Environmental Pollution 222 (2017) 486e494 Contents lists available at ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/locate/envpol Infants' indoor and outdoor residential exposure to benzene and respiratory health in a Spanish cohort* Amparo Ferrero a, b, *, Ana Esplugues c, a, b, Marisa Estarlich b, a, Sabrina Llop a, b, ~ iguez a, b Amparo Cases a, b, Enrique Mantilla d, Ferran Ballester a, c, b, Carmen In a Epidemiology and Environmental Health Joint Research Unit, FISABIOeUniversitat Jaume I, Universitat de Val encia, Avenida de Catalunya 21, 46020, Valencia, Spain b ndez Almagro, 3-5, 28029, Madrid, Spain Spanish Consortium for Research on Epidemiology and Public Health (CIBERESP), Melchor Ferna c n ~ ez, 13, 46010 Valencia, Spain Faculty of Nursing and Chiropody, Universitat de Val encia, Av Blasco Iba d gico, Charles R Darwin, 14, 46980 Paterna, Valencia, Spain Center for Mediterranean Environmental Studies, (CEAM), Parque Tecnolo a r t i c l e i n f o a b s t r a c t Article history: Received 11 October 2016 Received in revised form 20 November 2016 Accepted 21 November 2016 Available online January 2017 Benzene exposure represents a potential risk for children's health Apart from being a known carcinogen for humans (group according to IARC), there is scientific evidence suggesting a relationship between benzene exposure and respiratory problems in children But results are still inconclusive and inconsistent This study aims to assess the determinants of exposure to indoor and outdoor residential benzene levels and its relationship with respiratory health in infants Participants were 1-year-old infants (N ¼ 352) from the INMA cohort from Valencia (Spain) Residential benzene exposure levels were measured inside and outside dwellings by means of passive samplers in a 15-day campaign Persistent cough, low respiratory tract infections and wheezing during the first year of life, and covariates (dwelling traits, lifestyle factors and sociodemographic data) were obtained from parental questionnaires Multiple Tobit regression and logistic regression models were performed to assess factors associated to residential exposure levels and health associations, respectively Indoor levels were higher than outdoor ones (1.46 and 0.77 mg/m3, respectively; p < 0.01) A considerable percentage of dwellings, 42% and 21% indoors and outdoors respectively, surpassed the WHO guideline of 1.7 mg/m3 derived from a lifetime risk of leukemia above 1/100 000 Monitoring season, maternal country of birth and parental tobacco consumption were associated with residential benzene exposure (indoor and outdoors) Additionally, indoor levels were associated with mother's age and type of heating, and outdoor levels were linked with zone of residence and distance from industrial areas After adjustment for confounding factors, no significant associations were found between residential benzene exposure levels and respiratory health in infants Hence, our study did not support the hypothesis for the benzene exposure effect on respiratory health in children Even so, it highlights a public health concern related to the personal exposure levels, since a considerable number of children surpassed the abovementioned WHO guideline for benzene exposure © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Keywords: Infant Air pollution Benzene Respiratory tract infections Respiratory signs and symptoms Introduction A recent WHO report about “preventing disease through healthy environment” claimed that 12.6 million deaths in 2012, representing almost a quarter of the global deaths worldwide, were * This paper has been recommended for acceptance by David Carpenter * Corresponding author Epidemiology and Environmental Health Joint Research ncia, Avenida de Catalunya Unit, FISABIOeUniversitat Jaume I, Universitat de Vale 21, 46020, Valencia, Spain E-mail address: ferrero_ampsan@gva.es (A Ferrero) attributable to unhealthy environments (WHO, 2016) Of these, 1.7 million involved children under years of age, the main causes being respiratory tract infections and diarrheic diseases Regarding household and ambient air pollution, 4.3 and 3.7 million deaths can be attributable to these factors, respectively (Lim et al., 2012) Due to their behavioral patterns, children are considerably more exposed to air pollution than adults Inhalation rate is higher than in adults and infants' height places them much closer to the exhaust pipes of cars (Landrigan et al., 2004; Moya et al., 2004) Moreover, children's respiratory system is characterized by its high vulnerability since development of the lungs is not fully completed until around the third year of life (Schittny and Burri, 2007) During http://dx.doi.org/10.1016/j.envpol.2016.11.065 0269-7491/© 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) A Ferrero et al / Environmental Pollution 222 (2017) 486e494 the developmental period, certain toxic exposures such as tobacco smoke and air pollutants could have an influence on the normal development of the lungs This, in turn, could determine respiratory illnesses later in childhood and adulthood (Duijts, 2012; Shi et al., 2007) Benzene, a volatile organic compound (VOC) derived from petroleum, is a common solvent The main source of benzene is tobacco smoke, but its levels indoors are also increased after using gas stoves, and cleaning or painting products (Kotzias et al., 2005; Topp et al., 2004; WHO, 2010) Outdoor sources are traffic exhaust fumes, and emissions from industrial facilities and/or gas stations Levels of this pollutant in indoor air are usually higher than outdoors, since it is more commonly derived from household sources and activities Apart from being known carcinogens, according to the International Agency for Research on Cancer (IARC, 2012), some VOCs such as formaldehyde or benzene have been suggested to play a role in the development of respiratory diseases (Ferrero et al., 2014; Fuentes-Leonarte et al., 2009; IARC, 2012; WHO, 2010) They could irritate the airways or alter the multipathway inflammation mechanisms in epithelial lung cells after inhalation at normal concentrations indoors (ranging within mg/m3) (Mascelloni et al., 2015; Pariselli et al., 2009; Wang et al., 2014) However, the strongest and most consistent evidence on respiratory effects in epidemiological studies in children are for formaldehyde, while for benzene exposure evidence is still scarce and inconclusive (Bolden et al., 2015; Heinrich, 2011; Mendell, 2007; Roda et al., 2013) As part of the multicenter prospective INMA project (for its acronym in Spanish: Infancia y Medio Ambiente) (Guxens et al., 2012), a previous study assessed the relationship between ambient exposure to benzene and NO2 in residential dwellings in two developmental periods of children (prenatal and during 1st year of life), and respiratory health in children aged in four INMA cohorts (Aguilera et al., 2013) Exposure levels in both periods were estimated by means of land use regression (LUR) models and predicted for residential addresses LUR models were based on measurements in more than 50 representative locations spread over each study area during the mothers' pregnancy (years 2004e2007) Regarding respiratory health associations, benzene exposure in the prenatal period was not related to respiratory infections or symptoms In the INMA cohort of Valencia, one study has empirically measured and described residential exposure levels to VOCs (specifically to benzene, toluene, ethyl benzene and xylene, known as BTEX compounds) in indoor and outdoor air in half of the dwellings with an infant population during their first years of life (Esplugues et al., 2010) Here, we examined levels of residential exposure to benzene measured indoors and outdoors at the first year of life, as well as its relationship with respiratory health in infancy Thus, the purposes of this study were to assess (i) the environmental, lifestyle and individual factors associated to benzene exposure levels measured directly inside and outside the dwellings at age 1, and (ii) their relationship with respiratory symptoms and problems suffered by infants during their first year of life Methods 2.1 Study population The study population consisted of children participating in the INMA Project in Valencia (Spain) Their mothers were recruited during pregnancy between November 2003 and June 2005 at the first prenatal health examination at the reference hospital (Hospital n et al., 2005) At “La Fe” in Valencia) (Guxens et al., 2012; Ramo birth, 787 newborns were included in the INMA-Valencia cohort Of these, 708 participated at follow up at one year of age (2006e2007) 487 Participants in the present study were a subsample of 352 children with available measurements of any of the pollutants monitored at age (VOC or NO2) (Supplementary Fig 1) Written informed consent was obtained from parents and the study was approved by the ethics committee of the reference hospital The study area of the INMA-Valencia cohort covers around 1372 km2 including 34 municipalities with a wide range of sociodemographic and environmental traits The area is made up of a typically urban zone (city of Valencia), a metropolitan area containing the neighboring towns around the city of Valencia, a semi-urban area dedicated to industrial and agricultural activity, and a typically rural zone 2.2 Benzene exposure assessment Residential benzene exposure levels were measured inside and outside the dwellings during the children's first year of life (years 2006e2007) They were monitored in a 15-day campaign by means of passive samplers located in the living room of each dwelling and outside the home in a window or on a balcony The method for benzene level measurement has been described elsewhere (Esplugues et al., 2010a) Briefly, passive samplers were stainless steel tubes 8.9 cm in length and with an internal diameter of 0.5 cm containing a solid sorbent for chemical fixing (Tenax TA 60/80; Analytical Columns, New Addington, Croydon, England) Chemical analyses consisted in thermal desorption of the samples coupled to gas chromatography with flame ionization detection (TD/GC/FID) The detection limit (DL) was 1.04 mg/m3 Benzene levels for residences below the DL (36% indoors and 42% outdoors) were considered as DL/2 2.3 Respiratory outcomes At the first-year follow up, health information was obtained by means of a questionnaire administered to parents by trained personnel Questions on respiratory symptoms and diagnosis during the previous 12 months were based on a structured questionnaire from the AMICS study (Sunyer et al., 2001; Polk et al., 2004) Respiratory outcomes in the present study were (i) persistent cough (lasting more than weeks), (ii) low respiratory tract infections (LRTI) including doctor's diagnosis of bronchitis, bronchiolitis or pneumonia, and (iii) wheezing, considered as dichotomous variables Specific questions (in Spanish) and further information can be found in Supplementary Fig 2.4 Covariates Information related to environmental exposure or respiratory health was also obtained by a questionnaire administered to parents in different visits Information gathered during the first trimester of pregnancy were maternal age, educational level, maternal country of birth, familial history of allergies and zone of residence; during the third trimester they were maternal tobacco consumption during pregnancy and social class based on parental occupation; and at the first-year follow up, they were parental tobacco consumption, frequency of traffic in the nearest street, distance to an industrial area, number of rooms and rooms equipped with air conditioning, number of people living at home, daily ventilation rate, household size, attendance at day-care centers, type of cooking and heating, household pesticide use, habitual use of painting and cleaning products such as bleach, ammonia, hydrochloric acid, degreasers, solvent stain-removers, sprays for air fresheners and perfumed products In addition, NO2 exposure levels during pregnancy and at age were tested as confounder variables Details of the methodology for the estimation of residential ambient exposure levels to NO2 during 488 A Ferrero et al / Environmental Pollution 222 (2017) 486e494 pregnancy by means of LUR models have been presented elsewhere ~ iguez et al., 2009) At age 1, residential exposure levels to NO2 (In were monitored indoors and outdoors in a similar way to the present study (Esplugues et al., 2010b) 2.5 Statistical analysis A descriptive study was conducted for residential benzene exposure levels and for respiratory health Indoor and outdoor residential exposure levels and the ratio between indoor and outdoor levels (I/O ratio) were tabulated according to monitoring season and zone of residence, and their distributions were compared by means of a Kruskal-Wallis test Spearman correlations between benzene and NO2 exposure levels in different periods were also calculated For subsequent analysis, logarithmic transformations of residential exposure levels were calculated Concerning respiratory outcomes, accumulated incidences (AI) in the study population were calculated A detailed comparison of each respiratory outcome between the whole INMA-Valencia population and the study population (Chi2 test) as well as according to individual factors (logistic regression) is presented in Supplementary Material Table Factors associated to residential benzene exposure levels were obtained by means of Tobit regression models These models are appropriate when considering an explanatory continuous variable with left-censored distribution in the DL (Aurrekoetxea et al., 2015; ndez and Tobías Garce s, 2002) Multivariate models Bleda Herna were constructed considering factors (i) related to exposure in univariate analysis (p < 0.2), and (ii) with a significant likelihood ratio test in the multivariate models (p > 0.1) In order to adjust for seasonality of benzene levels, monitoring season was included in the models as a forced factor, regardless of its statistical significance Associations between residential exposure levels with respiratory health outcomes at age were assessed by means of logistic regression Variables related with the outcome were selected following previous studies, and confounders were obtained for their relationship with benzene exposure (those found to be significant in previous multivariate Tobit regression analyses) Factors included in multivariate models were selected by one of the following criteria: in a first step, factors related to respiratory outcomes were considered (p < 0.1 in likelihood ratio test); and in a second step, covariates modifying effects estimates of each exposure by more than 20% when included in the model Health associations were expressed as the odds ratio (OR) and the confidence intervals at 95% (CI 95%) derived for an increment of one unit of natural logarithm of residential benzene exposure levels (indoor and outdoor) during winter and the lowest during summer (Table 1) for both indoors and outdoors Children living in rural areas were generally exposed to higher benzene exposure levels indoors than those in urban and metropolitan areas By contrast, children from metropolitan and urban areas were exposed to higher levels outdoors The median I/O ratio was close to unity; however, it was higher during winter and spring, and in semi-urban, residential and rural areas 3.2 Factors associated to residential benzene exposure levels In the adjusted Tobit regression model (Table 2), residential benzene exposure levels indoors were significantly associated with season (levels were higher in all the seasons compared to summer), with maternal country of birth (they were higher in the children of non-Spanish women compared to Spanish), with exposure to parental tobacco smoke during age (they were higher for those children whose parents both smoked at home compared to those children whose parents did not smoke during their first year), with maternal age at delivery (they were lower in children from older mothers aged more than 35 years) and with type of heating (they were significantly higher in dwellings with wood or coal heating compared to those with central heating) Regarding residential exposure outdoors, results from the Tobit regression model showed that levels were significantly associated with season (they were higher during spring months compared to winter), with maternal country of birth (they were higher in dwellings of non-Spanish women compared to Spanish), with parental tobacco smoking at age (levels were higher in dwellings where only mothers smoked at home), with distance to an industrial area (levels were higher in dwellings further than 50 m from an industrial area compared to those closer) and with zone of residence (levels were lower in residential areas in both non-urban zones and in rural zones compared to urban areas) Results 3.1 Benzene exposure levels Residential benzene exposure levels were measured in the dwellings of 352 children, which represents more than a half of the population at age (Supplementary Fig 1) Valid data were obtained for 341 and 292 dwellings, indoors and outdoors, respectively Median levels (percentiles 5, 95) for indoors and outdoors were 1.46 (0.52, 8.63) mg/m3 and0.77 (0.52, 2.58) mg/m3, respectively (Fig 1) In general, correlations between residential exposure levels (indoors and outdoors) for benzene and NO2 were low, except for a correlation between residential NO2 indoors and outdoors, which was moderate-high (r ¼ 0.65) (Supplementary Table 1) According to season, the highest levels of benzene occurred Fig Benzene exposure levels measured indoors and outdoors of the residences in INMA-Valencia cohort at 1st year of life (2006e2007) Red continuous line indicates the boundary value established by Directive 2008/50/EC on Ambient air quality (5 mg/m3) Red discontinuous line indicates the WHO guideline concentration associated with an excess in lifetime risk of leukemia of 1/100 000 (1.7 mg/m3) In this graphic have not shown outliers for residential exposure indoors (35; 58 and 88.7 mg/m3) A Ferrero et al / Environmental Pollution 222 (2017) 486e494 Table Indoor and outdoor benzene exposure levels measured at the residences in the INMA-Valencia cohort at first year of life (2006e2007), according to monitoring season and zone of residence (expressed in mg/m3) Ratio of indoor and outdoor benzene levels (I/O ratio) is also presented Indoor levels N Monitoring season Summer Autumn Winter Spring Zone of residence Urban Metropolitan Semi-urban Residential areab Rural 341 107 85 49 100 341 27 150 86 58 20 Outdoor levels N Monitoring season Summer Autumn Winter Spring Zone of residence Urban Metropolitan Semi-urban Residential areab Rural 292 81 73 39 99 292 23 126 73 51 19 Ratio I/O N Monitoring season Summer Autumn Winter Spring Zone of residence Urban Metropolitan Semi-urban Residential areab Rural Total 290 81 72 39 98 290 23 126 73 50 18 290 p-valuea Median % (%): Percentage of dwellings with a indoors/outdoors ratio of benzene levels above %