Secondary Logo

Journal Logo

Reviews

Review of the effect of air pollution exposure from industrial point sources on asthma-related effects in childhood

Buteau, Stéphanea; Geng, Xiaohuib; Labelle, Remib; Smargiassi, Audreya,b,c,*

Author Information
doi: 10.1097/EE9.0000000000000077

Abstract

What this study adds

This study is the first to review and summarize findings of epidemiologic studies of the association between exposure to air pollution from industries and asthma-related outcomes in childhood. This study identifies substantial heterogeneity across design and results of selected studies. As well, many studies had important limitations that make causal inference difficult. This review stresses the importance that further well-conducted studies in terms of design and methods, particularly to assess exposure, are needed to improve risk estimates. Harmonization of methods may improve comparability across studies for future meta-analysis and help to shed light on putative agents and drivers of heterogeneity.

Introduction

There is considerable evidence that an increase in ambient air pollution is associated with both acute and chronic respiratory health outcomes.1,2 Children are a subgroup of particular interest; early life and childhood are likely critical exposure windows because of the immaturity of the immune system and the potential for developmental disruption.3–6 Children also typically experience a greater exposure to outdoor air pollution when compared with adults because of their higher respiratory volume relative to their body mass and because they tend to be more active and spend more time outdoors.3–6Several epidemiologic studies have reported positive associations between ambient air pollution and increase in the number of emergency department visits and hospitalizations for respiratory causes in children.7–13 Others suggest that outdoor air pollution may exacerbate acute respiratory infections in children, exacerbate respiratory and asthma symptoms as well as acute changes in lung function in children.14–17 While past evidence on the contribution of air pollution in the development of asthma have been mixed,18,19 findings from the most recent and comprehensive systematic review and meta-analysis on the subject support that childhood exposures to air pollutants, particularly traffic-related, is associated with the development of the disease in childhood.20

Although air pollution studies on the respiratory outcomes in children have mostly focused on ambient urban and traffic-related air pollutants,13,16,20–22 industrial emissions may importantly contribute to ambient air pollution experienced by some local populations. Ambient air pollution affected by industrial point-source emissions may differ to that of urban settings in terms of concentrations and composition, possibly yielding to differential toxicity.23 We sought to investigate the effect of air pollution from industrial point sources on asthma in childhood. We performed a structured review of epidemiologic studies of the associations between children’s exposure to air pollution from industrial point sources and the following outcomes: asthma, asthma-like symptoms, wheezing, and bronchiolitis. Specific objectives included providing a detailed description of characteristics of the selected studies, of estimates of association and, when appropriate, summarize estimates of effect using a meta-analysis of results.

Methods

Literature search

We searched EmBase, MedLine, “EBM Reviews/Cochrane,” and CINAHL (Cumulative Index to Nursing and Allied Health Literature) bibliographic databases using a combination of keywords representing the following concepts: (1) air pollution; (2) industries; (3) respiratory outcomes; and (4) children. More details about the search, including the complete list of keywords, is presented in eAppendix A; http://links.lww.com/EE/A67.

Our search was restricted to articles written in English published between January 1, 2000 and September 6, 2017. Our strategy consisted of screening all titles and abstracts to assess their eligibility according to the criteria listed below. We then obtained the full-text of papers meeting our selection criteria. In the next stage, we underwent full-text read of selected articles to determine whether they should be included in the review. We completed our literature search manually through inspection of the reference list of selected articles and relevant reviews. All the process was performed by two investigators (X.G. and R.L.) and in case of discrepancy a third investigator (A.S. or S.B.) was consulted to reach a consensus.

Selection criteria

Our review was restricted to studies that reported quantitative estimates of association (thus excluding studies that reported only a P value, as this statistic provides no information about the magnitude of the effect) between exposures to air pollution related to industrial sources and selected asthma-related respiratory outcomes in children. Specifically, selected outcomes were asthma, asthma-like symptoms, wheezing, and bronchiolitis. Wheezing and bronchiolitis were included as they produce clinical symptoms that may be similar to those of asthma and difficult to distinguish, especially in children 5 years and younger.24

In terms of the study population, consistently with in another review of childhood asthma and wheezing,25 the age range of interest was children 14 years of age and younger; thus, we excluded studies not reporting effects for participants 14 years and younger. However, we did retain a very few studies that reported findings for a group of participants that were mostly overlapping with our age group of interest, but also included some older participants as long as there were not adults (i.e., ≥18 years of age). (For example, in one study,26 the age range was 6–15 years, with 88% of participants 6–11 years of age; thus, this study was included.)

We excluded studies in which the exposure investigated was not assessed during the aforementioned age range of interest; thus, excluding studies examining the effects of prenatal air pollution exposures. Studies that reported associations for concentrations of air pollutants measured at monitoring sites or using personal monitoring were excluded if it was not clearly specified that the study population was living in close proximity or in areas affected by emissions from industrial point sources. Studies in which the exposure was estimated by models, such as atmospheric dispersion models, were included if it was made clear that the air pollution exposure related to emissions of industrial point sources.

Extraction of the data

Data extraction from the selected articles was performed independently by at least two investigators (from X.G., R.L., and S.B.). In case of discrepancies between the extracted information, another investigator (A.S.) reviewed the article to reach consensus. Information extracted included publication year, study design, location, sample size and age of study subjects, type of industries, method of exposure assessment, type of outcome and method of characterization, statistical analysis method, metric of exposure and effect size estimates, as well as covariates used for adjustments. When associations were reported for different age groups, we extracted the stratified results rather than for combined age groups. In instances where results were reported for the same outcome, metric of exposure and population in more than one article, we extracted the results from the latest publications (for particles with median diameter of >10 µm [PM10]: Howel et al27 vs. Pless-Mulloli et al28; for residential proximity: Pless-Mulloli et al29 vs. Pless-Mulloli et al28) or which reported the primary results,30 whereas another article reported sensitivity analyses.31

Method of analysis

With the aim to summarize effects of exposure to air pollution from industries on asthma-related outcomes by mean of a meta-analysis of results, we carefully considered a number of study design characteristics to determine whether study results can be pooled. Notably, we considered the type of outcomes, distinguishing prevalence from incidence, effects resulting from short-term exposures from those of long-term exposures, and whether the outcome was assessed from questionnaire or from administrative health data. Although we had no restriction in terms of study design, this was carefully considered to determine if studies could combined in a meta-analysis of results, as the parameters of the association being estimated may differ depending on the design used.32 The exposure assessment methods used in the selected studies (e.g., defining exposure as living within a municipality with an industry vs. within 2 or 10 km of the industry) as well as the functional representation of the exposure variable in the statistical analysis (e.g., categorizing residential proximity to industry vs. treating it on its native scale as a continuous variable) were also of particular concern. These differences may yield to measures of associations that are not quantitatively comparable or that cannot be expressed uniformly across studies.

We performed a meta-analysis of results when at least three studies were deemed to be comparable according to the outcome investigated, the outcome and exposure assessment methods, and the exposure metric used in the analysis. Meta-estimates of association were calculated where appropriate using random-effects models;33 thus, assuming that true effect size varies across studies. In instance where the air pollutant concentration (treated as continuous) was used as the metric of exposure, then effect size estimates and pooled estimates were expressed in terms of 10-unit increment in the air pollutant concentrations to facilitate comparisons. We quantified heterogeneity among estimates using the I2 statistic, representing the percentage of the total variability explained by differences between studies rather than sampling error.34 As rules of thumb, an I2 value of 25%, 50%, and 75% are often used to characterize low, moderate, and high heterogeneity. In presence of high heterogeneity, no definite conclusion should be drawn from the pooled effect estimate, even if a random-effects model is used. We did not perform meta-regression analysis to investigate specific factors contributing to heterogeneity because of the limited number of studies included in each meta-analysis.

Results

Selection of studies

Figure 1 shows the selection of studies included in this review. Our initial bibliographic search yielded 308 peer-reviewed articles. Three additional studies35–37 were identified throughout the inspection of the references of the articles selected and one study38 by the authors was also included. One hundred eleven articles underwent full-text review, whereas one article39 could not be retrieved. Thirty-nine articles fulfilled our inclusion criteria. This includes one study40 that we retained although the outcome investigated was hospitalization for asthma, bronchiolitis, bronchitis, and pneumonia, as it was reported in text that sensitivity analyses restricted to asthma yielded similar results. We included five studies that reported estimates of association not strictly for children 0–14 years of age; specifically these were reported for 0–15 years,41,42 5–15 years,43 6–15 years,26 and <17 years old.44 We identified two studies for which multiple articles were published; more precisely, there were two articles30,45 (another31 was excluded as it was a sensitivity analysis) on the Viadana study and three articles27–29 on the opencast coal mining study in England. These multiple articles were all retained for this review but considered as a single citation; therefore, this review included a total of 39 published articles from 36 unique studies.

Figure 1.
Figure 1.:
Flow chart presenting the selection of studies of the association between asthma-related outcomes and exposure to air pollution from industrial point sources.

Characteristics of the selected studies

Study settings and population

Table 1 summarizes the main characteristics of the selected studies, according to the geographic locations (eTable B1; http://links.lww.com/EE/A67, includes more information about the exposure metrics and covariates). Most studies were conducted in North America (n = 12), followed in decrease order by Europe (n = 10), Latin America and the Caribbean (n = 5), Middle East countries (n = 4), East, South, and Southeast Asia (n = 3), and southern Africa (n = 2).

Table 1
Table 1:
Main characteristics of the selected studies, ordered by geographic region and country.

The age range varied substantially across studies; some focused strictly on younger children <5 years of age,35,36,46–52 others on children 5 years and older,26,43,53–59 whereas some studies included both younger and older children.28–30,40,41,45,60–63 In one study,58 the exact age range was unclear; the methods section states that the study population is children 11–14 years old, but estimates of association are reported for children with mean age of 6.65 years (SD: 0.69 years).

Study design and outcome assessment

In terms of the outcome, it should be noted that we reported in Table 1 only information about the selected asthma related-outcomes, but other respiratory outcomes were investigated in the selected studies. In the 20 studies26–30,41,43,45,50,53–57,59–61,64–69 that assessed respiratory outcomes through questionnaires (including diaries, questionnaire survey or interviews), the outcome frequently investigated were the prevalence of asthma (n = 16) and of wheezing (n = 14). Questionnaires were generally derived from standardized ones, notably the International Study of Asthma and Allergies in Childhood (ISAAC). In 18 studies using a questionnaire, a cross-sectional design was used to investigated the effects of long-term exposure to air pollution from industries on asthma-related outcomes. The remaining two studies were longitudinal; one short-term panel investigating short-term effects54 and one with a longer follow-up investigating long-term effects.50 In addition, in the opencast coal mining study in England, both a cross-sectional and panel study analysis was used to investigate long- and short-term effects, respectively.27–29 Specifically, lifetime and period (2 and 12 months) prevalence of asthma, wheezing, and asthma-related symptoms were obtained from a questionnaire, whereas short-term effects were assessed from a daily dairy of respiratory symptoms collected over 6 weeks during which daily PM10 was concurrently measured. General Practitioner consultation records were also obtained in that study but this was not considered in the review because reasons for consultations were not reported specifically for asthma.

In 16 studies, asthma-related outcomes were assessed from administrative health data (i.e., general practitioner consultation records, hospital discharges, and emergency department visits). Four studies36,46–48 investigated the effects of short-term exposure to air pollution from industries on asthma-related outcomes. In these four studies, a case-crossover design was used and the outcome investigated was hospitalization for asthma,46,47 and hospitalization for asthma or bronchiolitis.36,48 In the remaining 12 studies using health administrative data, long-term effects were investigated using different type of study designs, including cohort (n = 339,47,69), nested case-control (n = 327,36,48), ecological (n = 344,46,67), case-control study (n = 141), longitudinal (n = 153) and cross-sectional (n = 155). Notably, one cohort study was a natural experiment, investigating the incidence of wheezing before and after the closure of a factory among birth cohorts living in close proximity and further away from the factory.52 Relation between exposure to industrial point sources and childhood asthma onset was investigated in one nested case-control35 and in one population-based birth cohort study.38

Exposure assessment

In terms of the methods used to assess children’s exposure to air pollution from industries, there was substantial variation across the selected studies. In 15 studies the proximity of the residence or school (e.g., Karakis et al63) to the point source, generally treated as ecological and binary, was used as a proxy measure for long-term exposure to industrial air pollution emissions. However, varying criteria were also used to define the categories of exposure; in some studies, the exposure was defined by the city or municipality of residence (or school) (such exposure assessment was referred to as “area-based” in Table 1), whereas in others it was determined according to residing within a specific distance from the industry. In the latter studies, different cutoff values, ranging from ~400 meters (quarter mile)42 to 20 km,63 were used to distinguish children exposed to those unexposed to emissions from the industry. In some other studies, the distance between the industrial air pollution point source and the child’s residence treated as a continuous variable was used as the metric of exposure.38,44,51,69 In the Viadana study, one of the surrogate measures used for exposure was based on the distance from home and school to the closest chipboard industries, weighted to account for the time spent at each location.30,31,45 In addition, a three-category variable, defined as the number (i.e., none, one, two) of wood factories within 2 km from home and school, was used.

Although in most studies measurements of air pollutants concentrations were available, these were mostly used for descriptive purposes. Specifically, 10 studies used measurements from fixed-site monitors as the metric of exposure. Pollutants investigated were sulfur dioxide (SO2),36,46–48,54,59,64,68 PM10,27,28,54,59 or of <2.5 µm (PM2.5),36,47,48,54 total suspended particles,56 and nitrogen dioxide (NO2).48 In the only study66 that made use of personal monitoring PM10 were collected for 24 hours and metals in filters were measured, including Mn, Ni, Fe, Cr, and Zn. Dispersion modeling was used to estimate exposure to ambient concentration of SO246,64 and total suspended particulate matter56 from industrial point sources. In another study, ambient PM2.5 concentration from the industry was estimated using positive matrix factorization.54

In other studies, varying metrics of exposure were constructed using information about emissions of industrial point sources, distance to the industry and meteorological data. Specifically, metrics used to assess exposure of children included tons of air pollutants emitted by industries nearby the residence,38 percentage of hours that the child’s residence was downwind of the industry,36,47 indicators of exposure combining data on emissions and residential proximity to industries,35 emissions and percent time that the child’s residence was downwind of the industry,48 emissions corrected for the residential proximity and wind direction,38 and residential proximity corrected for wind direction and wind speed.69

Associations between exposure to emissions of industries and selected asthma-related outcomes

In the next sections, we present findings of studies included in this review, according to the outcome assessment methods and the metric of exposure used. Specifically, we have analyzed separately studies in which the outcome was assessed from administrative health data from those using a questionnaire. In view of the varying outcomes investigated and metrics of exposure used, many studies were not comparable to each other; thus a meta-analysis of results was possible in a few instances and included a limited number of studies.

Studies using administrative health data

Main findings from all studies making use of administrative health data are reported in the eTable B2; http://links.lww.com/EE/A67. The findings were organized according to the type of exposure metrics and of outcome.

Association to air pollutant concentrations

Four short-term effects studies used ambient air pollutant concentrations as the metric of exposure to pollutants from main industrial emitters to investigate the association with hospitalization for asthma or bronchiolitis in childhood.36,46–48 All of these studies assessed exposures using measurements at fixed-site monitors, but in addition Smargiassi et al46 used dispersion modeling. Associations for SO2 and PM2.5 from these studies were pooled in a meta-analysis of results. Figures 2 and 3 present forest plot showing estimates from individual primary studies together with pooled estimates and heterogeneity as measured by the I2.

Figure 2.
Figure 2.:
Forest plot of the association between hospitalization for asthma or bronchiolitis. A, Same-day daily mean concentration of SO2 and (B) same-day hourly maximum concentration of SO2, from case-crossover studies using administrative health data. Effect size and 95% CI are expressed relative to a 10-ppb increase. Pooled estimates of effect size are indicated by black squares and 95% CI are represented by horizontal lines; size of black square around point estimate is proportional to weight in calculating pooled estimate.

For SO2 (Figure 2), the meta-analyses presented in Figure 2 include findings from Smargiassi et al46 estimated from dispersion modeling rather than measurements from fixed-site, as this method likely provide the better estimates of exposure. Pooled odds ratios (ORs) of hospitalization for asthma or bronchiolitis were 1.02 [95% confidence intervals (CI): 0.96, 1.08] and 1.01 (95% CI: 0.97, 1.05) for a 10-ppb increase in same-day daily mean and hourly maximum concentrations of SO2, respectively. In both instances, heterogeneity was considerable (I2 = 56% and 64%, respectively). When considering findings from Smargiassi et al46 derived from fixed-site monitors rather than dispersion modeling (eFigure B1 and B2; http://links.lww.com/EE/A67), pooled odds ratios were 1.00 (95% CI: 0.95, 1.05) and 1.00 (95% CI: 0.97, 1.03), whereas heterogeneity was reduced (I2 ranging between 34% and 38%).

For PM2.5 (Figure 3), pooled odds ratios were 1.02 (95% CI: 0.93, 1.10) and 1.01 (95% CI: 0.98, 1.03) for a 10-µg/m3 increase in the 24-hour daily mean and hourly maximum concentration. Heterogeneity across results was moderate (I2 = 56% and 32%).

Figure 3.
Figure 3.:
Forest plot of the association between hospitalization for asthma or bronchiolitis. A, Same-day daily mean concentration of PM2.5 and (B) same-day hourly maximum concentration of PM2.5, from case-crossover studies using administrative health data. OR and 95% CI are expressed relative to a 10-ppb increase. Pooled random-effect estimate of ORs is indicated by vertical points of diamonds and 95% CI are represented by horizontal points. Black squares represent individual effect size of primary studies and the bars the 95% CI; size of black squares is proportional to weight in calculating random effect summary estimate.

In addition, ambient NO2 was investigated in one study48; positive association with hospitalization for asthma or bronchiolitis was reported for the daily mean (OR per 7.4 ppb = 1.09; 95% CI: 0.65, 1.82) and hourly maximum concentration (OR per 14.6 ppb = 1.15; 95% CI: 0.64, 2.06), but confidence intervals were fairly large and included the null.

Association to residential proximity

In six studies, residential proximity to major but varied industries such as powers plants and smelters was treated as categorical (either binary or 3-level variables).40,42,58,62,70 All of these studies reported statistically significant associations between residential proximity and health service used for asthma, including hospitalization, emergency department visits, and clinical visits. A population-based birth cohort study found that the risk of asthma onset in children’s living within 7.5 km of a major industrial air pollutant emitter (defined as emitting >100 tons per year of either PM2.5 or SO2) was significantly greater than in those living further than 7.5 km, whereas within 7.5 km every 1-km increase was associated with a 2.2% (95% CI: 1.0%, 3.3%) decrease in the hazard of asthma onset.38 In the other studies that treated the distance to the industry as a continuous variable, every 1-km decrease was found to increase the risk of hospitalization for asthma by 7% (95% CI: −2%, 18%)51 and of asthma-related medical visits by 69% (95% CI: 50%, 91%).44

A meta-analysis of results was not possible because the number of studies investigating a similar outcome and using a similar metric of exposure was insufficient.

Association to tons emitted and wind exposure

In the two case-crossover studies36,47 that considered as the metric of exposure the percentage of daily time that the child’s residence was downwind of the point-source (in this case a smelter), associations were found with hospitalization for asthma or bronchiolitis.36,47 More precisely, Lewin et al47 found a positive association in children 2–4 years old (OR per 29% increment: 1.27; 95% CI: 1.03, 1.56), but no association in those <2 years old (OR per 21% increment: 1.00; 95% CI: 0.84, 1.20). As well, one cross-sectional study used a binary indicator of exposure based on wind direction from an industrial park housing various heavy industries, and found association with the prevalence of asthma (OR: 1.95; 95% CI: 1.01, 3.76).63

Three studies, one of short-term48 and two of long-term effects,49 used a metric of exposure combining tons emitted and distances. Brand et al48 found no association for hospitalization for asthma or bronchiolitis and daily exposure to air emissions from pulp mills, oil refineries, and metal smelters, using a case-crossover analysis. Using a nested case-control design, Karr et al49 found a positive association between hospitalization and emergency department visits for asthma or bronchiolitis (OR: 1.10; 95% CI: 1.06, 1.13) and the proximity-weighted sum of total regulated air pollutant emissions during the first year of life. In a population-based birth cohort study of long-term effects, childhood asthma onset was associated with the yearly tons emitted by industries weighted by the percentage of time downwind and the inverse distance.38

Association to other metrics of exposure

Using a natural-experimental design, Câra et al52 found that the closing of a iron, steel, and coke factory was associated with a significant decrease in the occurrence of wheezing among cohorts of children <2 years old.

Studies using questionnaire

Results of associations from all studies making using a questionnaire (or a diary) to assess the outcome are presented in eTable B3; http://links.lww.com/EE/A67. Studies were organized according to the type of exposure metrics and the type of outcome.

Association to air pollutant concentration

We identified six studies investigating the association between the selected respiratory outcomes and air pollutant concentrations. A meta-analysis of results was not possible because of substantial differences across study design, outcome investigated, and exposure assessment methods.

For PM10, long-term effects were investigated in two cross-sectional studies,59,66 whereas short-term effects were investigated in two panel studies.27,54 Notably, in one cross-sectional study, 24-hour personal PM10 exposure was found to be associated with lifetime asthma and asthma medication use in past 12 months; for an interquartile range (IQR) (38 g/m3), risk ratios (RR) were 1.12 (95% CI: 1.00, 1.21) and 1.21, (95% CI: 1.09, 1.35), respectively).66 In contrast, panel studies found no association between asthma reliever use and 24-hour daily mean PM10 exposure from fixed-site monitors at lag 0 and 1 day,27 and for 1 to 7 days averaging window of exposure.54

Prieto-Parra et al54 investigated wheezing and reliever use in association to exposure to ambient PM2.5 as well as ambient PM2.5 attributable to the copper smelter. For ambient PM2.5 they found stronger association for longer exposure; specifically associations where positive for 5- and 7-day averaging concentration [e.g., for wheezing and 7-day average exposure, RR per IQR (18.0 μg/m3) = 1.60; 95% CI: 1.15, 2.26], whereas for 1- and 3-day averaging concentration associations were null. These findings were not consistent with those for PM2.5 attributable to the copper smelter, as for this metric associations were negative and mostly comprising the null. Hrubá et al56 found positive associations between annual total suspended particulate matter concentrations from dispersion modeling and the lifetime prevalence of asthma and wheezing.

For SO2, asthma prevalence was found to be associated with exposure during the first year of life in one study.68 The two short-term effect studies investigating respiratory symptoms found no association. Specifically, one was a panel study54 investigating the effects daily exposures related to emissions of a copper smelter, whereas the other59 was a cross-sectional study contrasting children from a highly industrialized compared with a non-industrialized area, and using their 8-month average SO2 exposure at school.

Association to residential proximity

In 12 studies, exposure was based on residential proximity to various types of industries including petro-chemical, chipboard, and cement plants treated as binary covariate. From these, six studies41,43,45,53,57,59 reported positive statistically significant association with at least one of the selected outcomes. We pooled results from the 5 studies reporting an odds ratio for the association between residential proximity to industries and the lifetime prevalence of asthma and wheezing.

For asthma (Figure 4), the pooled odds ratio was 1.98 (95% CI: 0.87, 3.09), but heterogeneity across study results was substantial (I2 = 71%). Much larger effect was reported in the two studies.57,61 When removing these two studies (one on steel iron coke plant57 and the other unspecified61), the pooled odds ratio was 1.85 (95% CI: 0.78, 2.93), and heterogeneity remained substantial (I2 = 79%) (eAppendix Figure B3; http://links.lww.com/EE/A67, for the forest plot).

Figure 4.
Figure 4.:
Forest plot of the association between residential proximity to industries and the prevalence of asthma, from cross-sectional studies using questionnaire. Pooled random-effect estimate of ORs is indicated by vertical points of diamonds and 95% CI are represented by horizontal points. Black squares represent individual effect size of primary studies and the bars the 95% CI; size of black squares is proportional to weight in calculating random-effect summary estimate. The arrows indicate that the confidence interval extends beyond the range of the value display.

For wheezing (Figure 5), the pooled odds ratio was 1.33 (95% CI: 0.86, 1.79; I2 = 65%). However, two studies included in this meta-analysis used a definition of wheezing that is not entirely consistent with the others. Specifically, in one study59 wheezing was defined as chest sounding wheezy or whistling on most days and nights, whereas in the other study53 the occurrence of wheezing was limited to the past 12 months. When removing these two studies, the pooled OR was 1.29 (95% CI: 0.50, 2.08; I2 = 58%) (eFigure B4; http://links.lww.com/EE/A67, for the forest plot).

Figure 5.
Figure 5.:
Forest plot of the association between residential proximity to industries and the prevalence of wheezing, from cross-sectional studies using questionnaire. Pooled random effect estimate of ORs is indicated by vertical points of diamonds and 95% CI are represented by horizontal points. Black squares represent individual effect size of primary studies and the bars the 95% CI; size of black squares is proportional to weight in calculating random-effect summary estimate. The arrows indicate that the confidence interval extends beyond the range of the value display.

Other studies making use of indicators of exposure based on residential proximity included one cross-sectional study that found no association between the prevalence of asthma and asthma-like symptoms and the weighted average of minimum distances of each child’s home and school from the chipboard industries.30 In another study, exposure was represented by the percentage of a 250-m buffer from the child’s residence that was within 0.80 km of an unspecified industrial point source.50 In this study, positive association (OR: 1.30; 95% CI: 0.98, 1.52) was found between proximity to air pollution industrial point sources and parental reporting of asthma diagnosis, but not of wheezing.50

Association to wind and other related metrics of exposure

In one cross-sectional study a binary indicator of exposure based on wind direction from a petrochemical plant was used, and increased risk of prevalence of wheezing was reported (OR: 2.01; 95% CI: 1.01, 4.01).60 White et al69 found that distance weighted for wind direction and wind speed, but not simple distance from the refinery, was positively associated with asthma-related symptom prevalences.

Discussion

We reviewed and summarized 36 studies investigating the association between exposure to air pollution from industrial point source and asthma-related outcomes in childhood. Although individual studies mostly reported positive associations, there were some mixed results and our meta-analyses did not provide strong evidence about the effect of exposure to air pollution from industries on asthma outcomes in childhood. Specifically, pooled effect estimates from case-crossover studies using administrative health data suggest no effect of daily exposure to SO2 and PM2.5 on hospitalization for asthma and bronchiolitis in younger children (<5 years). Findings from cross-sectional studies using questionnaires suggest that residential proximity to industry may be associated with the prevalence of asthma (pooled OR: 1.98; 95% CI: 0.87, 3.09) and wheezing (pooled OR: 1.33; 95% CI: 0.86, 1.79), but heterogeneity across results was considerable (I2 = 71% and 65%).

While some previous reviews of the association between air pollution and asthma have focused on urban and traffic-related air pollution,15,18–20,71,72 this review is the first to focus on the effects of air pollution from industrial point sources. Despite that our quantitative analysis did not provide strong evidence of an association, we consider that this should not be interpreted as air pollution emitted by industries has no effect on asthma-related outcome in children. The body of evidence from epidemiologic studies strongly support the association between ambient air pollution and asthma exacerbation as well as asthma onset in childhood; therefore, it is very difficult to envision that emissions from industries would not contribute to such adverse respiratory effects. Rather, we consider that the mixed associations from primary studies and the lack of evidence from our meta-analyses are due to important limitations that we discuss below.

Notably, although we did not exclude studies based on quality assessment, we consider that many of the selected studies were not of high quality. Particularly, almost half of the studies used a cross-sectional design. This type of design provides a snapshot of the outcome and exposure at a specific point in time; therefore, prevalence can be measured but not incidence.73 The lack of information about temporality is an important limitation to infer causality.

The exposure assessment, which is a critical component of air pollution studies, was a major limitation in many of the selected studies. Assessing the contribution of point sources to air pollution experienced by individuals is challenging. Point source characteristics, including the amount of emissions, stack height, and plume properties, as well as other parameters influencing dispersion in the atmosphere such as meteorological conditions (e.g., wind speed and direction) and topography are among factors determining the contribution of point sources to ambient air pollution.74 Most studies that account for wind data have found positive association.36,38,44,47,60,69 Especially, findings from White et al69 showed that in adding wind adjustments (speed, direction, and proportion of time blown) to the distance may make an appreciable difference to the inference of association between point source emissions exposure and respiratory symptoms. However, in many studies (n = 15), the exposure was treated as binary based on the community of residence or using an arbitrarily cutoff distance, thus neglecting to account for factors influencing the spatial dispersion of emissions, such as wind direction. Furthermore, the exposure was time-invariant, thus assuming that the current residential location is an adequate proxy of the children long-term (or historical) exposure. Such exposure assessment is subject to substantial exposure misclassification, making it difficult to ascertain whether the observed associations were indeed attributable to air pollution and to what extent emissions from industrial point sources contributed to the observed effects. If misclassification in exposure were non-differential, this would bias effect estimates toward the null and could explain some of the mixed results. For the investigation of the short-term effects, the use of personal monitoring is typically recommended to better quantify the exposure of participants to air pollution, as it allows to account for mobility and time-activity patterns. However, distinguishing the contribution of industries from that of other sources remains very challenging, particularly for air pollutant emitted by multiple sources such as PM. In the only study that made use of personal monitoring, which was limited by a cross-sectional design, the contribution of PM attributable to industries was not determined.66 Alternatively, atmospheric dispersion modeling making use of reliable and sufficient data accounting for characteristics of point source, meteorological conditions and topography, can provide spatiotemporally refined estimates of exposure (both short and long term) to industrial emissions at participants’ residence when compared with fixed-site monitoring.75,76 For short-term exposure to SO2, stronger association with hospitalization for asthma and bronchiolitis was found when using dispersion modeling when compared with measurements at fixed-site monitors.46 Fixed-site monitors may not be specifically located to capture the influence of industrial point source emissions. The ability of a fixed-site monitoring station to represent the exposure of individuals will depend on its location, particularly in terms of distance to the point source and wind direction, and will likely be limited to individuals residing in very close proximity to the station. Additionally, the use of dispersion modeling can be refined when combined with other air pollution estimates, such as background regional ambient concentrations of air pollutant from satellite-based or land use regression models.76 However, because emissions data are rarely available on daily basis, the variation of estimates of pollution from dispersion modeling may be limited to the use of meteorological factors (e.g., wind direction and speed) when investigating the short-term effects of air pollution exposure.

Another limitation of this review relates to the substantial variation across study design and characteristics that made pooling findings generally not possible. These variations likely contributed to the statistical heterogeneity (reflected by the I2 statistic) observed in the meta-analyses conducted, which make interpretation of findings challenging. Particularly, the exposure assessment was a main source of heterogeneity, as a variety of exposure metrics were used. In studies in which the outcome was collected from a questionnaire, many used a binary indicator of exposure based on residential proximity to industries; however, varying distances or definitions were used across studies limiting our ability to compare results across studies. In addition to being subject to important exposure misclassification, such qualitative indicators of exposure do not allow to account for the quantitative magnitude and type of exposure in the analysis, making impossible to derive an exposure-response functions that would support causality.

The fact that studies were conducted in different countries of the world, where ambient air pollution concentration and emission standards may also substantially differ, is likely another source of heterogeneity. In addition, studies covered a wide range of industrial activities, emitting complex mixtures of pollutants that likely vary in composition and toxicity. Evidence from toxicologic, controlled human exposure, and epidemiologic studies suggests that adverse health effects of particulate matter likely depend on the size, composition, and solubility of the particulate matter.2 This was our motivation for extracting information about the types of industrial source, but it was not possible to consider this in our meta-analysis because of the very few number of studies included and the relatively large variety of industrial activities. To address possible drivers of heterogeneity, putative agents and differential toxicity by type of point source, a greater standardization of future study design and methods may be desirable, including greater harmonization of methods and definitions for the outcome and for the exposure, as well as harmonization of confounders included in the analysis. However, because the air pollution mixture may importantly differ across geographical location and depending on the type of industry, any evidence base as defined by non-heterogeneous meta-analysis may be difficult to achieve. In our meta-analysis of case-crossover studies of the association between air pollutants and hospitalization for asthma and bronchiolitis, statistical heterogeneity was substantial despite all studies36,46–48 used a similar design, were conducted in Canada, among children of similar age, using similar methods of exposure and outcome assessment, and of statistical analysis. This inherent heterogeneity implies that evidence may have to rely on findings from a very limited number of studies rather than meta-analyses, thus stressing the importance of having high quality and well-conducted studies.

In this review, a large number of the cross-sectional studies used questionnaire surveys to assess respiratory symptoms and doctor-diagnosed respiratory outcomes. Although studies mostly used standardized questionnaires, these are subject to recall bias; we may suspect that people leaving in proximity to industries will have a tendency to overreport, thus yielding to overestimated effect size. In addition, standardized questionnaires, including the ISAAC and the ATS, are subjected to between-country and between-language variation that have been shown to influence the results and may limit comparison of findings across studies.77,78 Administrative health data using ICD codes is also not free of possible outcome misclassification, but this method appears less prone to biases as compared with questionnaire. However, administrative health data are influenced by healthcare access, which is an issue in several countries worldwide. This may be an additional source of heterogeneity in this review.

Although our review focus on children, age groups varied across studies and this may also be a source of uncertainty and heterogeneity. The diagnosis of asthma is particularly difficult in children <5 years old due to developmental limitations, which may lead outcome misclassification, particularly if this diagnosis is from emergency room visits.79 Age is a risk factor for asthma and possibly an effect modifier of the association with air pollution from industries; however, this could not be addressed in this review because too few studies were comparable.

Conclusions and recommendations

This review highlighted substantial heterogeneity across study design and methods, limiting the conduct of a meta-analysis to a few instances and including very few studies. Results from meta-analyses suggested no evidence for short-term asthma-related effects, whereas for long-term effects there was an indication of an association; however, limitations in terms of design (i.e., cross-sectional) and exposure assessment (i.e., binary exposure based on residential proximity) preclude drawing definite conclusions. Further high quality and well-conducted studies are needed to improve our understanding of the effects of industrial air pollution emissions on asthma and other respiratory outcomes. Specific recommendations include the use of a longitudinal study design, of methods of outcome assessment beyond reporting of doctor-diagnosis (e.g., using prescribed medication from prescription registry and/or diagnosis codes), of refined exposure assessment methods (e.g., atmospheric dispersion modeling) that capture local influence of point sources and that may allow distinguishing the contribution of industries from that of other sources, of continuous rather than categorical exposure analysis and, the inclusion of all important confounders in the analyses.

Conflict of interest statement

The authors declare that they have no financial conflict of interest with regard to the content of this report.

REFERENCES

1. US EPA2008Washington, DCU.S. Environmental Protection Agency
2. US EPA2009Washington, DC
3. Bateson TF, Schwartz J. Children’s response to air pollutants.J Toxicol Environ Health A200871238–243
4. Dixon JK. Kids need clean air: air pollution and children’s health.Fam Community Health2002249–26
5. Gilliland FD, McConnell R, Peters J, Gong H Jr.. A theoretical basis for investigating ambient air pollution and children’s respiratory health.Environ Health Perspect1999107suppl 3403–407
6. Schwartz J. Air pollution and children’s health.Pediatrics20041134 suppl1037–1043
7. Delfino RJ, Wu J, Tjoa T, Gullesserian SK, Nickerson B, Gillen DL. Asthma morbidity and ambient air pollution: effect modification by residential traffic-related air pollution.Epidemiology20142548–57
8. Lavigne E, Villeneuve PJ, Cakmak S. Air pollution and emergency department visits for asthma in Windsor, Canada.Can J Public Health20121034–8
9. Samoli E, Nastos PT, Paliatsos AG, Katsouyanni K, Priftis KN. Acute effects of air pollution on pediatric asthma exacerbation: evidence of association and effect modification.Environ Res2011111418–424
10. Strickland MJ, Darrow LA, Klein M, et al. Short-term associations between ambient air pollutants and pediatric asthma emergency department visits.Am J Respir Crit Care Med2010182307–316
11. Weichenthal SA, Lavigne E, Evans GJ, Godri Pollitt KJ, Burnett RT. Fine particulate matter and emergency room visits for respiratory illness. Effect modification by oxidative potential.Am J Respir Crit Care Med2016194577–586
12. Zheng XY, Ding H, Jiang LN, et al. Association between air pollutants and asthma emergency room visits and hospital admissions in time series studies: a systematic review and meta-analysis.PLoS One201510e0138146
13. Lim H, Kwon HJ, Lim JA, et al. Short-term effect of fine particulate matter on children’s hospital admissions and emergency department visits for asthma: a systematic review and meta-analysis.J Prev Med Public Health201649205–219
14. Dales R, Chen L, Frescura AM, Liu L, Villeneuve PJ. Acute effects of outdoor air pollution on forced expiratory volume in 1 s: a panel study of schoolchildren with asthma.Eur Respir J200934316–323
15. Orellano P, Quaranta N, Reynoso J, Balbi B, Vasquez J. Effect of outdoor air pollution on asthma exacerbations in children and adults: systematic review and multilevel meta-analysis.PLoS One201712e0174050
16. Rodriguez-Villamizar LA, Magico A, Osornio-Vargas A, Rowe BH. The effects of outdoor air pollution on the respiratory health of Canadian children: a systematic review of epidemiological studies.Can Respir J201522282–292
17. Weinmayr G, Romeo E, De Sario M, Weiland SK, Forastiere F. Short-term effects of PM10 and NO2 on respiratory health among children with asthma or asthma-like symptoms: a systematic review and meta-analysis.Environ Health Perspect2010118449–457
18. Guarnieri M, Balmes JR. Outdoor air pollution and asthma.Lancet20143831581–1592
19. Koenig JQ. Air pollution and asthma.J Allergy Clin Immunol19991044 pt 1717–722
20. Khreis H, Kelly C, Tate J, Parslow R, Lucas K, Nieuwenhuijsen M. Exposure to traffic-related air pollution and risk of development of childhood asthma: a systematic review and meta-analysis.Environ Int20171001–31
21. Holguin F. Traffic, outdoor air pollution, and asthma.Immunol Allergy Clin North Am200828577–588, viii
22. Zhang S, Li G, Tian L, Guo Q, Pan X. Short-term exposure to air pollution and morbidity of COPD and asthma in East Asian area: a systematic review and meta-analysis.Environ Res201614815–23
23. Seagrave J, McDonald JD, Bedrick E, et al. Lung toxicity of ambient particulate matter from southeastern U.S. sites with different contributing sources: relationships between composition and effects.Environ Health Perspect20061141387–1393
24. Pedersen SE, Hurd SS, Lemanske RF Jr, et al.; Global Initiative for AsthmaGlobal strategy for the diagnosis and management of asthma in children 5 years and younger.Pediatr Pulmonol2011461–17
25. Hehua Z, Qing C, Shanyan G, Qijun W, Yuhong Z. The impact of prenatal exposure to air pollution on childhood wheezing and asthma: a systematic review.Environ Res2017159519–530
26. Herrera R, Radon K, von Ehrenstein OS, Cifuentes S, Muñoz DM, Berger U. Proximity to mining industry and respiratory diseases in children in a community in Northern Chile: a cross-sectional study.Environ Health20161566
27. Howel D, Darnell R, Pless-Mulloli T. Children’s respiratory health and daily particulate levels in 10 nonurban communities.Environ Res2001871–9
28. Pless-Mulloli T, Howel D, King A, et al. Living near opencast coal mining sites and children’s respiratory health.Occup Environ Med200057145–151
29. Pless-Mulloli T, Howel D, Prince H. Prevalence of asthma and other respiratory symptoms in children living near and away from opencast coal mining sites.Int J Epidemiol200130556–563
30. Rava M, Marcon A, Girardi P, et al. Proximity to wood factories and hospitalizations for respiratory diseases in children.Sci Total Environ2011410-41180–86
31. Rava M, Crainicianu C, Marcon A, et al. Proximity to wood industries and respiratory symptoms in children: a sensitivity analysis.Environ Int20123837–44
32. Buteau S, Goldberg MS. Methodological issues related to pooling results from panel studies of heart rate variability and its association with ambient air pollution.Environ Res2015140462–465
33. DerSimonian R, Laird N. Meta-analysis in clinical trials.Control Clin Trials19867177–188
34. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses.BMJ2003327557–560
35. Clark NA, Demers PA, Karr CJ, et al. Effect of early life exposure to air pollution on development of childhood asthma.Environ Health Perspect2010118284–290
36. Labelle R, Brand A, Buteau S, Smargiassi A. Hospitalizations for respiratory problems and exposure to industrial emissions in children.Environment and Pollution2015477–85
37. Portnov BA, Dubnov J, Barchana M. On ecological fallacy, assessment errors stemming from misguided variable selection, and the effect of aggregation on the outcome of epidemiological study.J Expo Sci Environ Epidemiol200717106–121
38. Buteau S, Doucet M, Tétreault LF, et al. A population-based birth cohort study of the association between childhood-onset asthma and exposure to industrial air pollutant emissions.Environ Int2018121pt 123–30
39. Biesiada M, Zejda JE, Skiba M. Air pollution and acute respiratory diseases in children: regression analysis of morbidity data.Int J Occup Med Environ Health200013113–120
40. Nirel R, Maimon N, Fireman E, Agami S, Eyal A, Peretz A. Respiratory hospitalizations of children living near a hazardous industrial site adjusted for prevalent dust: a case-control study.Int J Hyg Environ Health2015218273–279
41. Awasthi S, Tripathi P, Prasad R. Environmental risk factors for asthma in Lucknow: a case–control study.Clinical Epidemiology and Global Health20131115–123
42. Maantay J. Asthma and air pollution in the Bronx: methodological and data considerations in using GIS for environmental justice and health research.Health Place20071332–56
43. Kobrossi R, Nuwayhid I, Sibai AM, El-Fadel M, Khogali M. Respiratory health effects of industrial air pollution on children in North Lebanon.Int J Environ Health Res200212205–220
44. Loyo-Berríos NI, Irizarry R, Hennessey JG, Tao XG, Matanoski G. Air pollution sources and childhood asthma attacks in Catano, Puerto Rico.Am J Epidemiol2007165927–935
45. de Marco R, Marcon A, Rava M, et al. Proximity to chipboard industries increases the risk of respiratory and irritation symptoms in children: the Viadana study.Sci Total Environ2010408511–517
46. Smargiassi A, Kosatsky T, Hicks J, et al. Risk of asthmatic episodes in children exposed to sulfur dioxide stack emissions from a refinery point source in Montreal, Canada.Environ Health Perspect2009117653–659
47. Lewin A, Buteau S, Brand A, Kosatsky T, Smargiassi A. Short-term risk of hospitalization for asthma or bronchiolitis in children living near an aluminum smelter.J Expo Sci Environ Epidemiol201323474–480
48. Brand A, McLean KE, Henderson SB, et al. Respiratory hospital admissions in young children living near metal smelters, pulp mills and oil refineries in two Canadian provinces.Environ Int20169424–32
49. Karr CJ, Demers PA, Koehoorn MW, Lencar CC, Tamburic L, Brauer M. Influence of ambient air pollutant sources on clinical encounters for infant bronchiolitis.Am J Respir Crit Care Med2009180995–1001
50. Patel MM, Quinn JW, Jung KH, et al. Traffic density and stationary sources of air pollution associated with wheeze, asthma, and immunoglobulin E from birth to age 5 years among New York City children.Environ Res20111111222–1229
51. Aylin P, Bottle A, Wakefield J, Jarup L, Elliott P. Proximity to coke works and hospital admissions for respiratory and cardiovascular disease in England and Wales.Thorax200156228–233
52. Câra AC, Buntinx F, Van den Akker M, Dinant GJ, Manolovici C. Industrial air pollution and children’s respiratory health: a natural experiment in Călăraşi.Eur J Gen Pract200713135–143
53. Wichmann FA, Müller A, Busi LE, et al. Increased asthma and respiratory symptoms in children exposed to petrochemical pollution.J Allergy Clin Immunol2009123632–638
54. Prieto-Parra L, Yohannessen K, Brea C, Vidal D, Ubilla CA, Ruiz-Rudolph P. Air pollution, PM2.5 composition, source factors, and respiratory symptoms in asthmatic and nonasthmatic children in Santiago, Chile.Environ Int2017101190–200
55. Rovira E, Cuadras A, Aguilar X, et al. Asthma, respiratory symptoms and lung function in children living near a petrochemical site.Environ Res2014133156–163
56. Hrubá F, Fabiánová E, Koppová K, Vandenberg JJ. Childhood respiratory symptoms, hospital admissions, and long-term exposure to airborne particulate matter.J Expo Anal Environ Epidemiol20011133–40
57. Câra AC, Degryse J, van den Akker M, Dinant GJ, Manolovici C, Buntinx F. Impact of early childhood air pollution on respiratory status of school children.Eur J Gen Pract201016133–138
58. Chiang TY, Yuan TH, Shie RH, Chen CF, Chan CC. Increased incidence of allergic rhinitis, bronchitis and asthma, in children living near a petrochemical complex with SO2 pollution.Environ Int2016961–7
59. Naidoo RN, Robins TG, Batterman S, Mentz G, Jack C. Ambient pollution and respiratory outcomes among schoolchildren in Durban, South Africa.SAJCH20137127–134
60. Lopes de Moraes AC, Ignotti E, Netto PA, Jacobson LSV, Castro H, Hacon SS. Wheezing in children and adolescents living next to a petrochemical plant in Rio Grande do Norte, Brazil.J Pediatr201086337–344
61. Ripabelli G, Tamburro M, Sammarco ML, de Laurentiis G, Bianco A. Asthma prevalence and risk factors among children and adolescents living around an industrial area: a cross-sectional study.BMC Public Health2013131038
62. Alwahaibi A, Zeka A. Respiratory and allergic health effects in a young population in proximity of a major industrial park in Oman.J Epidemiol Community Health201670174–180
63. Karakis I, Kordysh E, Lahav T, et al. Life prevalence of upper respiratory tract diseases and asthma among children residing in rural area near a regional industrial park: cross-sectional study.Rural Remote Health200991092
64. Deger L, Plante C, Jacques L, et al. Active and uncontrolled asthma among children exposed to air stack emissions of sulphur dioxide from petroleum refineries in Montreal, Quebec: a cross-sectional study.Can Respir J20121997–102
65. Mirabelli MC, Wing S. Proximity to pulp and paper mills and wheezing symptoms among adolescents in North Carolina.Environ Res200610296–100
66. Rosa MJ, Benedetti C, Peli M, et al. Association between personal exposure to ambient metals and respiratory disease in Italian adolescents: a cross-sectional study.BMC Pulm Med2016166
67. Rusconi F, Catelan D, Accetta G, et al. Asthma symptoms, lung function, and markers of oxidative stress and inflammation in children exposed to oil refinery pollution.J Asthma20114884–90
68. Deng Q, Lu C, Norbäck D, et al. Early life exposure to ambient air pollution and childhood asthma in China.Environ Res2015143pt A83–92
69. White N, teWaterNaude J, van der Walt A, Ravenscroft G, Roberts W, Ehrlich R. Meteorologically estimated exposure but not distance predicts asthma symptoms in schoolchildren in the environs of a petrochemical refinery: a cross-sectional study.Environ Health2009845
70. Liu X, Lessner L, Carpenter DO. Association between residential proximity to fuel-fired power plants and hospitalization rate for respiratory diseases.Environ Health Perspect2012120807–810
71. Bowatte G, Lodge C, Lowe AJ, et al. The influence of childhood traffic-related air pollution exposure on asthma, allergy and sensitization: a systematic review and a meta-analysis of birth cohort studies.Allergy201570245–256
72. Lau N, Norman A, Smith MJ, Sarkar A, Gao Z. Association between traffic related air pollution and the development of asthma phenotypes in children: a systematic review.Int J Chronic Dis201820184047386
73. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology2008Philadelphia, PALippincott Williams & Wilkins
74. Hertel O, Goodsite ME. Chapter 1 Urban air pollution climates throughout the world.Air Quality in Urban Environments2009The Royal Society of Chemistry1–22
75. Cora MG, Hung YT. Air dispersion modeling: a tool for environmental evaluation and improvement.Environ Qual Manage20031275–86
76. Hodgson S, Nieuwenhuijsen MJ, Colvile R, Jarup L. Assessment of exposure to mercury from industrial emissions: comparing “distance as a proxy” and dispersion modelling approaches.Occup Environ Med200764380–388
77. Worldwide variation in prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and atopic eczema: ISAACThe International Study of Asthma and Allergies in Childhood (ISAAC) Steering Committee.Lancet (London, England)19983511225–1232
78. Osterman JW, Armstrong BG, Ledoux E, Sloan M, Ernst P. Comparison of French and English versions of the American Thoracic Society respiratory questionnaire in a bilingual working population.Int J Epidemiol199120138–143
79. Plante C, Goudreau S, Jacques L, Tessier F. Agreement between survey data and Régie de l’assurance maladie du Québec (RAMQ) data with respect to the diagnosis of asthma and medical services use for asthma in children.Chronic Dis Inj Can201434256–262

Supplemental Digital Content

Copyright © 2019 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of Environmental Epidemiology. All rights reserved.