Respiratory medication sales and urban air pollution in Brussels (2005 to 2011)

EI-03380; No of Pages 7 Environment International xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Environment International journal homepage: www.elsevier.com/locate/envint

Respiratory medication sales and urban air pollution in Brussels (2005 to 2011)☆ Lidia Casas a,f, Koen Simons b,c, Tim S. Nawrot a,d, Olivier Brasseur e, Priscilla Declerck e, Ronald Buyl c, Danny Coomans c, Benoit Nemery a, An Van Nieuwenhuyse b,⁎ a

Centre for Environment and Health, Department of Public Health and Primary Care, KU Leuven, Herestraat 49, 3000 Leuven, Belgium Unit Health and Environment, Scientific Institute of Public Health, Juliette Wytsmanstraat 14, 1050 Brussels, Belgium Department of Biostatistics and Medical Informatics, Public Health, Vrije Universiteit Brussel, Laarbeeklaan 103, 1090 Brussels, Belgium d Centre for Environmental Sciences, Hasselt University, Hasselt, Belgium e Department Laboratory and Air Quality, Brussels Environment, Gulledelle 100, 1200 Brussels, Belgium f ISGlobal, Centre for Research in Environmental Epidemiology (CREAL), Dr Aiguader 88, 08003 Barcelona, Spain b c

a r t i c l e

i n f o

Article history: Received 23 February 2016 Received in revised form 26 May 2016 Accepted 16 June 2016 Available online xxxx Keywords: Asthma COPD air pollution medication sales time-series analyses

a b s t r a c t Background: We investigated the associations between daily sales of respiratory medication and air pollutants in the Brussels-Capital Region between 2005 and 2011. Methods: We used over-dispersed Poisson Generalized Linear Models to regress daily individual reimbursement data of prescribed asthma and COPD medication from the social security database against each subject's residential exposure to outdoor particulate matter (PM10) or NO2 estimated, by interpolation from monitoring stations. We calculated cumulative risk ratios (RR) and their 95% confidence intervals (CI) for interquartile ranges (IQR) of exposure for different windows of past exposure for the entire population and for seven age groups. Results: Median daily concentrations of PM10 and NO2 were 25 μg/m3 (IQR = 17.1) and 38 μg/m3 (IQR = 20.5), respectively. PM10 was associated with daily medication sales among individuals aged 13 to 64 y. For NO2, significant associations were observed among all age groups except N84 y. The highest RR were observed for NO2, among adolescents, including three weeks lags (RR = 1.187 95%CI: 1.097–1.285). Conclusion: The associations found between temporal changes in exposure to air pollutants and daily sales of respiratory medication in Brussels indicate that urban air pollution contributes to asthma and COPD morbidity in the general population. © 2016 Published by Elsevier Ltd.

1. Background Exposure to urban air pollution increases overall mortality, as well as respiratory mortality and morbidity, as manifested by hospital admissions or emergency room visits (Atkinson et al., 2014). Although these are reliable indicators of acute exacerbations of respiratory conditions, they only capture the more severe cases of asthma and COPD. In contrast, medication use provides reliable information on a wider range of severity of exacerbations and affected population (Samet and Krewski, 2007). Panel studies indicate that increased levels of air pollutants are positively associated with asthma or COPD medication intake in asthmatic and COPD populations (Segala et al., 1998; Gent et al., 2009; van der Zee et al., 1999; Slaughter et al., 2003; Escamilla-Nuñez et al., 2008; Hiltermann et al., 1998; Silkoff et al., 2005). These observations have been confirmed by a few studies that used registries of medication sales or prescriptions in small/medium size French and Italian cities ☆ Take home message: Urban air pollution increases the sales of prescribed asthma and COPD medication from birth to young elderly ages. ⁎ Corresponding author. E-mail address: [email protected] (A. Van Nieuwenhuyse).

(Sofianopoulou et al., 2013; Pitard et al., 2004; Laurent et al., 2009; Zeghnoun et al., 1999; Vegni et al., 2005). In Belgium, the levels of ambient air pollutants in the last decade have been close to or above the European standards. The air pollution in Brussels is strongly influenced by traffic related pollutants (Fierens et al., 2011), and compliance with the European limit for NO2 remains problematic. Here, we present a time-series study in seven groups of age where we combined air pollution information with reimbursement data of prescribed asthma and COPD medication from the Belgian social security for the period 2005 to 2011 in the Brussels-Capital Region. We hypothesize that increases in the levels of air pollution result in increases in medication sales specific to age groups. 2. Methods 2.1. Study area, population, and period The Brussels-Capital Region occupies a territory of 161.4km2 comprising 19 municipalities in the center of Belgium. With slightly more than one million inhabitants, it is almost entirely urbanized and the most densely populated region in Belgium. Our study is based on daily

http://dx.doi.org/10.1016/j.envint.2016.06.019 0160-4120/© 2016 Published by Elsevier Ltd.

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

2

L. Casas et al. / Environment International xxx (2016) xxx–xxx

data on air pollutants and prescribed respiratory medications reimbursed in the Brussels-Capital Region, between the 1st of January 2005 and the 31st of December 2011 (i.e. 2556 days). The geographical unit of observation was the whole region. We considered seven population age groups defined based on the natural history of obstructive airway diseases (Martinez et al., 1995; Viegi et al., 2007): pre-school children and infants (0 to 5 y), school-age children (6 to 12 y), adolescents (13 to 18 y), young adults (19 to 39 y), adults (40 to 64 y), young elderly (65 to 84 y), and elderly population (≥85 y). 2.2. Exposure: environmental pollutants In the Brussels-Capital Region, the air concentrations of various pollutants are measured by the Brussels Environment Agency (http:// www.ibgebim.be) and transferred to the Belgian Interregional Environment Agency, where daily levels of pollutants are estimated for 4 × 4 km grids using a land use regression model. The interpolation is based on a detrended kriging interpolation model that uses land cover data obtained from satellite images (Janssen et al., 2008). It accounts for the local variations of pollutant concentrations at locations where no monitoring stations are available. During the study period, PM10 and NO2 were monitored in the Brussels-Capital Region. SO2 was also monitored, but because of the very low concentrations achieved in the last 10 years, we show the SO2 results as complementary information in the online supplement. 2.3. Health outcome: Sales of respiratory medication In Belgium, the entire residing population is enrolled in the social security system. Most health care expenditure (i.e. prescribed medication, medical procedures, etc.) is reimbursed by seven sickness funds and this information is centralized by the “Intermutualistisch Agentschap – Agence Inter-Mutualiste” (IMA-AIM). The IMA-AIM database contains detailed records of sales of prescribed drugs that are reimbursed for every person covered by the social security. It consists of records of medication purchase in pharmacies (i.e. one record per box of drugs) linked to the product code, the ATC (Anatomical Therapeutic Chemical) code (WHO Collaborating Center for Drug Statistics Methodology, 2013), the date of purchase, the encoded national social security number, and individual information (age, sex, and home address). The ATC classification system (WHO Collaborating Center for Drug Statistics Methodology, 2013) aims at standardizing chemical substance classifications to allow international comparisons. The active substances are divided into groups at 5 levels according to the target organ/system (1st level) and their therapeutic (2nd), pharmacological (3rd and 4th) and chemical properties (5th). Thus, each medication is related to a unique ATC-code. Due to the similarities in the physiology and symptomatology of asthma and COPD, the medication prescribed for both disorders is the same. Nevertheless, COPD is unlikely to affect people below 40 years old, so medication sales below this age are most likely to be taken by asthmatics. In our study, we included daily sales of prescribed asthma and COPD medication (R03) that were reimbursed, for residents of the Brussels-Capital Region during the study period (2005–2011), those were daily sales of boxes of drugs that we consider as indicators of medication consumption. We also separately considered the short-acting medications. A list of the included medication is provided in Table S1 in the online supplement. Cough and cold medications (R05) were also considered, but the findings relative to these drugs with unproven efficacy are only shown in the online supplement. The use of the reimbursement data was authorized by the Sectorial Committee of Social Security and Health in the Commission for the Protection of Privacy. 2.4. Potential confounders Daily meteorological conditions (temperature and relative humidity (RH)) and weekly influenza-like infections during the study period

were considered. We obtained daily minimum temperatures (°C) and daily averages of RH (%) from the Belgian Royal Meteorological Institute. Information on influenza-like infections is available from a national reporting system, based on weekly reports by general practitioners, with a binary indicator for influenza epidemics being defined by an epidemic threshold of 140 visits per 100,000 inhabitants per week during two consecutive weeks (Van Casteren et al., 2010). 2.5. Statistical analyses For each pollutant, we constructed a single daily exposure variable for the entire region: we weighted the 4x4km2 grid cells for which the daily exposure concentrations are given, by the total population per cell, to account for the proportion of exposed individuals in each cell. In Belgium, pharmacies are open Monday to Friday. We excluded from the main analyses medication sales performed on Saturdays, Sundays and official holidays, when pharmacies open only until noon (Saturdays) or take turns in providing drugs. From the sales records, we excluded individual purchases of more than five eligible products on a single day. Such atypical events (0.15% of the sales) were likely due to registration errors or inappropriate sales. Data were aggregated into ecological time series and dates were used to link the environmental data with the health data. We used over-dispersed Poisson Generalized Linear Model (GLM) and adjusted for time using cubic splines with seven degrees of freedom per year, for daily minimum temperature using natural cubic splines with six degrees of freedom, for average relative humidity using natural cubic splines with three degrees of freedom, for day of the week, and for influenza epidemics. We opted for single pollutant models because of the large correlation among pollutants. For each pollutant, we fitted unrestricted models with various lags from zero up until 21 days before the sale event. We estimated relative risks (RR) for an increase of the interquartile ranges (IQR) of the pollutant concentrations. The RR were calculated for specific lags (lag 0 to lag 21) and cumulative lags (lag 0 to 1, 0 to2, 0 to7, 0 to14, and 0 to 21) among the entire population as well as for the seven previously described age groups. In sensitivity analyses, we ran the same models including medications sold on Saturdays. Moreover, because influenza infections may (partly) mediate the association between the pollutants and medication sales (Xu et al., 2013), we also ran our models without adjusting for influenza epidemics. We used the R3.1.0 statistical package for all the statistical analyses (R, Development Core team, 2014). 3. Results The population covered by the social security system in the study region ranged from an average of 964,500 individuals in 2005 to 1,050,000 individuals in 2011. On average, the population was composed of 23.0% children and adolescents (≤ 18 y), 61.5%, adults and 15.5% elderly (≥ 65 y). The daily concentrations of PM10 and NO2, the average RH and the minimum temperature in Brussels-Capital Region during the study period are described in Table 1. The median concentrations of PM10 and NO2 were 25 μg/m3 (IQR = 17.1 μg/m3) and 38 μg/m3 (IQR = 20.5 μg/m3), respectively. The time-series plots of PM10 and NO2 concentrations (Fig. 1) show that the concentrations were higher during winter, and the highest concentrations of both pollutants were achieved during the winter of 2008. The seasonal variability observed in the Figures is consistent with the negative correlation between minimum temperature and both pollutants (Table S2 in the online supplement). The correlation between PM10 and NO2 concentrations during the study period was strong and positive (correlation coefficient = 0.7). The median SO2 concentrations during the study period were largely below the thresholds proposed by the WHO and the EC (median SO2 concentrations: 2.5 μg/m3; IQR = 2.3 μg/m3) and they were also

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

L. Casas et al. / Environment International xxx (2016) xxx–xxx

3

Table 1 Description of the concentrations (μg/m3) of PM10 and NO2, and the average relative humidity (RH, %), and minimum temperatures (tmin, °C) in Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011).

Minimum P25 Median Mean P75 Maximum WHO recommendations (annual means) European Community recommendations (annual means)

PM10

NO2

RH

tmin

6.0 18.6 25.0 29.0 35.7 119.5 20

8.2 28.4 38.0 39.6 48.9 124.8 40

31.9 68.5 77.6 75.6 84.2 99.8 –

−8.1 3.4 8.1 7.6 12.4 21.9 –

40

40

–

–

strongly correlated with PM10 and NO2 (Tables S2 and S3 in the online supplement). During the study period, daily sales of asthma and COPD medications ranged from 843 to 2780 units, with a median of 1754 (IQR = 334.5) sales per day on weekdays (i.e. Monday to Friday), excluding official holidays. The daily median number of sales of short-acting drugs was 763 (IQR = 168.5). Fig. 2 shows the time-series plot of daily sales of asthma and COPD medications. Overall, the daily sales were higher during winter than in summer. The daily sales (absolute and relative to 1000 inhabitants) of asthma and COPD medications per age group are described in Table 2. In absolute numbers, the median sales were highest among adults and the young elderly (40 to 84 y), followed by pre-school age children (0 to 5 y). The median daily sales per 1000 inhabitants were higher among the elderly population (≥65 y) (0.3 sales/1000 inhabitants). For short-acting drugs, we observed the same patterns for absolute numbers, whilst the median daily sales per 1000 inhabitants were higher among preschool children followed by the elderly population. Daily sales of reimbursed cough and cold preparations (Table S4) were lower than asthma and COPD medications (median: 247; IQR = 92). For asthma and COPD medications, the RRs and their 95%CI per IQR increase in PM10 or NO2 for the entire population are shown in Table 3. When the entire population was considered, we did not observe any statistically significant association between PM10 exposure and medication sales. In contrast, NO2 concentrations were significantly associated with daily medication sales for all time periods. The findings for the separate age groups are presented in Figs. 3 and 4 and the RR for specific lags in Tables S5 and S6 in the online supplement. Overall, the association estimates for NO2 were stronger and more frequently statistically significant than for PM10. In general, the estimates for PM10 and NO2 were highest among adolescents, and also when longer lags were used to calculate the cumulative RRs.

PM10 (Fig. 3) was significantly associated with daily sales among individuals aged 13 to 64 y. The strongest effect estimates were observed among adolescents (13 to 18 y) and the effect size was lower in younger (b13 y) and older (N18 y) age groups. These statistically significant associations were observed for lag 0 and lag 0–7 for adults (40 to 64 y), and for lag 0–14 and 0–21 among individuals aged 13 to 64 y. Regarding NO2 (Fig. 4), concentrations were significantly associated with daily sales among all age groups except for those ≥ 85 y. As for PM10, the strongest estimates were observed among adolescents with lower effect size in younger and older age groups. In the young elderly (65 to 84 y), significant associations were observed for lag 0 and lag 0–1. In the younger age groups, the effects were stronger with longer lag periods. The analyses including only daily sales of short-acting asthma and COPD drugs showed similar trends (Table 4). Concentrations of SO2 were significantly associated with asthma and COPD medication and short acting medication, in particular for ages between 6 and 64 y (Table S7 in the online supplement). Daily sales of cough and cold preparations were also significantly associated with concentrations of PM10, NO2 and SO2 (Table S8 in the online supplement). Sensitivity analyses were conducted with inclusion of sales on Saturdays (Tables S9 and S10), and without adjusting for influenza epidemics (Tables S11 and S12). Although there are a few differences in statistical significance, the magnitude and direction of the effect estimates do not differ from those shown in Tables 3 and 4, and Figs. 3 and 4. 4. Discussion We demonstrated that concentrations of PM10 and NO2 are associated with daily sales of asthma and COPD medications in Brussels, the administrative capital of the European Union, and a densely populated area situated in a hot-spot of air pollution within Europe (EEA, 2015).

Fig. 1. Time-series plot of concentrations of PM10 and NO2 (μg/m3), in Brussels-Capital Region (Belgium) from the 1st of January 2005 to the 31st of December 2011.

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

4

L. Casas et al. / Environment International xxx (2016) xxx–xxx

Fig. 2. Time-series plot of daily sales of asthma and COPD medication (all and short-acting) in Brussels-Capital Region (Belgium) from the 1st of January 2005 and the 31st of December 2011.

With observations of around one million residents over a period of seven years, this is the largest study that has investigated the relationship between daily respiratory medication sales and air pollution. To date, only few studies have used registry data of respiratory medication sales to investigate the health effects of air pollution (Pitard et al., 2004; Laurent et al., 2009; Zeghnoun et al., 1999; Vegni et al., 2005). Compared with hospital admissions and mortality, information on prescribed medication sales can be considered a good indicator of the burden of respiratory disease in the community because it captures the less severe and more frequent asthma and/or COPD exacerbations (Samet and Krewski, 2007). Previous studies of medication sales were limited in terms of size of the population and years of follow-up, thus not allowing an investigation of associations in specific age groups. These previous studies also included lags up to 10 days at most. The inclusion of longer lags may capture delayed sales due to home medication stocks, which is not uncommon for chronic patients. Moreover, previous

studies used air pollution and medication sales information of ten to twenty years ago, whereas our observations span the period of 2005 up to 2011. When investigating respiratory obstructive disorders considering age is important. Asthma-like symptoms during pre-school age are often transient and mostly caused by viral infections. When the symptoms persist or start at school age, the probability of developing asthma is high. Nevertheless, asthma onset may also occur during adolescence or early adulthood (Martinez et al., 1995; Ducharme et al., 2014). In contrast, COPD is rare below the age of 40 (Viegi et al., 2007; Mannino and Buist, 2007). Moreover, age is an effect modifier of the association between air pollution and health, with higher vulnerability among older individuals (Bell et al., 2013). Thanks to the large number of individuals included in our study we were able to investigate the association between daily concentrations of air pollutants and daily medication sales over many different stages in the life course. For childhood, we differentiated pre-school

Table 2 Daily sales of asthma and COPD medication in Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011, by age group. R03

0 to 5 years old Daily sales Daily sales/1000 inhab. 6 to 12 years old Daily sales Daily sales/1000 inhab. 13 to 18 years old Daily sales Daily sales/1000 inhab. 19 to 39 years old Daily sales Daily sales/1000 inhab. 40 to 64 years old Daily sales Daily sales/1000 inhab. 65 to 84 years old Daily sales Daily sales/1000 inhab. ≥85 years old Daily sales Daily sales/1000 inhab.

Short R03

Min

P25

Median

Mean

P75

Max

Min

P25

Median

Mean

P75

Max

23 0.02

150 0.13

199 0.17

220 0.19

267 0.23

888 0.76

12 0.01

110 0.62

148 0.14

167 0.13

202 0.09

731 0.17

14 0.01

54 0.05

70 0.06

70 0.06

85 0.07

167 0.15

1 b0.01

20 0.08

28 0.03

29 0.02

37 0.02

97 0.03

5 0.01

29 0.03

37 0.04

38 0.04

45 0.05

91 0.10

1 b0.01

11 0.05

14 0.02

15 0.02

18 0.01

43 0.02

91 0.02

159 0.04

181 0.04

183 0.04

206 0.05

336 0.07

35 0.01

71 0.04

82 0.02

83 0.02

94 0.02

186 0.02

284 0.07

494 0.12

545 0.13

551 0.14

604 0.15

906 0.22

127 0.03

196 0.10

217 0.05

219 0.05

240 0.05

389 0.06

306 0.17

540 0.30

586 0.33

590 0.33

635 0.35

964 0.54

121 0.07

189 0.23

210 0.12

212 0.12

234 0.11

415 0.13

40 0.11

86 0.24

99 0.28

103 0.29

115 0.32

299 0.84

17 0.05

38 0.46

45 0.13

47 0.13

53 0.11

163 0.15

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

L. Casas et al. / Environment International xxx (2016) xxx–xxx Table 3 Estimated cumulative relative risks (RR)⁎ and their 95% confidence intervals of sales of asthma and COPD medication per IQR increase in the concentration of PM10 and NO2 in the Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011.

lag 0 lag 0 to 1 lag 0 to 2 lag 0 to 7 lag 0 to 14 lag 0 to 21

PM10(IQR = 17.1 μg/m3)

NO2 (IQR = 20.5 μg/m3)

1.004 (0.998–1.010) 1.001 (0.994–1.007) 0.998 (0.991–1.005) 1.003 (0.994–1.013) 1.005 (0.992–1.019) 1.003 (0.987–1.020)

1.014 (1.005–1.022) 1.014 (1.004–1.024) 1.013 (1.002–1.024) 1.025 (1.010–1.040) 1.040 (1.019–1.061) 1.040 (1.013–1.068)

⁎ Adjusted for daily minimum temperature, average relative humidity, day of the week and influenza epidemics. Bold indicates p-value b 0.05.

age from primary-school age and adolescence. We had two adult groups using a cut-off of 40 y to distinguish medication users with moderate to high probability of COPD (≥40 y) from those with very low likelihood of having COPD (b 40 y). In the elderly, we differentiated between the young elderly (65 to 85 y) and the older elderly (N85 y). We observed increased sales of respiratory medication in relation to high PM10 or NO2 concentrations from birth to young elderly ages. For both pollutants, estimates were stronger for younger ages and no association was observed among the ≥85 y age group (Bell et al., 2013). Older patients with established respiratory conditions may be more likely to be hospitalized compared with younger populations or may live in institutions and not buy the medication themselves. Unfortunately, information on actual medication consumption is not available in our study. Furthermore, the elderly population may be less exposed to outdoor air pollution due to chronic conditions that limit their mobility.

5

Most previous studies on medication sales did not calculate risks for specific age groups (Zeghnoun et al., 1999; Vegni et al., 2005) or the number of groups included was limited to two (Laurent et al., 2009). Pitard et al. (2004) performed separate analyses in 4 groups of age. Their RRs followed similar patterns as those reported in our study. They observed statistically significant associations of NO2 and/or SO2 with daily sales of asthma and COPD medication and cough and cold preparations from childhood to the age of 64. Nevertheless, their age groups were rather broad and did not fully differentiate critical age periods for vulnerability and natural history of respiratory disorders. Pharmacies in Belgium are open on Saturdays until noon, and people who work during the week may choose Saturdays to buy their medication. In our study, we a priori excluded Saturdays from the main analyses because medication sales on Saturdays followed a different distribution. Sensitivity analyses including Saturday sales showed increased statistical significance of a few effect estimates, but no substantial differences in the direction or magnitude of the estimates, and the confidence intervals became narrower. This suggests that exclusion of Saturdays only decreased statistical power. Our models were also adjusted a priori for influenza epidemics. Previous studies have suggested that air pollution is associated with respiratory infections (MacIntyre et al., 2014), and infections may increase the risk of asthma and COPD exacerbations (Altzibar et al., 2015; Wedzicha and Seemungal, 2007). Thus, influenza may be considered as a mediator. Sensitivity analyses excluding influenza epidemics from the models did not show changes in the effect estimates. Moreover, ambient pollen concentrations have been described as potential confounders in the association between respiratory health and air pollution (Guilbert et al., 2016), however sensitivity analyses including daily aerial pollen concentrations from Poaceae in our models did not modify the effect estimates.

Fig. 3. Estimated cumulative relative risks (RR)* and their 95% confidence intervals of sales of asthma and COPD medication per IQR (17.1 μg/m3) increase in the concentration of PM10 in Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011, by age groups. Estimates are adjusted for daily minimum temperature, average relative humidity, day of the week and influenza epidemics.

Fig. 4. Estimated cumulative relative risks (RR)* and their 95% confidence intervals of sales of asthma and COPD medication per IQR (20.5 μg/m3) increase in the concentration of NO2 in Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011, by age groups Estimates are adjusted for daily minimum temperature, average relative humidity, day of the week and influenza epidemics.

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

6

L. Casas et al. / Environment International xxx (2016) xxx–xxx

Table 4 Estimated cumulative relative risks (RR)⁎ and their 95% confidence intervals of sales of short acting asthma and COPD medication per IQR increase in the concentration of PM10 and NO2 in Brussels-Capital Region (Belgium) between the 1st of January 2005 and the 31st of December 2011, by age groups. Cumulative RR (95% CI) lag 0

lag 0 to 1

lag 0 to 2

lag 0 to 7

lag 0 to 14

lag 0 to 21

PM10(IQR = 17.1 μg/m ) 0 to 5 0.996 (0.983–1.009) 6 to 12 0.994 (0.974–1.014) 13 to 18 1.005 (0.981–1.029) 19 to 39 0.997 (0.986–1.008) 40 to 64 1.010 (1.001–1.018) 65 to 84 1.002 (0.994–1.010) ≥85 1.004 (0.986–1.022) All ages 1.002 (0.995–1.009)

0.992 (0.978–1.007) 0.992 (0.970–1.014) 1.010 (0.983–1.037) 1.002 (0.990–1.014) 1.005 (0.996–1.014) 0.999 (0.990–1.008) 0.996 (0.976–1.016) 0.999 (0.992–1.006)

0.989 (0.973–1.004) 0.982 (0.958–1.006) 1.004 (0.976–1.034) 1.003 (0.990–1.016) 1.006 (0.996–1.016) 0.995 (0.986–1.005) 0.992 (0.970–1.014) 0.997 (0.989–1.005)

0.991 (0.970–1.012) 0.990 (0.957–1.023) 1.006 (0.967–1.046) 1.018 (1.000–1.036) 1.020 (1.006–1.034) 1.002 (0.989–1.015) 1.001 (0.972–1.031) 1.005 (0.994–1.016)

0.983 (0.955–1.012) 1.002 (0.957–1.048) 1.038 (0.984–1.095) 1.031 (1.007–1.056) 1.029 (1.010–1.047) 0.996 (0.979–1.014) 1.012 (0.972–1.053) 1.008 (0.993–1.023)

0.978 (0.942–1.016) 0.985 (0.929–1.045) 1.049 (0.979–1.123) 1.040 (1.010–1.072) 1.023 (1.000–1.047) 0.995 (0.973–1.018) 0.987 (0.938–1.038) 1.005 (0.986–1.025)

NO2 (IQR = 20.5 μg/m3) 0 to 5 1.018 (0.999–1.037) 6 to 12 1.016 (0.987–1.046) 13 to 18 1.018 (0.984–1.054) 19 to 39 1.000 (0.985–1.016) 40 to 64 1.016 (1.004–1.028) 65 to 84 1.009 (0.997–1.020) ≥85 0.993 (0.968–1.019) All ages 1.012 (1.003–1.022)

1.020 (0.999–1.042) 1.016 (0.983–1.051) 1.038 (0.999–1.079) 1.004 (0.986–1.021) 1.015 (1.002–1.028) 1.007 (0.994–1.020) 0.992 (0.963–1.022) 1.012 (1.002–1.023)

1.027 (1.004–1.052) 1.015 (0.978–1.053) 1.054 (1.009–1.101) 1.008 (0.989–1.028) 1.016 (1.001–1.031) 1.005 (0.990–1.020) 0.983 (0.951–1.016) 1.014 (1.001–1.026)

1.049 (1.016–1.083) 1.052 (1.000–1.106) 1.065 (1.004–1.130) 1.041 (1.014–1.068) 1.036 (1.016–1.058) 1.013 (0.993–1.033) 0.987 (0.944–1.033) 1.032 (1.015–1.049)

1.070 (1.024–1.118) 1.096 (1.023–1.174) 1.144 (1.055–1.240) 1.079 (1.042–1.118) 1.065 (1.036–1.095) 1.008 (0.981–1.036) 1.001 (0.942–1.065) 1.050 (1.027–1.074)

1.044 (0.985–1.107) 1.055 (0.964–1.154) 1.165 (1.049–1.295) 1.109 (1.059–1.161) 1.074 (1.036–1.113) 1.010 (0.975–1.046) 0.988 (0.913–1.069) 1.046 (1.016–1.077)

3

⁎ Adjusted for daily minimum temperature, average relative humidity, day of the week and influenza epidemic. Bold indicates p-value b 0.05.

Our study has also a number of limitations. The use of air pollution measurements from local monitoring stations facilitates the inclusion of a large number of individuals and time periods. However, it lacks precision regarding the actual personal exposure. Our approach assumed that the relevant exposure for all subjects would be their outdoor residential exposure, although changes in exposure may occur during commuting to work or school. Nevertheless, most individuals living in Brussels Capital Region work and go to school within the region, and temporal changes in air pollution can be considered to be distributed equally within this area. Another issue to consider when studying the effects of outdoor air pollution is the strong correlation that exists between specific pollutants. Not unexpectedly, PM10, SO2 and NO2 in our study were strongly correlated, thus complicating the interpretation of single pollutant models, on the one hand, and making it impossible to disentangle the effects of each separate pollutant, on the other. Thus, we interpret the associations found with the generally low concentrations of SO2 to be indicative of a proxy effect of other pollutants, rather than an effect of SO2 per se. In addition, we had information on ozone (O3) concentrations during study period. The concentrations of this pollutant were strongly but negatively correlated with the other studied pollutants. As a consequence, we obtained RR below 1 (data not shown). These results are, therefore, most likely due to the effect of the low concentrations of PM10 or NO2 when O3 concentrations are high than to an O3 effect. Finally, whether NO2 should similarly be considered as an indicator of traffic-related pollution, rather than as a toxic agent causing respiratory injury, remains to be established, as in most circumstances, NO2 may serve as a surrogate for all traffic related combustion products (Brunekreef and Holgate, 2002). In conclusion, our findings show that high concentrations of urban air pollutants are associated with increased sales of prescribed asthma and COPD medication. For NO2, significant associations are observed in all age groups from birth to young elderly ages, even at concentrations below the European thresholds. This pharmaco-epidemiologic study of medications prescribed for common respiratory conditions provides evidence of adverse effects of current levels of urban air pollution on respiratory health. Acknowledgments This work was supported by Innoviris (Brussels institute for research and innovation), grant no. PRFB 2014-121. The authors thank Mrs.

Ragna Préal for the advice and continuous support in the exploitation of the Health Insurance Database. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.envint.2016.06.019. References Altzibar, J.M., Tamayo-Uria, I., De Castro, V., Aginagalde, X., Albizu, M.V., Lertxundi, A., Benito, J., Busca, P., Antepara, I., Landa, J., Mokoroa, O., Dorronsoro, M., 2015. Epidemiology of asthma exacerbations and their relation with environmental factors in the Basque Country. Clin. Exp. Allergy 45, 1099–1108. Atkinson, R.W., Kang, S., Anderson, H.R., Mills, I.C., Walton, H.A., 2014. Epidemiological time series studies of PM2.5 and daily mortality and hospital admissions: a systematic review and meta-analysis. Thorax 69, 660–665. Bell, M.L., Zanobetti, A., Dominici, F., 2013. Evidence on vulnerability and susceptibility to health risks associated with short-term exposure to particulate matter: a systematic review and meta-analysis. Am. J. Epidemiol. 178, 865–876. Brunekreef, B., Holgate, S.T., 2002. Air pollution and health. Lancet (Lond. Engl.) 360, 1233–1242. Ducharme, F.M., Tse, S.M., Chauhan, B., 2014. Diagnosis, management, and prognosis of preschool wheeze. Lancet 383, 1593–1604. EEA, 2015. Air Quality in Europe — 2015 Report. Copenhagen. Escamilla-Nuñez, M.-C., Barraza-Villarreal, A., Hernandez-Cadena, L., Moreno-Macias, H., Ramirez-Aguilar, M., Sienra-Monge, J.-J., Cortez-Lugo, M., Texcalac, J.-L., del RioNavarro, B., Romieu, I., 2008. Traffic-related air pollution and respiratory symptoms among asthmatic children, resident in Mexico City: the EVA cohort study. Respir. Res. 9, 74. Fierens, F., Vanpoucke, C., Adriaenssens, S., Trimpeneers, E., Peeters, O., Brasseur, O., de Vos, T., Maetz, P., 2011. ANNUAL REPORT Air Quality in Belgium 2011. Gent, J.F., Koutrakis, P., Belanger, K., Triche, E., Holford, T.R., Bracken, M.B., Leaderer, B.P., 2009. Symptoms and medication use in children with asthma and traffic-related sources of fine particle pollution. Environ. Health Perspect. 117, 1168–1174. Guilbert, A., Simons, K., Hoebeke, L., Packeu, A., Hendrickx, M., De Cremer, K., Buyl, R., Coomans, D., Van Nieuwenhuyse, A., 2016. Short-term effect of pollen and spore exposure on allergy morbidity in the Brussels-capital region. EcoHealth. Hiltermann, T.J., Stolk, J., van der Zee, S.C., Brunekreef, B., de Bruijne, C.R., Fischer, P.H., Ameling, C.B., Sterk, P.J., Hiemstra, P.S., van Bree, L., 1998. Asthma severity and susceptibility to air pollution. Eur. Respir. J. 11, 686–693. Janssen, S., Dumont, G., Fierens, F., Mensink, C., 2008. Spatial interpolation of air pollution measurements using CORINE land cover data. Atmos. Environ. 42, 4884–4903. Laurent, O., Pedrono, G., Filleul, L., Segala, C., Lefranc, A., Schillinger, C., Rivière, E., Bard, D., 2009. Influence of socioeconomic deprivation on the relation between air pollution and beta-agonist sales for asthma. Chest 135, 717–723. MacIntyre, E.A., Gehring, U., Mölter, A., Fuertes, E., Klümper, C., Krämer, U., Quass, U., Hoffmann, B., Gascon, M., Brunekreef, B., Koppelman, G.H., Beelen, R., Hoek, G., Birk, M., de Jongste, J.C., Smit, H.A., Cyrys, J., Gruzieva, O., Korek, M., Bergström, A., Agius, R.M., de Vocht, F., Simpson, A., Porta, D., Forastiere, F., Badaloni, C., Cesaroni, G., Esplugues, A., Fernández-Somoano, A., Lerxundi, A., et al., 2014. Air pollution and

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019

L. Casas et al. / Environment International xxx (2016) xxx–xxx respiratory infections during early childhood: an analysis of 10 European birth cohorts within the ESCAPE project. Environ. Health Perspect. 122, 107–113. Mannino, D.M., Buist, A.S., 2007. Global burden of COPD: risk factors, prevalence, and future trends. Lancet (Lond. Engl.) 370, 765–773. Martinez, F.D., Wright, A.L., Taussig, L.M., Holberg, C.J., Halonen, M., Morgan, W.J., 1995. Asthma and wheezing in the first six years of life. The group health medical associates. N. Engl. J. Med. 332, 133–138. Pitard, A., Zeghnoun, A., Courseaux, A., Lamberty, J., Delmas, V., Fossard, J.L., Villet, H., 2004. Short-term associations between air pollution and respiratory drug sales. Environ. Res. 95, 43–52. Samet, J., Krewski, D., 2007. Health effects associated with exposure to ambient air pollution. J. Toxic. Environ. Health A 70, 227–242. Segala, C., Fauroux, B., Just, J., Pascual, L., Grimfeld, A., Neukirch, F., 1998. Short-term effect of winter air pollution on respiratory health of asthmatic children in Paris. Eur. Respir. J. 11, 677–685. Silkoff, P.E., Zhang, L., Dutton, S., Langmack, E.L., Vedal, S., Murphy, J., Make, B., 2005. Winter air pollution and disease parameters in advanced chronic obstructive pulmonary disease panels residing in Denver, Colorado. J. Allergy Clin. Immunol. 115, 337–344. Slaughter, J.C., Lumley, T., Sheppard, L., Koenig, J.Q., Shapiro, G.G., 2003. Effects of ambient air pollution on symptom severity and medication use in children with asthma. Ann. Allergy Asthma Immunol. 91, 346–353. Sofianopoulou, E., Rushton, S.P., Diggle, P.J., Pless-Mulloli, T., 2013. Association between respiratory prescribing, air pollution and deprivation, in primary health care. J. Public Health (Oxf.) 35, 502–509.

7

Van Casteren, V., Mertens, K., Antoine, J., Wanyama, S., Thomas, I., Bossuyt, N., 2010. Clinical surveillance of the influenza a(H1N1)2009 pandemic through the network of sentinel general practitioners. Arch. Public Heal. BioMed Central 68, 62. van der Zee, S., Hoek, G., Boezen, H.M., Schouten, J.P., van Wijnen, J.H., Brunekreef, B., 1999. Acute effects of urban air pollution on respiratory health of children with and without chronic respiratory symptoms. Occup. Environ. Med. 56, 802–812. Vegni, F.E., Castelli, B., Auxilia, F., Wilkinson, P., 2005. Air pollution and respiratory drug use in the city of Como, Italy. Eur. J. Epidemiol. 20, 351–358. Viegi, G., Pistelli, F., Sherrill, D.L., Maio, S., Baldacci, S., Carrozzi, L., 2007. Definition, epidemiology and natural history of COPD. Eur. Respir. J. 30, 993–1013. Wedzicha, J.A., Seemungal, T.A., 2007. COPD exacerbations: defining their cause and prevention. Lancet 370, 786–796. WHO Collaborating Center for Drug Statistics Methodology, 2013. ATC/DDD index [internet]. [cited 2015 Dec 7].Available from: http://www.whocc.no/atc_ddd_index/. Xu, Z., Hu, W., Williams, G., Clements, A.C.A., Kan, H., Tong, S., 2013. Air pollution, temperature and pediatric influenza in Brisbane, Australia. Environ. Int. 59, 384–388. Zeghnoun, A., Beaudeau, P., Carrat, F., Delmas, V., Boudhabhay, O., Gayon, F., Guincètre, D., Czernichow, P., 1999. Air pollution and respiratory drug sales in the City of Le Havre, France, 1993-1996. Environ. Res. 81, 224–230.

Please cite this article as: Casas, L., et al., Respiratory medication sales and urban air pollution in Brussels (2005 to 2011), Environ Int (2016), http://dx.doi.org/10.1016/j.envint.2016.06.019