart from haemorrhagic stroke, the strongest association between NO 2 concentrations and number of hospital admissions is observed at lag 0. For haemorrhagic stroke, the association is strongest with a delay of 2 days. All associations calculated without stratification are statistically significant and range Generalized additive model was applied to analyse the data with a quasi-Poisson regression.

Wallonia, Belgium
Data accessibility Data is with this article.

Value of the data
Extensive data on temporal variations of nitrogen dioxide concentrations. Extensive data on the lag effect of nitrogen dioxide on different cardiovascular disease across age groups.
The data on short-term effect of air pollution on health could be useful for government and health workers to make decisions that could reduce the risk of disease among the population.
The result from the analysis can be compared with other environmental studies.

Data
Many studies have shown a short-term association between NO 2 and cardiovascular disease. However, few data are available on the delay between exposure and a health-related event. The aim of the present study is to determine the strength of association between NO 2 and cardiovascular health in Wallonia for the period 2008-2011.

Geographical area
The present study was conducted in Wallonia from 1 January 2008 to 31 December 2011. Wallonia is the southern region of Belgium, with an area of 16 844 km 2 and 3 525 000 inhabitants in 2011.

Environmental data
Concentrations of NO 2 and temperature for the period between 2008 and 2011 were obtained from the ISSeP (Institut scientifique de service public). Temperature data were collected from five measuring stations while NO 2 concentrations were recorded by 10 stations spread across the study area. NO 2 was analysed by chemoluminescence (Horiba, APEA-370). Correlation between monitoring stations was analysed using Spearman's rank correlation. More than 75% of the correlation coefficients between stations were higher than 0.75 ( Table 1). The daily concentrations of each pollutant were averaged from the available results. Averaged pollution data were used as surrogates of individual exposure.
Nitrogen dioxide concentrations showed a strong seasonal pattern with higher levels in the cold period than in the warm period (Table 2). A minimum of 13.1 µg/m 3 in July and a maximum of 26.9 µg/m 3 in January were observed. The coldest months are December, January and February and the hottest months are June, July and August Temperature and nitrogen dioxide were negatively correlated in the cold period and positively in the warm period Table 3.

Hospital admissions data for cardiovascular diseases
The analyses presented below relate to patients admitted in an hospital between 25 and 104 years of age over time. The daily counts of hospital admissions for cardiovascular disease were taken from the'Résumé Hospitalier Minimum' (RHM) for 42 hospitals within the study region. The RHM is a mandatory register kept by Belgian hospitals containing patient data (e.g.: year of birth, gender, place of residence) and stay data (e.g.: admission date). Clinical admission diagnoses were registered using the ICD-9 codes (International Classification of Disease, 9 th version). The daily counts of hospital admissions for CVD were graded: arrhythmia (ICD9: 426 and 427), acute myocardial infarction (AMI, Table 1 Correlations between monitoring stations.

CH01
CH03 CH04 SG01 LG03 LG06  (Table 4). Forty-five percent of patients were women and 66.5% were 65 and older. Heart rhythm disorders account for the majority of hospital admissions for cardiovascular disease, that is, more than 50,000 cases.

Data analysis
A time-series design was used to assess the association between short-term exposure to air pollution and hospital admissions for CVD. The generalized additive model was applied to analyse the data with a quasi-Poisson regression to account for overdispersion [2]. The day of the week was modelled using an indicator variable. Adjustment for temperature was performed with a natural spline as a smoothing function with three degrees of freedom to overcome the non-linear effect of temperature [3]. Seasonality and long-term trend were also modelled using a natural spline with three degrees of freedom per year [2,3]. NO 2 was added in the model as a linear term without a delay (lag 0) and with a delay (lag 1 to 6). The lagged variables were introduced in the model separately. The effect of the delay was analysed for the association between NO 2 and CVD stratified over age and for the association between NO 2 and the hospital admissions for various cardiovascular pathologies. For each subgroup analysed, the lag giving the strongest association was selected. Residuals and partial autocorrelation were checked graphically to ensure the goodness of the model. The sum of the absolute values of the partial autocorrelation function was calculated for each degree of freedom. The analyses were performed without stratification (overall analysis) and with stratification on gender and age in three subgroups: 25-54, 55-64 and Z 65 years. Excess relative risk (ERR) and 95% confidence intervals (CI) were calculated using a Poisson regression model and R software 2.15.0 (The R Foundation for Statistical Computing) with the mgcv and spline packages. The excesses relative risks for an increase of 10 µg/m 3 of NO 2 are presented in the tables.

Sensitivity analysis
Sensitivity analysis was performed to check the robustness of the model. Analysis using different degrees of freedom of the two natural splines was performed to estimate the effects on the strength of association. This sensitivity analysis was conducted using total daily admissions for CVD. The number of degrees of freedom giving the lowest sum of the absolute values of the partial autocorrelation function was selected.
The association of NO 2 and AMI hospital admissions decreases slightly when the number of degrees of freedom of the smoothing function used for seasonal adjustment increases. The sum of the absolute values of the partial autocorrelation function was lowest for 3 df (Table 5).

The lag effect
The effect of age on the lag pattern of the association between NO 2 concentrations and CVD is shown in Fig. 1. Regardless of age, the association is strongest taking into account the measurement of pollution on the day of the event (lag 0). The lag patterns for the four cardiovascular diseases are presented in Fig. 2. For all ages, the strongest association between NO 2 concentrations and the number of hospital admissions is observed at lag 0 (Fig. 2), except for haemorrhagic stroke where the strongest association occurs at lag 2. Relative risks (RR with 95% confidence intervals) for the association between NO 2 and different cardiovascular diseases per 10 µg/m3 increase of NO 2 concentration obtained with the single lag model. Analysis performed for all age groups.

Excess of risk
The associations between NO 2 concentrations and the various pathologies are presented in Table 6. All associations calculated without stratification are statistically significant and range from an ERR of 2.8% for myocardial infarction to 4.9% for haemorrhagic strokes. Apart from haemorrhagic stroke where excess relative risk is 7.3% for women and 2.3% for men, gender has very little effect on the association between NO 2 and cardiovascular disease. However, age has a more marked impact. For heart rhythm disorders and haemorrhagic stroke, extreme age groups are the most susceptible to NO 2 . For myocardial infarction, the strongest association is found for the 55-64 years age group. Temperature has a modifying effect on the association between NO 2 and hospital admissions for CVD as well as on heart rhythm disorders and myocardial infarction. For these pathologies, the effect of NO 2 is far more pronounced when the temperature is above 16.3°C (P75). The season has a similar effect on CVD and heart rhythm disorders but has no effect on myocardial infarction.

Transparency document. Supporting information
Transparency data associated with this article can be found in the online version at http://dx.doi. org/10.1016/j.dib.2017.12.056.