1 Introduction
Asthma is a common chronic respiratory disease worldwide. In 2015, 0.40 million people died from asthma, and the number of patients with asthma increased by 12.6% [
1]. Ambient particulate matter pollution is the fourth risk factor contributing to deaths and disability-adjusted life years in China [
2]. Asthma and ambient pollution constitute a substantial disease burden.
Many studies demonstrated the influence of ambient air pollution on asthma [
3], especially in children. Air pollution can increase the risk of asthma [
4–
7] and affect the disease stability and healthy quality of life of children with asthma [
8]. Air pollution also negatively affects the lung function of children with asthma [
6,
7,
9–
12] and similarly impacts the lung function of adolescents [
13,
14]. In adults with asthma, air pollution exposure increase the use of asthma medication [
15], the frequency of acute exacerbation, the number of emergency visits, the frequency of hospital admissions [
16–
18], and mortality [
19].
However, few studies [
20–
22] have illustrated the relationship between air pollution exposure and lung function in adults with asthma. Current evidence focuses on childhood asthma. Meanwhile, the children and adult studies mentioned above focused on the effects of air pollution on large airway function; data on small airway function are limited, even though dysfunction of the small airways is common in patients with asthma [
23–
27] and healthy people [
28]. Thus, prospective longitudinal cohort studies in adults with asthma are necessary to reveal the relationship between air pollution exposure and lung function, including the large and small airways, in severely polluted developing areas.
In this prospective cohort study, adults with asthma were recruited at Peking Union Medical College Hospital between December 2015 and September 2017 in China. The participants were followed at 3-month intervals during the next 2 years to investigate whether or not long-term exposure to high-level air pollution (particles less than 2.5, 10 µm in diameter (PM2.5, PM10), nitrogen dioxide (NO2), ozone (O3), sulfur dioxide (SO2), and carbon monoxide (CO)) is associated with large airway function (forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC, FEV1/FVC), small airway lung function (forced expiratory flow at 50% and 75% of vital capacity % predicted (FEF50, FEF75) and forced expiratory flow between 25% and 75% of vital capacity % predicted (FEF25-75)) decline. The effect of different lung function levels was also analyzed to explore the disparity of different severity grades.
2 Methods
2.1 Study design
In this prospective study, participants with asthma were evaluated every 3 months for 2 years. Lung functions were measured at each visit, and air pollution data were collected. All participants were residents of Beijing, China. The inclusion criterion was confirmed asthma in accordance with the diagnostic criteria in the 2015 Global Initiative for Asthma (GINA) [
29]. House address and demographic characteristics were collected at baseline.
The protocol was approved by the Institutional Review Board of Peking Union Medical College Hospital (JS-914). All participants signed an informed consent prior to this study.
2.2 Lung functions
Spirometry (pre- and post-bronchodilator (pre-/post-BD) administration) was conducted at each visit in accordance with the 2005 American Thoracic Society and European Respiratory Society standards [
30]. Up to eight tracings were obtained every day. Participants inhaled 500 μg of salbutamol after finishing the pre-BD administration, and post-BD administration was conducted 20 min later. We used pre- and post-BD lung function % predicted as predictive variables in accordance with Global Lung Initiative reference equations to avoid confounding factors, such as sex, height, weight, and age, and we defined these variables as pre-/post-FEV
1, pre-/post-FVC, pre-/post-FEV
1/FVC, pre-/post-FEF
50, pre-/post-FEF
75, and pre-/post-FEF
25-75. Spirometry was performed on the FlowScreen Spirometer. The FolwScreen spirometer is a pneumotach, and it was calibrated by the 3L calibrator daily at low, moderate, high rates, respectively.
At baseline, some lung function tests were performed on the JAEGER Spirometer; those tests were not included in this paper to ensure consistency.
2.3 Pollution exposure
The daily average exposure captured by 35 monitors in Beijing was collected from the data issued by the China National Environmental Monitoring Centre hourly. CO, NO
2, SO
2, PM
2.5, and PM
10 were 24 h averages, and O
3 was an 8 h largest moving average. For long-term exposure, the past 90 days of chronic exposure (lag90) were evaluated by the air pollutant weighted average of the 90 days before the visit. The individual exposure of the participants was estimated on the basis of pollution concentration data from 35 monitors by using the inverse distance weight (IDW) method [
19] with a power parameter of 2, according to the home address of the participants. The daily air pollutant exposure of each participant was estimated by calculating the inverse distance (1/
d2) weighted average of each day concentration at the nearest four monitoring stations. The unit of all pollutants was μg/m
3, except that of CO, which was mg/m
3.
2.4 Statistics analysis
The effects of pollution were determined in all participants and also by three lung function levels based on the lung function level of a single visit: severe, post bronchodilator forced expiratory volume in 1 s/ forced vital capacity (post-FEV1/FVC)<50%; moderate, 50%≤post-FEV1/FVC<70%; and mild, post-FEV1/FVC≥70%. The lung function at different visits and severity grade were compared using single-factor ANOVA. A linear mixed-effect model was applied to predict changes in lung function with each unit change in exposure concentration, and each subject had a unique intercept to adjust for individual disparities. The effect was explored in univariate analysis, and then the univariate results were used in a multivariate model. Age, sex, cooking, smoking, inhaled corticosteroids (ICS) and long-acting β-agonist (LABA), temperature, humidity, day of week, education, chronic obstructive pulmonary disease (COPD) complications, and season were included in the multivariate model to adjust for confounding factors. The multiple pollution model failed to converge because of the collinearity of the pollutants. The model with the best R2 was the final predictive model, and pollution was the best predictive factor. Interaction terms of pollution exposure and severity were included in the model, and the significance of the model was tested through ANOVA. All data were managed using IBM SPSS statistics version 20. Statistical analysis was performed in R programming language version 3.6.3 through the lmerTest package.
3 Results
3.1 Basic characteristics
Two of 117 patients with asthma withdrew after receiving diagnoses of Churg–Strauss syndrome and allergic bronchopulmonary aspergillosis. Three participants who left Beijing frequently during the 2 years of observation were not included in the analysis. Visits of participants who left Beijing 10 days or more in the past 90 days were also excluded. Finally, during the two years, 112 participants in our study finished 827 visits. Table 1 shows the main characteristics of the participants. The mean age of the participants was 49.8 years; 53.1% were male and 26.1% had complications of COPD. The severe group had the highest regular ICS availability (93.6%), whereas the mild group had the lowest.
The lung function and air pollution concentration for the last 90 days are shown in Table 2. The FVC of all participants was normal, ensuring the accuracy of small airway function. In patients with severe and moderate asthma, the large and small airway functions had a wide decline. In patients with mild asthma, even with optimal lung function and normal FEV1/FVC, the small airway function failed to eliminate this decline. During the 2 years of observation, the variation in lung function was not significant (all P>0.1) (Table S1). All pollutant concentrations positively correlated with each other (except O3, which negatively correlated with other pollutants) (Table 3). PM2.5 displayed a stronger correlation with CO and PM10 (coefficients were 0.85 and 0.83, respectively), which may explain the collinearity of the pollutants.
3.2 Associations between air pollution and lung function
3.2.1 PM2.5 and CO exposure are negatively associated with the large airway function of patients with asthma
In all participants, PM2.5 and CO were negatively associated with pre-FEV1 and pre-FVC, and CO was also negatively associated with post-FEV1 and post-FVC (Fig. 1). Other chronic effects were nearly all insignificant (Table S2).
3.2.2 PM2.5, CO, and SO2 are negatively associated with the large and small airway functions of the severe group
In the severe asthma group, PM2.5, SO2, and CO were negatively associated with nearly all FEF25-75, FEF50, and FEF75; the effect was weakened after bronchodilator administration, except SO2 (Fig. 2). PM2.5, CO, and SO2 exposure were negatively associated with pre-FEV1 and pre-FVC, but the association was not maintained after bronchodilator administration (Fig. 3). Other pollutants, such as NO2, had a slight reversible effect on small airway function (Table S3).
3.2.3 Air pollution has slight associations with the lung function of the moderate group
In the moderate group, only O3 was negatively associated with post-FEV1/FVC and post-FEF50 (Table S4).
3.2.4 Air pollution shows wide negative associations with the lung function of the mild group
In the mild group, CO, NO2, PM2.5, and PM10 were negatively associated with both pre- and post-bronchodilator large airway functions. All the effects slightly alleviated after bronchodilator administration (Fig. 4). SO2 presented negative impacts on large and small airway functions, and the effect on large airway function alleviated after bronchodilator administration, but that on small airway function was enhanced (Fig. 5). Other pollutants, such as O3, also presented some impacts (Table S5).
3.2.5 Asthma patients react differently to air pollution based on lung function levels, and the effects of SO2 are more serious
Fig. 6 presents all effects of air pollution exposure on lung function. CO and PM2.5 negatively affected the large airway function of all participants. SO2, CO, and PM2.5 presented more significant effects than the other pollutants. CO and PM2.5 presented more effects on large airway function, and effects on small airway function were more prominent in the severe group than in the other groups. SO2 affected the large and small airway functions more evenly and widely in the severe and mild groups than in the moderate group. This result suggests that the effect of SO2 may be more prominent in adults with asthma and that the small airways are more sensitive to SO2 than the large airways. The small airways were the main vulnerable region in the severe group. However, in the mild group, the large airways were considerably more vulnerable to air pollution than the small airways. The moderate group experienced the slightest impact. ANOVA test results of interaction terms for air pollution and group were significant, but some effects of air pollution on lung function were not significant. Table S6 shows the best predictive air pollution factor of each lung function change and its adjusted association of lung function and chronic exposure in all visits and each group.
4 Discussion
Results showed that the inverse association of CO and PM2.5 with large airway function in adults with asthma and the effect of SO2 were more prominent compared with other pollutants in adults with asthma. Moreover, patients with asthma react differently to air pollutants based on their lung function levels. The small airway functions of patients with asthma were widely impaired, even for those with a relatively normal central airway function.
This prospective longitudinal panel study systematically elaborated the negative associations of chronic air pollution exposure with lung function, including large and small airway functions, in adults with asthma in China for 2 years. In our study, CO and PM
2.5 presented an overall negative effect on large airway function. Previous studies shared a similar outcome with ours regarding CO [
12] and PM
2.5 [
11,
21,
31–
33] exposure on large airway function, but our research added knowledge of the long-term impact of PM
2.5 and CO on small airway function in asthmatic adults with poor lung function. The effects seemed reversible partly after bronchodilators. Previous trials showed that annual long-term and 3-day short-term SO
2 exposure is negatively associated with FEV
1/FVC and FEF
25-75, respectively, in children with asthma [
6,
9], and single-day SO
2 exposure is associated with FEV
1/FVC in adults with asthma [
20]. In our research, SO
2 negatively affected the large and small airway functions of the severe and mild groups, and the small airways seemed more sensitive to SO
2 than the large airways in adults with asthma. Reduction of SO
2 pollution may be a critical strategy in the management of asthma.
The effects of NO
2 and PM
10 in our study failed to provide new evidence regarding NO
2 [
31] and PM
10 [
21] compared with other related studies; these other studies were limited to studies of large airway effects. In participants with moderate asthma, only chronic O
3 exposure was inversely associated with lung function, and the association was consistent with previous evidence [
12]. In participants with mild asthma, O
3 exposure was positively associated with lung function. This association was not in accordance with previous results. However, negative correlations were found between O
3 and other pollutants, which prompted that inverse collinearity may lead to the positive effect of O
3 and that the negative effects of other pollutants were significant.
The different asthma subgroups were distinctively affected. The small airways were the main vulnerable region in the severe group, and the lung function in the mild group was affected more widely. In the participants with mild lung function, the large airways were more vulnerable to air pollution than the small airways. By contrast, the adults with moderate asthma experienced the slightest impact. With the deterioration of lung function, the effect on the large airways was attenuated and that on the small airways was enhanced. The severe asthma group had the highest regular ICS availability (93.6%), whereas the mild group had the lowest ICS availability. This result suggests that ICS can lighten the effect of air pollution on the large airways but not on the small airways. Asthma patients with different lung function levels react differently to air pollution exposure, indicating that post-FEV1/FVC is a reasonable objective marker to distinguish the special subgroups of patients with asthma vulnerable to air pollution. The effect of air pollution presented a U-type effect in our study, and the moderate group presented the weakest effect, whereas obvious and wide effects were found in the severe and mild groups.
The small airways play a key role in the prevalence of asthma [
23], and small airway dysfunction leads to asthma persistence and terrible control [
26]. A multinational prospective cohort study showed that small airway dysfunction is present in all severities of asthma, particularly in the severe type [
27]. In our panel, the small airway functions of the patients with asthma were impaired widely, even in individuals with a relatively normal large airway function, emphasizing the high prevalence of small airway dysfunction in adults with asthma in China.
Our 2-year prospective longitudinal panel showed that chronic air pollution exposure exerted a significantly negative impact on the large and small airway functions of adults with asthma in China. However, this study has some limitations. First, few associations were found in the moderate asthma subgroup, which violated the hypothesis that the air pollution effect coincides with the severity of asthma. Both asthmatic participants with and without complicated COPD were included in the study even though complicated COPD did not show significant effects. This result should be clarified in future studies without including patients with complicated COPD. Second, as in the same city with severe air pollution, the levels of the six air pollutants in Beijing were tightly correlated. Therefore, the single effects of each pollutant were difficult to distinguish definitively because of the inevitable multicollinearity. However, the collinearity of the pollutants could not influence the prediction of lung function variation. We determined the optimal predictive factors based on the R2 of the model. A wearable air quality monitor may better distinguish the single effect of each pollutant, eliminating the collinearity and ambient environmental influence [
34].
5 Conclusions
CO and PM2.5 exposure were negatively associated with the large airway function of adults with asthma. Reaction modes of air pollutants varied based on different lung function levels. The negative effects of SO2 were widely observed in patients with severe and mild asthma but not in patients with moderate asthma.