The effects of outdoor air pollution on the respiratory health of Canadian children : A systematic review of epidemiological studies

1School of Public Health, University of Alberta, Edmonton, Alberta; 2Department of Public Health, School of Medicine, Universidad Industrial de Santander, Bucaramanga, Colombia; 3Department of Pediatrics; 4Department of Emergency Medicine, University of Alberta, Edmonton, Alberta Correspondence: Dr Laura A Rodriguez-Villamizar, Department of Emergency Medicine, University of Alberta, 7.40 University Terrace, 8440-112 Street Northwest, Edmonton, Alberta T6G 2T4. Telephone 780-492-5489, fax 780-407-3982, e-mail lrodrigu@ualberta.ca Outdoor air pollution is a global problem with serious effects on human health (1). In fact, it has been estimated that in 2011, approximately 80% of the world’s population was exposed to air pollution levels that exceeded WHO guidelines (2,3). Air pollution is a complex mixture of compounds that vary greatly depending on the emission sources. Typically, the so-called criteria air pollutants (CAP, which include particulate matter [PM], ozone [O3], lead [Pb], carbon monoxide [CO], sulphur oxides [SOx] and nitrogen oxides [NOx]), are monitored in surveillance air-quality networks. Interestingly, PM itself represents a complex mixture of particles of various sizes and concentrations of soil, metals, organics, inorganics, elemental carbon, ions and endotoxins, among other contaminants (4). Recently, the PM2.5 (PM size ≤2.5 μm in aerodynamic diameter) has been the focus of most outdoor air pollution and health studies due to its ability to penetrate the lung tissue and induce local and systemic effects (4). Based on findings for lung and bladder cancer, the International Agency for Research on Cancer recently classified outdoor air pollution, as a whole, as a group 1 carcinogen (5). In addition, well-documented associations exist between outdoor air pollution and other health conditions including asthma, cardiovascular diseases, respiratory infections, adverse birth outcomes and additional cancers, such as leukemia (1,6,7). Children are considered to be highly susceptible to the effects of air pollution due to the immaturity of their immune system, the potential for developmental disruption, greater amount of time spent outdoors and, therefore, higher exposure levels, and a relatively high volume of air exchange relative to body mass (8,9). In fact, outdoor air pollution consistently shows an adverse effect on childhood respiratory health, especially on asthma outcomes, with a total estimated health care cost (among 34 countries, including Canada) of approximately US$1.7 trillion in 2010 (10,11). Asthma is one of the top 10 causes of years lost due to disability in male children worldwide (12). The effects of outdoor air pollution on asthma and other respiratory conditions have been the subject of study involving many adult and children populations in Canada and LA Rodriguez-Villamizar, A Magico, A Osornio-Vargas, BH Rowe. The effects of outdoor air pollution on the respiratory health of Canadian children: A systematic review of epidemiological studies. Can Respir J 2015;22(5):282-292.

elsewhere.Air pollution levels in Canada are relatively low and most Canadian cities experience extreme low temperatures.Thus, Canadian studies offer a unique opportunity to examine the effects of more moderate doses of air pollution compared with those experienced in many other nations (13).In addition, Canada boasts one of the highest percentage of foreign-born citizens (14), being a society of mixed languages, cultures and genetic diversity.Recent findings suggest that the influence of genetic diversity on the population's susceptibility to air pollution is an important factor that should be considered in this field (15).In 2007, a systematic review of air pollution and children's health in Canada analyzed the results of epidemiological studies published between January 1989 and December 2004 (16).From 11 studies over a 15-year period, the review identified associations between respiratory health effects and at least some CAP measurements.These associations were, however, often weaker than those reported in studies conducted in other countries.This was believed to be due to the lower levels of air pollution in Canada, the lower number of hours spent outdoors during the colder Canadian winters, as well as reduced levels of outdoor air pollution infiltrating into homes, which could act to reduce personal exposure to outdoor air pollution.
The objective of the present study was to conduct a comprehensive systematic review of the literature reporting the effects of outdoor air pollution on the respiratory health of children in Canada.We focused on the literature published during the past 10 years to update the previous review, identify new findings on types of associations between air pollutants and childhood respiratory health, and evaluate differences in those associations across Canadian cities.

METHODS
An a priori systematic literature review protocol was developed.The research question addressed in the present review was: what is the effect of outdoor air pollution exposure on respiratory conditions in Canadian children?Respiratory conditions included respiratory symptoms, lung function measurements and the use of health services due to respiratory disease.

Search strategy
To increase sensitivity, the search strategy used in the previous review (16) was modified.Specifically, four electronic bibliographic databases (MEDLINE, CINAHL, Scopus and CAB abstracts) were searched (Appendix 1).In general, databases were searched with a combination of terms and derived key words including variation to the following basic terms: "air pollution", "outdoor air pollution", "asthma", "respiration disorders", "respiratory health", "respiratory symptoms", "child", "adolesc", "youth" and "Canada".In the MEDLINE search, the names of 16 specific Canadian cities were included to increase search sensitivity.The search strategy was not restricted by language or publication type.A Google Scholar web search was conducted and references of relevant studies were scanned and selected as a complementary search strategy.

Study selection and data extraction
The criteria for selecting studies included: any observational analytic design; publication date between January 1, 2004 and November 30, 2014; population included and reported data for children up to 18 years of age residing in Canada; exposure(s) included any nonbiological outdoor air pollutant whether measured directly or inferred (ie, by proximity to roadways), with special interest in the CAP (CO, NO 2 , SO 2 , O 3 , and PM 10 and PM 2.5 ); and outcomes included health services use (HSU), lung function measurement or self-reported respiratory symptoms.Studies that included a subset or cohort of Canadian children in which the data for the Canadians were not presented separately were excluded.Two reviewers (LR-V and AM) independently screened the identified articles' titles and abstracts to select the articles for full review, and reviewed citations that were found to be potentially relevant for inclusion.A third reviewer (AO-V) resolved disagreement.Articles selected for full review were screened in a second round to confirm that the inclusion criteria were met.Agreement was measured using kappa (κ) statistics.Data extraction was performed by two reviewers (LR-V and AM) and summarized in standardized tables.

Quality assessment
Study quality was assessed using the Newcastle-Ottawa Scale (NOS), which uses an eight-item rating system to evaluate the method of selection of participants, the exposure/outcome assessment, and comparability among study groups (17).Comparability was evaluated by controlling for potential confounders in terms of study design and the type of health effects under evaluation.The Cochrane Non-Randomized Studies Methods Working Group recommends the use of the NOS, although the study of its psychometric properties remains in progress (18).The NOS quality scores range from 0 to 9 (0 to 4 = poor quality; 5 to 7 = moderate quality; 8 to 9 = high quality).The NOS has specific formats for cohort and case-control studies only.The cohort study form was used to evaluate noncohort longitudinal studies and the case control form to evaluate case-crossover and cross-sectional studies.Two reviewers (LR-V and AM) independently performed the quality assessment of the included studies and disagreements were discussed and resolved by consensus.

Data analysis
Descriptive results of the included studies are provided.While quantitative analyses using pooled measures and random effect models were planned, they could not be conducted due to heterogeneity among study populations (children's age), outcomes and study designs.Kappa statistics and 95% CIs were generated using Stata version 11.1 (StataCorp, USA).

Search results
The present review follows the PRISMA recommendations (19).As indicated in Figure 1, the systematic search identified 162 studies.After removing duplicates, initial screening with inclusion/exclusion criteria and full-text review, 27 studies were included.Studies were excluded for a variety of reasons, primarily because they did not report results on Canadian children.Reviewer agreement was substantial for identifying potentially relevant studies (disagreement 22%; κ=0.73 [95% CI 0.70 to 0.75]) and excellent for identifying included/excluded studies in full-text review (κ=0.91 [95% CI 0.78 to 1.00]).

Study characteristics
The 27 studies that met the selection criteria varied in design, study location, number and type of air pollutants considered, age of children population and respiratory outcome.Tables 1 and 2 summarize the main characteristics and results of the individual studies grouped according to respiratory outcome examined .

Continued on next page
two used Alberta data (43,45) and one study used data from 11 Canadian cities (25).At least one of the two Ontario cities -Windsor and Hamilton -were represented in 13 of the articles analyzed.The most common study designs used were longitudinal (n=7), cross-sectional (n=7) and case-crossover (n=6).All studies included children <15 years of age, except for one that included children up to 17 years of age (20).Twelve (44%) studies received funding support from provincial or federal government agencies.Eighteen studies focused on exposures to at least one of the CAP, two used proximity to roads as the exposure source, two used Air Quality Health Index (AQHI) as the tested exposure, and four examined CAP as well as proximity to roads or industrial facilities.One study measured exposure to total suspended particles (TSP) in Hamilton, and other polycyclic aromatic hydrocarbon (PAH) and volatile organic compounds in Montreal (Quebec).Of the 23 studies tracking CAP or TSP, the majority (n=13) used fixed ambient air monitoring stations to assess exposures, two modelled exposures, seven used a combination of measurements and models, and one used personal monitoring devices.Regarding respiratory outcomes, 14 studies relied on health services use for respiratory conditions as the main health outcome (Table 1), while five used lung function measures based on spirometry; four used self-report of respiratory symptoms; three used a combination of lung function measures and symptoms; and one calculated incidence of asthma diagnosis (23) (Table 2).

Quality assessment
The overall methodological quality assessment of the studies was moderate, with a mean (± SD) NOS score of 6.04±1.3from a maximum score of 9. Figure 2 summarizes the methodological quality of the studies based on the NOS items.All studies clearly defined outcomes and most (96.3%) of the studies controlled for potential confounding variables; however, the nonexposed group definition, the follow-up criteria (especially for noncohort longitudinal studies), and the report of nonparticipation (in cross-sectional studies) were not met in >50% of the articles.

HSU as a result of air pollution exposure
Nine studies focused on HSU for the treatment of asthma symptoms.A variety of age ranges and pollutants had positive associations with asthma-related HSU (Figure 3A).Higher adverse effects were reported consistently during the warm season in different cities.The highest percentage increase of emergency department (ED) visits for asthma (up to 36%) was reported in one-day lag exposures to CO, SO 2 and NO 2 in Windsor during the warm season (32), and the highest percentage increase of asthma hospitalization (up to 42%) was reported in children exposed to SO 2 in refinery areas in Montreal (41).Studies examining NO 2 produced the most consistent results, showing a positive association with HSU at different ages (from 0 to 17 years of age).There was also evidence that the effect of NO 2 was exacerbated among female children of low socioeconomic status (20,34).Two pollutants (CO, O 3 ) gave conflicting results among studies; however, it is important to note that the studies generally differed according to age and city studied, as well as the study period.Two studies using the AQHI as an air pollution indicator also showed positive associations with HSU.
Five studies examined HSU for general respiratory conditions (which could include, but were not limited to, asthma) and infections.As with HSU for asthma, several positive associations were demonstrated (Figure 3B).In this case, CO, NO 2 and PM 2.5-10 were positively associated with HSU for respiratory ailments.Conflicting results were demonstrated for O 3 , PM 10 and SO 2 , depending on the city and age of the children studied.

Air pollution effects on lung function and asthma symptoms/diagnosis
A variety of pollutants was tested for an association with increased asthma symptoms (AS) or decreased lung function (LF) (Figure 3C).LF was typically assessed using forced vital capacity (FVC) and/or forced expiratory volume in 1 s (FEV 1 ) measures.The most consistent null effects were reported for O 3 , when either LF or AS were examined in children, in Windsor and Hamilton.Associations with reduced LF were also apparent for PAH (Montreal) and TSP (Hamilton).No associations with reduced lung function were seen with benzene (Montreal) or PM 2.5-10 (Windsor); and no increased AS were apparent with increased PM 10 exposure (Hamilton).While proximity to roadways had no association with AS in Hamilton, it did show an effect on LF in three Windsor-based studies (21,26,27).Inconsistent results were observed for NO 2 , PM 2.5 and SO 2 , even though the contradictory studies often examined similar health effects and age groups in the same cities.For NO 2 , most LF data failed to demonstrate an association (Montreal and Windsor), while one showed the opposite (Windsor), as did one study using AS (Hamilton).There was no clear pattern in the PM 2.5 data, with three Windsor articles identifying effects on LF and/or inflammation, and two from Montreal and Windsor showing no effect on LF.For SO 2 , four studies using either LF, AS, or LF and inflammation failed to demonstrate an effect in Hamilton and Windsor, while a single study in Montreal detected an association with AS.Two studies showed effect modification of NO 2 on asthma symptoms by chronic psychosocial stress and other allergic disease (20,27).
Only one study examined increases in asthma diagnosis, as a consequence of exposure to CAPs, wood smoke and black carbon (soot), or industrial and road proximity in children <5 years of age from southwest BC (23).Positive associations were seen with CO, NO, NO 2 , PM 10 , SO 2 , black carbon and industrial proximity (Table 2).

DISCUSSION
The present systematic review summarizes the evidence available from epidemiological studies exploring the association of outdoor air pollution on Canadian children's respiratory in the past 10 years.From 2004 to 2014, 27 new studies were identified; all but one (40) reconfirm the adverse effects of outdoor air pollution on respiratory symptoms, lung function and HSU at different CAP concentrations, almost all of which were below United States and Canadian (available only for PM 2.5 and O 3 ) standards (47).The present review also identified that the increase in respiratory-related ED visits and hospitalizations were demonstrated in higher proportions than the outpatient visits, and that those effects are even higher in places near industrial facilities or refinery areas in Windsor and Montreal, respectively.The findings showed more consistent associations of adverse respiratory outcomes for traffic-related exposures of PM and NO 2 , especially related to health services use.Some studies also report differential effects of gases and particles on female and socially disadvantaged children.
Our review updates a similar study covering 1989 to 2004 ( 16), enabling us to compare publications and describe trends in the research related to the effect of air pollution on Canadian children over the past 25 years.There are differences, mainly in terms of the number of publications, type of study designs, exposure measurement and study locations.Compared with publications from 1989 to 2004, the publications in the past decade were: more than twofold more frequent, over a shorter period of time studied; conducted mainly in Windsor and Hamilton rather than in Toronto and Vancouver; most commonly used cross-sectional, longitudinal and case-crossover study designs rather than time-series analysis; and increasingly introduced air pollutant exposure assessment using model-based small area estimations (eg, landuse regression models), in addition to data from fixed ambient air monitoring stations.These changes could be explained by several factors, such as the increase in funding for the study of environmental-related health conditions in Canadian children (48), the increased access to high quality administrative data, growing societal concern for the potential health effects of industrial development around cities (12 of 27 studies between 2004 and 2014 were supported by provincial or federal Figure 2) Quality assessment of the 27 included studies using the Newcastle-Ottawa Scale, which assess three main groups of criteria consisting of a total of eight items.All criteria are assessed in a binary fashion (0 or 1), except for 'comparability of cases/controls', which was scored as 0, 1 or 2. Percentages indicate the total score of all the articles in the indicated categories out of the total possible score government agencies), and the development of epidemiology, spatial and statistical methods applicable to air pollution research (49).
Conversely, similarities across time included the preference for reporting asthma-related outcomes and the use of CAP concentrations as the metric to assess outdoor air pollution exposure.These similarities can be explained by the persistent high prevalence of asthma in Canadian children (50), the availability of high-quality administrative health data, especially for acute asthma-related conditions (51) and detailed CAP data from the air quality monitoring surveillance systems in many densely populated Canadian cities (52).
National air pollution surveillance data have shown that the concentration of CAP gases decreased slightly over time in Canada (53), and mean Canadian CAP levels are lower than those in most of the major cities of the world (54)(55)(56)(57)(58).Even at current Canadian levels, however, they are associated with adverse health effects in children, as well as cardiovascular, respiratory and gastrointestinal effects in adults (mainly linked to SO 2 and PM) (59,60).Moreover, several of the studies referenced herein suggest heterogeneous effects of the current levels of gases and particles on children according to sex, socioeconomic status and seasonality; however, none of them included an analysis of a b c  Continued on next page potential mechanisms explaining those differences, which may vary across populations (61).

Limitations
The methodological quality of the studies included in the present review was relatively homogeneous, with a mean 'moderate quality' score according to the NOS tool.This 'moderate' score is lower than expected for most included articles because the NOS tool is designed only for classic cohort or case-control studies.Thus, we believe that properly assessing the quality of observational studies related to environmental health remains challenging.
To avoid selection bias in the present review, two independent reviewers conducted the screening and selection processes.Although efforts to undertake meta-analyses failed due to the high qualitative heterogeneity in the included studies, qualitative summary tables and graphs were developed to examine trends.Having more homogeneous studies in terms of children's age in studies would allow quantitative analysis in future reviews.Another limitation of the review was that unpublished studies in the formal scientific literature were not identified; therefore, there may be a chance of publication bias.However, the included studies showed both positive and null effects of different CAP for different cities.In fact, the common mixture of positive and null effects in many studies may be explained by the fact that air pollutants effects differ not only according to type of pollutant (alone or in combination), but also by the different physical and sociodemographic conditions of the places and populations under study.Finally, the scope of the present review was limited to the childhood population and the results and conclusions may be not generalizable to adults, even in Canada.

CONCLUSIONS
The present review provides researchers, clinicians and environmental health authorities with a current summary of the evidence linking the adverse effects of outdoor air pollution to children's respiratory health in Canada.Further studies should fill knowledge and methodological gaps that are related but not restricted to: deepening the understanding of the 'why' of the differences in the observed adverse effects for some pollutants and socioeconomic conditions across cities; exploring the combined effect of various air pollutants; expanding the study of the health effects of non-CAP air toxicants that are emitted in Canada, such as PAH or volatile organic compounds; strengthening the advances in epidemiological, spatial, statistical and social analysis as applied to air pollution studies, aiming for a more integrated approach between the physical and social environment; and developing and validating a tool for assessing the methodological quality of observational studies commonly used in environmental health studies, other than cohort and case control studies.

ACKNOWLEDGEMENTS:
The authors thank Mrs Maria Tan, library consultant at the John W Scott Health Sciences Library at University of Alberta for her support in defining the specific search strings.Author contributions: LR-V designed the study, undertook the literature searches, study assessment, data extraction and analysis.AM conducted study assessment, data extraction and analysis.AO-V contributed as third reviewer in the selection process and contributed to the data interpretation.BR contributed to the study design and data interpretation.All authors contributeed to writing and editing the manuscript, and approved the final version for publication.("Canada" or "Vancouver" or "Edmonton" or "Calgary" or "Victoria" or "Regina" or "Saskatchewan" or "Winnipeg" or "Toronto" or "Hamilton" or "Windsor" or "Mississauga" or "Montreal" or "Halifax" or "Ottawa" or "Kingston" or "London").mp.[mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier] 15. 13 or 14 16. 5 and 9 and 12 and 15

Figure 1 )
Figure 1) PRISMA flow diagram for selection of studies Figure 3) Respiratory health outcomes according to air pollutant and children's age range.The age groups of children tested by each reference were plotted for each air pollutant measure tested.Bars indicate either a statistically significant (P<0.05,black) or nonsignificant (P>0.05,grey) association between the indicated pollutant and the health effect, for each age group studied by the indicated study.A The incidence of emergency department (ED) visits and hospitalizations for asthmatic symptoms and pollution exposure.*Effect was present or stronger in populations with low socioeconomic status.# ED visits and hospitalizations for bronchiolitis were also included as relevant health outcomes in the study.B Associations between pollution exposure and health services use for conditions other than asthma.C Lung function measurements and symptoms of affected lung function (eg, wheezing and other asthma-like symptoms) were tested for associations with differing levelsof pollutant exposure.*Lung function was measured; + Lung inflammation was measured; # Asthmatic symptoms were measured.AQHI Air Quality Health Index; CO Carbon monoxide; NO 2 Nitrogen dioxide; O 3 Ozone; PAHs Polycyclic aromatic hydrocarbons; PM 2.5 /PM 10 Particulate matter size ≤2.5 µm/≤10 µm in aerodynamic diameter; SO 2 Sulphur dioxide; TRS Total reduced sulphur; TSP Total suspended particles

FUNDING:
The study was supported by the Emergency Medicine Research Group (EMeRG) affiliated with the Department of Emergency Medicine, and by the Department of Pediatrics, University of Alberta.Dr Rodriguez-Villamizar is supported by COLCIENCIAS Colombia.Adam Magico's postdoctoral research is supported by a Hair Massacure grant to Dr Osornio-Vargas.Dr Rowe's research is supported by a Canadian Institutes of Health Research through a Tier I Canada Research Chair in Evidence-based Emergency Medicine from the Government of Canada (Ottawa, Ontario)

TAble 1 -continued Characteristics and results of included studies with health services use outcomes
OR of hours downwind of a smelter on same-day hospitalization of 2-4-year-old children: 1.27 (95% CI 1.03-1.56),and PM 2.5 exposure: 1.22 (95% CI 1.03-1.11)*Parts per billion for gasses, μg/m 3 for particulate matter (PM).AQHI Air Quality Health Index; BC British Columbia; CO Carbon monoxide; ED Emergency department; IQR Interquartile range; NO 2 Nitrogen dioxide; NS Not stated; O 3 Ozone; OASIS Ontario Asthma Surveillance Information System; RR Relative risk; SES Socioeconomic status; SO 2 Sulphur dioxide; TEOM Tapered element oscillating microbalance; TRS Total reduced sulphur

TAble 2 -continued Characteristics and results of included studies with respiratory symptoms, lung function measures and incidence of asthma diagnosis outcomes Author (reference); year, location; study period Study design, study population and size Pollutants (mean or median levels*) and methods assessing exposure Respiratory outcome Study findings Adjustment for confounding factors
Parts per billion for gasses, μg/m 3 for particulate matter (PM) and total suspended particles (TSP).CO Carbon monoxide; ED Emergency department; eNO Exhaled nitric oxide; ETS Environmental tobacco smoke; FEF 25-75% Forced expiratory flow between 25% and 75% of the forced vital capacity (FVC); FeNO Fractional exhaled nitric oxide; FEV 1 Forced expiratory volume in 1 s; IDW Inverse distance weighted; IQ Interquartile; IQR IQ range; LUR Land use regression; MEF 75% , maximum expiratory flow after 75% of FVC has been exhaled; NO Nitrogen monoxide; NO 2 Nitrogen dioxide; NS Not stated; O 3 Ozone; PAH Polycyclic aromatic hydrocarbon; PEF Peak expiratory flow; PF Pulmonary function; PR Prevalence ratio; RR Relative risk; SES Socioeconomic status; SO 2 Sulphur dioxide; TBARS Thiobarbituric acid reactive substances; VOC Volatile organic compound *