Skip to main content
NCBI home page
Environment & Health logoLink to Environment & Health
. 2026 Jan 23;4(6):1176–1184. doi: 10.1021/envhealth.5c00439

Ambient Air Pollution and Hospital Admissions of AECOPD in 10 Regions of China: A Self-Controlled Study Based on a Cohort

Lu Chen †, Yuxuan Zhao †, Jun Lv †,‡,§,∥, Dianjianyi Sun †,‡,§, Pei Pei ‡, Ling Yang ⊥, Yiping Chen ⊥, Huaidong Du ⊥, Shichun Yan #, Dan Schmidt ⊥, Junshi Chen ∇, Zhengming Chen ⊥, Liming Li †,‡,§, Canqing Yu †,‡,§,*
PMCID: PMC13288233  PMID: 42344593

Abstract

Gaps exist concerning the associations of PM2.5–10, gaseous pollutants, and composite pollutants with the acute exacerbation of chronic obstructive pulmonary disease (AECOPD). The interaction effects of lifestyle factors, indoor air pollution, and meteorological conditions with air pollutants have been underexplored. This time-stratified case-crossover study was nested within the China Kadoorie Biobank cohort study of middle-aged and elderly participants from five urban and five rural areas in China spanning from 2004 to 2018. 10,712 participants were included, each with a mean of 2.3 episodes of AECOPD hospitalizations during the observation period. A composite air pollution score was derived through principal component analysis. Short-term exposure to particulate matter and NO2 was positively associated with hospital admissions for AECOPD, with the largest temperature- and humidity-adjusted odds ratio (95% confidence interval) of 1.101 (1.073, 1.129) at a lag of 1 day for per 1 standard deviation increase in air pollution score. The effect of O3 appeared to be conflicting. The associations were further corroborated by the self-controlled case-series design. Unfavorable body shapes, cold season, and low humidity could exacerbate the influence of air pollution. These findings reinforce the evidence of links between air pollution and AECOPD and implicate the management of body weight and cold and dry circumstances.

Keywords: air pollution, chronic obstructive pulmonary disease, acute exacerbation, case-crossover design, self-controlled case-series design

graphic file with name eh5c00439_0005.jpg

graphic file with name eh5c00439_0003.jpg

Introduction

Ambient air pollution, primarily attributable to particulate matter (PM) with an aerodynamic diameter of ≤2.5 μm (PM2.5), was estimated to result in 4.2 million premature deaths of cardiovascular and respiratory diseases and cancer worldwide in 2019. It constitutes a major environmental health concern, especially within the World Health Organization (WHO) South-East Asia and Western Pacific Regions.

Emerging evidence shows that air pollution may be associated not only with noncommunicable diseases but also with acute attacks of diseases, including the acute exacerbation of chronic obstructive pulmonary disease (AECOPD). − However, few studies distinguish the effects of PM2.5–10 from PM10, and the independent role of PM2.5–10 remains poorly understood. Also, the findings regarding the associations of gaseous pollutants with AECOPD were inconsistent. − This discrepancy may be attributed to limited variations in air pollution concentrations, especially in developed countries or regions, given that many studies were confined to single administrative divisions or specific geographical areas. , To enhance our understanding of the impact of air pollution on AECOPD, a study investigating and comparing multiple locations is merited. While the comprehensive assessment of health impacts from multiple air pollutants has garnered increasing recognition in environmental epidemiology, , to date, no study has systematically examined the combined short-term effects of multipollutant exposure on AECOPD, to the best of our current scientific understanding.

Most prior studies relied on aggregate data, whereas data at the individual level render an opportunity to examine nuanced effect modifications by personal characteristics. Although some researchers have explored potential effect modifiers (i.e., sociodemographic characteristics and the season of hospital admissions) through stratified analyses, few studies have considered the effect modifications by lifestyle factors and indoor air pollution. Furthermore, the interaction effects of meteorological conditions with air pollutants have been underexplored, and the existing findings are conflicting.

In the present study, we employed a self-controlled design to examine associations of individual and composite air pollutants with AECOPD across 5 urban and 5 rural regions in China based on the large-scale cohort data from the China Kadoorie Biobank (CKB) study, spanning from 2004 to 2018. We further investigated the interaction effects of multiple pollutants and effect modifiers on AECOPD.

Methods

Study Population

The design of CKB has been elaborated upon elsewhere. , Briefly, the baseline survey of CKB was initiated during 2004–2008, enrolling ∼0.5 million participants aged 30–79 from 5 urban and 5 rural areas. The China Kadoorie Biobank (CKB) complies with all of the required ethical standards for medical research on human subjects. Ethical approvals were granted and have been maintained by the Chinese Centre for Disease Control and Prevention (Beijing, China: 005/2004) and the University of Oxford (UK: 025–04). Written informed consent was procured from each participant involved in the study after a detailed introduction of the research at baseline.

Baseline Data Collection

During the baseline survey, about 1800 study clinics were set near participants’ residences to ensure convenience (most within a 1 km radius). Qualified staff conducted interviews with participants to collect information about sociodemographic characteristics, lifestyle factors, indoor air pollution, and personal medical history. Detailed information on smoking included smoking status (never, ever, or currently) and the primary reason for smoking cessation in quitters. The main fuel types for cooking and heating were classified into 2 major categories: solid and nonsolid. The frequency of direct exposure to second-hand smoke (SHS) either at home or in the workplace was inquired.

Physical measurements, including height, weight, waist circumference (WC), forced expiratory volume in 1 s (FEV1), and forced vital capacity (FVC), were performed by trained technicians using calibrated instruments. Body mass index (BMI) was derived as the ratio of weight (kg) to the square of the baseline height (m). FEV1 and FVC were assessed through two successful expiratory maneuvers, and a larger measurement was adopted.

Exposure and Covariate Assessment

Daily concentrations of PM2.5, PM10, 8h O3, and NO2 were retrieved from the ChinaHighAirPollutants data sets. PM2.5–10 level was estimated by subtracting PM2.5 from PM10. The air pollution score was calculated as the first principal component (PC) by principal components analysis (PCA), and standardized with a mean of 0 and a standard deviation (SD) of 1. Additionally, daily 2-m temperature and specific humidity were extracted from NASA GESDISC DATA. Participants were assigned ambient exposure assessments based on the study clinics where they were recruited. The details of assessments are described in the Supporting Information.

Outcome Ascertainment

Incident outcome cases were primarily identified by electronic linkage, via a unique personal identification number, to the inpatient health insurance (HI) database. Hospitalizations that occurred during the follow-up and had a discharge diagnosis of AECOPD (coded as J44.0-J44.1 by the International Classification of Diseases, 10th revision) were extracted from the HI data. By December 31, 2018, 24,409 incident AECOPD cases involving 10,727 participants were documented since the baseline survey.

Study Design

A time-stratified case-crossover design was employed to estimate the associations of exposure to air pollutants with hospital admissions for AECOPD. The design was initially proposed by Maclure to study the effects of transient exposures on the abrupt onset of events and then generalized to the context of air pollution and acute events. The self-matched design compares each case’s exposure levels during hazard periods relevant to the causation of the outcome with one or more control periods. , This method eliminates control selection bias and confounding led by time-invariant characteristics, thus enhancing efficiency. , The time-stratified approach selects control periods based on the day-of-week (DOW) within the same year and month as case days, namely, the day on which the AECOPD hospital admission occurred, thereby accounting for the long-term and seasonal trend, as well as the influence of DOW, ensuring minimal bias.

Statistical Analysis

Associations of exposure to individual air pollutants with hospital admissions for AECOPD were estimated using conditional logistic regression models. According to previous studies, models were adjusted for the natural cubic splines of the 3-day average temperature and humidity, from the index date to the two preceding days, with degrees of freedom (df) of 6 and 3, respectively. , Considering the self-matched design, adjustments were not made for relatively stable factors in the short term, such as sociodemographic factors, lifestyles, and health status. Instead, the specific modifying effects of these characteristics on associations were thoroughly examined, leveraging the detailed profile of the cohort participants. To uncover potential lag patterns, exposures of single-day lag over 0–7 days were included in the models, and the single-day lag yielding the largest effect estimate was then selected as the main indicator for quartile and subgroup analyses.

Also, we conducted subgroup analyses to explore whether the association of air pollution with AECOPD hospitalization was consistent across various subpopulations. These subpopulations were stratified by sociodemographic characteristics, lifestyle factors, indoor air pollution, disease status, season of hospital admission, and meteorological conditions. The interaction effects were examined using likelihood ratio tests by comparing models with and without interaction terms for air pollutants and strata variables. To maximize the potential for discovering novel relationships, we opted not to apply multiple testing corrections when exploring potential effect modifiers.

To ensure the robustness of our findings, we conducted several sensitivity analyses. First, we adopted the self-controlled case-series (SCCS) design to confirm the relationship between exposure to air pollutants and hospital admissions for AECOPD, further considering the heterogeneity across the 10 cities. The observation period from enrollment to censoring for each case was split day by day. Also, the occurrence of events was regressed on the air pollutants on a continuous scale each day. The advantages of SCCS resemble the case-crossover design. By contrast, SCCS builds on the principles of a cohort study, focusing on intrapersonal comparisons of each case’s outcome risk during exposure periods to the risk during baseline periods, thereby yielding estimates of relative incidence. , The SCCS model requires certain assumptions to produce valid and unbiased estimates. One assumption requires that recurrences of AECOPD events are independent; therefore, only first hospital admissions of AECOPD were included, and participants with self-reported or screened COPD at baseline were excluded. Another criterion to be met is that the event does not affect subsequent exposure, and thus deaths within 28 days of AECOPD events were further excluded. Here, a two-stage analysis was conducted to examine both city-specific and combined associations. At the first stage, conditional logistic models with adjustments for age effect, temperature, humidity, month, and DOW were conducted at the individual regional level. At the second stage, meta-analysis was used to combine the effect estimates from the 10 cities to obtain the short-term health effects of air pollution at the pooled level. Second, pollutants were further mutually adjusted to eliminate potential confounding by coexposures, except for the pair of PM2.5 and PM10, considering the high correlation (Table S1). Third, the models were adjusted for the natural cubic splines of the 7-day average temperature and humidity. Fourth, relative humidity was included in the models instead of specific humidity. Fifth, the air pollution score was calculated using the equation: air pollution score=∑i=1n(PCiEi)∑i=1nEi , where PC i was the PC for which the corresponding eigenvalue (Ei ) was ≥1. Sixth, recurrent cases of AECOPD within 28 days of the last acute exacerbation were excluded from the analysis. The robustness of the findings was directly tested according to the difference in estimates of ORs and statistical significance for ORs.

We used Stata (ver. 17.0) and the SCCS package in R software (ver. 4.3.2) for statistical analyses. Statistical significance was defined as the two-sided P-values <0.05.

Results

Basic Characteristics

The primary analysis included 10,712 participants, each with a mean of 2.3 episodes of AECOPD hospitalizations during the observation period (Figure S1). Of these participants, the mean (SD) age was 61.6 (8.7) years, and 49.0% were female. Compared with the whole CKB participants, AECOPD cases included in the study were generally older, predominantly male, resided in rural areas, had lower educational attainment, smoked regularly, and primarily used solid fuels for cooking. Notably, 58.1% of these hospital admissions occurred during the cold season (October–March next year) (Table ).

1. Baseline Characteristics of AECOPD Cases during the Study Period .

variables AECOPD cases (N = 10,712) whole CKB cohort (N = 512,724)
age, years 61.6 (8.7) 52.0 (10.7)
female 5245 (49.0%) 302,519 (59.0%)
urban residents 2014 (18.8%) 226,191 (44.1%)
middle school or above 2189 (20.4%) 252,359 (49.2%)
agriculture/factory worker 7236 (67.6%) 286,319 (55.8%)
currently smoking 5263 (49.1%) 150,804 (29.4%)
BMI, kg/m2 22.4 (3.7) 23.7 (3.4)
WC, cm 78.0 (10.5) 80.3 (9.8)
solid fuel for cooking 5542 (51.7%) 185,266 (36.1%)
solid fuel for heating 2999 (28.0%) 187,805 (36.6%)
exposure to SHS 6–7 d/w 5110 (47.7%) 212,970 (41.5%)
self-reported physician-diagnosed asthma 335 (3.1%) 2806 (0.5%)
cold (October–March next year) season at hospital admission 14,092 (58.1%)  
Open in a new tab
a

Continuous and categorical characteristics are presented as mean (SD) and count (percentage), respectively. Abbreviations: AECOPD: acute exacerbation of chronic obstructive pulmonary disease; BMI: body mass index; WC: waist circumference; SHS: second-hand smoke; SD: standard deviation.

b

Ex-smokers who had stopped smoking for illness were categorized as current smokers.

c

All recurrent cases included.

The distributions of air pollutants showed prominent regional disparities and seasonal variations. Haikou recorded the lowest levels of particulate matter and NO2; the pollution from particulate matter was the most severe in Henan; and Zhejiang was high in O3 and NO2. The particulate pollution and NO2 levels peaked in the winter and hit bottom in the summer, whereas the opposite trend in fluctuation was found for O3 (Figure S2). The concentration of O3 was negatively correlated with PM2.5, PM10, and NO2. Generally, low to moderate correlations were observed among these air pollutants, except for the correlation between PM2.5 and PM10 concentrations (r = 0.948) (Table S1).

The difference between the case and control days reached the maximum at lag 1, 1, 1, 2, and 0 days for PM2.5, PM10, PM2.5–10, O3, and NO2, respectively. The mean (SD) concentrations of air pollutants on the case days with the maximum difference were 58.1 (33.2) μg/m3 for PM2.5, 92.3 (46.0) μg/m3 for PM10, 34.1 (18.5) μg/m3 for PM2.5–10, 81.9 (33.3) μg/m3 for O3, and 36.0 (15.5) μg/m3 for NO2. On the control days, the corresponding values were 57.3 (33.1), 90.6 (44.7), 33.3 (16.9), 81.4 (33.3), and 35.7 (15.3) μg/m3, respectively. The temperature on the case days approximated to that on the control days. Specific humidity on the case days was slightly lower than that on the control days (Table ).

2. Distribution of Air Pollutants and Meteorological Conditions on the Case and Control Days .

indicator case/control day lag 0 lag 1 lag 2 lag 3 lag 4 lag 5 lag 6 lag 7
PM2.5, μg/m3 case day 57.9 (32.5) 58.1 (33.2) 58.2 (33.7) 58.1 (33.2) 58.0 (33.0) 57.9 (33.1) 57.7 (32.9) 57.6 (33.0)
control day 57.4 (33.1) 57.3 (33.1) 57.4 (33.1) 57.5 (32.8) 57.6 (33.0) 57.7 (33.2) 57.6 (33.0) 57.6 (33.2)
PM10, μg/m3 case day 91.5 (44.3) 92.3 (46.0) 92.4 (46.8) 92.1 (45.5) 91.6 (44.8) 91.5 (45.1) 91.4 (44.7) 91.2 (44.8)
control day 90.7 (44.9) 90.6 (44.7) 90.8 (44.7) 90.9 (44.6) 91.1 (44.7) 91.1 (44.9) 91.0 (44.8) 91.1 (45.2)
PM2.5–10, μg/m3 case day 33.7 (17.0) 34.1 (18.5) 34.2 (19.2) 34.0 (18.1) 33.6 (17.1) 33.6 (17.3) 33.7 (17.3) 33.5 (17.2)
control day 33.3 (17.2) 33.3 (16.9) 33.4 (16.8) 33.4 (17.0) 33.5 (17.0) 33.5 (17.0) 33.4 (17.1) 33.5 (17.3)
O3, μg/m3 case day 81.5 (33.5) 81.9 (33.4) 81.9 (33.3) 81.7 (33.3) 81.9 (33.0) 81.6 (33.2) 81.4 (33.1) 81.2 (33.0)
control day 81.6 (33.2) 81.6 (33.2) 81.4 (33.3) 81.4 (33.3) 81.5 (33.4) 81.5 (33.4) 81.5 (33.2) 81.3 (33.2)
NO2, μg/m3 case day 36.0 (15.5) 36.1 (15.5) 36.1 (15.6) 36.1 (15.6) 36.0 (15.6) 36.0 (15.6) 35.9 (15.5) 35.9 (15.4)
control day 35.7 (15.3) 35.8 (15.3) 35.8 (15.4) 35.8 (15.3) 35.9 (15.4) 36.0 (15.4) 36.0 (15.5) 35.9 (15.4)
temperature, °C case day 13.9 (9.0) 13.9 (9.1) 13.9 (9.1) 13.8 (9.1) 13.8 (9.1) 13.8 (9.1) 13.8 (9.1) 13.8 (9.1)
control day 13.9 (9.1) 14.0 (9.1) 13.9 (9.1) 14.0 (9.1) 14.0 (9.1) 14.0 (9.1) 14.0 (9.1) 13.9 (9.1)
specific humidity, g/10 kg case day 92.8 (54.0) 92.6 (54.1) 92.3 (54.2) 92.2 (54.3) 92.2 (54.3) 92.2 (54.2) 92.1 (54.2) 92.2 (54.2)
control day 93.1 (54.2) 93.1 (54.2) 93.1 (54.2) 93.1 (54.2) 93.1 (54.1) 93.1 (54.1) 93.1 (54.1) 93.1 (54.2)
Open in a new tab
a

Distributions are presented as mean (SD). Abbreviation: PM2.5: particulate matter with an aerodynamic diameter <2.5 μm; PM10: particulate matter with an aerodynamic diameter <10 μm; PM2.5–10: particulate matter with an aerodynamic diameter of 2.5–10 μm; O3: ozone; NO2: nitrogen dioxide; SD: standard deviation; P25: 25th percentile; P50: 50th percentile; P75: 75th percentile.

Air Pollution and Hospital Admissions of AECOPD

The second PC was strongly correlated to O3, whereas the first PC primarily represented other pollutants (PM2.5, PM2.5–10, NO2) than O3. The air pollution score contributed 54.1–54.5% of the total variance over lag 0–7 days (Table S2; other data not shown). As illustrated in Figure and Table S3, the association curves of PM2.5, PM10, PM2.5–10, and NO2 and air pollution score with AECOPD hospitalizations displayed an inverted J-shape over a lag period of 0–7 days. The risk of hospital admissions for AECOPD increased at lag 0 days. It persisted for the following 4–5 days, with the largest temperature- and humidity-adjusted odds ratio (95% confidence interval, CI) of 1.016 (1.009, 1.023), 1.016 (1.011, 1.021), 1.043 (1.031, 1.054), 1.054 (1.035, 1.074), and 1.101 (1.073, 1.129) at lag 1, 1, 1, 0, and 1 days, for a 10 μg/m3 increase in PM2.5, PM10, PM2.5–10, and NO2, and 1 SD increase in air pollution score, respectively. For O3, a 10 μg/m3 increase initially reduced the risk of hospital admissions at lag 0 days, followed by a rise in risk, with the highest odds ratio (95% CI) of 1.009 (1.002, 1.016) at lag 4 days. Overall, when the air pollutants were divided into quarters, the risk progressively increased across Q2, Q3, and Q4 compared to Q1. All of the estimates for Q4 vs Q1 were statistically significant.

1.

1

Open in a new tab

Odds ratios of the adjusted lag-effect between air pollutants and hospital admissions of AECOPD. Abbreviations: AECOPD: acute exacerbation of chronic obstructive pulmonary disease; PM2.5: particulate matter with an aerodynamic diameter <2.5 μm; PM10: particulate matter with an aerodynamic diameter <10 μm; PM2.5–10: particulate matter with an aerodynamic diameter of 2.5–10 μm; O3: ozone; NO2: nitrogen dioxide; CI: confidence interval. Models were adjusted for the natural cubic splines of the 3-day average temperature and humidity (lag 0–2 days) with 6 and 3 degrees of freedom. The centers of squares represent point estimates for a 10 μg/m3 increase in air pollutants or a 1 SD increase in air pollution score, and the lower and upper ends of whiskers represent the lower and upper limits of 95% CI, respectively. The sizes of squares are proportional to precision (1/squares of the standard errors of estimates).

Subgroup Analyses

The stratified analysis by body shape showed that there was an increased risk of AECOPD hospitalizations associated with PM2.5 and PM10 among the underweight or obese participants (BMI < 18.5 kg/m2, BMI ≥ 28.0 kg/m2, or WC ≥ 90 cm for male/≥85 cm for female) compared to those with favorable body shape (P int = 0.029 for PM2.5 and 0.041 for PM10). Further stratification analysis revealed that O3 was positively associated with AECOPD hospitalizations only among solid heating fuel users. Also, NO2 and air pollution scores were associated with increased risks of AECOPD hospitalizations only during the cold season and under low-humidity conditions. Concerning other characteristics, including sociodemographic characteristics, smoking status, cooking fuel type, SHS exposure, and asthma diagnosis, the associations of air pollution with hospital admissions for AECOPD were consistent across different subpopulations (all P int > 0.05) (Table S4).

Sensitivity Analyses

The associations of air pollution with AECOPD hospitalizations were generally robust to sensitivity analyses (Figure and Table S5). The SCCS design examined the relationship between exposure to air pollutants at lag 0 days and hospital admissions of AECOPD. The findings were consistent with those in the case-crossover design, and estimates were comparable across regions, although estimates were not statistically significant in some regions (heterogeneity I 2 = 0–48%). Considering the moderate correlation between particulate matter and NO2, reductions in the estimates after mutual adjustments were anticipated and deemed acceptable. However, after controlling for other pollutants considered, the effect of PM2.5 at lag 1 day and O3 at lag 4 day becomes nonsignificant, and the protective effect of O3 at lag 0 day becomes more pronounced (not shown; OR = 0.991; 95% CI = 0.983, 0.998). After adjusting for longer durations of temperature and humidity, the risks brought by air pollutants diminished and even to the null for O3. Compared with the first PC, the weighted air pollution score took more O3 into account and contributed 80.8% of the total variance, and the estimated effect was slightly lower at lag 1 day.

2.

2

Open in a new tab

Relative incidence of air pollutants for the first hospital admissions of AECOPD. Abbreviations: AECOPD: acute exacerbation of chronic obstructive pulmonary disease; PM2.5: particulate matter with an aerodynamic diameter <2.5 μm; PM10: particulate matter with an aerodynamic diameter <10 μm; PM2.5–10: particulate matter with an aerodynamic diameter of 2.5–10 μm; O3: ozone; NO2: nitrogen dioxide; CI: confidence interval; RR: risk ratio. The I 2 statistic was used to evaluate the heterogeneity of the effect size between cities, with I 2 > 50% as the existence of heterogeneity. Data on NO2 was available since January 1, 2008, and therefore cases occurring prior to January 1, 2008 were excluded.

Discussion

Based on a large cohort with linkage to inpatient hospitalization records in China, this individual-level case-crossover study demonstrated positive associations of short-term exposure to PM, especially PM2.5–10, NO2, and air pollution score, with hospital admissions for AECOPD, which could last up to 4–5 days. The findings for O3 were conflicting and susceptible to the influence of other factors. Participants with unfavorable body shapes might be more susceptible to PM2.5 and PM10. Additionally, the cold season and low humidity could potentially exacerbate the influence of air pollution.

The short-term detrimental impacts of PM in triggering acute respiratory events have been well documented in prior studies. , One meta-analysis study, which included 14 studies with effect estimates ranging from 0.98 to 1.12, found the risk of AECOPD events, including hospitalization, emergency room visit, and outpatient visit, significantly increased with increments in PM2.5 at lag 1 day. Similar results were observed for PM10. Evidence has shown that PM can induce oxidative stress injury in lung cells. Both animal models and human cohort studies have linked PM to impaired lung function, accompanied by characteristic inflammatory alterations. , The deleterious components of PM, including metals, polyaromatic hydrocarbons, and mineral dust, may contribute to such regulation. , Also, our study further strengthened the AECOPD risk posed by PM2.5–10, which was underexplored compared to PM2.5. Although PM2.5–10 generally cannot reach the alveoli, it can enter the bronchi and bronchioles and directly harm the lungs. Future studies can distinguish the effects of different particles on chronic bronchitis-type and emphysema-type COPD.

The effect of NO2 and O3 on AECOPD has been less thoroughly explored. A few studies indicated that high concentrations of ambient NO2 and O3 could cause inflammation of the bronchial mucosa, downregulate the bronchoconstriction threshold, and further increase the risk of acute injury to the lung tissue. , Nonetheless, the findings remained mixed, − and most of the case-crossover studies exhibited no associations of O3 with AECOPD. , In our study, O3 did not have harmful effects until a lag of 2 days. A mortality study based on the Medicare cohort of all beneficiaries aged ≥65 years in the contiguous U.S. from 2000 to 2015 also found that the effects of O3 and NO2 were reversed in beneficiaries of different characteristics. Evidence has shown that NO x is not only a major precursor of O3 in the photochemical reactions but a quencher of O3 through NO x titration (NO + O3 → NO2 + O2). Therefore, a negative correlation between NO2 and O3 was common, especially in heavily polluted areas, as observed in our study. However, controlling for other pollutants strengthened the protective effect of O3 at lag 0 days and diminished the harmful effect at lag 4 days. It is hard to disentangle our findings, and further investigation is required.

Few studies have explored the impact of obesity on the association of air pollution with AECOPD. Our study found that underweight or obese individuals were more susceptible to PM2.5 and PM10. Low weight was accompanied by a paucity of muscular mass and diminished muscle strength, which were established factors in managing respiratory diseases. By contrast, excessive fat accumulation, particularly in the abdomen, indicated the oversecretion of bioactive substances, such as inflammatory factors, resulting in structural changes and dysfunction in the body, as well as limited the capacity and endurance of gas exchange. A recent study evaluated air pollution-related mortality based on ∼19,000 deceased veterans with prior COPD residing in 25 US metropolitan regions during 2016–2019 and also found that obesity notably increased vulnerability.

The 2025 Global Initiative for Chronic Obstructive Lung Disease (GOLD) report underscored the potential interaction between climate and air quality. However, the existing evidence is limited and elusive. A study in urban areas of Chengdu, China, from 2015 to 2016, demonstrated stronger associations with COPD hospital admissions at lower temperatures for ambient PM2.5 and PM10, but not for NO2 and O3. A retrospective ecological study, utilizing data on daily emergency hospital admissions to 15 major hospitals in Hong Kong for COPD during 2000–2004, revealed NO2 and O3 had a greater effect on COPD admissions during the cold season, while no significant difference was identified regarding PM2.5 and PM10. The conflicting findings concerning interactions between meteorological conditions and pollutants on AECOPD may partially reflect bias in the personal actual exposure.

To our knowledge, this study constitutes the first epidemiological evidence demonstrating a significant dose–response relationship between a PCA-derived composite air pollution index and AECOPD risk, thereby quantifying synergistic pollutant interactions at environmentally relevant exposure levels. Furthermore, the study entails self-controlled analyses that eliminate confounding and selection bias through time-invariant characteristics. A previous CKB study showed that of the participants who attended both the 2004–2008 baseline survey and the 2013–2014 resurvey, the lifestyles remained largely unchanged, further supporting the robustness of our case-crossover design spanning only one month. The findings from the two types of self-controlled design adopted further corroborated each other. More than 10 years of follow-up, broad coverage of 10 regions, detailed personal characteristics, and >97% national medical insurance rates also enriched and consolidated the study.

Inevitably, several limitations must be acknowledged. First, participants were assigned exposure levels based on the coordinates of study clinics near their residences, possibly leading to nondifferential misclassification and biasing the association to the null. Exposure during commutes and work was not collected, which may lead to measurement error. However, due to the relatively small variation in pollutant concentrations within regions, this is unlikely to affect our results significantly. Second, indoor ambient exposure may vary between days. In particular, differences in behaviors may affect personal exposure, such as avoiding outdoor activities and using air cleaners on highly polluted days. Nonetheless, this would typically cause an underestimation of the associations. Third, only inpatient admissions for AECOPD events were included in the study. Consequently, mild AECOPD cases without the need for hospitalization admissions were not included, and their association cannot be investigated. Fourth, data on education and occupation were collected only at baseline; thus, we could not assess the impact of their potential changes over time on the observed associations. However, resurveys in a subpopulation have shown that these factors have changed little, and the correlations are moderate to high.

Conclusion

These findings reinforce the evidence of a link between air pollution and the AECOPD established in regional and local studies, which serves as a solid theoretical basis for implementing efficient interventions and policies for the mitigation of air pollution. This study also implicates the importance of body weight and other enviromental conditions for the management of AECOPD.

Supplementary Material

eh5c00439_si_001.pdf (425KB, pdf)

Acknowledgments

The most important acknowledgment is to the participants in the study and the members of the survey teams in each of the 10 regional centers, as well as to the project development and management teams based in Beijing, Oxford, and the 10 regional centers.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/envhealth.5c00439.

  • Supplementary methods; Pearson correlation coefficients among air pollutants and meteorological conditions; eigenvectors and eigenvalues from the principal component analysis eigen decomposition; odds ratios of the adjusted lag-effect between air pollutants and hospital admissions of AECOPD; associations between air pollutants and hospital admissions of AECOPD in the sensitivity analysis; inclusion and exclusion flowchart of participants and records; and daily trends of air pollutants across the 10 regions during 2004–2018 (PDF)

C.Y. conceptualized and designed the paper. L.L., Z.C., and J.C., as members of the CKB steering committee, designed and supervised the conduct of the whole study, obtained funding, and together with J.L., D.S., P.P., L.Y., Y.C., H.D., S.Y., and D.S. acquired the data. L.C. drafted the manuscript. L.C. and Y.Z. analyzed the data. C.Y. contributed to the interpretation of the results and critical revision of the manuscript for important intellectual content. All authors contributed to and approved the final manuscript. C.Y. is the study guarantor.

This work was supported by the National Major Science & Technology Projects (2023ZD0510101 and 2023ZD0510100) and the National Natural Science Foundation of China (82192901, 82192904, 82192900, and 82388102). The CKB baseline survey and the first resurvey were supported by the Kadoorie Charitable Foundation in Hong Kong. The long-term follow-up has been supported by Wellcome grants to Oxford University (212946/Z/18/Z, 202922/Z/16/Z, 104085/Z/14/Z, and 088158/Z/09/Z) and grants (2016YFC0900500) from the National Key R&D Program of China, the National Natural Science Foundation of China (81390540, 91846303, and 81941018), and the Chinese Ministry of Science and Technology (2011BAI09B01).

The authors declare no competing financial interest.

References

  1. World Health Organization (WHO) . Ambient (Outdoor) Air Pollution. https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health (accessed May 08, 2025).
  2. Niu Y., Niu H., Meng X., Zhu Y., Ren X., He R., Wu H., Yu T., Zhang Y., Kan H., Chen R., Yang T., Wang C.. Associations Between Air Pollution and the Onset of Acute Exacerbations of COPD: A Time-Stratified Case-Crossover Study in China. Chest. 2024;166:998–1009. doi: 10.1016/j.chest.2024.05.030. [DOI] [PubMed] [Google Scholar]
  3. Wang Z., Zhou Y., Zhang Y., Huang X., Duan X., Chen D., Ou Y., Tang L., Liu S., Hu W., Liao C., Zheng Y., Wang L., Xie M., Zheng J., Liu S., Luo M., Wu F., Deng Z., Tian H., Peng J., Yang H., Xiao S., Wang X., Zhong N., Ran P.. Association of change in air quality with hospital admission for acute exacerbation of chronic obstructive pulmonary disease in Guangdong, China: A province-wide ecological study. Ecotoxicol. Environ. Saf. 2021;208:111590. doi: 10.1016/j.ecoenv.2020.111590. [DOI] [PubMed] [Google Scholar]
  4. Liang L., Cai Y., Barratt B., Lyu B., Chan Q., Hansell A. L., Xie W., Zhang D., Kelly F. J., Tong Z.. Associations between daily air quality and hospitalisations for acute exacerbation of chronic obstructive pulmonary disease in Beijing, 2013–17: an ecological analysis. Lancet Planet. Health. 2019;3(6):e270–e279. doi: 10.1016/S2542-5196(19)30085-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Strosnider H. M., Chang H. H., Darrow L. A., Liu Y., Vaidyanathan A., Strickland M. J.. Age-Specific Associations of Ozone and Fine Particulate Matter with Respiratory Emergency Department Visits in the United States. Am. J. Respir. Crit. Care Med. 2019;199(7):882–890. doi: 10.1164/rccm.201806-1147OC. [DOI] [PubMed] [Google Scholar]
  6. Li J., Sun S., Tang R., Qiu H., Huang Q., Mason T. G., Tian L.. Major air pollutants and risk of COPD exacerbations: a systematic review and meta-analysis. Int. J. Chron. Obstruct. Pulmon. Dis. 2016;11:3079–3091. doi: 10.2147/COPD.S122282. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Oakes M., Baxter L., Long T. C.. Evaluating the application of multipollutant exposure metrics in air pollution health studies. Environ. Int. 2014;69:90–99. doi: 10.1016/j.envint.2014.03.030. [DOI] [PubMed] [Google Scholar]
  8. Yu L., Liu W., Wang X., Ye Z., Tan Q., Qiu W., Nie X., Li M., Wang B., Chen W.. A review of practical statistical methods used in epidemiological studies to estimate the health effects of multi-pollutant mixture. Environ. Pollut. 2022;306:119356. doi: 10.1016/j.envpol.2022.119356. [DOI] [PubMed] [Google Scholar]
  9. Global Initiative for Chronic Obstructive Lung Disease (GOLD) . Global Strategy for Prevention, Diagnosis and Management of COPD: 2025 Report, 2024.
  10. Chen Z., Lee L., Chen J., Collins R., Wu F., Guo Y., Linksted P., Peto R.. Cohort profile: the Kadoorie Study of Chronic Disease in China (KSCDC) Int. J. Epidemiol. 2005;34(6):1243–1249. doi: 10.1093/ije/dyi174. [DOI] [PubMed] [Google Scholar]
  11. Walters R. G., Millwood I. Y., Lin K., Schmidt Valle D., McDonnell P., Hacker A., Avery D., Edris A., Fry H., Cai N., Kretzschmar W. W., Ansari M. A., Lyons P. A., Collins R., Donnelly P., Hill M., Peto R., Shen H., Jin X., Nie C., Xu X., Guo Y., Yu C., Lv J., Clarke R. J., Li L., Chen Z.. Genotyping and population characteristics of the China Kadoorie Biobank. Cell Genomics. 2023;3(8):100361. doi: 10.1016/j.xgen.2023.100361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Wei, J. High-resolution and High-quality Ambient Air Pollutants Dataset for China, 2024.
  13. Gelaro, R. ; McCarty, W. ; Suárez, M. J. ; Todling, R. ; Molod, A. ; Takacs, L. ; Randles, C. A. ; Darmenov, A. ; Bosilovich, M. G. ; Reichle, R. ; Wargan, K. ; Coy, L. ; Cullather, R. ; Draper, C. ; Akella, S. ; Buchard, V. ; Conaty, A. ; da Silva, A. M. ; Gu, W. ; Kim, G.-K. ; Koster, R. ; Lucchesi, R. ; Merkova, D. ; Nielsen, J. E. ; Partyka, G. ; Pawson, S. ; Putman, W. ; Rienecker, M. ; Schubert, S. D. ; Sienkiewicz, M. ; Zhao, B. . The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) NASA Goddard Earth Sciences Data and Information Services Center (GES DISC): Greenbelt, MD, USA; 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Maclure M.. The Case-Crossover Design: A Method for Studying Transient Effects on the Risk of Acute Events. Am. J. Epidemiol. 1991;133(2):144–153. doi: 10.1093/oxfordjournals.aje.a115853. [DOI] [PubMed] [Google Scholar]
  15. Jaakkola J. J.. Case-crossover design in air pollution epidemiology. Eur. Respir J. Suppl. 2003;21:81s–85s. doi: 10.1183/09031936.03.00402703. [DOI] [PubMed] [Google Scholar]
  16. Tobias A., Kim Y., Madaniyazi L.. Time-stratified case-crossover studies for aggregated data in environmental epidemiology: a tutorial. Int. J. Epidemiol. 2024;53(2):dyae020. doi: 10.1093/ije/dyae020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lei J., Chen R., Liu C., Zhu Y., Xue X., Jiang Y., Shi S., Gao Y., Kan H., Xuan J.. Fine and coarse particulate air pollution and hospital admissions for a wide range of respiratory diseases: a nationwide case-crossover study. Int. J. Epidemiol. 2023;52(3):715–726. doi: 10.1093/ije/dyad056. [DOI] [PubMed] [Google Scholar]
  18. Whitaker H. J., Paddy Farrington C., Spiessens B., Musonda P.. Tutorial in biostatistics: the self-controlled case series method. Stat. Med. 2006;25(10):1768–1797. doi: 10.1002/sim.2302. [DOI] [PubMed] [Google Scholar]
  19. Petersen I., Douglas I., Whitaker H.. Self controlled case series methods: an alternative to standard epidemiological study designs. BMJ. 2016;354:i4515. doi: 10.1136/bmj.i4515. [DOI] [PubMed] [Google Scholar]
  20. Schwartz J. D., Yitshak-Sade Ma., Zanobetti A., Di Q., Requia W. J., Dominici F., Mittleman M. A.. A self-controlled approach to survival analysis, with application to air pollution and mortality. Environ. Int. 2021;157:106861. doi: 10.1016/j.envint.2021.106861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Bishoi B., Prakash A., Jain V. K.. A Comparative Study of Air Quality Index Based on Factor Analysis and US-EPA Methods for an Urban Environment. Aerosol Air Qual. Res. 2009;9(1):1–17. doi: 10.4209/aaqr.2008.02.0007. [DOI] [Google Scholar]
  22. Li N., Ma J., Ji K., Wang L.. Association of PM2.5 and PM10 with Acute Exacerbation of Chronic Obstructive Pulmonary Disease at lag0 to lag7: A Systematic Review and Meta-Analysis. COPD. 2022;19(1):243–254. doi: 10.1080/15412555.2022.2070062. [DOI] [PubMed] [Google Scholar]
  23. Hamad S. H., Schauer J. J., Antkiewicz D. S., Shafer M. M., Kadhim A. K.. ROS production and gene expression in alveolar macrophages exposed to PM(2.5) from Baghdad, Iraq: Seasonal trends and impact of chemical composition. Sci. Total Environ. 2016;543(Pt A):739–745. doi: 10.1016/j.scitotenv.2015.11.065. [DOI] [PubMed] [Google Scholar]
  24. Li Q., Sun J., Chen X., Li S., Wang Y., Xu C., Zhao J., Zhu Z., Tian L.. Integrative characterization of fine particulate matter-induced chronic obstructive pulmonary disease in mice. Sci. Total Environ. 2020;706:135687. doi: 10.1016/j.scitotenv.2019.135687. [DOI] [PubMed] [Google Scholar]
  25. Li J., Yao Y., Shang B., Xie Y., Yin H., Song Y., Ma J.. Ambient air pollution, low-grade inflammation, and lung function: Evidences from the UK Biobank. Ecotoxicol. Environ. Saf. 2024;284:116998. doi: 10.1016/j.ecoenv.2024.116998. [DOI] [PubMed] [Google Scholar]
  26. Zhou L., Li P., Zhang M., Han B., Chu C., Su X., Li B., Kang H., Ning J., Zhang B., Ma S., Su D., Pang Y., Niu Y., Zhang R.. Carbon black nanoparticles induce pulmonary fibrosis through NLRP3 inflammasome pathway modulated by miR-96 targeted FOXO3a. Chemosphere. 2020;241:125075. doi: 10.1016/j.chemosphere.2019.125075. [DOI] [PubMed] [Google Scholar]
  27. Monteiro A., Carvalho V., Oliveira T., Sousa C.. Excess mortality and morbidity during the July 2006 heat wave in Porto, Portugal. Int. J. Biometeorol. 2013;57:155–167. doi: 10.1007/s00484-012-0543-9. [DOI] [PubMed] [Google Scholar]
  28. Lin M. T., Kor C. T., Chang C. C., Chai W. H., Soon M. S., Ciou Y. S., Bin Lian I., Chang C. C.. Association of meteorological factors and air NO(2) and O(3) concentrations with acute exacerbation of elderly chronic obstructive pulmonary disease. Sci. Rep. 2018;8(1):10192. doi: 10.1038/s41598-018-28532-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ross B. A., Doiron D., Benedetti A., Aaron S. D., Chapman K., Hernandez P., Maltais F., Marciniuk D., O’Donnell D. E., Sin D. D., Walker B. L., Tan W., Bourbeau J.. Short-term air pollution exposure and exacerbation events in mild to moderate COPD: a case-crossover study within the CanCOLD cohort. Thorax. 2023;78(10):974–982. doi: 10.1136/thorax-2022-219619. [DOI] [PubMed] [Google Scholar]
  30. Wang H., Huang C., Tao W., Gao Y., Wang S., Jing S., Wang W., Yan R., Wang Q., An J., Tian J., Hu Q., Lou S., Pöschl U., Cheng Y., Su H.. Seasonality and reduced nitric oxide titration dominated ozone increase during COVID-19 lockdown in eastern China. npj Clim. Atmos. Sci. 2022;5(1):24. doi: 10.1038/s41612-022-00249-3. [DOI] [Google Scholar]
  31. Lan Y., Ke Y., Sun D., Pei P., Yang L., Chen Y., Du H., Lv S., Barnard M., Chen J., Chen Z., Lv J., Li L., Yu C.. Associations of Muscle-Related Metrics With Respiratory Disease in Chinese Adults: A Prospective Cohort Study. J. Cachexia, Sarcopenia Muscle. 2025;16(1):e13650. doi: 10.1002/jcsm.13650. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Goyal A., Nimmakayala K. R., Zonszein J.. Is There a Paradox in Obesity? Cardiol. Rev. 2014;22(4):163. doi: 10.1097/CRD.0000000000000004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Choe E. K., Kang H. Y., Lee Y., Choi S. H., Kim H. J., Kim J. S.. The longitudinal association between changes in lung function and changes in abdominal visceral obesity in Korean non-smokers. PLoS One. 2018;13(2):e0193516. doi: 10.1371/journal.pone.0193516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Aron J., Baldomero A. K., Rau A., Fiecas M. B., Wendt C. H., Berman J. D.. Individual Risk Factors of PM(2.5) Associated With Wintertime Mortality in Urban Patients With COPD. Chest. 2024;165(4):825–835. doi: 10.1016/j.chest.2023.10.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Qiu H., Tan K., Long F., Wang L., Yu H., Deng R., Long H., Zhang Y., Pan J.. The Burden of COPD Morbidity Attributable to the Interaction between Ambient Air Pollution and Temperature in Chengdu, China. Int. J. Environ. Res. Public Health. 2018;15(3):492. doi: 10.3390/ijerph15030492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Ko F. W. S., Tam W., Wong T. W., Chan D. P., Tung A. H., Lai C. K., Hui D. S.. Temporal relationship between air pollutants and hospital admissions for chronic obstructive pulmonary disease in Hong Kong. Thorax. 2007;62(9):780–785. doi: 10.1136/thx.2006.076166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Hu Y., Sun Z., Yu C., Guo Y., Pei P., Yang L., Chen Y., Du H., Sun D., Pang Y., Tian X., Gilbert S., Avery D., Chen J., Chen Z., Li L., Lv J.. Association between pneumonia hospitalisation and long-term risk of cardiovascular disease in Chinese adults: A prospective cohort study. EClinicalMedicine. 2023;55:101761. doi: 10.1016/j.eclinm.2022.101761. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

eh5c00439_si_001.pdf (425KB, pdf)

Articles from Environment & Health are provided here courtesy of American Chemical Society

RESOURCES