Dimensional and Chemical Characterization of Airborne Particles in Schools : Respiratory Effects in Children

This work reports the detailed characterization of child exposure to particles in three naturally-ventilated Italian schools. Two polluted urban schools and a rural one are considered in this study. Dimensional and chemical analyses of particles were performed by measuring particle number concentrations, particle size distributions, OC/BC concentrations and relative inorganic and organic fractions, both indoors and outdoors. Respiratory function, exhaled Nitric Oxide (eNO) and Prick Skin tests were also performed on 75 children attending the three schools in order to evaluate the effects of such exposure on the children’s respiratory systems. Exposure to particles at the urban schools was found higher than at the rural one, from both dimensional and chemical perspectives. Indoor particle number concentrations ranged from 1.95 × 10 to 3.49 × 10 part./cm, whereas indoor black carbon concentrations varied between 1.9–13.9 μg/m, with the lowest levels of both measured at the rural school. In contrast to the dimensional and chemical results, the respiratory function and eNO test results were statistically similar among children attending all the three schools: the lung function values were within 20% of the European Community for Steel and Coal predicted ones, with the eNO data for healthy children ranging from 10.1 to 11.5 ppm. Medical outcomes indicate that these levels of school exposure did not significantly affect the children’s respiratory outcomes.


INTRODUCTION
An increasing interest in children exposure to particles at schools has been documented in the scientific literature over the last decade.Indeed, a number of papers that aimed to measure particle concentrations in classrooms, both in terms of mass and number, were carried out (Fromme et al., 2007;Zollner et al., 2007;Diapouli et al., 2008;Guo et al., 2008;Tippayawong et al., 2009;Mullen et al., 2011;Zhang and Zhu, 2012).In fact, the exposure at schools was considered crucial in the evaluation of long-and short-term effects of air pollution on this specific population group, i.e., children (Burtscher and Schüepp, 2012).

Why Schools?
Schools represent an appealing microenvironment in terms of air pollution since children spend a lot of time therein.In fact, the International Review of Curriculum and Assessment Frameworks Internet Archive (INCA, 2009) estimated that children living in developed countries spend from 175 to 220 days at school for 5-8 hours per day.Moreover, schools are often placed near highly-trafficked urban roads leading to high exposure since children attend school during daytime hours on weekdays, i.e., when the traffic intensity is maximum.Thus, particles can penetrate indoor causing health effects (Janssen et al., 2003): indeed, a number of studies reported PM 10 concentrations in schools exceeding WHO guidelines (Lee and Chang, 2000;Mi et al., 2006;World Health Organization, 2006;Ekmekcioglu and Keskin, 2007;Diapouli et al., 2008;Branis et al., 2009;Mejía et al., 2011), other studies recognized lowered lung function, inflammation and asthma of children living or attending schools near major roadways (Brunekreef et al., 1997;Dales et al., 2008;Kulkarni and Grigg, 2008).Therefore, school environments could be a major contributor to children exposure to air pollutants.
Anyway, particle number concentration levels were usually found lower than the corresponding outdoor ones when no indoor sources are in operation.Average particle concentrations ranging from 0.6 × 10 3 part./cm 3to about 3.0 × 10 4 part./cm 3 were measured in naturally ventilated schools with no indoor sources (Fromme et al., 2007;Guo et al., 2008;Weichenthal et al., 2008;Mullen et al., 2011).The high variation in total particle concentration can be due to both different outdoor background concentrations and spatial variability inside the school itself.
Exposure at school strongly changes when indoor sources are in operation.As example, indoor-to-outdoor concentration ratios higher than 1 were found either when fan-heaters were in operation or during food preparation activity (through both cooking and microwave), smoking, classroom cleaning, painting, gluing, and drawing (Guo et al., 2008;Morawska et al., 2009;Mullen et al., 2011;Zhang and Zhu, 2012) leading to particle concentrations higher than 1.0 × 10 5 part./cm 3 .
Concerning the chemical composition of particles sampled at schools, an interesting study was performed by Oeder et al. (2012).They combined PM elemental composition with toxicity testing of PM sampled both indoor and outdoor at six naturally ventilated elementary schools in Munich, Germany recognizing different indoor PM chemical composition from outdoor one.In fact, PM 10 sampled indoor shown higher proportion (than outdoor) of resuspended crustal materials (Si, Al), previously penetrated from outdoor, mainly because of children movement.Also emission of skin flakes from children themselves was documented indoor.From a toxicity point of view, indoor PM composition was found more cytotoxic than outdoor-derived one, even if no particular attention to the distance of schools from major roads was paid in such measurement discussions.

Why Children?
As regards the choice of population group, children is the most biologically vulnerable group (Brent et al., 2004;Makri et al., 2008) because even when children are exposed to the same aerosol concentrations, they can receive particle doses larger than adults mainly due to their increased breathing rate and more frequent mouth breathing compared to adults (Ginsberg et al., 2005;Heinrich and Slama, 2007;Bateson and Schwartz, 2008).Moreover, when children and adults particle doses are compared, the absolute dose could represent a misleading parameter as it does not take into account the different air-tissue interface (alveolar surface area exposed to air) in the alveolar region of the lung.In fact, after their birth children's alveolar surface area keeps growing (2.8 m 2 at birth) reaching a maximum value at the age of 20 (about 75 m 2 ; Dunnill, 1962;Thurlbeck, 1982).Therefore, children exposure to lower absolute particle concentrations can anyhow lead to a higher deposited particle "density" (i.e., dose normalized to the actual available air-tissue interface) in the alveolar region.Buonanno et al. (2012a) showed that such normalized deposition trend is completely different from the absolute one: in particular, infants received the highest normalized particle number deposition among the considered population age groups as it is higher or equal to the one experienced by working adults (age 19-65).For that reason, infants and children are exposed to the highest normalized doses in the period of their life when it could be more dangerous from a physiological point of view.This should be considered since research has shown that particle exposure of developing lungs can permanently affect lungs themselves.In particular early life exposure to ultrafine particles (UFPs, particle smaller than 100 nm) can result in persistent alterations in distal airway architecture that is characterized by an initial decrease in airway cell proliferation (Lee et al., 2010).

Aims of the Work
A number of studies found that school environment may lead to inflammations, asthma worsening, allergic reactions as well as lung respiratory function impairment (Janssen et al., 2003;Kim et al., 2004;Moshammer et al., 2006;Holguin et al., 2007;Dales et al., 2008;Zhao et al., 2008;McConnell et al., 2010).Both medical and air-quality experts tried to find a link between children respiratory symptoms and particle levels in schools whose they were exposed to.Medical-based studies are usually able to perform accurate respiratory measurements (typically respiratory function, exhaled nitric oxide) but, typically, indoor particle concentrations experienced by children were only estimated from traffic data and distance from roadways (Janssen et al., 2003;Moshammer et al., 2006;Holguin et al., 2007;Dales et al., 2008;Zhao et al., 2008;McConnell et al., 2010).
Air-quality-based studies can evaluate the actual particle concentrations whose children are exposed to (Brunekreef et al., 1997;Mejía et al., 2011); anyway, a lack of understanding still exists in a correct correlation between children exposure to particle in schools and adverse respiratory effects.
To this purpose the international "Ultrafine particle from traffic emission on children health (UPTECH)" project was conceived to provide epidemiology results concerning the long-term effects of the exposure to ultrafine particles emitted by motor vehicles on children health in schools.Details on the UPTECH project are available online (http://www.ilaqh.qut.edu.au/Misc/UPTECH%20Study%20Design.htm).The present study has been performed as part of the UPTECH study in order to compare the actual exposure of children to airborne particles at schools to their long-term respiratory effects.In particular, the paper is focused on i) the accurate characterization of children exposure to particles at three schools in Cassino (Italy), and ii) the respiratory response of children attending such schools.Therefore, both dimensional and chemical analyses of particles were performed at different spatial scales: particle number concentrations, particle size distributions, OC/BC concentrations as well as the relative inorganic (i.e., elemental composition) and organic (such as PAH, aldehyde, ketone content) fractions were measured.As regards medical tests, respiratory function (i.e., spirometry) test, aimed to evaluate the lung function, exhaled Nitric Oxide (eNO) test, aimed to evaluate airway inflammations, and Prick Skin Test (PST), to quantify the allergic subjects, were performed on 75 children attending the three schools.

Sampling Site
Three naturally ventilated public schools (named S 1 , S 2 and S 3 ) were selected in the area of Cassino (Central Italy), and investigated from December 2010 to December 2011.Schools placed both in urban and rural area were considered: in particular, the schools S 1 and S 2 are placed in the urban area of Cassino, whereas school S 3 is located in a rural area (about 5 km away from urban area of Cassino).Moreover, S 3 is sited in a court not directly facing streets and where only one school bus can go into.Therefore, S 1 and S 2 can be considered similar in terms of exposure to traffic-related particles, whereas, S 3 presents a different exposure in terms of traffic density and degree of urbanization.A summary of the schools' characteristics and experimental campaign is reported in Table 1.School buildings analyzed were built 30-60 years ago, they are naturally ventilated and equipped with a radiator heating system.Classrooms are 37-47 m 2 marble pattern flooring.They are supplied with standard school tables and chairs, and a blackboard with chalk at the front.About 16-27 children occupied the classrooms during the experimental analyses.Air quality data were collected at different sampling locations commonly used by students both indoors (I S ) and outdoors (O S ), within the school sites.In particular, one outdoor site (O S ) and three indoor sites (I SA , I SB , I SC ) were considered.
The outdoor site (O S ) was set at the school entrance, typically facing the closest busy road to represent traffic effects on school air quality as best as possible.Different classrooms were selected as indoor sites (I SA , I SB , I SC ) to evaluate the spatial variation inside the schools.The instruments for indoor sampling were placed inside the rooms on a school table (0.8 m above the ground) in the proximity of students' seating area.During the indoor measurement period, windows and doors were kept closed, as it represents the ventilation condition typically encountered during cold periods.The whole measurement campaign was performed during school hours (08.30a.m.-01.30p.m. for S 1 and S 2 ; 08.30 a.m.-04.30p.m. for S 3 ) over a two week period in each school.
Moreover, a background sampling point was set in the urban area of Cassino and was located on the rooftop of the University of Cassino and Southern Lazio's main building.This location was chosen to monitor continuously the background particle concentration from October 2010 to July 2012.

Materials and Methods
In order to relate children exposure to particles at schools to their feasible adverse respiratory response, particle dimensional and chemical analyses were performed along with respiratory function tests on 75 children attending the three schools, whose age is 10.1 ± 1.1 years.

Particle Dimensional Characterization
Particle dimensional analyses were performed through several instruments both indoor and outdoor at the height of 0.8 m above the floor.At the outdoor site (O S ) particle concentrations, number distributions and black carbon (BC) concentrations were continuously monitored during the whole school time.To this purpose: i) a butanol-Condensation Particle Counter (CPC 3775, TSI Inc.) was used to monitor total particle number concentration down to 4 nm with 1-s frequency sampling; ii) a Scanning Mobility Particle Sizer spectrometer (SMPS 3936, TSI Inc.) was used to measure particle number distributions in the range of 0.015-0.700μm with a 64 channels per decade size resolution and a 180-s time resolution; iii) an Aethalometer (AE51, Magee Scientific), operating through particle light absorption technique, was employed to detect black carbon concentration with 1-s time resolution.CPC 3775 and SMPS 3936 were connected to mains power, otherwise, the AE51 ran on batteries.
At the indoor sites I SA , I SB , I SC , total particle number and BC concentrations were measured.Therefore, i) water-  Condensation Particle Counters (WCPC 3781 and 3786, TSI Inc.) were used to monitor total particle number concentration (down to 6.5 and 2.5 nm for WCPC 3781 and 3786, respectively) with 1-s frequency sampling; ii) Aethalometers (AE51, Magee Scientific) were employed to detected BC concentration with 1-s time resolution.At the background site, a further CPC 3775 was used to continuously measure total particle number concentration.The instrument, located on the University building's rooftop, was protected from rain and wind.Particle number concentrations were measured with a 30-s time resolution for three years.In addition, NO x urban concentration data were derived from the fixed-site monitoring station measurements performed by the Lazio Regional Agency for Prevention and the Environment (ARPA; www.arpal azio.net)according to the European Directive 2008/50/EC (EC Directive, 2008) threshold limit values.

Particle Chemical Characterization
The inorganic composition was investigated by a nuclear not-destructive technique: the Instrumental Neutron Activation Analysis (INAA).It is considered a reference analytical technique because all the experimental stages are totally traceable without any physical-chemical sample treatment so the positive and/or negative artifacts formation was reduced (Avino et al., 2008;Capannesi et al., 2009).Even if it is not a routinely analytical technique, its high sensitivity and multi-elemental character allows to reach good Limit of Detection (LOD) and accuracy for many elements (Avino et al., 2006): this chance is very important when it is necessary to find out elements at trace or ultratrace levels, especially in toxicological studies.A detailed description of the analytical procedure is reported in our previous studies (Avino et al., 2006(Avino et al., , 2008)).Briefly, samples, blank and standards, are irradiated at a neutron flux of 2.6 × 10 12 n/cm 2 /s for 30 h in the "Lazy Susan" rotary rack and at 2.68 × 10 13 n/cm 2 /s for 1 min in the central channel of the Triga Mark II nuclear reactor of the ENEA-Casaccia Laboratories.After irradiation, γ-ray spectrometry measurements of different duration were carried out for analyzing Al, As, Br, Ca, Cd, Cl, Cr, Cu, Fe, Hg, K, Mn, Ni, S, Se, Si, Ti, V and Zn according their nuclear data.In particular, a Ge(HP) Ortec detector (FWHM 1.68 keV at 1332 keV) was used.For the analysis (as well Quality Assurance and Quality Control, QA/QC), primary (As, Cd, Co, Cr, Cs, Fe, Hg, La, Ni, Sb, Se, Sm and Zn; Carlo Erba, Milan, Italy) and secondary (United States Geochemical Survey, USGS, n. 1, 4, 6 and Coal Fly Ash, NIST, n. 1633a) standards were used.A small amount of each sample underwent PIXE analysis to detect Pb levels.The analytical determination procedure is deeply investigated in previous papers (Blasi et al., 1990;Avino et al., 2011).
The organic fraction was chosen in accordance with the possible effects on the human health.In particular, for the PAH composition, we have focused our attention on benz The whole procedure is reported in Avino et al. (2002) and Buonanno et al. (2010).External calibrations with deuterated PAHs mixed standards were performed: calibrations curves were linearly fitted with correlation coefficients always higher than 0.9985 for all PAHs.Limits of detection (LODs between 1.4 pg/m 3 and 163 pg/m 3 with a signal-to-noise ratio of 3:1) and limits of quantification (LOQs between 4.8 pg/m 3 and 624 pg/m 3 ; signal-to-noise ratio of 10:1) were calculated (Knoll, 1985).Furthermore, the organic fraction composition was also investigated in terms of aldehydes (acrolein, crotonaldehyde, formaldehyde, acetaldehyde, benzaldehyde) and ketones (acetone): the content was determined using a gas chromatograph with flame ionization detector (GC-FID, Fisons, Milan, Italy), equipped with an HP-PLOT Q column (15 m × 0.53 mm ID × 30 μm thickness, J&W Scientific) following the procedure reported in Chan et al. (2010).
Finally, the carbonaceous fractions, such as Organic Carbon, OC, and Elemental Carbon, EC, was determined in the samples collected in the three different schools.The Total Carbon (TC) determination and the relative OC/EC separation can be performed by various analytical methods that exploit the different optical, thermal or chemical particle properties (Avino et al., 2001): in this study a methodology based on the particle thermal properties allowing the OC/EC separation, was used.In fact, the analyzer (mod.5004, Rupprecht & Patashnik, US), equipped with a no-dispersive infrared detector (NDIR), measures the CO 2 amount released by the oxidation at different temperatures of the particulate matter sampled on a special collector (Avino et al., 2001).

Auxiliary Data: Traffic and Microclimatic Data
In order to quantify the exposure to airborne particles at school due to traffic-related emissions, traffic intensity characteristics of streets close to the schools was monitored through video cameras during the whole measurement campaign.In Table 2 traffic density data recorded during the three experimental campaigns are reported.In particular, worst traffic conditions at S 1 and S 2 were recognized when school starts and finishes (08:30 a.m. and 01:30 p.m.).Similar average and peak traffic densities were measured at S 1 and S 2 , even if, the related traffic emissions could be different since all the roads surrounding S 1 were mainly crossed by cars (50% of diesel cars; only three school buses were used to transport children to this school), whereas one of the roads faced by S 2 was characterized by a notnegligible percentage of heavy duty (HD) vehicles (7.7%; typically buses; Buonanno et al., 2011).As previously pointed out, S 3 was not directly exposed to traffic, anyway the closest street traffic density is negligible as well as not telling peak traffic conditions were documented since a number of children usually go to school by walk and by bus (see Table 3).
As regards microclimatic condition measurements, a David Vantage Pro weather station was placed on the rooftop of University of Cassino and Southern Lazio building recording temperature, relative humidity, and wind speed/direction with a 15 min resolution.In Table 2 microclimatic conditions measured during the three experimental campaigns are reported.In terms of wind direction, a detectable prevailing wind direction was only recognized during the experimental campaign at the school S 2 : SSE, which is from the road towards the school.This is a key point as S 2 faces a road showing heavy-traffic conditions probably leading to different children exposure in different classrooms.Therefore, measuring particle concentrations at several indoor sites is particularly interesting for S 2 .

Respiratory Function Measurements and Population Study Description
In order to evaluate the children respiratory function both pulmonary function (spirometry) and exhaled Nitric Oxide (eNO) tests were performed on 75 children attending the three schools (16, 28 and 31 children attending S 1 , S 2 , and S 3 , respectively) by certified respiratory therapists.
As a general concept, pulmonary function test (spirometry) is a physiological test measuring air volumes and flow rates inhaled or exhaled by an individual: it reflects lung structure, airway caliber, and lung size (Pellegrino et al., 2005).In particular, a spirometry test can measure both the forced vital capacity (FVC) and the forced expiratory volume (FEV 1 ).The former is the volume delivered during an expiration made as forcefully and completely as possible starting from full inspiration; the latter, is the air volume delivered in the first second of an FVC manoeuvre.The maximum instantaneous flow achieved during a FVC maneuver (FEF MAX ); the forced expiratory flow related to some portion of the FVC curve (FEF X% ).
In the present work, spirometry tests were performed for the selected children at the schools through a computerized spirometer using European Community for Steel and Coal criteria (Quanjer et al., 1983) (Medgraphics, Cardiorespiratory Diagnostics, St. Paul, Minn, USA).The test procedure is based on a full expiratory and inspiratory loop as a single manoeuvre.In particular, each child was asked to i) take a rapid full inspiration from room air through the mouth, ii) perform an expiration with maximum force in a breathing tube just inserted in his mouth, iii) take a quick maximum inspiration.In this way, FEV 1 (L), FEF MAX (L/s), and FEF 25-75% (mean forced expiratory flow between 25% and 75% of the FVC, also known as the maximum mid-expiratory flow, L/s) were measured.Several maneuvers were performed for each child in order to achieve three acceptable flowvolume loops (American Thoracic Society, 2005).Results, expressed as percentage of European Community for Steel and Coal predicted values (Quanjer et al., 1983), represent the average of three acceptable measurements.
Concerning Exhaled Nitric Oxide (eNO) test, it allows to evaluate the increase of NO production in inflamed tissues since it is due to the increased expression of inducible nitric oxide synthase (iNOS) present in inflammatory cells and epithelial cells (American Thoracic Society, 2005; Baraldi and de Jongste, 2001).In the present experimental campaign measures of eNO were performed using a handheld electrochemical analyzer (NObreath ® , Bedfont Scientific Ltd., Rochester, Kent, UK) according to the American Thoracic Society and European Respiratory Society guidelines (American Thoracic Society, 2005).In particular, children were asked to inhale ambient air to near total lung capacity and then exhaled for 10 s at a constant flow rate of 50 mL/s through a disposable mouthpiece into the device.All measurements were performed between 9:00 a.m. and 12:00 p.m. Children did not eat or perform any strenuous physical activity during the last 60 minutes before testing was carried out.Moreover, the medical tests were performed during days showing no statistically different background concentrations as reported in the result section.
A Prick Skin Test (PST) was also performed by allergists on the 75 children under investigation to quantify the allergic subjects among the selected population group.The test was performed according to the standardized ISAAC II protocol (Weiland et al., 2004).The following allergens were monitored: dermatophagoides farinae, dermatophagoides pteronyssinus, cladosporium clarosporoides, alternaria tenuis, penicillium mix, cat dander, dog dander, cypress pollen, mixed grasses, olive, lolium perenne, wall pellitory, plantaginaceae, German cockroach (Stallergenes, Italy).Dermatophagoides farinae and dermatophagoides pteronyssinus allergens are reported in the following as house dust mites (HDMs).
Moreover, in order to describe the study population, to measure potential confounders and effect modifiers relevant the analysis such as housing conditions, socioeconomic status, exposure to environmental tobacco smoke and ethnicity, a questionnaire was developed following the International Study of Asthma and Allergies in Childhood (ISAAC) guidelines.As regards medical aspects, parents were asked to answer questions concerning the medical history of their children (since they were babies) in terms of allergies, asthma, bronchitis, cough, breathing, and cardiovascular health.Furthermore, time and children activities were recorded by children supervised by their parents in a two-day activity diary.

Data Analysis
The values were expressed as means ± standard deviation.When the measurement distributions were Gaussian, differences between groups were tested by analysis of variance (ANOVA).When a significant difference between groups was observed, intergroup comparisons were made using Student's t test.A p value < 0.01 (99% confidence level) was considered significant.

Population Study Characteristics
Characteristics of the population study were obtained through the ISAAC questionnaire.In Table 3 the main results in terms of children exposure apart from school are reported.Concerning children exposure during transportation at the school, questionnaires documented that children attending the two urban schools (S 1 and S 2 ) mainly live in urban areas (50% and 44% for S 1 and S 2 , respectively), even if a not-negligible percentage of them came from both rural (37%) and suburban areas (13% and 19% for S 1 and S 2 , respectively).This leads to high percentage of children going to school by car.On the contrary, children attending S 3 merely live in rural areas.Thus, different routes are usually travelled by children: the greatest part of children attending S 3 pass through suburban route, whereas, children going to S 1 and S 2 typically pass through urban and mixed routes characterized by heavy traffic conditions.In order to exhaustively evaluate the influence of potential confounders, indoor exposure characteristics should be considered analyzing indoor particle sources.To this purpose, questionnaire results reported in Table 3 shows the percentage of children potentially exposed to secondhand tobacco smoke, heating sources, and cooking.No significant differences amongst children attending the three different schools were revealed in terms of potential exposure to such indoor particle sources.Anyway, the authors point out that, even if the indoor exposures are potentially similar, the actual exposure could be different and no measurable only through questionnaires; otherwise individual exposures and doses should be measured through personal monitors (Buonanno et al., 2012b).
As regard the preexisting respiratory disorders of analyzed children, in Table 3 are reported the percentage data of allergic and asthmatic children in the three schools on the basis of PST and questionnaire results.The percentage of asthmatic children was found to be 25%, 11% and 16% for S 1 , S 2 , and S 3 , respectively, whereas the fraction of atopic children (showing positive response to at least one of 14 allergens tested) were 38%, 54% and 32% for S 1 , S 2 , and S 3 .Children reporting no preexisting respiratory disorders as asthma, allergies, and allergic rhinitis (reported in Table 3 as healthy children) are 44%, 32%, and 29% in S 1 , S 2 , and S 3 , respectively.

Dimensional Characterization
Results in terms of particle concentrations at different spatial scales analyzed (urban background, school outdoor, school indoor) are here reported.As regards urban background concentrations, in Fig. 1 total particle number concentration data are reported and compared to NO x concentrations.Particle number concentrations represent monthly average of three-year data (October 2010-July2012) collected at background site (rooftop of the University of Cassino and Southern Lazio's main building).In particular, in Fig. 1(a) data are classified as function of the "average" day (Weekday, Saturday, Sunday), whereas in Fig. 1(b), monthly average particle number concentrations (obtained as daily weighted average of monthly Weekday, Saturday and Sunday values) are compared to monthly average NO x concentrations measured at fixed-site monitoring station of Cassino by ARPA.
Altogether, Fig. 1 shows that higher particle concentrations occur during cold periods typically on weekdays (maximum concentration on Weekday in January, 5.1 × 10 4 part./cm 3 ).Such seasonal variability was also recognized in terms of NO x concentration: this is probably due to the temperature inversion phenomena typically occurring in winter time.Moreover, both outdoor NO x and particle concentrations are characteristics of traffic emissions, this is the reason why their average monthly trends are significantly correlated (Fig. 1(b)).As regards the average particle concentrations shown in Fig. 1(a), Weekday concentrations are higher mainly because of the higher traffic density typical of working days, Saturday and Sunday concentrations are, on average, 26% and 55% lower than Weekday ones, respectively.The Student's t test was applied to the and hourly average data measured at background site during the twoweek period of school measurements (Jan 10 th -24 th for S 1 , 2011; Feb 7 th -21 st for S 2 , 2011; Dec 5 th -19 th for S 3 , 2011) in order to evaluate if statistically significant day-by-day differences were recognized in children background exposure attending the same school.The p values for all the intergroup comparisons were higher than 0.01; then children of the same school are exposed to no statistically different background concentrations.This is a crucial aspect since the respiratory function measurements were randomly performed during the two-week period from 9.00 a.m. to 12.00 p.m.: then all the children attending the same school underwent to medical analysis without differences in short-term exposure.
Concerning particle concentration at school scale, in Table 4 data measured during the experimental campaigns for each school are reported: they represent the average (and standard deviation) of particle concentration measurement results obtained over a two-week period from 08.30 a.m. to 01.30 p.m. for S 1 and S 2 , and from 08.30 a.m. to 04.30 p.m. for S 3 .Outdoor particle concentrations (O S ) were measured higher than the background ones for urban schools (S 1 and S 2 ), whereas, S 3 (located in a rural area) showed an outdoor concentration lower than urban background value (indoorto-background concentration ratio, I/B, higher than 1).Concerning indoor scale, the average indoor particle concentration (average of I SA , I SB , and I SC measurements) were found in the range 2.04 × 10 4 -3.49× 10 4 part./cm 3 which is in agreement with other experimental data obtained for school with no indoor sources (Fromme et al., 2007;Guo et al., 2008;Weichenthal et al., 2008;Mullen et al., 2011).As expected, since no indoor sources were in operation during the experimental campaigns, average indoor concentrations were measured lower than outdoor one (ranging from 2.77 × 10 4 to 4.72 × 10 4 part./cm 3 ), leading to indoor-to-outdoor concentration (I/O) ratios lower than 1 for each school.
Summarizing, indoor particle number concentration values are associated to traffic-related outdoor particle concentrations when indoor sources are not present, in fact, the school buildings are able to reduce children exposure to such particles of 30-40% with respect to outdoor measured exposure due to particles low penetration efficiency.Since the main particle source is urban traffic, it is interesting to compare indoor and outdoor BC concentrations; indeed, BC is recognized as a proxy of diesel exhaust particles (Schneider et al., 2008;Kim et al., 2011).Outdoor BC concentrations are in the range 3.2-16.3μg/m 3 , similar values were measured in other European cities (Schneider et al., 2008;Vanderstraeten et al., 2011); as expected, BC concentrations were higher at urban schools (S 1 and S 2 ) than rural (S 3 ) both indoor and outdoor.In particular, the highest BC particle concentration was measured at S 2 , mainly because it faces a busy road with heavy duty vehicles (S 2 ), leading in this way to a higher indoor BC penetration.Correspondingly, BC concentrations measured at S 3 indoor and outdoor scales are five-and seven-fold lower than S 2 values, respectively.Overall, outdoor BC concentrations were higher than indoor ones for all the schools analyzed leading to I/O BC ratios lower than 1.In addition, rural-to-urban differences in terms of BC concentrations were larger than the corresponding particle concentration data highlighting a greater spatial heterogeneity in BC compared to UFPs.This is also confirmed by the wider deviation of the BC data with respect to particle ones.
As shown children exposure to particles varied significantly amongst the three schools investigated, moreover, school periods and hours are different for each school.Thus, in order to quantify the actual children overexposure to particle at school with respect to background concentrations, in Table 4 normalized school exposure concentration data (I S,normalized ) are reported.This parameter was evaluated as: S,school time S,normalized school year school time where I S,school time represents the average indoor particle number concentration ("Indoor particle concentration" column of Table 4) at school-scale evaluated during school hours (08.30a.m.-01.30p.m. for S 1 and S 2 ; 08.30 a.m.-04.30p.m. for S 3 ) over the two-week period in each school; B school time represents the average background particle number concentration ("Background particle concentration" column of Table 4) evaluated during school hours (08.30a.m.-01.30p.m. for S 1 and S 2 ; 08.30 a.m.-04.30p.m. for S 3 ) over the two-week period in each school; B school year is the annual average total particle number concentration ("School-year background" column of Table 4) evaluated during school hours (08.30a.m.-01.30p.m. for S 1 and S 2 ; 08.30 a.m.-04.30p.m. for S 3 ) by considering only the school months (from September to May) on the basis of three-year data (October 2010-July2012) collected at background site.Such exposure data range from 1.07 × 10 4 to 3.67 × 10 part./cm 3 , as expected S 3 showed the lowest exposure level, whilst, the higher exposure was evaluated at S 2 .In Fig. 2 particle size distributions measured at urban schools (S 1 and S 2 ) were reported.Distributions measured at different hours were compared in order to highlight the influence of the traffic: school starting (08.30-09.30a.m.), school finishing (12.30-01.30p.m.) and school time (10.30a.m.-12.00 p.m). Particle distributions showed a main mode at about 100 nm, typical of aged urban aerosol, at any considered time.Nevertheless, a second mode was measured in the range 20-60 nm which is due to fresh particle emission (Kittelson et al., 2006;Wehner et al., 2009) from diesel and gasoline fuelled vehicles in the proximity of the schools.As expected such mode is mainly detectable in the morning rush hour (08.30-09.30a.m.), whereas, it completely disappears during the 10:00 a.m.-12:00 p.m. period when the traffic intensity is reduced.

Chemical Characterization
The Quality Control/Quality Assurance (QC/QA) of the inorganic data determined by INAA, fundamental issue for evaluating the analytical methodology used, was performed through an intercomparison campaign for 16 elements promoted by International Atomic Energy Agency (IAEA) on air filter samples between 130 different worldwide laboratories using both spectrochemical, electrochemical and nuclear analytical techniques (Avino et al., 2008).
Basically, the concentration levels of all the elements are quite similar to the content of an urban profile (Table 5).In the last column the element content of PM 2.5 sampled in urban area (downtown Rome) is reported: data evidence low levels of As and Ni in all the locations, very low levels of toxic elements such as Cr, Cu, Ni and Se in the third locations (sub-rural place) where Hg is close to the LOD.On the contrary, it is evidenced a marked Br value in school S 3 characteristic of an important rural activity.58.0 * = p(S 1 -S 2 ) < 0.01; ∆ = p(S 1 -S 3 ) < 0.01; ▲ = p(S 2 -S 3 ) < 0.01.Furthermore, in order to evaluate the origin of such elements an analysis was carried out studying the Enrichment Factors (EFs).The EFs were calculated as reported in Avino et al. (2008) using Fe as normalizing trace element (Mason and Moore, 1982).Elements with EF values significantly higher than 1 can be considered not originating from local soil background and may be attributable to long-range transport phenomena from other natural sources and/or anthropogenic sources (if present), or, in some cases, to a possible preferential uptake.The obtained EF data showed that most of the EFs are below 1 except for Se and Br.Se was measured equal to 2.9, 3.9 and 1.7 in S 1 , S 2 and S 3 , respectively, evidencing a possible presence of other sources.Br data were measured equal to 5.7, 7.7 and 37.0 at S 1 , S 2 and S 3 confirming the presence of rural activity mainly at S 3 .
About organic compounds, some hazardous species in organic fraction (i.e., PAHs, aldehydes and ketones) were determined as reported in Table 6.PAHs were determined to identify the relative contributions of anthropogenic and natural sources.Basically, the PAH levels determined in the two urban schools are lower than the ones determined in the urban area of Rome characterized by intense anthropogenic activities (last column in Table 6); the samples collected at sub-rural place show a very low concentrations.In particular, in this site 8 PAHs show concentration levels below the LODs.The total PAHs account for about 0.15% of OC tot in samples collected at S 1 and S 2 and 0.08% at S 3 .As regards aldehydes and ketones, their values are quite similar to urban areas whereas it should be evidenced the absence of acrolein, crotonaldehyde and benzaldehyde and the low levels of formaldehyde, acetaldehyde and acetone in the samples collected at S 3 .
Finally, an investigation on the carbonaceous fraction, i.e., Total Carbon (TC), content was performed.In the samples collected in S 1 and S 2 it is made up of 34-41% of Elemental Carbon and 59-66% of Organic Carbon: both ranges are narrow considering the same pollution sources and the same territory where the sampling occurred.Samples collected at S 3 the level of OC fraction, characterized by a most complex fraction due to the plenty of compounds, is more than three-times the EC fraction, reaching 77% of the particulate matter composition.Summarizing, chemical analysis results shown not statistically significant differences (p > 0.01) in element composition, organic and inorganic fractions measured at the two urban schools (S 1 and S 2 ), whereas at rural school (S 3 ) were measured lower values and statistically different when compared to S 1 and S 2 ones as reported from p-value data in Table 5 and Table 6.
With regard to exhaled nitric oxide (eNO), geometric average concentrations ranging from 14.2 to 17.2 ppb were measured (16.2 ± 1.7 ppb); such data can be considered similar to the ones typically found in healthy children (Barroso et al., 2008;Olivieri et al., 2008).A statistically negligible difference in average data was observed amongst schools (p > 0.01), indicating no correlations amongst airways inflammation and children exposure at school.Since allergic sensitizations and asthma symptoms are known to influence eNO values (Byrnes et al., 1997;Sordillo et al., 2011;Yao et al., 2011), in Table 5 eNO results are also segregated for allergic, asthmatic and healthy children, in order to deepen the possible effect of school exposure on children airway inflammation.Allergic sensitizations and allergic rhinitis increase children eNO, in fact, average values were: 43.9 ± 25.2 ppb for HDM allergic children, 32.9 ± 13.6 ppb for atopic children; 21.5 ± 5.4 ppb for children with allergic rhinitis.Concerning asthmatic children, eNO values were found higher than non-asthmatic ones in S 1 and lower than non-asthmatic in S 2 and S 3 : this not clear trend was also recognized by Jackson et al. (2009) in their cohort study focused on the evaluation of environmental factors involved in the development of asthma and allergic diseases in children of different ages.Lowest average eNO data were measured in healthy children (11.0 ± 0.8 ppb): the negligible variability amongst eNO data in healthy children attending the three schools seems to confirm the absence of school exposure influence on airway inflammation data.
In conclusion, both pulmonary function and exhaled nitric oxide tests were negative correlated to the actual exposure of children to particles (both from a dimensional and chemical point of view) at school clearly contrasting the results found by similar researches (Janssen et al., 2003;Moshammer et al., 2006;Dales et al., 2008;McConnell et al., 2010) where a negative effects of children exposure in schools located near roads was recognized.However, the authors point out that a number of previous studies reporting respiratory measurements are mostly based on outdoor pollutant measurements, and indoor exposure models based on traffic data: this approach evidently can lead to incorrect estimates of actual children exposure at school.Finally, in order to generalize our finding, future developments of the present work should contemplate testing on larger number of children.

CONCLUSIONS
In the present paper the correlation between children exposure to particle at schools and adverse respiratory effects was deepened through experimental campaign carried out in three schools characterized by different locations and distances from urban traffic.In particular, two urban and one rural schools were analyzed.Particle dimensional and chemical characterizations were performed as well as children medical tests (spirometry, eNO and Prick Skin Test).
Indoor exposure to particle sources was recorded through an exposure questionnaire which showed that no significant differences amongst children attending the three different schools were revealed in terms of potential exposure to indoor particle sources.
As regards the exposure at school, children attending the two urban schools were exposed to particle concentrations from two-to almost four-time larger than the urban school ones.Results of particle chemical composition revealed quite similar element composition, organic and inorganic fractions for the two urban school, whereas the rural school is characterized by lower levels of toxic elements (Cr, Cu, Ni, Se, Hg), higher levels of Br (characteristic of rural activities) and negligible concentrations of PAHs, aldehydes and ketones.Summarizing, the exposure profiles of children attending urban and rural schools were found quite different both from a dimensional and chemical point of view.Anyway, on the basis of both pulmonary function and exhaled nitric oxide test results, the different actual exposure at school does not imply different children adverse respiratory effects.In fact, spirometry parameter values (FEV 1 , FEF MAX and FEF 25-75% ) were statistically similar amongst children attending the three schools; furthermore, such values were comparable to the European Community for Steel and Coal predicted values.Besides, also no statistically significant differences in eNO data were found in children analyzed in the three schools when potential confounder data (asthmatic and allergic children) are excluded.Concluding, the paper shows that the three Italian schools analyzed have not influence on children respiratory responses; thus, attention should be paid to other microenvironments that mainly contribute to the daily individual exposure of children to airborne particles.

Outdoor
s Indoor: I SA , I SB I SA , I SB classrooms have windows facing streets with similar traffic conditions.Dec 5 th -19 th , 2011

Fig. 1 .
Fig. 1.Statistics of urban particle number and NO x concentrations measured at background site and ARPA fixedmonitoring station, respectively: a) Weekday, Saturday and Sunday monthly average of three-year data; b) comparison between weekly three-year average particle number (daily weighted average of Weekday, Saturday and Sunday values) and NO x concentrations (monthly average).

Table 2 .
Auxiliary data measured during experimental campaigns at schools: traffic densities and microclimatic conditions.

Table 3 .
Characterization of the children population in terms of exposure and preexisting respiratory disorders.

Table 5 .
Synoptic table (mean value ± standard deviation) of elements concentration (ng/m 3 ) determined by INAA in PM samples collected in the three different locations.

Table 6 .
Mean value and standard deviation of PAH, aldehyde and ketone concentrations.

Table 7 .
Results of spirometry and exhaled Nitric Oxide (eNO) tests for children attending the selected schools (eNo values are reported in terms of geometric mean ± standard deviation).