Coal burning-derived SO₂ and traffic-derived NO₂ are associated with persistent cough and current wheezing symptoms among schoolchildren in Ulaanbaatar, Mongolia

We used a cross-sectional study design to investigate the effects of short-term outdoor exposure to NO₂ and SO₂ on the respiratory symptoms of children during the 2-year period from 2015 to 2016.

The participants were 1200 children (age 6–12 years) from five public elementary schools. The schools were located in three urban districts (Sukhbaatar, Khan-Uul, and Bayanzurkh) and a suburban district (Nalaikh) in Ulaanbaatar. The districts were named A, B, C, and D, respectively. Figure 1 shows the locations of the study areas in Ulaanbaatar. District A is located in the center of the city. Two public schools in district A were included in this study. District B is located adjacent to district A, and industrial activity takes place in this district. District C is located at the eastern end of the urban area, and most residents live in “gers”. For comparative analysis, we selected suburban area D, which is located 40 km southeast of the urban area. It is the nearest suburban area to the capital and has increased rapidly in size, as internal migrants have moved from rural areas; therefore, approximately 70% of the residents live in “gers”.

In each of these districts, the targeted school was close to traffic major road within 200 m. In the urban areas, outdoor air pollutant levels were measured by ambient air quality-monitoring stations. However, such systems have not yet been installed in district D (the suburban area; Fig. 1).

We estimated the distance between each subject’s residence and the nearest major road using the GeoNET 9.7.107 GPS navigator for mobile mapping software. Then, we categorized the subjects into ≤ 100  m and > 100 m groups, depending on the distance from their residence to the nearest major road. Also, we measured the daily traffic volume at the major road nearest to each school four times in 2015. The mean daily traffic volume was determined using a vehicle counter (Tempomat CRM, IV, Radarlux).

The survey about respiratory symptoms was based on a questionnaire adapted from the American Thoracic Society Epidemiologic Standardization Project questionnaire [19]; i.e., the ATS DLD 78 C questionnaire for children, which has been used in previous studies [8, 20‐24]. The questionnaire was translated into Mongolian and then distributed to the participants via their schools. It was completed by their parents at home after obtaining written parental consent. The participants were given a week to return the questionnaire. All of the questionnaires were given ID numbers prior to the analysis.

The items included in the questionnaire were age, sex, residential address, whether the subject was bottle-fed in infancy, how long they had lived at their current residence, the heating system used in their home, their parents’ smoking habits at home, and their history of respiratory disease and current respiratory symptom status. The questionnaire covered the following respiratory symptoms: persistent cough, persistent phlegm, current wheezing, and asthma-like symptom. Respiratory symptoms were recorded as being present if a participant answered, “Yes”, to a relevant question. The participants were provided with explanations about each respiratory symptom [see Additional file 1: Table S1].

We selected NO₂ and SO₂ as indicators of outdoor air pollution. The ambient air concentrations of NO₂ and SO₂ were measured near the schools. The concentrations of these pollutants were measured in February, April, August, and November in 2015 and in January, April, July, and October in 2016. The two field blanks used for the concentration analysis and the filters for NO₂ and SO₂ were exchanged five times each at intervals of approximately 48 h. The samplers were installed for 10 days (14,400 ± 5 min). We measured the mean air temperature and relative humidity during the same period as the NO₂ and SO₂ measurements were obtained. The levels of NO₂ and SO₂ were assessed using co-located Ogawa passive samplers (Ogawa & Co., Ltd., Kobe, Japan), which were used in previous studies [4, 25‐28]. The passive samplers were installed at two sites, which were located ≤ 100 m and > 100 m from the nearest major road, respectively, at a height of approximately 2.5 m.

All statistical analyses were performed using the PASW Statistics software, v.18 (Chicago, IL, USA). All questionnaire data were analyzed based on the ID numbers assigned to the questionnaires. The associations between respiratory symptoms and personal and environmental factors were analyzed statistically via multinomial logistic regression analysis. The respiratory symptoms were used as dependent variables, and potential confounding factors were used as independent variables. The following variables were considered to be potential confounding factors: gender, age, how long the subject had lived at their current residence (years), history of asthma diagnosed by a doctor, history of allergies or respiratory disease before the age of 2 years, history of pneumonia, being bottle-fed in infancy, parental smoking habits, and the distance between the subject’s residence and the nearest major road. We also analyzed the associations between the concentrations of air pollutants and respiratory symptoms; i.e., logistic regression was used to calculate odds ratio (OR) and 95% confidence interval (CI) values. P-values of <  0.05 (two-tailed) were considered statistically significant. Adjusted OR were estimated and scaled to represent a 1-ppb increase in the NO₂ or SO₂ concentration, based on the mean 2-year concentrations and traffic volume data for each district. The paired t-test was used to examine the significance of differences in the seasonal concentrations of SO₂ and NO₂, and changes in the NO₂ concentration due to the distance from the main road. We evaluated the statistical significance of differences in the frequency of respiratory symptoms between the residential areas using one-way analysis of variance (ANOVA). P-values of <  0.05 (two-tailed) were considered to be statistically significant.

The NO₂ collection filters (14.5 mmφ) were placed in 25-ml glass vials. Then, 8 ml distilled water was added to the filter-containing vials, and each vial was subjected to shaking for extraction. After 30 min of extraction, the vials were refrigerated at 4–5 °C for 30 min. Then, 2 ml of sulfanilamide solution and 1-naphthyl-ethylenediamine dihydrochloride (NEDA; 0.56 g dissolved in 100 ml distilled water, stored at − 4 °C) solution were added in a 10:1 ratio. The resultant mixture was simultaneously shaken and cooled for 30 min. After that, the vials were returned to room temperature. The absorbance of the samples was determined with a spectrophotometer at a wavelength of 545 nm.

Filter papers were put in 25-ml glass vials containing 8 ml distilled water. Then, the vials were shaken for extraction for about 30 min, before the filter papers were collected. Next, hydrogen peroxide solution (H₂O₂, 1.75%) was added, and the vials were shaken slowly for 10 min, before the mixture was equilibrated at room temperature and analyzed using ion chromatography. SO₂ concentrations were analyzed via ion chromatography at the laboratory of Ogawa & Co., Ltd. (Kobe, Japan). The same procedure was used for all blank samples and standard solutions. We calculated the ambient concentrations of NO₂ and SO₂ using the following equations:

α_NO2, α_SO2: ppb concentration conversion coefficient (ppb min/ng).

W _NO2: quantity (ng) collected in NO₂ collection elements.

W _SO2: quantity (ng) collected in SO₂ collection elements.

α_NO2 was calculated based on the temperature, humidity, and exposure time during the measurement period. We used meteorological data from the private Weather Underground Station Network, which directly collects data from weather stations around the world [29]. The abovementioned analytical methods for NO₂ and SO₂ concentrations were based on the Ogawa protocol.

During the measurement period, the mean daily temperatures varied among the seasons. In winter (February 2015 and January 2016), spring (April 2015 and 2016), summer (August 2015 and July 2016), and fall (November 2015 and October 2016), the mean daily temperature was − 22.1 °C, 8.5 °C, 22.0 °C, and − 8.2 °C and the mean relative humidity value was 65%, 35%, 52%, and 54%, respectively. Table 1 summarizes the distribution of mean air pollutant concentrations in the four districts of Ulaanbaatar during the 2-year period from 2015 to 2016. In each study area, the concentrations of SO₂ and NO₂ exceeded the ambient air quality standard values for Mongolia [annual mean concentration of SO₂: 10 μg/m³ (3.8 ppb), annual mean concentration of NO₂: 30 μg/m³ (16 ppb); MNS 4585:2007]. In districts A–D, the 2-year mean concentrations of SO₂ were 5.2 ppb, 4.4 ppb, 10.1 ppb, and 7.1 ppb, respectively, and the maximum concentrations of SO₂ were 36 ppb, 39 ppb, 63 ppb, and 46 ppb, respectively. The minimum concentration of SO₂ was 0.0 ppb in each district. Higher SO₂ concentrations were detected in districts C and D than in districts A and B.

For NO₂, the mean annual concentration was 30.0 ppb, 27.7 ppb, 30.0 ppb, and 18.2 ppb in districts A–D, respectively, and the maximum concentrations were 86 ppb, 74 ppb, 99 ppb, and 68 ppb, respectively. Higher NO₂ concentration were detected in districts A, B, and C than in district D. Figure 2 shows seasonal comparisons of outdoor SO₂ concentrations between the four districts. High SO₂ concentrations were detected in each district in January and February. During this period, the mean SO₂ concentration was 14.3 ± 9.0 ppb, 12.3 ± 10.2 ppb, 21.9 ± 10.3 ppb, and 19.6 ± 11.6 ppb in districts A–D, respectively. There were significant differences in the SO₂ concentration between each district, except between districts A and B (p < 0.01) [see Additional file 1: Table S3]. In April, the mean SO₂ concentrations in each district were roughly similar (the mean SO₂ concentrations of districts A–D decreased to 2.0 ± 1.5 ppb, 3.0 ± 1.9 ppb, 2.6 ± 1.8 ppb, and 2.1 ± 3.2 ppb, respectively). Only the SO₂ concentrations of districts A and B differed significantly (p < 0.05). In August, the mean SO₂ concentration decreased to 1.0 ppb in each district. However, in November, the mean SO₂ concentration increased again to 5.0 ± 2.5 ppb, 2.9 ± 1.7 ppb, 9.6 ± 8.3 ppb, and 8.0 ± 4.6 ppb in districts A–D, respectively. There were significant differences between the SO₂ concentrations of all districts, except for between the SO₂ concentrations of districts C and D.

The results indicated that in all districts, the mean NO₂ concentration exceeded the ambient air quality standard value in all months, except for August. Figure 3 shows a seasonal comparison of outdoor NO₂ concentrations between the districts. The concentration of NO₂ was high in each district in February, especially in the urban districts (A, B, and C). The mean NO₂ concentration was 54.6 ± 17.1 ppb, 48.1 ± 15.1 ppb, 56.3 ± 21.4 ppb, and 40.5 ± 14.1 ppb in districts A–D, respectively. There were significant differences between the NO₂ concentrations of all districts, except for between those of districts A and C (p < 0.01) [see Additional file 1: Table S4].

In April, the mean NO₂ concentration decreased to 20.0 ± 3.9 ppb, 21.6 ± 5.4 ppb, 17.5 ± 7.3 ppb, and 9.1 ± 1.9 ppb in districts A–D, respectively. There were significant differences between the NO₂ concentrations of each district except for between those for districts A and B, and districts A and C (p < 0.01). In August, the mean NO₂ concentrations of the three urban districts were roughly similar and had decreased to 16.5 ± 3.8 ppb, 16.5 ± 6.0 ppb, 16.9 ± 7.4 ppb, and 6.1 ± 1.8 ppb, respectively. The mean NO₂ concentration of district D differed significantly from those of districts A, B, and C (p < 0.000). However, the mean NO₂ concentration of each district increased again in November. The mean NO₂ concentration was 28.7 ± 8.5 ppb, 24.7 ± 9.2 ppb, 28.7 ± 8.0 ppb, and 17.4 ± 8.1 ppb in districts A–D, respectively. There were significant differences between the mean NO₂ concentrations of all districts, except between those of districts A and C (p < 0.01). In all districts, the mean NO₂ concentration in each season exceeded the ambient air quality standard value (except for the NO₂ concentrations of district D in April and August, Fig. 3).

In addition, in each district we examined the NO₂ concentrations detected at sites located ≤ 100 m and > 100 m from the nearest major road (hereafter referred to as the ≤ 100 m and > 100 m sites). Table 2 shows the differences in the NO₂ concentration depending on the distance from the nearest major road. In districts A, B, and C, significantly higher NO₂ concentrations were detected at the ≤ 100 m sites than at the > 100 m sites. The mean values obtained for the ≤ 100 m and > 100 m sites in each district were as follows: district A, 31.8 ppb vs. 29.1 ppb (p < 0.05); district B, 34 ppb vs. 23 ppb (p < 0.01); and district C, 33 ppb vs. 27 ppb (p < 0.01), respectively. However, there were no significant differences between the values obtained for the ≤ 100 m and > 100 m sites in the suburban district (district D) (17.5 ppb vs. 17.8 ppb; p < 0.15).

Mapping software was used to determine the location of each subject’s residence based on their home address, and the distance between each subject’s home and the nearest major road was manually determined for all participants [see Additional file 1: Figure S1]. All elementary schools were located within 200 m of the nearest major road. In the urban districts (A, B, and C), the examined major roads had 2–3 times higher traffic volumes than the major road in the suburban district. The mean ± standard deviation daily traffic volume was 16,842 ± 2211, 16,985 ± 1788, 19,473 ± 2686, and 7187 ± 736 vehicles in districts A–D, respectively.

The survey was carried out in 2015. The questionnaire was completed by 1130 children, and 21 children were excluded because they were living in other districts or did not fill out more than half of the questionnaire. Table 3 shows the personal and environmental data obtained via the questionnaire alongside the prevalence of respiratory symptoms by district. The ATS DLD questionnaire was correctly completed by 1109 (92.4%) participants. Among the subjects aged 6–12 years, the most common respiratory symptom was persistent cough (28.5%). In addition, 20.2% of the subjects reported persistent phlegm, 5.8% had current wheezing, and 0.7% reported asthma-like symptom. Furthermore, the following findings were obtained regarding the frequency of other background factors: being bottle-fed in infancy: 18.4%, a history of respiratory disease before the age of 2 years: 26%, a history of diagnosed asthma: 2.4%, a history of diagnosed pneumonia: 6.2%, history of residence year for > 3 years: 65.6%, parents smoke at home: 43.1%, coal being used for heating at home: 33.5%, and living ≤ 100 m from the nearest major road: 37.4%. In particular, in districts C and D, high prevalence of subjects had persistent cough (32% and 37%, respectively). However, the prevalence of asthma-like symptoms was low.

Table 4 shows the associations between the outdoor SO₂ or NO₂ concentration and the prevalence of respiratory symptoms identified using a logistic regression model adjusted for confounding variables (n = 1109). Persistent cough was associated with the outdoor concentration of SO₂ (OR 1.12, 95% CI 1.04–1.22; per 1 ppb increase), whereas current wheezing was associated with the NO₂ concentration (OR 1.33, 95% CI 1.01–1.75). Conversely, no statistically significant associations were found between the outdoor SO₂ or NO₂ concentration and other symptoms. Also, traffic volume was not associated with respiratory symptoms, such as with persistent cough (OR 0.87, 95% CI 0.73–1.03), persistent phlegm (OR 0.95, 95% CI 0.80–1.14), or current wheezing (OR 0.77, 95% CI 0.58–1.03). The associations between personal or environmental factors and respiratory symptoms were examined using multinomial regression analysis. Current wheezing was found to be significantly associated with being bottle-fed in infancy (OR 2.04, 95% CI 1.14–3.66), and persistent phlegm was demonstrated to be associated with a history of respiratory disease before the age of 2 years (OR 1.58, 95% CI 1.10–2.25). Also, a history of pneumonia was shown to be associated with persistent phlegm (OR 1.90, 95% CI 1.07–3.36) and current wheezing (OR 4.01, 95% CI 1.87–8.60) [see Additional file 1: Table S5]. Furthermore, the prevalence of persistent cough was significantly increased among the children that lived within 100 m of a major road (OR 1.86, 95% CI 1.37–2.51). However, we could not analyze the associations between current asthma-like symptom and confounding factors, as the prevalence of this condition was very low.

This study confirmed that Mongolian schoolchildren have high prevalence rates of persistent cough (28.5%), persistent phlegm production (20.2%), current wheezing (5.8%), and asthma-like symptoms (0.7%). When the prevalence was examined by district, the prevalence of persistent cough was high in the suburban compared to the urban district which has similar many gers dwellings (37% vs. 32%, respectively). Conversely, the concentration of SO₂ was higher in district C than in the suburban (district D). The survey indicated that there are several possible reasons for this first, more children in district D (72%) than in district C (47%) lived in a household in which coal was used. So, it is considered that these children were affected by emissions from coal. Another possible reason is that the children in district D had lived in the same place for a longer period than those in district C; i.e., 74% of the children in district D had lived in the same residence place for > 3 years, whereas the equivalent figure for district C was 62% (data shown in Table 3). In a previous study, in which the subjects had a similar age distribution to those in our study, it was concluded that outdoor air pollution is the major source of black carbon (BC) and fine particulate matter (PM_2.5) exposure for children living in Ulaanbaatar. Also, it was reported that the children living in a district containing many “gers” were exposed to the highest levels of outdoor air pollution [37]. In our study, persistent cough was significantly more common in the areas containing many “gers” (districts C and D) than in the areas containing large numbers of apartments (districts A and B, p < 0.001) [see Additional file 1: Table S6]. In particular, the prevalence rate of persistent cough was significantly higher in the suburban district (district D) than in the urban areas (districts A, B, and C, p < 0.000). Even in the suburbs, the use of coal has led to a high prevalence of persistent cough in children. We were reported that on an isolated island without major artificial sources of air pollutants, changes in the concentrations of air pollutants acutely influenced the pulmonary function of healthy subjects [38]. On the other hand, the increased prevalence of respiratory symptoms may also affect the burning of solid waste at the open dumpsites in the gers districts. There still has a problem with solid waste, coal ash and other rubbish, and also low-income households burn old tires, plastic bottles [39, 40]. Das B et al. reported that open burning of municipal solid waste is a poorly characterized and frequently underestimated source of air pollution. It is can trigger health impacts such as acute and chronic respiratory disease [41]. Boadi K. O et al. reported also that inadequate solid waste facilities result in indiscriminate burning and burying of solid waste. There is an association between waste burning and the incidence of respiratory health symptoms among children [42].

In the statistical analysis, we confirmed that the prevalence of persistent cough among children was associated with the concentration of SO₂. Some Chinese studies have reported that children exhibit enhanced sensitivity to the harmful effects of air pollution. In addition, it was confirmed that genetic susceptibility and/or exposure to common environmental factors have a strong influence on respiratory health [8, 43].

In our study, the current wheezing symptom among children associated an increase in NO₂ concentration (data shown in Table 4). Nicolai T et al.reported that the frequency of current wheezing among children increased with the NO₂ concentration [44]. Also, in our study, the children who were bottle-fed in infancy displayed an increased prevalence of current wheezing. Other studies have demonstrated that being breastfed was associated with a lower risk of lung function problems and reduced children’s susceptibility to the respiratory effects of pollutants [45, 46].

The prevalence of persistent cough and phlegm were reported to be more common among children in Mongolia than among children in the north and north-eastern cities of China and Thailand, but the frequencies of current wheezing and asthma-like symptom were similar in all of these places [8, 21, 22, 43]. It is considered that other than the geographical features mentioned above and the temperature inversion layer phenomenon that occurs in winter, meteorological features also have a great influence, as Ulaanbaatar has a longer heating period than China; i.e., it lasts for 8 months from mid-September to mid-May [47]. In winter, when coal consumption increases, the concentrations of air pollutants are expected to rise markedly, which might induce respiratory symptoms in children. However, a previous Japanese study found that persistent cough and phlegm were less common among Japanese children than among Mongolian children, although current wheezing and asthma-like symptom were more common among Japanese children [22]. One possible reason for this is that since the 1970s, the atmospheric concentration of SO₂ in Japan has reduced owing to the introduction of emission-control technologies. On the other hand, NO₂ concentrations decreased from 1970 to 1985, but increased from 1985 to 1995. NO₂ concentrations are notably higher at roadside air-monitoring stations [48].

We conducted a survey of respiratory symptoms among children and measured the air pollution levels in Ulaanbaatar, including in a suburban area. No previous studies have examined the prevalence of respiratory symptoms among children in suburban areas. Furthermore, outdoor air pollutants were being measured at ambient air quality monitoring stations in the urban areas of Ulaanbaatar. However, such systems have not been installed in the suburban areas surrounding the city. Moreover, traffic volume was assessed in a suburban area for the first time in this study.

In this study, we only measured the levels of NO₂ and SO_2. The obtained SO₂ concentrations were lower than those obtained via national air quality monitoring because they were measured at sites located within 200 m of major roads. Also, this study did not measure the levels of particulate matter, and the personal exposure levels of each subject were not evaluated. In addition, this study did not use continuous monitoring data, and the measurement of air pollutants was conducted over a short period of time. The concentrations of air toxins, such as diesel exhaust particles or surrogates, such as black carbon or soot, should be more widely monitored.