- Email: [email protected]
Winter air pollution and infant bronchiolitis in Paris
ARTICLE IN PRESS
Environmental Research 106 (2008) 96–100 www.elsevier.com/locate/envres
Winter air pollution and infant bronchiolitis in Paris$ Claire Se´galaa,, David Poizeaua, Mounir Mesbahb, Sylvie Willemsa, Manuel Maidenbergc a
SEPIA-Sante´, 18bis rue du Calvaire, 56310 Melrand, France b LSTA, Universite´ Pierre et Marie Curie, Paris, France c R.E.S.P.I.R.E.R., Paris, France
Received 25 January 2007; received in revised form 5 May 2007; accepted 18 May 2007 Available online 21 June 2007
Abstract Respiratory syncytial virus (RSV) is one of the most common respiratory pathogens in infants and young children. It is not known why some previously healthy infants, when in contact with RSV, develop bronchiolitis whereas others have only mild symptoms. Our study aimed to evaluate the possible association between emergency hospital visits for bronchiolitis and air pollution in the Paris region during four winter seasons. We included children under the age of 3 years who attended emergency room services for bronchiolitis (following standardized definition) during the period 1997–2001. Two series of data from 34 hospitals, the daily number of emergency hospital consultations (n ¼ 50 857) and the daily number of hospitalizations (n ¼ 16 588) for bronchiolitis, were analyzed using alternative statistical methods; these were the generalized additive model (GAM) and case-crossover models. After adjustments for public holidays, holidays and meteorological variables the case-crossover model showed that PM10, BS, SO2 and NO2 were positively associated with both consultations and hospitalizations. GAM models, adjusting for long-term trend, seasonality, holiday, public holiday, weekday and meteorological variables, gave similar results for SO2 and PM10. This study shows that air pollution may act as a trigger for the occurrence of acute severe bronchiolitis cases. r 2007 Elsevier Inc. All rights reserved. Keywords: Air pollution; Respiratory disease; Bronchiolitis; Case crossover; GAM; Infant
1. Introduction The short-term effects of air pollution on humans are assessed using time-series studies and panel studies. Timeseries studies have shown broadly consistent associations between air pollutants and related outcomes such as total mortality, cardiorespiratory mortality and hospital admissions. Whereas most studies have focused on the general population and elderly subjects, there have been few on children (Pope, 2000; WHO Task Group, 2004; Health Effect Institute, 2003; Se´gala, 1999). Conversely, there have been numerous panel studies on mostly asthmatic children. Panel studies measure health effects, such as symptoms, on $ Funding: The study was supported by the ‘‘Conseil National Scientifique’’ of ANTADIR. Corresponding author. Fax: +33 2 97 28 81 10. E-mail address: [email protected] (C. Se´gala).
0013-9351/$ - see front matter r 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.envres.2007.05.003
an individual basis (Desqueyroux and Momas, 1999). In epidemiology, only few studies focused on the role of air pollution as a risk factor for specific respiratory infections, such as respiratory syncytial virus (RSV) bronchiolitis (Zamorano et al., 2003; Karr et al., 2006). However, experimental data have shown that air pollutants affect lung immune responses and inflammatory reactions and that these effects may underlie the increased risk for respiratory infections (Becker and Soukup, 1999; Gilmour et al., 2001; Harrod et al., 2003; Lambert et al., 2003). RSV is one of the most common respiratory pathogens in infants and young children. Most infants develop only mild or no symptoms when in contact with RSV. However, some infants develop RSV bronchiolitis, which is characterized by expiratory wheezing and respiratory distress. This condition presents a major public health problem, as the winter peak periods of RSV virus activity often result in numerous hospital consultations and hospitalizations.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
Very few studies have assessed bronchiolitis mortality. A US study reported between 171 and 510 bronchiolitis associated deaths per year between 1979 and 1997 among children under 5 years of age (Law et al., 2002). The aim of our study was to evaluate the association between emergency hospital visits for bronchiolitis and short-term exposure to air pollution in the Paris region during four winter seasons, using our alternative approach. Our null hypothesis was that air pollutants trigger acute severe bronchiolitis. 2. Materials and methods 2.1. Study subjects We included children under the age of 3 years who attended emergency room services for bronchiolitis during the period 1997–2001. Daily bronchiolitis data were obtained from the ‘‘Epide´miologie et Receuil des Bronchiolites en Urgence pour la Surveillance’’ (ERBUS) database (Thelot et al., 1998). This contains data, available every year, for the period between 15th October and 15th January from 43 hospitals within greater Paris. Physicians reviewed the clinical records of all daily visits and selected those with diagnoses matching the standard definition of bronchiolitis: respiratory dyspnea and/or sibilants and wheezing for children under the age of 3 years during the surveillance period. Two series of data were available: the daily number of emergency hospital consultations and the daily number of hospitalizations for bronchiolitis. In 1999–2000, data collection was temporarily stopped in some hospitals. Therefore, only the 34 hospitals providing complete data were retained for the analysis.
2.2. Environmental data Air pollutants were routinely measured by the AIRPARIF monitoring network. We retained data from the urban background monitoring sites only. These sites are representative of ambient air pollution in the Paris region and are not directly affected by local sources of pollution. The following air pollution data were recorded: sulfur dioxide (SO2) concentration (measured at 30 stations using ultraviolet fluorescence); nitrogen dioxide (NO2) concentration (measured at 21 stations using chemiluminescence); levels of suspended particles with an aerodynamic diameter less than 10 mm (PM10) (measured at 9 stations by tapered element oscillating microbalance—TEOM) and black smoke (BS; suspended black particulate) (measured at 21 stations using reflectometry—French standard method, NF-X 43-005). SO2, NO2 and PM10 were measured hourly and the mean value was calculated for each day. BS was measured as 24-h mean levels. The overall mean of the mean daily readings at all stations was calculated according to APHEA methods for analyzing correlations between stations and the replacement of missing values (Katsouyanni et al., 1996). The daily mean of 12 weather variables was determined at the Paris weather station (Me´te´o-France). A principal component analysis (Benzecri, 1992) allowed us to select the five least correlated meteorological variables. These were daily minimum temperature, mean humidity, precipitation, wind and atmospheric pressure.
2.3. Statistical methods Since the mid-1990s the generalized additive model (GAM) has been the most popular model for assessing, over time, the relationship between air pollution and hospitalization data. It uses nonparametric smoothing to control for time varying factors such as trend, season and weather variables. However, this model is complex and has been the subject to recent criticism (Dominici et al., 2002; Ramsay et al., 2003). Therefore, we used an alternative approach: the case-crossover design (Maclure, 1991).
97
We first carried out a descriptive analysis of environmental and bronchiolitis data and then compared these data between the four winter seasons. We then assessed the association between the daily visit data and air pollution using two different statistical methods. The case-crossover design is an adaptation of the case–control design, in which cases serve as their own controls (Maclure, 1991). The exposure of subjects to air pollutants at the time of the visit (case period) was compared with that at other times (control periods). We used a bidirectional design to avoid a time-trend bias (Bateson and Schwartz, 1999). To evaluate the contemporaneous effect and the cumulative effect on several days of pollutant concentrations, two exposure windows of 2-day mean (days 0 and 1) and 5-day mean (from day 0 to 4) were examined. Thus, we selected four control times for each case, and, for example, for the 2-day mean: day 7–8 and 14–15 before the visit, and 7–8 and 14–15 days after the visit were chosen. These matched pairs (strata) were analyzed with SAS, using conditional logistic regression. The analysis was carried out on 291 strata due to missing values in some case period or control period variables. The model was created stepwise, introducing public holidays, holidays and meteorological variables. The appropriate lag of meteorological variables was decided on the basis of Akaike’s information criteria. Also, we applied the GAM model to examine the association between the daily visit data and air pollution using a Poisson regression. This model is an extension of standard linear regression model, in which several predictors can be related to the dependent variable (the visit data) using a nonlinear smoothing function (Hastie and Tibshirani, 1990). According to recent recommendations (Health Effect Institute, 2003), GAM models should be run with penalized regression splines (P-splines) using R software. Long-term trend, seasonality, holiday, public holiday, weekday and meteorological variables adjustments were introduced into the model based on the visual inspection of residuals, predicted series and partial autocorrelation of residuals graphs. The forms of the curves of regression between bronchiolitis data and single pollutant were investigated using smoothing functions. If there was any linearity in the associations, the pollutants were introduced as a linear term in the model. For both models, the results were expressed as relative risk (odds ratio in the case-crossover design), with 95% confidence intervals (CI), for an increase of 10 mg/m3 in pollutant concentration.
3. Results We found a strong correlation between the levels of the pollutants during the 4 years studied. Temperature, rain and wind were negatively correlated with the pollutants, whereas atmospheric pressure was positively correlated with them (Tables 1 and 2). Several environmental variables differed for the four winter seasons. The mean concentrations of the four pollutants were lower and the mean minimal temperature was higher during the 2000–2001 winter season than during the three previous winter seasons (Table 3). The daily number of emergency hospital consultations ranged from 35 to 299 (n ¼ 139/day) during the study period and the daily number of hospitalizations for bronchiolitis from 9 to 117 (n ¼ 45/day). A seasonal peak occurred each December. As data are available only between mid-October and mid-January, each winter season is aggregated with the following one. We adjusted, using the case-crossover model, the associations between the daily variations in bronchiolitis consultations and hospitalizations and the urban air pollutant concentrations (Table 4). The four pollutants were positively associated with both consultations and hospitalizations.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
98
Table 1 Descriptive statistics of daily environmental variables during the study period
3
BS (mg/m ) NO2 (mg/m3) PM10 (mg/m3) SO2 (mg/m3) Hum. (%) Temp. (1C) Rain (mm) Pres. (hPa) Wind (m/s)
Mean7sd
(min.–max.)
Mean (sd)
16.9712.7 50.8716.7 22.7710.4 10.577.4 63.5714.5 7.373.4 2.174.5 101579.4 11.374.1
(3.4–113.0) (13–169.9) (6.3–99.1) (2.4–71.9) (28–94) (1.4–17.2) (0–47) (976.9–1040.2) (3–26)
Visits 4.1 Hospital 1.4 16.6 SO2 (mg/m3) BS (mg/m3) 21.5 55.0 NO2 (mg/m3) PM10 (mg/m3) 24.2 Temp. (1C) 5.1 Hum. (%) 81.6 Rain (mm) 2.3 Wind (m/s) 12.4 Pres. (hPa) 1013.9
BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind.
Table 2 Correlation coefficients between daily environmental variables BS
NO2
BS 1 0.83 1 NO2 PM10 0.87 0.74 SO2 0.76 0.78 Hum. 0.14 0.08 Temp. 0.29 0.34 Rain 0.31 0.34 Pres. 0.41 0.46 Wind 0.75 0.67
Table 3 Comparison of bronchiolitis data and environmental variables between the four winter seasons
PM10
SO2
Hum. Temp. Rain Pres. Wind
1 0.73 0.04 0.27 0.44 0.49 0.63
1 0.11 1 0.66 0.18 1 0.43 0.35 0.31 1 0.52 0.18 0.31 0.51 1 0.49 0.16 0.21 0.36 0.31 1
BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind. p40.05.
The 5-day mean was more consistently associated with bronchiolitis than the 2-day mean for all pollutants, but for SO2. For an increase of 10 mg/m3 of pollutant, the number of consultations increased by between 3% and 12% and the number of hospitalizations by between 2% and 12%. The strongest associations were for SO2 and PM10. Our analysis also showed independent associations with three meteorological factors. The risk of bronchiolitis increased with cold temperature, high humidity and strong wind. There was no association with barometric pressure and rain. The GAM model was also used to adjust the associations between air pollutants and daily bronchiolitis consultations and hospitalizations (Table 5). With consultations, only the 5-day mean of PM10 and SO2 showed a linear effect that can be expressed in terms of relative risk; with hospitalizations, it was the case only with the 5-day mean of SO2. An increase of 10 mg/m3 in PM10 is associated with an increase of 4% in the number of consultations. An increase of 10 mg/m3 in SO2 is associated with an increase of 6% in the number of consultations and an increase of 11% in the number of hospitalizations.
1997–1998
1998–1999
1999–2000
(5.5) 3.9 (4.2) 4.2 (1.8) 1.3 (1.6) 1.4 (9.0) 14.7 (9.0) 14.3 (13.5) 18.4 (13.4) 20.6 (13.5) 55.6 (14.6) 57.8 (10.4) 21.7 (11.6) 24.2 (4.0) 5.0 (4.3) 4.3 (9.0) 82.3 (7.1) 79.6 (4.2) 2.06 (4.0) 2.1 (4.7) 12.2 (4.4) 12.3 (10.1) 1018.4 (8.4) 1019.7
2000–2001
(4.6) 4.1 (1.6) 1.3 (7.0) 10.0 (18.0) 14.5 (17.4) 47.1 (13.6) 18.6 (3.7) 6.5 (8.4) 80.4 (4.1) 2.9 (4.3) 12.9 (11.2) 1009.2
(4.1) (1.6) (3.4) (7.9) (9.6) (5.6) (3.4) (8.1) (4.4) (4.3) (9.9)
Visits: daily number of emergency hospital consultations for bronchiolitis; Hospital: daily number of hospitalizations for bronchiolitis; BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind. Table 4 Adjusted effects [OR (95% CI)]a of an increase of 10 mg/m3 of air pollutants on daily bronchiolitis consultations and hospitalizations with case-crossover model Consultations OR (95% CI)
Hospitalizations OR (95% CI)
PM10
lag 0–1 lag 0–4
0.93 (0.98–1.01) 1.06 (1.04–1.08)
1.01 (0.99–1.03) 1.06 (1.03–1.10)
BS
lag 0–1 lag 0–4
0.97 (0.96–0.98) 1.03 (1.01–1.05)
0.93 (0.95–1.02) 1.02 (0.99–1.04)
SO2
lag 0–1 lag 0–4
1.08 (1.06–1.11) 1.12 (1.09–1.15)
1.10 (1.06–1.15) 1.12 (1.07–1.16)
NO2
lag 0–1 lag 0–4
0.99 (0.98–1.01) 1.03 (1.02–1.05)
1.01 (0.98–1.03) 1.04 (1.02–1.07)
a Case-crossover model adjusted for public holidays, holidays and weather variables. po0.05.
Table 5 Adjusted effects [OR (95% CI)]a of an increase of 10 mg/m3 of air pollutants on daily bronchiolitis consultations and hospitalizations with GAM model
PM10 SO2
lag 0–4 lag 0–4
Consultations OR (95% CI)
Hospitalizations OR (95% CI)
1.04 (1.02–1.07) 1.06 (1.03–1.09)
– 1.11 (1.05–1.17)
a
GAM model adjusted for long-term time trend, public holidays, holidays, weekday and weather variables. po0.05.
4. Discussion This study shows short-term relationships between air pollutants and the daily numbers of emergency hospital consultations and hospitalizations for bronchiolitis.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
It is not known why some previously healthy infants, when in contact with RSV, develop bronchiolitis whereas others have only mild symptoms (Sigurs et al., 1995). Only few studies, with conflicting results thus far, have investigated the relationship between air pollution and bronchiolitis. A Spanish study (Zamorano et al., 2003) had negative results, but did not use optimal statistical analysis. A recent study, using case-crossover analysis, evaluated air pollutants effect on 19 901 infants in California between 1995 and 2000 with an hospital discharge record for bronchiolitis in the first year of life (Karr et al., 2006). The authors found little support for a link between acute increases in ambient air pollution and infant bronchiolitis, except modestly increased risk for PM2.5 exposure for the most premature infants born at gestational ages between 25 and 29 weeks. A prospective cohort study of Chilean infants aged 4 months to 1 year focused on ‘‘wheezy bronchitis’’ and acute exposure to ambient air pollution (Pino et al., 2004) and the authors estimated that for each 10 mg/m3 increase of PM2.5, the risk for receiving a diagnosis of wheezy bronchitis increased by 5% (95% CI: 0–9%). No consistent associations were detected with NO2. An Italian study reported increased risks of bronchiolitis diagnosis in the first 2 years of life when parents reported lorry traffic ‘‘sometimes’’ (OR ¼ 1.52; 95% CI: 1.05–2.18) or ‘‘often’’ near their residence (OR ¼ 1.74; 95% CI: 1.09–2.77) compared with those who reported no lorry traffic (Ciccone et al., 1998). There is experimental evidence to support the hypothesis of an effect of air pollution on bronchiolitis. Authors model primarily exposure to pollution soon after infection and demonstrate potentiation of the disease process. Becker and Soukup (1999) suggest that possible increases in viral clinical symptoms associated with NO2 may result from effects of NO2 on host defenses that prevent the spread of the virus. Lambert et al. (2003) demonstrate a synergistic effect of ultrafine carbon black particles on RSV infection. Harrod et al. (2003) assessed the impact of inhaled diesel engine emissions (DEE) on the in vivo host response to RSV. They showed that DEE exposure modulates the host lung defense to respiratory viral infections, with increased levels of inflammatory mediators and mucous cell metaplasia, and increased RSV gene expression being seen. The health effects of other environmental factors on bronchiolitis, such as passive smoking, have been shown in several studies (McConnockie and Roghmann, 1986; Benigno et al., 1991). In particular, a recent study has shown that sudden heavy exposure to cigarette smoke may predispose an infant to acute bronchiolitis (Gurkan et al., 2000). Maternal smoking during pregnancy is also considered to be a main risk factor for the subsequent development of bronchiolitis (Cano Fernandez et al., 2003). However, the avoidance of exposure to tobacco smoke, cold air and air pollutants seems beneficial to long-term recovery from RSV bronchiolitis (Jeng and Lemen, 1997). Other identified risk factors include: being male (O’Kelly and Hillary, 1991), having a family history of asthma (McConnockie and Roghmann, 1986; Ruiz-Charles et al., 2002), having older siblings (McConnockie and
99
Roghmann, 1986), being premature (Ruiz-Charles et al., 2002) and living in an area of social and material deprivation (Spencer et al., 1996). In Paris, the local traffic and the high percentage of diesel engines contribute to the emissions of pollutants. With regard to the common source and the high correlation between these pollutants, we did not examine the effect of one pollutant while controlling for another and studied pollutants may be considered as indicators of a complex mixture of pollutants. Furthermore, any health effect of pollution exposure is not necessarily contemporaneous and, in our study, most of the significant associations are displayed with the 5-day means of pollutants. This result is consistent with exposure occurring during the most likely times of virus acquisition, incubation and replication/initial clinical recognition (Karr et al., 2006). We have shown that the results of the case-crossover design and GAM models were in agreement, with elevated emergency hospital consultations and hospitalizations for bronchiolitis risks being associated with air pollutant exposure. Both models assumed ecologic exposure and examined short-term effects. In the GAM model, each day is considered as producing an independent count, whereas in the case-crossover model each case of bronchiolitis is considered as an independent event. As the case-crossover model controls for time trends and seasonal patterns by design, this avoids concerns about complex mathematical modeling and adequacy of seasonal control. Moreover, the case-crossover design eliminates confounding by fixed characteristics of the individuals (such as age and sex) and reduces the potential confounding role of temperature and other weather variables. This study has several limitations. The clinical characteristics of RSV infection, especially in neonates, are often indistinguishable from those of other viral respiratory tract infections. In our data set, cases of RSV infection were confirmed by laboratory methods only at the beginning of the surveillance period of bronchiolitis (Thelot et al., 1998). Therefore, RSV infection in subsequent cases was diagnosed on the basis of manifestations suggestive of RSV infection (following a standardized definition). It is known that RSV infection accounts for up to 90% of the bronchiolitis cases that occur in infancy (Domachowske and Rosenberg, 1999; Hall, 2001). Also, with our data set it was not possible to distinguish between a first case of bronchiolitis and a recurrent case. This study suggests that air pollutants may act as a trigger for the occurrence of acute severe bronchiolitis cases. This finding should be supported by conducting case–control studies with individual data. Acknowledgment The authors would like to thank the ‘‘De´le´gation a` l’Information Me´dicale et a` l’Epide´miologie’’ of the ‘‘Assistance Publique-Hoˆpitaux de Paris’’ for providing access to the bronchiolitis data.
ARTICLE IN PRESS 100
C. Se´gala et al. / Environmental Research 106 (2008) 96–100
References Bateson, T.F., Schwartz, J., 1999. Control for seasonal variation and time trend in case-crossover studies of acute effects of environmental exposures. Epidemiology 10, 539–544. Becker, S., Soukup, J.M., 1999. Effect of nitrogen dioxide on respiratory viral infection in airway epithelial cells. Environ. Res. 81, 159–166. Benigno, V., Cusimano, R.A., Colanino, G., Basile, A., Varia, F., La Grutta, S., 1991. Is appearance of bronchiolitis affected by environmental and genetic factors? Pediatr. Med. Chir. 13, 155–157. Benzecri, J.P., 1992. Correspondence Analysis Handbook. Marcel Dekker, New York. Cano Fernandez, J., Zabaleta Caminio, C., de la Torre Montes de Neira, E., Yep Chullen, G.A., Melendu Crespo, J.M., Sainchez Bayle, M., 2003. Pre and postnatal tobacco exposure and bronchiolitis. Ann. Pediatr. (Barcelona) 53, 115–120. Ciccone, G., Forastiere, F., Agabiti, N., SIDRIA Collaborative Group, et al., 1998. Road traffic and adverse respiratory effects in children. Occup. Environ. Med. 55, 771–778. Desqueyroux, H., Momas, I., 1999. Pollution atmosphe´rique et sante´: une synthe`se des e´tudes longitudinales de panel publie´es de 1987–1998. Rev. Epidemiol Sante´ Publique 47, 361–375. Domachowske, J.B., Rosenberg, H.F., 1999. Respiratory syncytial virus infection: immune response, immunopathogenesis, and treatment. Clin. Microbiol. Rev. 12, 298–309. Dominici, F., McDermott, A., Zeger, S.L., Samet, J.M., 2002. On the use of generalized additive models in times-series studies of air pollution and health. Am. J. Epidemiol. 156, 193–203. Gilmour, M.I., Daniels, M., McCrillis, R.C., Winsett, D., Selgrade, M.K., 2001. Air pollutants enhanced respiratory disease in experimental animals. Environ. Health Perspect. 109 (Suppl. 4), 619–622. Gurkan, F., Kiral, A., Dagli, E., Karakoc, F., 2000. The effect of passive smoking on the development of respiratory syncytial virus bronchiolitis. Eur. J. Epidemiol. 16, 465–468. Hall, C., 2001. Respiratory syncytial virus and parainfluenza virus. N. Engl. J. Med. 344, 1917–1928. Harrod, K.S., Jaramillo, R.J., Rosenberger, C.L., et al., 2003. Increased susceptibility to RSV infection by exposure to inhaled diesel engine emissions. Am. J. Respir. Cell Mol. Biol. 28, 451–463. Hastie, T.J., Tibshirani, R.J., 1990. Generalized Additive Models. Chapman & Hall, London. Health Effect Institute, 2003. Revised analyses of time-series studies of air pollution and health. Special Report, May. Jeng, M.J., Lemen, R.J., 1997. Respiratory syncytial virus bronchiolitis. Am. Fam. Physician 55, 1139–1146. Karr, C., Lumley, T., Shepperd, K., et al., 2006. A case-crossover study of wintertime ambient air pollution and infant bronchiolitis. Environ. Health Perspect. 114, 277–281.
Katsouyanni, K., Schwartz, J., Spix, C., et al., 1996. Short-term effects of air pollution on health: a European approach using epidemiologic time-series data: the APHEA protocol. J. Epidemiol. Community Health 50 (Suppl. 1), S2–S8. Lambert, A.L., Mangum, J.B., Delorme, M.P., Everitt, J.I., 2003. Ultrafine carbon black particles enhance respiratory syncytial virusinduced airway reactivity, pulmonary inflammation, and chemokine expression. Toxicol. Sci. 73, 339–348. Law, B.J., Carbonel-Estrany, X., Simoes, E.A.F., 2002. An update on respiratory syncytial virus epidemiology: a developed country perspective. Respir. Med. 96, S1–S7. Maclure, M., 1991. The case-crossover design: a method for studying transient effects on the risk of acute events. Am. J. Epidemiol. 133, 144–153. McConnockie, K.M., Roghmann, K.J., 1986. Parental smoking, presence of older siblings and family history of asthma increase risk of bronchiolitis. Am. J. Dis. Child. 140, 806–812. O’Kelly, E.A., Hillary, I.B., 1991. Epidemiology of respiratory syncytial virus infection among infants over three winter seasons. Ir. J. Med. Sci. 160, 12–16. Pino, P., Walter, T., Oyarzun, M., Villegas, R., Romieu, I., 2004. Fine particulate matter and wheezing illnesses in the first year of life. Epidemiology 15, 702–708. Pope, C.A., 2000. Epidemiology of fine particulate air pollution and human health: biologic mechanisms and who’s at risk? Environ. Health Perspect. 108, 713–723. Ramsay, T.O., Burnett, R.T., Krewski, 2003. The effect of concurvity in generalized additive models linking mortality to ambient particule matter. Epidemiology 14, 18–23. Ruiz-Charles, M.G., Catillo-Renan, R., Bermadez-Felizardo, F., 2002. Risk factors associated with bronchiolitis in children under 2 years of age. Rev. Invest. Clin. 54, 125–132. Se´gala, C., 1999. Health effects of urban outdoor air pollution in children. Current epidemiological data. Pediatr. Pulmonol. 18, 6–8. Sigurs, N., Bjarnason, R., Sigurbergson, F., Kjellman, B., Bjorksten, B., 1995. Asthma and immunoglobulin E antibodies after respiratory syncytial virus bronchiolitis: a prospective cohort study with matched controls. Pediatrics 95, 500–505. Spencer, N., Logan, S., Scholey, S., Gentle, S., 1996. Deprivation and bronchiolitis. Arch. Dis. Child. 74, 50–52. Thelot, B., Be´nichou, J., Cheron, G., Chevalier, B., Begue´, P., Bourdillon, A., 1998. Surveillance e´pide´miologique hospitalie`re de la bronchiolite du nourisson par le re´seau ERBUS. Rev. Epidemiol Sante´ Publique 46, 277–288. WHO Task Group, 2004. Meta-analysis of time-series studies and panel studies of particulate matter (PM) and ozone (O3). WHO Regional Office for Europe, Copenhagen, Denmark. Zamorano, A., Marquez, S., Aringuiz, J.L., Bedegral, P., Sanchez, I., 2003. Association of acute bronchiolitis with climate factors and environmental contamination. Rev. Med. Chile 131, 1117–1122.
Environmental Research 106 (2008) 96–100 www.elsevier.com/locate/envres
Winter air pollution and infant bronchiolitis in Paris$ Claire Se´galaa,, David Poizeaua, Mounir Mesbahb, Sylvie Willemsa, Manuel Maidenbergc a
SEPIA-Sante´, 18bis rue du Calvaire, 56310 Melrand, France b LSTA, Universite´ Pierre et Marie Curie, Paris, France c R.E.S.P.I.R.E.R., Paris, France
Received 25 January 2007; received in revised form 5 May 2007; accepted 18 May 2007 Available online 21 June 2007
Abstract Respiratory syncytial virus (RSV) is one of the most common respiratory pathogens in infants and young children. It is not known why some previously healthy infants, when in contact with RSV, develop bronchiolitis whereas others have only mild symptoms. Our study aimed to evaluate the possible association between emergency hospital visits for bronchiolitis and air pollution in the Paris region during four winter seasons. We included children under the age of 3 years who attended emergency room services for bronchiolitis (following standardized definition) during the period 1997–2001. Two series of data from 34 hospitals, the daily number of emergency hospital consultations (n ¼ 50 857) and the daily number of hospitalizations (n ¼ 16 588) for bronchiolitis, were analyzed using alternative statistical methods; these were the generalized additive model (GAM) and case-crossover models. After adjustments for public holidays, holidays and meteorological variables the case-crossover model showed that PM10, BS, SO2 and NO2 were positively associated with both consultations and hospitalizations. GAM models, adjusting for long-term trend, seasonality, holiday, public holiday, weekday and meteorological variables, gave similar results for SO2 and PM10. This study shows that air pollution may act as a trigger for the occurrence of acute severe bronchiolitis cases. r 2007 Elsevier Inc. All rights reserved. Keywords: Air pollution; Respiratory disease; Bronchiolitis; Case crossover; GAM; Infant
1. Introduction The short-term effects of air pollution on humans are assessed using time-series studies and panel studies. Timeseries studies have shown broadly consistent associations between air pollutants and related outcomes such as total mortality, cardiorespiratory mortality and hospital admissions. Whereas most studies have focused on the general population and elderly subjects, there have been few on children (Pope, 2000; WHO Task Group, 2004; Health Effect Institute, 2003; Se´gala, 1999). Conversely, there have been numerous panel studies on mostly asthmatic children. Panel studies measure health effects, such as symptoms, on $ Funding: The study was supported by the ‘‘Conseil National Scientifique’’ of ANTADIR. Corresponding author. Fax: +33 2 97 28 81 10. E-mail address: [email protected] (C. Se´gala).
0013-9351/$ - see front matter r 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.envres.2007.05.003
an individual basis (Desqueyroux and Momas, 1999). In epidemiology, only few studies focused on the role of air pollution as a risk factor for specific respiratory infections, such as respiratory syncytial virus (RSV) bronchiolitis (Zamorano et al., 2003; Karr et al., 2006). However, experimental data have shown that air pollutants affect lung immune responses and inflammatory reactions and that these effects may underlie the increased risk for respiratory infections (Becker and Soukup, 1999; Gilmour et al., 2001; Harrod et al., 2003; Lambert et al., 2003). RSV is one of the most common respiratory pathogens in infants and young children. Most infants develop only mild or no symptoms when in contact with RSV. However, some infants develop RSV bronchiolitis, which is characterized by expiratory wheezing and respiratory distress. This condition presents a major public health problem, as the winter peak periods of RSV virus activity often result in numerous hospital consultations and hospitalizations.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
Very few studies have assessed bronchiolitis mortality. A US study reported between 171 and 510 bronchiolitis associated deaths per year between 1979 and 1997 among children under 5 years of age (Law et al., 2002). The aim of our study was to evaluate the association between emergency hospital visits for bronchiolitis and short-term exposure to air pollution in the Paris region during four winter seasons, using our alternative approach. Our null hypothesis was that air pollutants trigger acute severe bronchiolitis. 2. Materials and methods 2.1. Study subjects We included children under the age of 3 years who attended emergency room services for bronchiolitis during the period 1997–2001. Daily bronchiolitis data were obtained from the ‘‘Epide´miologie et Receuil des Bronchiolites en Urgence pour la Surveillance’’ (ERBUS) database (Thelot et al., 1998). This contains data, available every year, for the period between 15th October and 15th January from 43 hospitals within greater Paris. Physicians reviewed the clinical records of all daily visits and selected those with diagnoses matching the standard definition of bronchiolitis: respiratory dyspnea and/or sibilants and wheezing for children under the age of 3 years during the surveillance period. Two series of data were available: the daily number of emergency hospital consultations and the daily number of hospitalizations for bronchiolitis. In 1999–2000, data collection was temporarily stopped in some hospitals. Therefore, only the 34 hospitals providing complete data were retained for the analysis.
2.2. Environmental data Air pollutants were routinely measured by the AIRPARIF monitoring network. We retained data from the urban background monitoring sites only. These sites are representative of ambient air pollution in the Paris region and are not directly affected by local sources of pollution. The following air pollution data were recorded: sulfur dioxide (SO2) concentration (measured at 30 stations using ultraviolet fluorescence); nitrogen dioxide (NO2) concentration (measured at 21 stations using chemiluminescence); levels of suspended particles with an aerodynamic diameter less than 10 mm (PM10) (measured at 9 stations by tapered element oscillating microbalance—TEOM) and black smoke (BS; suspended black particulate) (measured at 21 stations using reflectometry—French standard method, NF-X 43-005). SO2, NO2 and PM10 were measured hourly and the mean value was calculated for each day. BS was measured as 24-h mean levels. The overall mean of the mean daily readings at all stations was calculated according to APHEA methods for analyzing correlations between stations and the replacement of missing values (Katsouyanni et al., 1996). The daily mean of 12 weather variables was determined at the Paris weather station (Me´te´o-France). A principal component analysis (Benzecri, 1992) allowed us to select the five least correlated meteorological variables. These were daily minimum temperature, mean humidity, precipitation, wind and atmospheric pressure.
2.3. Statistical methods Since the mid-1990s the generalized additive model (GAM) has been the most popular model for assessing, over time, the relationship between air pollution and hospitalization data. It uses nonparametric smoothing to control for time varying factors such as trend, season and weather variables. However, this model is complex and has been the subject to recent criticism (Dominici et al., 2002; Ramsay et al., 2003). Therefore, we used an alternative approach: the case-crossover design (Maclure, 1991).
97
We first carried out a descriptive analysis of environmental and bronchiolitis data and then compared these data between the four winter seasons. We then assessed the association between the daily visit data and air pollution using two different statistical methods. The case-crossover design is an adaptation of the case–control design, in which cases serve as their own controls (Maclure, 1991). The exposure of subjects to air pollutants at the time of the visit (case period) was compared with that at other times (control periods). We used a bidirectional design to avoid a time-trend bias (Bateson and Schwartz, 1999). To evaluate the contemporaneous effect and the cumulative effect on several days of pollutant concentrations, two exposure windows of 2-day mean (days 0 and 1) and 5-day mean (from day 0 to 4) were examined. Thus, we selected four control times for each case, and, for example, for the 2-day mean: day 7–8 and 14–15 before the visit, and 7–8 and 14–15 days after the visit were chosen. These matched pairs (strata) were analyzed with SAS, using conditional logistic regression. The analysis was carried out on 291 strata due to missing values in some case period or control period variables. The model was created stepwise, introducing public holidays, holidays and meteorological variables. The appropriate lag of meteorological variables was decided on the basis of Akaike’s information criteria. Also, we applied the GAM model to examine the association between the daily visit data and air pollution using a Poisson regression. This model is an extension of standard linear regression model, in which several predictors can be related to the dependent variable (the visit data) using a nonlinear smoothing function (Hastie and Tibshirani, 1990). According to recent recommendations (Health Effect Institute, 2003), GAM models should be run with penalized regression splines (P-splines) using R software. Long-term trend, seasonality, holiday, public holiday, weekday and meteorological variables adjustments were introduced into the model based on the visual inspection of residuals, predicted series and partial autocorrelation of residuals graphs. The forms of the curves of regression between bronchiolitis data and single pollutant were investigated using smoothing functions. If there was any linearity in the associations, the pollutants were introduced as a linear term in the model. For both models, the results were expressed as relative risk (odds ratio in the case-crossover design), with 95% confidence intervals (CI), for an increase of 10 mg/m3 in pollutant concentration.
3. Results We found a strong correlation between the levels of the pollutants during the 4 years studied. Temperature, rain and wind were negatively correlated with the pollutants, whereas atmospheric pressure was positively correlated with them (Tables 1 and 2). Several environmental variables differed for the four winter seasons. The mean concentrations of the four pollutants were lower and the mean minimal temperature was higher during the 2000–2001 winter season than during the three previous winter seasons (Table 3). The daily number of emergency hospital consultations ranged from 35 to 299 (n ¼ 139/day) during the study period and the daily number of hospitalizations for bronchiolitis from 9 to 117 (n ¼ 45/day). A seasonal peak occurred each December. As data are available only between mid-October and mid-January, each winter season is aggregated with the following one. We adjusted, using the case-crossover model, the associations between the daily variations in bronchiolitis consultations and hospitalizations and the urban air pollutant concentrations (Table 4). The four pollutants were positively associated with both consultations and hospitalizations.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
98
Table 1 Descriptive statistics of daily environmental variables during the study period
3
BS (mg/m ) NO2 (mg/m3) PM10 (mg/m3) SO2 (mg/m3) Hum. (%) Temp. (1C) Rain (mm) Pres. (hPa) Wind (m/s)
Mean7sd
(min.–max.)
Mean (sd)
16.9712.7 50.8716.7 22.7710.4 10.577.4 63.5714.5 7.373.4 2.174.5 101579.4 11.374.1
(3.4–113.0) (13–169.9) (6.3–99.1) (2.4–71.9) (28–94) (1.4–17.2) (0–47) (976.9–1040.2) (3–26)
Visits 4.1 Hospital 1.4 16.6 SO2 (mg/m3) BS (mg/m3) 21.5 55.0 NO2 (mg/m3) PM10 (mg/m3) 24.2 Temp. (1C) 5.1 Hum. (%) 81.6 Rain (mm) 2.3 Wind (m/s) 12.4 Pres. (hPa) 1013.9
BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind.
Table 2 Correlation coefficients between daily environmental variables BS
NO2
BS 1 0.83 1 NO2 PM10 0.87 0.74 SO2 0.76 0.78 Hum. 0.14 0.08 Temp. 0.29 0.34 Rain 0.31 0.34 Pres. 0.41 0.46 Wind 0.75 0.67
Table 3 Comparison of bronchiolitis data and environmental variables between the four winter seasons
PM10
SO2
Hum. Temp. Rain Pres. Wind
1 0.73 0.04 0.27 0.44 0.49 0.63
1 0.11 1 0.66 0.18 1 0.43 0.35 0.31 1 0.52 0.18 0.31 0.51 1 0.49 0.16 0.21 0.36 0.31 1
BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind. p40.05.
The 5-day mean was more consistently associated with bronchiolitis than the 2-day mean for all pollutants, but for SO2. For an increase of 10 mg/m3 of pollutant, the number of consultations increased by between 3% and 12% and the number of hospitalizations by between 2% and 12%. The strongest associations were for SO2 and PM10. Our analysis also showed independent associations with three meteorological factors. The risk of bronchiolitis increased with cold temperature, high humidity and strong wind. There was no association with barometric pressure and rain. The GAM model was also used to adjust the associations between air pollutants and daily bronchiolitis consultations and hospitalizations (Table 5). With consultations, only the 5-day mean of PM10 and SO2 showed a linear effect that can be expressed in terms of relative risk; with hospitalizations, it was the case only with the 5-day mean of SO2. An increase of 10 mg/m3 in PM10 is associated with an increase of 4% in the number of consultations. An increase of 10 mg/m3 in SO2 is associated with an increase of 6% in the number of consultations and an increase of 11% in the number of hospitalizations.
1997–1998
1998–1999
1999–2000
(5.5) 3.9 (4.2) 4.2 (1.8) 1.3 (1.6) 1.4 (9.0) 14.7 (9.0) 14.3 (13.5) 18.4 (13.4) 20.6 (13.5) 55.6 (14.6) 57.8 (10.4) 21.7 (11.6) 24.2 (4.0) 5.0 (4.3) 4.3 (9.0) 82.3 (7.1) 79.6 (4.2) 2.06 (4.0) 2.1 (4.7) 12.2 (4.4) 12.3 (10.1) 1018.4 (8.4) 1019.7
2000–2001
(4.6) 4.1 (1.6) 1.3 (7.0) 10.0 (18.0) 14.5 (17.4) 47.1 (13.6) 18.6 (3.7) 6.5 (8.4) 80.4 (4.1) 2.9 (4.3) 12.9 (11.2) 1009.2
(4.1) (1.6) (3.4) (7.9) (9.6) (5.6) (3.4) (8.1) (4.4) (4.3) (9.9)
Visits: daily number of emergency hospital consultations for bronchiolitis; Hospital: daily number of hospitalizations for bronchiolitis; BS: black smoke; NO2: nitrogen dioxide; PM10: suspended particles with an aerodynamic diameter of o10 mm; SO2: sulfur dioxide; Temp.: minimal temperature; Hum.: relative humidity; Rain; Pres.: atmospheric pressure; Wind. Table 4 Adjusted effects [OR (95% CI)]a of an increase of 10 mg/m3 of air pollutants on daily bronchiolitis consultations and hospitalizations with case-crossover model Consultations OR (95% CI)
Hospitalizations OR (95% CI)
PM10
lag 0–1 lag 0–4
0.93 (0.98–1.01) 1.06 (1.04–1.08)
1.01 (0.99–1.03) 1.06 (1.03–1.10)
BS
lag 0–1 lag 0–4
0.97 (0.96–0.98) 1.03 (1.01–1.05)
0.93 (0.95–1.02) 1.02 (0.99–1.04)
SO2
lag 0–1 lag 0–4
1.08 (1.06–1.11) 1.12 (1.09–1.15)
1.10 (1.06–1.15) 1.12 (1.07–1.16)
NO2
lag 0–1 lag 0–4
0.99 (0.98–1.01) 1.03 (1.02–1.05)
1.01 (0.98–1.03) 1.04 (1.02–1.07)
a Case-crossover model adjusted for public holidays, holidays and weather variables. po0.05.
Table 5 Adjusted effects [OR (95% CI)]a of an increase of 10 mg/m3 of air pollutants on daily bronchiolitis consultations and hospitalizations with GAM model
PM10 SO2
lag 0–4 lag 0–4
Consultations OR (95% CI)
Hospitalizations OR (95% CI)
1.04 (1.02–1.07) 1.06 (1.03–1.09)
– 1.11 (1.05–1.17)
a
GAM model adjusted for long-term time trend, public holidays, holidays, weekday and weather variables. po0.05.
4. Discussion This study shows short-term relationships between air pollutants and the daily numbers of emergency hospital consultations and hospitalizations for bronchiolitis.
ARTICLE IN PRESS C. Se´gala et al. / Environmental Research 106 (2008) 96–100
It is not known why some previously healthy infants, when in contact with RSV, develop bronchiolitis whereas others have only mild symptoms (Sigurs et al., 1995). Only few studies, with conflicting results thus far, have investigated the relationship between air pollution and bronchiolitis. A Spanish study (Zamorano et al., 2003) had negative results, but did not use optimal statistical analysis. A recent study, using case-crossover analysis, evaluated air pollutants effect on 19 901 infants in California between 1995 and 2000 with an hospital discharge record for bronchiolitis in the first year of life (Karr et al., 2006). The authors found little support for a link between acute increases in ambient air pollution and infant bronchiolitis, except modestly increased risk for PM2.5 exposure for the most premature infants born at gestational ages between 25 and 29 weeks. A prospective cohort study of Chilean infants aged 4 months to 1 year focused on ‘‘wheezy bronchitis’’ and acute exposure to ambient air pollution (Pino et al., 2004) and the authors estimated that for each 10 mg/m3 increase of PM2.5, the risk for receiving a diagnosis of wheezy bronchitis increased by 5% (95% CI: 0–9%). No consistent associations were detected with NO2. An Italian study reported increased risks of bronchiolitis diagnosis in the first 2 years of life when parents reported lorry traffic ‘‘sometimes’’ (OR ¼ 1.52; 95% CI: 1.05–2.18) or ‘‘often’’ near their residence (OR ¼ 1.74; 95% CI: 1.09–2.77) compared with those who reported no lorry traffic (Ciccone et al., 1998). There is experimental evidence to support the hypothesis of an effect of air pollution on bronchiolitis. Authors model primarily exposure to pollution soon after infection and demonstrate potentiation of the disease process. Becker and Soukup (1999) suggest that possible increases in viral clinical symptoms associated with NO2 may result from effects of NO2 on host defenses that prevent the spread of the virus. Lambert et al. (2003) demonstrate a synergistic effect of ultrafine carbon black particles on RSV infection. Harrod et al. (2003) assessed the impact of inhaled diesel engine emissions (DEE) on the in vivo host response to RSV. They showed that DEE exposure modulates the host lung defense to respiratory viral infections, with increased levels of inflammatory mediators and mucous cell metaplasia, and increased RSV gene expression being seen. The health effects of other environmental factors on bronchiolitis, such as passive smoking, have been shown in several studies (McConnockie and Roghmann, 1986; Benigno et al., 1991). In particular, a recent study has shown that sudden heavy exposure to cigarette smoke may predispose an infant to acute bronchiolitis (Gurkan et al., 2000). Maternal smoking during pregnancy is also considered to be a main risk factor for the subsequent development of bronchiolitis (Cano Fernandez et al., 2003). However, the avoidance of exposure to tobacco smoke, cold air and air pollutants seems beneficial to long-term recovery from RSV bronchiolitis (Jeng and Lemen, 1997). Other identified risk factors include: being male (O’Kelly and Hillary, 1991), having a family history of asthma (McConnockie and Roghmann, 1986; Ruiz-Charles et al., 2002), having older siblings (McConnockie and
99
Roghmann, 1986), being premature (Ruiz-Charles et al., 2002) and living in an area of social and material deprivation (Spencer et al., 1996). In Paris, the local traffic and the high percentage of diesel engines contribute to the emissions of pollutants. With regard to the common source and the high correlation between these pollutants, we did not examine the effect of one pollutant while controlling for another and studied pollutants may be considered as indicators of a complex mixture of pollutants. Furthermore, any health effect of pollution exposure is not necessarily contemporaneous and, in our study, most of the significant associations are displayed with the 5-day means of pollutants. This result is consistent with exposure occurring during the most likely times of virus acquisition, incubation and replication/initial clinical recognition (Karr et al., 2006). We have shown that the results of the case-crossover design and GAM models were in agreement, with elevated emergency hospital consultations and hospitalizations for bronchiolitis risks being associated with air pollutant exposure. Both models assumed ecologic exposure and examined short-term effects. In the GAM model, each day is considered as producing an independent count, whereas in the case-crossover model each case of bronchiolitis is considered as an independent event. As the case-crossover model controls for time trends and seasonal patterns by design, this avoids concerns about complex mathematical modeling and adequacy of seasonal control. Moreover, the case-crossover design eliminates confounding by fixed characteristics of the individuals (such as age and sex) and reduces the potential confounding role of temperature and other weather variables. This study has several limitations. The clinical characteristics of RSV infection, especially in neonates, are often indistinguishable from those of other viral respiratory tract infections. In our data set, cases of RSV infection were confirmed by laboratory methods only at the beginning of the surveillance period of bronchiolitis (Thelot et al., 1998). Therefore, RSV infection in subsequent cases was diagnosed on the basis of manifestations suggestive of RSV infection (following a standardized definition). It is known that RSV infection accounts for up to 90% of the bronchiolitis cases that occur in infancy (Domachowske and Rosenberg, 1999; Hall, 2001). Also, with our data set it was not possible to distinguish between a first case of bronchiolitis and a recurrent case. This study suggests that air pollutants may act as a trigger for the occurrence of acute severe bronchiolitis cases. This finding should be supported by conducting case–control studies with individual data. Acknowledgment The authors would like to thank the ‘‘De´le´gation a` l’Information Me´dicale et a` l’Epide´miologie’’ of the ‘‘Assistance Publique-Hoˆpitaux de Paris’’ for providing access to the bronchiolitis data.
ARTICLE IN PRESS 100
C. Se´gala et al. / Environmental Research 106 (2008) 96–100
References Bateson, T.F., Schwartz, J., 1999. Control for seasonal variation and time trend in case-crossover studies of acute effects of environmental exposures. Epidemiology 10, 539–544. Becker, S., Soukup, J.M., 1999. Effect of nitrogen dioxide on respiratory viral infection in airway epithelial cells. Environ. Res. 81, 159–166. Benigno, V., Cusimano, R.A., Colanino, G., Basile, A., Varia, F., La Grutta, S., 1991. Is appearance of bronchiolitis affected by environmental and genetic factors? Pediatr. Med. Chir. 13, 155–157. Benzecri, J.P., 1992. Correspondence Analysis Handbook. Marcel Dekker, New York. Cano Fernandez, J., Zabaleta Caminio, C., de la Torre Montes de Neira, E., Yep Chullen, G.A., Melendu Crespo, J.M., Sainchez Bayle, M., 2003. Pre and postnatal tobacco exposure and bronchiolitis. Ann. Pediatr. (Barcelona) 53, 115–120. Ciccone, G., Forastiere, F., Agabiti, N., SIDRIA Collaborative Group, et al., 1998. Road traffic and adverse respiratory effects in children. Occup. Environ. Med. 55, 771–778. Desqueyroux, H., Momas, I., 1999. Pollution atmosphe´rique et sante´: une synthe`se des e´tudes longitudinales de panel publie´es de 1987–1998. Rev. Epidemiol Sante´ Publique 47, 361–375. Domachowske, J.B., Rosenberg, H.F., 1999. Respiratory syncytial virus infection: immune response, immunopathogenesis, and treatment. Clin. Microbiol. Rev. 12, 298–309. Dominici, F., McDermott, A., Zeger, S.L., Samet, J.M., 2002. On the use of generalized additive models in times-series studies of air pollution and health. Am. J. Epidemiol. 156, 193–203. Gilmour, M.I., Daniels, M., McCrillis, R.C., Winsett, D., Selgrade, M.K., 2001. Air pollutants enhanced respiratory disease in experimental animals. Environ. Health Perspect. 109 (Suppl. 4), 619–622. Gurkan, F., Kiral, A., Dagli, E., Karakoc, F., 2000. The effect of passive smoking on the development of respiratory syncytial virus bronchiolitis. Eur. J. Epidemiol. 16, 465–468. Hall, C., 2001. Respiratory syncytial virus and parainfluenza virus. N. Engl. J. Med. 344, 1917–1928. Harrod, K.S., Jaramillo, R.J., Rosenberger, C.L., et al., 2003. Increased susceptibility to RSV infection by exposure to inhaled diesel engine emissions. Am. J. Respir. Cell Mol. Biol. 28, 451–463. Hastie, T.J., Tibshirani, R.J., 1990. Generalized Additive Models. Chapman & Hall, London. Health Effect Institute, 2003. Revised analyses of time-series studies of air pollution and health. Special Report, May. Jeng, M.J., Lemen, R.J., 1997. Respiratory syncytial virus bronchiolitis. Am. Fam. Physician 55, 1139–1146. Karr, C., Lumley, T., Shepperd, K., et al., 2006. A case-crossover study of wintertime ambient air pollution and infant bronchiolitis. Environ. Health Perspect. 114, 277–281.
Katsouyanni, K., Schwartz, J., Spix, C., et al., 1996. Short-term effects of air pollution on health: a European approach using epidemiologic time-series data: the APHEA protocol. J. Epidemiol. Community Health 50 (Suppl. 1), S2–S8. Lambert, A.L., Mangum, J.B., Delorme, M.P., Everitt, J.I., 2003. Ultrafine carbon black particles enhance respiratory syncytial virusinduced airway reactivity, pulmonary inflammation, and chemokine expression. Toxicol. Sci. 73, 339–348. Law, B.J., Carbonel-Estrany, X., Simoes, E.A.F., 2002. An update on respiratory syncytial virus epidemiology: a developed country perspective. Respir. Med. 96, S1–S7. Maclure, M., 1991. The case-crossover design: a method for studying transient effects on the risk of acute events. Am. J. Epidemiol. 133, 144–153. McConnockie, K.M., Roghmann, K.J., 1986. Parental smoking, presence of older siblings and family history of asthma increase risk of bronchiolitis. Am. J. Dis. Child. 140, 806–812. O’Kelly, E.A., Hillary, I.B., 1991. Epidemiology of respiratory syncytial virus infection among infants over three winter seasons. Ir. J. Med. Sci. 160, 12–16. Pino, P., Walter, T., Oyarzun, M., Villegas, R., Romieu, I., 2004. Fine particulate matter and wheezing illnesses in the first year of life. Epidemiology 15, 702–708. Pope, C.A., 2000. Epidemiology of fine particulate air pollution and human health: biologic mechanisms and who’s at risk? Environ. Health Perspect. 108, 713–723. Ramsay, T.O., Burnett, R.T., Krewski, 2003. The effect of concurvity in generalized additive models linking mortality to ambient particule matter. Epidemiology 14, 18–23. Ruiz-Charles, M.G., Catillo-Renan, R., Bermadez-Felizardo, F., 2002. Risk factors associated with bronchiolitis in children under 2 years of age. Rev. Invest. Clin. 54, 125–132. Se´gala, C., 1999. Health effects of urban outdoor air pollution in children. Current epidemiological data. Pediatr. Pulmonol. 18, 6–8. Sigurs, N., Bjarnason, R., Sigurbergson, F., Kjellman, B., Bjorksten, B., 1995. Asthma and immunoglobulin E antibodies after respiratory syncytial virus bronchiolitis: a prospective cohort study with matched controls. Pediatrics 95, 500–505. Spencer, N., Logan, S., Scholey, S., Gentle, S., 1996. Deprivation and bronchiolitis. Arch. Dis. Child. 74, 50–52. Thelot, B., Be´nichou, J., Cheron, G., Chevalier, B., Begue´, P., Bourdillon, A., 1998. Surveillance e´pide´miologique hospitalie`re de la bronchiolite du nourisson par le re´seau ERBUS. Rev. Epidemiol Sante´ Publique 46, 277–288. WHO Task Group, 2004. Meta-analysis of time-series studies and panel studies of particulate matter (PM) and ozone (O3). WHO Regional Office for Europe, Copenhagen, Denmark. Zamorano, A., Marquez, S., Aringuiz, J.L., Bedegral, P., Sanchez, I., 2003. Association of acute bronchiolitis with climate factors and environmental contamination. Rev. Med. Chile 131, 1117–1122.