Air pollution, inhaled therapy and COPD exacerbations in Yunlin, Taiwan: a 2016–2024 single-centre retrospective cohort with environmental and carbon-footprint analyses Authors: Chung-Yu Chen, Ling-Yu Huang, Bor-Wen Cheng Published: 2025 ABSTRACT AbstractBackgroundAir pollution exacerbates chronic obstructive pulmonary disease (COPD), increasing hospitalisation and exacerbation rates. While inhaled therapy is a cornerstone of COPD management, pressurised metered-dose inhalers (pMDIs) significantly contribute to greenhouse gas (GHG) emissions. This study evaluates the interplay between air pollution exposure, inhaled therapy selection and COPD exacerbations to identify strategies that optimise clinical outcomes while reducing environmental impact.MethodsA retrospective observational cohort study was conducted using COPD patient records (ICD-10: J44.0, J44.1, J44.8, J44.9) from the National Taiwan University Hospital Yunlin Branch (2016–2024). Patient visits, including outpatient, emergency and inpatient admissions, were analysed alongside air pollution data (PM2.5, PM10, NO2, O3, CO) from the Environmental Protection Administration. Regression models assessed the impact of inhaled therapies, short-acting β2-agonists (SABAs), long-acting β2-agonists (LABAs), long-acting muscarinic antagonists (LAMAs) and triple therapy (ICS/LABA/LAMA), on exacerbation rates and inhaler-associated carbon emissions.ResultsHigher PM2.5 levels correlated with a 0.97% increase in COPD exacerbation rates (B=0.0291, p=0.0001). Triple therapy was significantly associated with reduced exacerbations (B=−0.0035, p=0.0003), whereas higher SABA use was associated with markers of poorer COPD control (B=7.145, p<0.001). The uptake of non-pMDIs, such as dry powder inhalers and soft mist inhalers, rose in 2020–2021 as hospitalisations declined. In 2022–2024, pMDI use and estimated emissions increased, while emergency room (ER) visits were unchanged and hospitalisations declined modestly. A 42% increase in non-pMDI prescriptions in 2020–2021 was associated with a decline in hospitalisations and emissions. However, a subsequent shift back to pMDIs in 2022–2024 coincided with an increase in exacerbations and increased carbon footprint.ConclusionIn this single-centre retrospective study, higher ambient PM2.5 was associated with higher COPD acute exacerbation (AE) rates, and greater use of triple therapy correlated with lower hospitalisations. Our inhaler carbon-footprint estimates quantify GHG differences between device types but were not linked to AE and should inform sustainability discussions rather than clinical effectiveness.Keywords: COPD Exacerbations, Inhaler devices WHAT IS ALREADY KNOWN ON THIS TOPIC Air pollution increases chronic obstructive pulmonary disease (COPD) exacerbations, and triple-inhaled therapy reduces this risk. However, pressurised metered-dose inhalers (pMDIs) contribute significantly to carbon emissions, and the combined impact of pollution and inhaler type on outcomes remains underexplored. Air pollution increases chronic obstructive pulmonary disease (COPD) exacerbations, and triple-inhaled therapy reduces this risk. However, pressurised metered-dose inhalers (pMDIs) contribute significantly to carbon emissions, and the combined impact of pollution and inhaler type on outcomes remains underexplored. WHAT THIS STUDY ADDS Higher PM2.5 levels were linked to more COPD exacerbations, while triple therapy reduced hospitalisations. A shift to low-carbon inhalers reduced emissions and exacerbations, reverting to pMDIs increased emissions without improving outcomes. Higher PM2.5 levels were linked to more COPD exacerbations, while triple therapy reduced hospitalisations. A shift to low-carbon inhalers reduced emissions and exacerbations, reverting to pMDIs increased emissions without improving outcomes. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Findings support integrating air quality and inhaler sustainability into COPD care, promoting non-pMDI options when appropriate to improve outcomes and reduce emissions. Findings support integrating air quality and inhaler sustainability into COPD care, promoting non-pMDI options when appropriate to improve outcomes and reduce emissions. Introduction Air pollution is a well-established risk factor for acute exacerbations of chronic obstructive pulmonary disease (AECOPD), contributing to increased hospitalisations, disease progression and mortality. Exposure to pollutants such as particulate matter (PM2.5, PM10), nitrogen dioxide (NO2), ozone (O3) and sulfur dioxide (SO2) has been shown to impair lung function, induce systemic inflammation and increase oxidative stress, all of which contribute to COPD exacerbations.1,3 Epidemiological studies indicate that a 10 µg/m³ increase in PM2.5 is associated with a 2.5% rise in emergency department visits and hospitalisations for AECOPD.1 Furthermore, NO2 levels above 16.65 ppb and O3 levels exceeding 35.65 ppb have been significantly correlated with a higher risk of severe exacerbations.2 Long-term exposure to polluted environments accelerates lung function decline and increases exacerbation frequency, emphasising the urgent need for pollution mitigation strategies in COPD management.4 Optimised inhaled therapy plays a critical role in reducing AECOPD risk and improving patient outcomes. Dual therapy, typically a combination of a long-acting β2-agonist (LABA) and a long-acting muscarinic antagonist (LAMA), has demonstrated significant benefits, including improved lung function, reduced dyspnoea and lower exacerbation rates compared with monotherapy.4 Triple therapy, incorporating an inhaled corticosteroid (ICS) alongside LABA and LAMA, is particularly effective in patients with high exacerbation risk, elevated eosinophil counts or persistent symptoms despite dual therapy.4 Clinical trials suggest that ICS/LABA/LAMA combinations reduce exacerbations by 20–30% compared with dual therapy, particularly in eosinophilic COPD phenotypes.4 Personalised inhaler therapy selection, integrating exacerbation history, biomarkers and environmental considerations, is essential for optimising COPD management while reducing the ecological impact of inhalers.4 The environmental impact of inhaled therapy is a growing concern, with pressurised metered dose inhalers (pMDIs) contributing significantly to greenhouse gas (GHG) emissions due to hydrofluoroalkane (HFA) propellants, each emitting CO2 equivalent to a 180-mile car journey.5 6 While dry powder inhalers (DPIs) and soft mist inhalers (SMIs) offer lower-carbon alternatives, they may not suit all patients due to inhalation technique and disease severity.7 8 Transitioning to DPIs or reformulating pMDIs with low-global warming potential (GWP) propellants like HFA-152a could cut emissions by up to 89%.8 9 Improving respiratory care, reducing misdiagnosis, rescue inhaler overuse and poor disease control can further lower emissions without compromising outcomes.10 However, forced switching without patient-specific considerations may reduce adherence and worsen disease control. A balanced approach integrating clinical effectiveness and sustainability is essential.6 As regulations evolve, life-cycle carbon assessments, inhaler recycling programmes and evidence-based prescribing will be key to achieving net-zero respiratory care while ensuring optimal patient outcomes.11 12 Despite the well-documented impact of air pollution on COPD, the interplay between environmental pollutants and inhaled medication effectiveness remains underexplored. While inhaled therapy is essential, the choice of inhaler influences both clinical outcomes and environmental sustainability. Little is known about how inhaler device choice modulates the clinical outcomes of COPD in polluted environments, especially in East Asian populations. Although propellants in pMDIs (eg, HFA-134a, HFA-227ea) are not classified as ambient air pollutants, they are potent GHGs that contribute to climate change. Climate-driven phenomena, such as increased temperatures, wildfires and altered atmospheric circulation, can indirectly raise PM2.5 and other pollutant levels, potentially worsening respiratory disease outcomes. This study had two aims (1) to quantify associations between ambient air pollution and COPD outcomes and (2) describe inhaler-related GHG emissions using published life-cycle emission factors. Any health effects of inhaler GHGs would be indirect and long-term via climate forcing and were not evaluated in our outcome models. Methods Study design and setting This retrospective study analyses COPD healthcare utilisation at the Thoracic Medicine Center, National Taiwan University Hospital Yunlin Branch, Taiwan. It includes patients with COPD exacerbations requiring emergency or inpatient care from 1 January 2016 to 31 December 2024. The study evaluates the impact of air pollution and inhaled medication types on COPD exacerbation frequency. Independent variables include air pollution levels (PM2.5, PM10, SO2, CO, NO2, O3) and inhaled medications (LABA, LAMA, ICS, dual and combination therapies). The dependent variable is COPD exacerbation frequency, assessed through outpatient visits, emergency department visits and hospitalisations. Participants COPD cases were identified using International Classification of Diseases (ICD)-10 codes (J44.0, J44.1, J44.8, J44.9) and categorised by care setting. These diagnostic codes were confirmed via electronic medical records, ensuring accuracy and consistency. Exclusion criteria included enrolment in asthma-specific programmes, long-term bedridden status, nursing home residency, severe neuromuscular or musculoskeletal impairments, chronic mechanical ventilation dependence, severe cognitive impairment, unconsciousness or psychiatric illness. Data sources Air pollution monitoring and Air Quality Index construction Air quality data were sourced from Yunlin Environmental Protection Administration monitoring stations, excluding industrial sites. Key pollutants (PM2.5, PM10, SO2, CO, NO2, O3) were extracted for analysis. Air pollution data were systematically recorded from 1 January 2016 to 31 December 2024. Daily pollutant sub-indices from each station were used to calculate county-level daily means, which were subsequently aggregated into monthly averages for analysis. Patient encounters were not geocoded or linked to individual residential or occupational addresses. Ambient air quality was summarised using Taiwan’s official Air Quality Index (AQI), which is calculated from sub-indices for SO2, CO, NO2, O3, PM10 and PM2.5 based on 1-hour, 8-hour or 24-hour averaged concentrations, with each pollutant’s sub-index derived by linear interpolation within predefined concentration breakpoints.13 The overall AQI is defined as the maximum of these pollutant-specific sub-indices (online supplemental table 1). Medication capture and device-type classification Medication use was ascertained from the hospital pharmacy information system. For each calendar month, we extracted the number of dispensed inhaler prescriptions by pharmacological class (short-acting β2-agonists (SABA), LABA, LAMA, ICS/LABA, LABA/LAMA, ICS/LABA/LAMA) and by device type (pMDI vs non-pMDI, including DPIs and SMIs). Each dispensed inhaler device (canister or DPI/SMI unit) was counted once, irrespective of the number of actuations per device. The carbon footprint of inhalers was assessed using life-cycle assessment across four phases: manufacturing, distribution, usage and disposal.14 Manufacturing emissions stem from raw material extraction, production and assembly, while distribution emissions vary by inhaler type and weight. The usage phase is the largest contributor, especially for pMDIs, which contain high-GWP HFA propellants. The NHS estimates pMDI emissions based on actuations, with some devices emitting up to 48 kgCO₂e, whereas DPIs and SMIs produce 20–30 times fewer emissions.14 15 The disposal phase considers landfill, incineration or recycling, with NHS promoting inhaler recycling to reduce environmental impact.16 Carbon accounting was conducted independently of outcome models. CO₂e estimates were not entered as exposures and were not linked to AE, emergency room (ER) visits or hospitalisations, which likely contribute substantially to total COPD-related healthcare emissions. As a result, our analysis quantifies differences between inhaler device types but does not provide a full net-carbon balance for prevented exacerbations or hospitalisations. Statistical analysis We used multivariable regression models to estimate the association between air pollution indicators and COPD outcomes and evaluate inhaler class and device correlates of utilisation-based outcomes. Time-series and panel regressions used monthly observations with calendar month and year fixed effects to account for secular trends and seasonality. Primary pollutant effects are reported per 10 µg/m³ (PM2.5, PM10) or per 10 ppb (NO2, O3) increase to facilitate interpretation alongside prior literature. Descriptive statistics summarised continuous variables as means±SD and categorical variables as percentages. A multivariate regression model assessed relationships between air pollution indicators (PM2.5, PM10, NO2, O3, CO), inhaler types and COPD exacerbation rates. Pearson correlation analysis measured association strength, while time-series analysis examined trends in air pollution, inhaler utilisation and exacerbation rates. Independent t-tests and ANOVA compared exacerbation rates and inhaler-related carbon emissions across periods, with χ2 tests analysing categorical differences. Model performance was evaluated using R² and adjusted R², with residual diagnostics confirming regression validity. Statistical significance was set at p<0.05. Trend analysis explored shifts in inhaler prescription patterns and their correlation with hospitalisation rates and carbon footprint. Exploratory predictor ranking identified key environmental and pharmacological factors influencing COPD exacerbations to optimise clinical recommendations. Statistical analyses were conducted using SPSS 25.0 to evaluate associations between air pollution levels, inhaler prescriptions and COPD exacerbations. Results Impact of air pollution on COPD acute exacerbations In single-pollutant regressions, monthly COPD exacerbation rate (AE, %) rose with higher pollutant levels, PM2.5 (R² = 0.196; p<0.001; AE = 3.007 + 0.0291×PM2.5), PM10 (R² = 0.136; p=0.001), NO2 (R² = 0.074; p=0.021) and SO2 (R² = 0.072; p=0.023); O3 and CO were not significant. AQI was also associated with AE (R² = 0.097; p=0.008). In the multipollutant model, only PM2.5 remained independently associated (p=0.005), underscoring its dominant contribution (figure 1A–F). Consistently, pollutant levels were associated with AE-related ER visits (AQI, PM2.5, PM10, NO2, SO2; all p<0.05) and hospitalisations (AQI, PM2.5, PM10; all p<0.05). Annual trends in air quality indicators in Yunlin County, Taiwan, from 2016 to 2024 are presented in online supplemental figure 1. Association of air pollutants with chronic obstructive pulmonary disease (COPD) exacerbation rates. This figure illustrates the association between air pollution exposure and COPD acute exacerbation (AE) rates (%), with scatter plots for individual pollutants and a forest plot summarising regression coefficients. Each orange circle represents an observed data point, while the blue regression line depicts the best-fit linear model with a 95% CI. (A) PM2.5 (y=3.007 + 0.0291 × PM2.5, r=0.44, p<0.001); (B) PM10 (y=3.146 + 0.0411 × PM10, r=0.37, p=0.001); (C) NO2 (y=3.541 + 0.0591 × NO2, r=0.27, p=0.021); (D) SO2 (y=3.092 + 0.284 × SO2, r=0.27, p=0.023); (E) Air Quality Index (AQI) (y=3.263 + 0.0216 × AQI, r=0.31, p=0.008) all show moderate positive correlations with COPD exacerbation rates, indicating that higher pollutant levels are significantly associated with increased exacerbations. (F) The forest plot presents regression coefficients and 95% CIs for pollutants, with PM2.5 being the only variable showing a significant positive association (p<0.05), reinforcing its role in exacerbation risk. CO, carbon monoxide; NO2, nitrogen dioxide; O3, ozone; PM2.5, PM10, particulate matter; SO2, sulfur dioxide. Impact of pharmacological interventions on COPD exacerbations From 2016 to 2024, monthly SABA prescriptions had a relatively stable mean of 2840.56 dispensed devices (SD: 192.58), showing a slight negative correlation with time (r=−0.34). ICS/LABA showed moderate growth (mean: 4594.11; SD: 433.84) but with a weaker positive correlation (r=0.18). LAMA usage exhibited the most significant decline (mean: 2857; SD: 821.21; r=−0.83). LABA/LAMA displayed a strong upward trend (mean: 3138.67; SD: 1084.15; r=0.90). The most pronounced increase was seen in ICS/LABA/LAMA (mean: 728.67; SD: 853.49; r=0.94) (figure 2A). (A) Annual trends in medication usage. The statistical analysis of medication trends from 2016 to 2024 highlights a shift towards combination therapies. SABA usage remained stable (mean: 2840.56; r=−0.34), while ICS/LABA showed moderate growth (mean: 4594.11; r=0.18). LAMA declined significantly (mean: 2857; r=−0.83), whereas LABA/LAMA (mean: 3138.67; r=0.90) and ICS/LABA/LAMA (mean: 728.67; r=0.94) increased sharply, indicating a strong preference for dual and triple therapy. (B) The regression model of triple therapy associated with COPD hospitalisation rates. Triple therapy (ICS/LABA/LAMA) is significantly associated with a reduction in COPD hospitalisation rates (R2 = 0.1714, F=14.4847, p=0.0003). The regression model (y=1.5981 – 0.003568 ×TRIPLE) indicates that for each unit increase in triple therapy use, the hospitalisation rate decreases by 0.003568, reinforcing its protective effect in COPD management. AE, acute exacerbation; COPD, chronic obstructive pulmonary disease; ICS, inhaled corticosteroid; LABA, long-acting β2-agonist; LAMA, long-acting muscarinic antagonist; SABA, short-acting β2-agonist. Pharmacological interventions had opposing effects on COPD exacerbation rates. SABA use (B=7.145, p<0.001) was associated with a significant increase in exacerbations. SABA use was linked to higher ER visits (B=6.772, p=0.009) and increased hospitalisations (B=21.257, p<0.001). In contrast, triple therapy (ICS/LABA/LAMA) is significantly associated with a reduction in COPD hospitalisation rates. The regression model suggests that each unit increase in triple therapy use leads to a 0.003568 reduction in hospitalisation rate (p=0.0003) (figure 2B). Exploratory predictor ranking and correlation analysis of COPD AE Exploratory predictor ranking identified PM2.5, PM10 and SABA usage as the strongest predictors of COPD exacerbations. In contrast, triple therapy (ICS/LABA/LAMA) showed a protective effect (figure 3A). The analysis examined COPD AE in relation to air pollution indicators (PM2.5, PM10, AQI, NO2, SO2, CO) and medication usage (SABA, LAMA, ICS/LABA, TRIPLE, DUAL). PM2.5 and PM10 exhibited moderate positive correlations with COPD AE. NO2 and AQI showed mild to moderate correlations. SABA usage correlated positively with worsening COPD symptoms, while triple therapy demonstrated a protective effect, potentially reducing AECOPD risk (figure 3B). (A) Feature importance in predicting COPD AE. The feature importance analysis identifies key variables contributing to COPD AE prediction. Among them, PM2.5 and PM10 emerge as the most significant factors, emphasising the detrimental impact of poor air quality on COPD exacerbations. SABA usage is also ranked as an important predictor. This likely reflects reverse causality and confounding by indication, in that patients experiencing more frequent or severe exacerbations use more SABA as rescue therapy. Thus, SABA use should be interpreted as a marker of poorer control rather than a causal driver of COPD AE. Reducing exacerbations would be expected to reduce SABA use and, where SABAs are delivered by pMDIs, to lower the associated inhaler-related CO2 emissions. (B) Associations between COPD AE, air pollution and medication use. This analysis examines the relationship between COPD AE, air pollution indicators (PM2.5, PM10, AQI, NO2, SO2, CO) and medication usage (SABA, LAMA, ICSLABA, TRIPLE, DUAL). PM2.5 and PM10 show moderate positive correlations with COPD AE, suggesting that higher air pollution levels contribute to increased exacerbations. Other pollutants, including NO2 and AQI, display mild to moderate correlations, further reinforcing the environmental influence on COPD AE. Similarly, SABA usage demonstrates a moderate positive correlation, indicating that frequent use of SABA is linked to worsening COPD symptoms. In contrast, triple therapy (ICS/LABA/LAMA) exhibits a protective effect, highlighting its potential in reducing COPD AE risk. AE, acute exacerbation; AQI, Air Quality Index; CO, carbon monoxide; COPD, chronic obstructive pulmonary disease; ICS, inhaled corticosteroid; LABA, long-acting β2-agonist; LAMA, long-acting muscarinic antagonist; NO2, nitrogen dioxide; O3, ozone; PM2.5, PM10, particulate matter; pMDI, pressurised metered-dose inhaler; SABA, short-acting β2-agonist; SO2, sulfur dioxide. Trends in COPD care, inhaler use and carbon emissions Outpatient visits increased from 12 254 (2016) to 19 553 (2024), while AE rates declined from 8.39% to 4.04%, with the sharpest drop in 2020 (3.65%) (figure 4A). Prescription patterns shifted towards combination therapies and sustainable inhalers. Period A (2016–2018): ICS/LABA and LAMA dominated, with a non-MDI to MDI ratio of 1.45. Period B (2019–2021): Trelegy Ellipta (non-MDI) introduction in 2020 increased LABA/LAMA prescriptions, raising the ratio to 2.06 (+42%). Period C (2022–2024): the entry of Trimbow (MDI) and Breztri (MDI) shifted usage back towards MDIs, lowering the ratio to 1.38. MDIs, driven by hydrofluorocarbon (HFC) propellants, contributed significantly to emissions, rising from 92.4 M kg CO2e (2016) to 144.5 M kg CO2e (2024) (+56%), with MDIs responsible for ~80% of emissions. Despite increased MDI use, ER visits remained unchanged (p=0.12), while hospitalisation rates declined slightly (p=0.03), suggesting no major improvement in COPD control (figure 4B). Trends in outpatient visits, AE rates and the environmental impact of MDI prescriptions in COPD management. From 2016 to 2024, outpatient visits increased from 12 254 to 19 553, while AE rates declined from 8.39% to 4.04%, with the sharpest drop in 2020 (3.65%) (figure 3A). MDIs, accounting for ~80% of inhaler-related CO2 emissions in 2024, drove a 56% rise in emissions (92.4 M kg CO2e in 2016 to 144.5 M kg CO2e in 2024) due to HFC propellants. Despite increased MDI prescriptions, COPD-related ER visits remained unchanged (p=0.12), while hospitalisations declined slightly but significantly (p=0.03). This modest improvement in outcomes could not, however, be clearly attributed to the increase in MDI use in this ecological analysis (figure 3B). AE, acute exacerbation; COPD, chronic obstructive pulmonary disease; ER, emergency room; HFC, hydrofluorocarbon; MDI, metered-dose inhaler; OPD, outpatient department. Between 2019 and 2021, mean MDI usage was 32.63%±1.80%, rising to 41.83%±3.69% (p<0.001) in 2022–2024. COPD hospitalisations declined from 1.42%±0.68% to 1.13%±0.39% (p=0.03). However, MDI prescriptions rose significantly from 4757 (2019–2021) to 6357 (2022–2024) (p<0.05), increasing carbon emissions from 100.2 M kg CO2e to 132.0 M kg CO2e (p<0.01). Despite the increase in MDI use, ER visit rates remained stable, indicating that higher MDI prescriptions did not significantly reduce AEs requiring emergency care. While hospitalisation rates showed a modest decline, the impact was insufficient to suggest a major improvement in overall COPD control (table 1). MDIs accounted for 80% of inhaler-related CO2e in 2024; however, device-mix trends are ecological and were not analysed as predictors of AE after adjusting for pollution. Because device mix was evaluated at the ecological clinic level rather than as a stable patient-level exposure, these temporal associations between MDI share, AE rates and emissions should be interpreted as hypothesis-generating and do not constitute causal comparisons between specific inhaler device types Table 1Comparison of MDI usage, COPD hospitalisation and carbon emissions between 2019–2021 and 2022–2024Year2019–20212022–2024P valueMDI usage (%) (mean±SD)32.63±1.8041.83±3.69<0.001COPD hospitalisation rate (%) (mean±SD)1.42±0.681.13±0.390.03MDI prescriptions47576357<0.05Carbon emissions (kg CO2e)100 249 928132 047 452<0.01COPD, chronic obstructive pulmonary disease; MDI, metered-dose inhaler. Comparison of MDI usage, COPD hospitalisation and carbon emissions between 2019–2021 and 2022–2024 COPD, chronic obstructive pulmonary disease; MDI, metered-dose inhaler. To assess robustness, an additional exploratory sensitivity analysis excluded the COVID-19 pandemic period (March 2020–December 2023). Trends were re-estimated using pre-pandemic data (2016–2019) and post-pandemic data (2024). Outpatient visits continued to increase, while AE rates maintained a declining trend, consistent with the full dataset. ICS/LABA/LAMA prescriptions rose sharply in 2024, aligning with post-pandemic shifts towards triple therapy. CO2 emissions from inhalers increased by 44% compared with the pre-pandemic average, with MDIs contributing 80%. Despite this rise in MDI use, COPD-related ER visits remained statistically unchanged (p=0.11), and hospitalisations declined modestly but significantly (p=0.03), indicating that although COPD hospitalisations declined modestly over the study period, this improvement could not be clearly attributed in timing or magnitude to the rise in MDI use. Discussion This study underscores the intricate relationship between air pollution, inhaled therapies and COPD exacerbations, highlighting the dual priorities of clinical efficacy and environmental sustainability. Elevated levels of PM2.5 were strongly correlated with increased COPD exacerbations, reinforcing the impact of environmental exposures on disease progression. The strong association between SABA use and AE rates in our models most likely reflects confounding by indication, with higher rescue use occurring in patients who are already experiencing more frequent exacerbations or unstable disease, rather than a direct harmful effect of SABAs. While triple therapy (ICS/LABA/LAMA) proved effective in reducing exacerbation rates, the growing reliance on pMDIs contributed to both higher carbon emissions. The ecological trends in AEs, ER visits and hospitalisations did not demonstrate any additional improvement in COPD control that could be linked to the increased use of pMDIs. The modest decline in COPD hospitalisations over the study period is clinically important, but given the ecological design and concurrent changes in air quality, triple-therapy uptake and pandemic-related factors, we cannot attribute this improvement to MDI use per se. These findings emphasise the urgent need for a balanced prescribing strategy that integrates low-carbon inhaler alternatives (eg, DPIs) without compromising therapeutic efficacy, ensuring that COPD management aligns with global environmental sustainability goals. PM2.5 exposure is a major environmental risk factor for AECOPD, with studies linking elevated levels to increased exacerbations, hospitalisations and ER visits.17 A meta-analysis found that a 10 µg/m³ rise in PM2.5 significantly increases COPD-related hospital visits, with effects lasting up to 7 days.18 Mechanisms include oxidative stress, airway inflammation and immune dysregulation, leading to heightened airway hyper-responsiveness and infection risk.19 PM2.5 exposure also contributes to lung function decline, elevated inflammatory markers (eg, C-reactive protein, neutrophils) and increased susceptibility to pulmonary fibrosis.17 20 Epidemiological studies suggest that each 10 µg/m3 increase in PM2.5 is linked to a 2.5% rise in AECOPD-related ER visits and hospitalisations.1 In our single-pollutant model (AE (%) = 3.007 + 0.0291 × PM2.5), a 10 µg/m3 increase corresponds to a 0.291-percentage-point absolute rise in the monthly AE rate, which, across our observed range (3.65–8.39%), translates to a 3.5–8.0% relative increase. This modestly exceeds prior estimates, likely because our outcome captures all AEs (outpatient, ER and inpatient) at a monthly level. These findings underscore the need for air quality regulations, patient education on pollution avoidance and optimised inhaler therapies to reduce PM2.5-induced COPD exacerbations. Triple inhaled therapy (ICS/LABA/LAMA) is highly effective in COPD management, offering superior exacerbation reduction, improved lung function and better symptom control than dual therapy. Studies show that patients on triple therapy experience fewer hospitalisations and exacerbations, highlighting its role in high-risk patients.21 A Japanese retrospective cohort study found that early triple-therapy initiation after an exacerbation delayed recurrence and reduced hospital readmissions.22 Modelling studies suggest early escalation preserves lung function, enhances quality of life and lowers all-cause mortality.23 Major trials, including TRIBUTE and IMPACT, consistently demonstrate its efficacy in reducing moderate-to-severe exacerbations compared with dual therapy.24 Post hoc analyses indicate the greatest benefits in patients with frequent exacerbations and higher eosinophil counts, supporting precision medicine approaches.25 Real-world studies confirm that early triple therapy prevents disease progression and reduces exacerbation-related mortality.26 However, pneumonia risk remains a concern, requiring careful patient selection.27 Inhaler choice significantly impacts clinical outcomes and adherence. Single-inhaler triple therapy improves adherence over multiple-inhaler therapy by simplifying treatment.28 29 DPIs offer better adherence and disease control than metered-dose inhalers (MDIs) due to ease of use and fewer inhalation errors.30 31 MDIs require hand–breath coordination, leading to higher critical errors and poorer drug delivery—>75% of pMDI users make at least one mistake, increasing exacerbation risk.31,34 Such technique errors not only compromise medication delivery but also increase the risk of exacerbations, highlighting the importance of patient education and device suitability. DPIs and SMIs eliminate coordination difficulties, improving adherence and reducing errors.35 36 DPIs are also linked to lower exacerbation rates and better lung function than MDIs.37 Studies comparing Fluticasone furoate/Umeclidinium/Vilanterol (FF/UMEC/VI) (DPI) and Budesonide/Glycopyrronium/Formoterol (BUD/GLY/FORM) (MDI) found that FF/UMEC/VI led to fewer exacerbations and greater lung function improvements.36 37 While inhaler training helps, MDIs still pose challenges in correct usage.38 Shifting towards DPIs over MDIs when appropriate could enhance adherence, reduce exacerbations and lower environmental impact, supporting a more sustainable COPD treatment strategy.39 40 The environmental impact of inhaler-related carbon emissions has raised concerns about sustainable respiratory treatments.12 Pressurised metered-dose inhalers (pMDIs), which use HFC propellants, contribute significantly to GHG emissions, accounting for 3–4% of healthcare-related emissions.41 In contrast, DPIs and SMIs have a carbon footprint nearly one-tenth that of pMDIs, offering a viable alternative for reducing emissions.41 Shifting from high-GWP pMDIs to lower-emission alternatives could cut inhaler-related emissions by up to 89%.42 43 Patients transitioning from pMDI-based to DPI-based maintenance therapy reduced their inhaler carbon footprint by >50% while maintaining stable COPD and asthma control.30 Given the lower environmental impact and clinical benefits of DPIs and SMIs in adherence and efficacy, prioritising non-pMDI inhalers could enhance patient outcomes while reducing carbon emissions.33,35 Health organisations, including the NHS and ERS, advocate for adopting lower-carbon inhaler alternatives.14 44 Despite their benefits, barriers to DPI adoption persist. Many patients and providers remain unaware of inhalers’ carbon impact, and prescribing habits favour MDIs due to familiarity.10 45 Misdiagnosis of COPD and asthma leads to unnecessary inhaler use, adding to avoidable emissions. Adherence to guideline-directed therapy and inhaler technique training could reduce reliever overuse, prevent hospitalisations and lower emissions.41 The healthcare sector plays a key role in climate action, with initiatives like the NHS Net Zero Strategy targeting carbon neutrality by 2040, including reductions in inhaler-related emissions.14 Innovations such as low-GWP HFA-152a inhalers and biodegradable materials are being explored to minimise environmental impact without compromising care.9 46 Some countries have successfully cut MDI use by >50%, demonstrating large-scale transition feasibility.12 Integrating environmental considerations into prescribing and patient education on device switching and proper inhaler use is essential for sustainable COPD management.10 Our carbon accounting did not include the footprint of ER visits or hospital admissions, which can be substantial at the level of energy use, diagnostics, procedures and inpatient care. Recent reviews of hospital care pathways suggest that in some settings emissions from acute care episodes may exceed those from chronic medication use.47 Accordingly, any reductions in COPD hospitalisations associated with optimised maintenance therapy are likely to yield net environmental benefits that are not captured by our inhaler-only CO2e estimates. As global temperatures rise with increasing carbon emissions, patients with COPD and asthma are exposed to worsening air quality and more frequent extreme weather events.7 Pollutants such as PM2.5 and NO2 aggravate respiratory disease and are associated with higher hospitalisation rates.11 Paradoxically, inhalers used to treat these conditions also contribute to GHG emissions, perpetuating a cycle in which environmental degradation further harms respiratory health. Air pollution, particularly PM2.5, PM10, NO2 and O3, is a major risk factor for chronic respiratory diseases.2 48 49 Although not directly examined in this study, climate change driven by GHG emissions and industrial activity is recognised to exacerbate air pollution through mechanisms such as wildfires, extreme weather and altered atmospheric dynamics.50 51 These processes may elevate PM2.5 and NO2 levels and thereby indirectly increase COPD exacerbation risk. Long-term exposure to air pollution accelerates lung function decline and heightens the risk of COPD exacerbations.52 53 Because many air pollutants share sources with GHG emissions, sustainable healthcare strategies, such as shifting from pressurised metered-dose inhalers to lower-emission DPIs, can reduce the carbon footprint of care while maintaining effective respiratory disease management.51,54 These findings highlight a potential vicious cycle whereby climate-driven rises in PM2.5 and related pollutants worsen COPD control, leading to more exacerbations and increased use of rescue and maintenance inhalers, including pMDIs. In turn, the resulting GHG emissions may further intensify climate change and indirectly deteriorate air quality. At the same time, severe exacerbations and hospital admissions themselves carry a substantial carbon footprint, so preventing them is likely to yield both clinical and environmental benefits across the care pathway. Accordingly, our inhaler CO2e estimates should not be seen as a reason to restrict clinically indicated therapy but rather as emphasising the need first to optimise disease control and then, where multiple options are clinically appropriate, to preferentially select lower-carbon devices (eg, DPIs or SMIs). Our findings support practical levers such as device-neutral formularies that prefer lower-GWP options when clinically suitable, inhaler take-back, recycling schemes and targeted reimbursement to lower patient cost-sharing for low-carbon devices. Outside Asia, programmes in high-income and middle-income settings show that transitions to DPIs or SMIs can be achieved without sacrificing outcomes when paired with adherence/technique training and equitable access policies.55 In parallel, nanoformulation-enabled delivery is an emerging avenue that may enhance dose efficiency and facilitate integration into non-propellant platforms, potentially reducing life-cycle impact while sustaining clinical efficacy.56 COPD AE rates in Yunlin declined between 2016 and 2024, with the most pronounced single-year reduction observed in 2020, the first year of the COVID-19 pandemic. Taiwan’s non-pharmaceutical interventions (including universal masking, border control and enhanced school hygiene) substantially reduced circulation of non-COVID respiratory viruses in 2020–2021,57 leading to marked declines in influenza and enterovirus activity and lower non-COVID respiratory virus positivity.58 At the same time, healthcare utilisation fell (outpatient visits by 19% and inpatient admissions by 10% in early 2020), which would reduce documented AEs independent of changes in underlying COPD biology. To address this, we modelled calendar month and year as fixed effects and conducted an additional sensitivity analysis excluding March 2020–December 2023. The main patterns remained, and we therefore interpret the 2020 AE nadir as being partly attributable to pandemic-related shifts in pathogen circulation and care-seeking behaviour rather than solely to treatment or programmatic effects. More broadly, COPD mortality in Taiwan has declined steadily from 2002 to 2022, with the steepest reductions during 2016–2022, indicating longer-term improvements that predate the pandemic.59 This study has several limitations. Its retrospective design limits causal inference despite multivariate adjustments. Data from a single centre may reduce generalisability to broader populations. Inhaler use was assessed via prescriptions, which may not reflect actual adherence or technique. Air pollution exposure was estimated from regional monitoring data rather than individual-level measurements. Carbon emissions were based on standard estimates and may not account for local variation in inhaler production or disposal. Clinical biomarkers were not included, limiting phenotype-specific analysis. Additionally, COVID-19 may have influenced healthcare behaviours during part of the study period. Conclusion In this single-centre retrospective cohort, higher ambient PM2.5 concentrations were associated with higher COPD AE rates, whereas greater use of guideline-directed maintenance therapy, particularly triple therapy (ICS/LABA/LAMA), was associated with lower hospitalisation rates. Our inhaler carbon-footprint analysis, based on published life-cycle emission factors, showed that pMDIs accounted for most of the estimated inhaler-related CO₂e but these emissions were not modelled as predictors of exacerbations, ER visits or hospitalisations and therefore should inform sustainability discussions rather than conclusions about clinical effectiveness. These findings suggest that optimising inhaled maintenance regimens to improve disease control is the primary priority, and that, when several inhaler options are clinically appropriate, lower-carbon devices can be considered to align COPD care with environmental goals. When clinically appropriate, non-pMDI inhalers such as DPIs and SMIs should be prioritised, given their favourable environmental profile and evidence from prior studies supporting their effectiveness, while recognising that our data were not designed to test patient-level effects of specific device types and cannot determine whether one device is superior to another for individual patients. Breaking this cycle requires coordinated action from healthcare providers, policymakers and patients to reduce air pollution, optimise COPD management and transition to low-carbon inhalers. Implementing air quality regulations, promoting correct inhaler use and integrating environmental considerations into prescribing practices are key steps in mitigating health and climate risks. Addressing this interconnected crisis demands a holistic approach to ensure long-term respiratory health and global sustainability. Future COPD management strategies must integrate both patient-centric care and environmentally responsible prescribing to address the dual burden of chronic respiratory disease and climate change. Supplementary material Acknowledgements We sincerely thank Ms Tzu-Yu Huang, Ms Hung-Chueh Peng and Mr Hsing-Che Wang for their invaluable support during the data collection process. Their contributions in providing database resources and assisting with manuscript refinement were pivotal to the successful completion of this work. Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Ethics approval: The study was approved by the Ethics Committee of National Taiwan University Hospital Yunlin Branch (approval no. 202406085RINA). The requirement for informed consent was waived as the study involved retrospective data collection and analysis from an existing database. Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. References Associated Data Supplementary Materials Data Availability Statement All data relevant to the study are included in the article or uploaded as supplementary information.