Asthma exacerbation risk after short-term air pollution exposure varies by age, sex, severity, and eosinophil count: East London cohort study

Abstract

Background

Air pollution is associated with asthma mortality and hospital admissions, yet few studies have focused on exacerbations managed in the community or which subgroups of asthma patients are most vulnerable. Identifying these groups is crucial for targeted public health interventions and clinical guidance.

Methods

We developed Poisson generalised regression models examining associations between short-term exposure to air pollutants (NO₂, PM_2.5 and PM₁₀, and O₃) and courses of oral corticosteroids (OCS) prescribed in primary care for exacerbations in a cohort of asthma-registered adults (18–80 years) in East London, between March 2019-February 2023. We analysed three time frames: pre-, during, and post-COVID-19 pandemic, and performed stratified analyses by gender, age, asthma management step, and blood eosinophil levels.

Results

For every interquartile range (IQR) increase in the preceding week’s average exposure, the risk of OCS prescription increased by 5% for NO₂, 2% for PM₁₀, and 2% for PM_2.5. Risk for OCS prescriptions were 2% lower during the COVID-19 period and 28% higher post-COVID-19 compared to pre-COVID-19. Stratified analyses showed that male and older adults had higher risks for NO₂ associated exacerbations with same-day exposure but lower risks for one-week exposure, compared to females and younger adults. Risks were also slightly higher in patients on step 4/5 asthma medications and in those with eosinophilia, with similar patterns observed for PM₁₀ and PM_2.5.

Conclusion

Short-term exposure to NO₂, PM₁₀, and PM_2.5 increases the risk of mild asthma exacerbations, however risk and lag were affected by age, gender, and patient subgroups, emphasizing the need for tailored public health strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12931-026-03512-w.

Background

Asthma is a prevalent chronic disease impacting approximately 300 million individuals worldwide. In the UK approximately 12% of the population has been diagnosed with asthma, with London having the highest asthma prevalence [1]. This is particularly evident in the east of the city, where hospitalisation for asthma is 14% higher than the London average [2].

Multiple interacting factors including demographic and lifestyle [3], genetic predisposition [4], and environmental factors [5], influence exacerbations and (loss of) asthma control. In primary care, air pollution is increasingly recognised as an important environmental factor influencing patterns of morbidity and mortality. Research indicates that there is a risk of increased adverse respiratory symptoms in adults and children even in areas with relatively low levels of pollution [6, 7]. A study in Brussels showed a significant association between ambient nitrogen dioxide, NO₂, exposure and daily sales of asthma and chronic obstructive pulmonary disease (COPD) medication [8]. Another study in Bradford, UK showed that air pollution, in particular, that derived from road transport, leads to a large childhood asthma burden [9]. Several systematic reviews and meta-analyses have examined the association between short- and long-term air pollution exposures with asthma exacerbations [10, 11]. They reported adverse associations between short-term NO₂ and particulate matter with diameter 10 microns or less (PM₁₀) exposure with asthma symptoms, and long-term NO₂, PM₁₀ and particulate matter with diameter 2.5 microns or less (PM_2.5) exposures with records of emergency department and/or hospital admissions because of asthma attacks. These associations are biologically plausible as PM₁₀, PM_2.5, NO₂ and ozone (O₃) have all been shown to cause allergic and non-allergic inflammation [12–14], induce oxidative stress [15, 16], stimulate airway nerves [17], and induce airway hypersensitivity and mucus hypersecretion [18]; ultimately resulting in airway remodelling [19].

Asthma exacerbations vary in severity, with severe exacerbations generally requiring an emergency department visit and likely hospitalisation [20]. An extensive body of studies has investigated the association between air pollution and mortality rates and hospital admissions related to asthma [21–27]. However, few studies have investigated less severe consequences of air pollution exposure, such as respiratory consultations with general practitioners (GP) and prescriptions of oral corticosteroid courses (OCS) in primary care. Even fewer studies have examined whether particular subgroups of patients with asthma may be at higher risk from air pollution exposure despite this being an important question when considering patient care and public health messaging during periods of elevated pollution [28].

Considering the substantial health and economic impacts of less severe exacerbations treated in the primary care setting, our study aimed to investigate the relationship between these exacerbations and short-term air pollution exposure, examining both same day and one-week average exposure. Our hypothesis was that short-term exposures to traffic related air pollutants in contemporary London triggers ‘late-phase’ asthmatic airway responses that will be associated with a worsening of asthma symptoms, as reflected in increased prescription of OCS in primary care over the following week. Leveraging a cohort of asthma-registered adults across all 183 practices in East London, we examined this association during three distinct time periods: pre-, during, and post-COVID-19 pandemic. The pandemic provided a unique sensitivity experiment as the reductions in traffic and increased time spent indoors should have led to decreased exposure to outdoor traffic pollutants, potentially resulting in fewer asthma exacerbations and weaker associations with ambient pollution.

We additionally conducted analyses in which the asthma cohort was stratified by gender, age, asthma medication step and most recent blood eosinophil count – demographic and clinical characteristics readily available in primary care to identify variation in the susceptibility of patient subgroups to short-term pollutant exposures.

Methods

Study design and setting

Adults with Doctor-diagnosed asthma (18–80 years old) based on the UK Quality and Outcomes Framework (QOF) data [29], in five clinical commissioning groups (CCGs) in East London including Newham, Tower Hamlets, City & Hackney, Havering and Waltham Forest were included in our cohort. We excluded people with polymyalgia rheumatic (PMR), giant cell arteritis and temporal arteritis, rheumatoid arthritis, ulcerative colitis and multiple sclerosis, since these patients may be prescribed oral corticosteroids for those alternate conditions. We extracted demographic factors including age, gender, ethnicity, smoking status and Body Mass Index (BMI), prescribed medicine including oral steroid courses, asthma management step (latest recorded) and eosinophil count (latest recorded) from the entire cohort.

Medical records were extracted anonymously on secure N3 terminals from EMIS Web and managed according to UK NHS information governance requirements.

Outcome and COVID-19 effect

The daily count of prescribed oral corticosteroid courses (OCS) for four years, from March 2019 until February 2023 was the outcome measure in this study. OCS was selected as a marker of a significant asthma exacerbation treated in the primary care setting. Prescription of oral steroids through primary care at weekends should be zero at many primary care centres, hence, we removed weekend prescription data.

To investigate the effects of COVID-19 pandemic, we defined a time varying variable with three levels: pre-COVID (1st March 2019–28th Feb. 2020), peri-COVID (1st March 2020–28th Feb. 2022), and post-COVID (1st March 2022–28th Feb. 2023).

Seasonal and day-of-week effect

To consider seasonality, we defined a categorical covariate with three levels: Spring (March, April and May), Summer (June, July and August) and Winter (September to February). Seasonality was modelled using three categories to reflect the clear patterns observed in adult asthma, where exacerbations rise sharply in September and remain elevated throughout the winter months. We also defined a five-level variable for day of the week (week days).

Air pollution exposure

We extracted daily air pollution measurements for four years (March 2019 until February 2023) from twelve monitoring stations (roadside and urban background) in East London (Figure S1), and we used average of these stations as daily air pollution exposure in east London. The measurements included NO₂, PM_2.5, PM₁₀, and O₃. Percentage of missing values (averaged over all stations) for the whole study period was 5%, 14%, 22% and 10% for NO₂, PM₁₀, PM_2.5 and O₃, respectively. We implemented multiple imputation by chained equation (MICE) [30] for imputing missing values.

Statistical analysis

To investigate the temporal association between air pollution exposure and daily OCS, we developed a Poisson generalised linear regression model. We estimated incident rate ratios (IRRs) with 95% confidence intervals (CI) for single-pollutant and bi-pollutants models. Our model included pollutant exposure on the day of potential OCS prescription, 1 day before (lag 1), 2 days before (lag 2) and average one week prior to day of potential OCS prescription. We used quartiles of ambient pollution concentrations in all models rather than absolute ambient concentrations.

We also developed stratified models based on gender (male vs. female), age group (adult: 18–60 yrs vs. older adult: >60 yrs), asthma management step (step 1/2/3 vs. step 4/5) and most recent blood eosinophil count (< 0.3 × 10⁹ cell/L vs. ≥ 0.3 × 10⁹/L). Asthma management step was defined using the most recent record in the primary care notes, following the UK BTS/SIGN stepwise model. Steps 1–5 reflect escalating treatment from reliever-only therapy to high-dose inhaled corticosteroids (ICS) with multiple add-on controllers or biologics. A brief summary of each step is provided in Supplementary Table S1. For analysis, we grouped steps 1–3 (lower treatment intensity: low–medium dose ICS ± LABA) and steps 4–5 (higher treatment intensity: high-dose ICS with additional controllers or maintenance oral corticosteroids). This grouping distinguishes mild–moderate from more severe asthma and is consistent with UK primary care practice.

All statistical analyses were performed using Stata (version 17.0).

Results

The cohort of this study included 71,620 individuals with asthma who met our selection criteria out of a total population of 5,791,542 individuals registered with one of the five CCGs (total of 183 practices) in East London at the end of February 2023. The total number of OCS prescriptions in four years for this cohort was 88,732. Characteristics of this cohort are presented in Table 1.

Table 1.

Characteristics of the study cohort

Variable	Category	N (%)
Gender	Male	30,519 (43)
	Female	41,101 (58)
Age	Adult (18–60 years old)	57,923 (81)
	Older adult (> 60 years old)	13,697 (19)
BMI	≤ 25 kg/m2	10,711 (15)
	> 25 kg/m2	27,234 (38)
	Unknown	33,675 (47)
Ethnicity	White	36,157 (50)
	Asian or Asian British	7498 (10)
	Black or Black British	15,898 (22)
	Mixed	2736 (4)
	Other Ethnic Group	3690 (5)
	Unknown	5641 (9)
Blood eosinophilic count	≥ 0.3 × 109/L	19,881 (28)
	< 0.3 × 109/L	12,806 (18)
	Unknown	38,933 (54)
Asthma management step	Step 1	4139 (6)
	Step 2	11,650 (16)
	Step 3	4647 (6)
	Step 4	581 (0.9)
	Step 5	47 (0.1)
	Unknown	50,425 (71)

Seasonal and day-of-week effects

Estimated IRRs for seasonal effects of spring, summer and winter on daily OCS prescription were similar in the four single-pollutant models for NO₂, PM₁₀, PM_2.5 and O₃; during summer the risk of asthma exacerbation decreased (for example IRR, 0.96; 95% CI, 0.94 − 0.097 in NO₂ model) compared to the spring, while it increased during winter (IRR, 1.28; 95% CI, 1.25–1.31 in NO₂ model). Similarly, estimated IRRs for “day-of-week” covariate showed consistent patterns across the four single-pollutant models. Compared to Monday, prescriptions on Tuesday didn’t change, however, prescriptions on Wednesday, Thursday and Friday were decreased by 9%, 18% and 10%, respectively (in NO₂ model).

Effects of COVID-19 pandemic

We investigated the impact of COVID-19 on the relationship between air pollution and daily OCS prescription, using a time-varying covariate. The estimated IRRs and 95% CIs for this covariate were similar in the single-pollutant models for NO₂, PM₁₀, PM_2.5 and O₃. For example, in the PM₁₀ model, the risk of asthma exacerbation, after adjusting for other factors, showed a slight decrease during the COVID-19 period, with a 2% lower risk (IRR, 0.98; 95% CI, 0.96–1.00.96.00) compared to the pre-COVID-19 period. Conversely, following the COVID-19 period this risk notably increased, with approximately a 28% higher risk (IRR, 1.28; 95% CI, 1.26–1.30) compared to the pre-COVID-19 period.

Air pollution exposure

Time series of NO₂, PM₁₀, PM_2.5 and O₃, over three study periods (pre-, peri- and post-COVID) are shown in Fig. 1. Mean, median and standard deviation (SD) of daily air pollution measurements (average of twelve monitoring stations in East London) for each study period are presented in Table S2. For NO₂, the mean concentration decreased significantly by 25%, from 37.15 µg/m³ pre-COVID to 28.01 µg/m³ during COVID, and further to 26.52 µg/m³ post-COVID, with corresponding medians and standard deviations reflecting a similar trend. PM₁₀ levels also declined slightly by 6%, with the mean dropping from 18.40 µg/m³ pre-COVID to 17.30 µg/m³ during COVID and then to 17.05 µg/m³ post-COVID, alongside decreasing medians and relatively stable standard deviations. PM_2.5 exhibited a slight reduction in mean concentration (3%) from 10.68 µg/m³ pre-COVID to 10.33 µg/m³ during COVID and 9.96 µg/m³ post-COVID, with the median decreasing slightly and the standard deviation remaining consistent. In contrast, O₃ levels increased by 11%, with the mean rising from 32.75 µg/m³ pre-COVID to 36.52 µg/m³ during COVID and further to 38.34 µg/m³ post-COVID, while the median showed fluctuations and the standard deviation increased. These changes indicate a general improvement in NO₂, PM₁₀, and PM_2.5 pollution levels during and after COVID-19.

Fig. 1.

Time series of daily pollutants with fitted moving average across the three study period for (a) NO₂, (b) PM₁₀, (c) PM_2.5 and (d) O₃

Pearson correlation coefficients of NO₂, PM₁₀, PM_2.5 and O₃ for three study periods (pre-, peri- and post-COVID) are presented in Table S3. Based on these results, NO₂ positively correlated with PM₁₀ during the study periods, while the correlation is higher in the post-COVID period (0.71), compared to the peri-COVID (0.55), and pre-COVID (0.66). Similarly, NO₂ and PM_2.5 were positively correlated during the study periods.

Single and bi-pollutant models

Estimated IRRs and 95% CIs for same-day pollutant exposure are summarized in Table 2, and for average of one-week exposure (single-pollutant and bi-pollutants models) in Table 3. We found negligible associations between per-IQR increase in the same-day exposure and risk of prescribing OCS: IRR per-IQR = 1.01, 0.99, 0.99 and 0.96 for NO₂, PM_2.5, PM₁₀ and O₃, respectively, although associations were significantly positive for the highest quartiles of NO₂ exposure compared to lowest quartile. However, we found positive significant associations between average of one-week NO₂, PM₁₀ and PM_2.5 exposure and risk of OCS prescription: per-IQR increase in the air pollution levels, the risk of asthma exacerbation increased by 5%, 2% and 2% for NO₂, PM₁₀ and PM_2.5, respectively. For ozone (O₃), a negative association was identified for one-week exposure, reflectively its inverse correlation with NO₂. Examination of quartile-specific risks showed pollutant-specific exposure–response patterns rather than uniformly graded increases across pollutants. For NO₂, the one-week average model demonstrated elevated risk beginning in the second quartile (22.12 µg/m³) and peaking in the highest quartile (33.0 µg/m³), whereas PM_2.5 showed meaningful increases only in the highest quartile (11.92 µg/m³), suggesting a potential threshold above which risk increases more sharply.

Table 2.

Incident rate ratios (IRRs) for same-day exposure with 95% CIs for single-pollutant and multi-pollutants models

Pollutant			Quartiles of Air pollution					P-for-trend	Per IQR (µg/m3)
			Q1	Q2	Q3	Q4	IQR
NO2		Quartile range (µg/m3)	12.57–22.11	22.12–26.88	26.89–32.99	33.00–60.58.00.58	10.88
	IRRs* (95% CI)	single pollutant	Ref	0.99 (0.97–1.01)	1.04 (1.01–1.06)	1.04 (1.02–1.07)		< 0.001	1.01 (0.98–1.04)
		bi-pollutants PM10 & NO2	Ref	1.00 (0.97–1.02)	1.05 (1.02–1.07)	1.07 (1.04–1.10)
		bi-pollutants PM2.5 & NO2	Ref	0.99 (0.97–1.02)	1.04 (1.02–1.07)	1.05 (1.02–1.08)
PM10	IRRs* (95% CI)	Quartile range (µg/m3)	8.79–13.55	13.56–15.82	15.83–19.18	19.19–46.65	5.72
		single pollutant	Ref	0.98 (0.96–1.00.96.00)	0.99 (0.98–1.01)	0.98 (0.96–1.00.96.00)		< 0.001	0.99 (0.99–1.03)
		bi-pollutants PM10 & NO2	Ref	0.96 (0.94–0.98)	0.98 (0.96–0.99)	0.95 (0.93–0.97)
PM2.5	IRRs* (95% CI)	Quartile range (µg/m3)	3.94–7.30	7.31–8.82	8.83–11.91	11.92–37.53	4.59
		single pollutant	Ref	1.00 (0.98–1.02)	1.00 (0.98–1.02)	1.03 (1.01–1.05)		< 0.001	0.99 (0.98–1.02)
		bi-pollutants PM2.5 & NO2	Ref	0.99 (0.97–1.01)	1.00 (0.97–1.02)	1.02 (0.99–1.04)
O3	IRRs* (95% CI)	Quartile range (µg/m3)	6.80–28.75.80.75	28.76–37.33	37.34–46.15	46.16–76.06	17.29
		single pollutant	Ref	0.99 (0.97–1.01)	0.99 (0.97–1.01)	0.91 (0.88–0.93)		< 0.001	0.96 (0.92–0.97)

* adjusted for air pollution at Lag 1, Lag 2 and average of one-week, seasonality, day-of-week and COVID effect

Table 3.

Incident rate ratios (IRRs) for average of one-week lag with 95% CIs for single-pollutant and multi-pollutants models

Pollutant			Quartiles of Air pollution					P-for-trend	Per IQR (µg/m3)
			Q1	Q2	Q3	Q4	IQR
NO2		Quartile range (µg/m3)	12.57–22.11	22.12–26.88	26.89–32.99	33.00–60.58.00.58	10.88
	IRRs (95% CI)	single pollutant	Ref	1.04 (1.02–1.06)	1.03 (1.00–1.05.00.05)	1.06 (1.02–1.09)		< 0.001	1.05 (1.03–1.07)
		bi-pollutants PM10 & NO2	Ref	1.04 (1.01–1.06)	1.03 (1.00–1.05.00.05)	1.05 (1.02–1.09)
		bi-pollutants PM2.5 & NO2	Ref	1.05 (1.02–1.07)	1.03 (1.00–1.06.00.06)	1.07 (1.03–1.10)
PM10	IRRs (95% CI)	Quartile range (µg/m3)	8.79–13.55	13.56–15.82	15.83–19.18	19.19–46.65	5.72
		single pollutant	Ref	1.05 (1.03–1.07)	1.05 (1.03–1.08)	1.06 (1.04–1.09)		< 0.001	1.02 (1.01–1.03)
		bi-pollutants PM10 & NO2	Ref	1.05 (1.03–1.08)	1.06 (1.03–1.08)	1.04 (1.01–1.07)
PM2.5	IRRs (95% CI)	Quartile range (µg/m3)	3.94–7.30	7.31–8.82	8.83–11.91	11.92–37.53	4.59
		single pollutant	Ref	1.03 (1.01–1.05)	1.01 (0.99–1.03)	1.05 (1.02–1.07)		< 0.001	1.02 (1.01–1.03)
		bi-pollutants PM2.5 & NO2	Ref	1.02 (1.00–1.04.00.04)	1.00 (0.97–1.02)	1.01 (0.99–1.04)
O3	IRRs (95% CI)	Quartile range (µg/m3)	6.80–28.75.80.75	28.76–37.33	37.34–46.15	46.16–76.06	17.29
		single pollutant	Ref	0.92 (0.90–0.94)	0.95 (0.92–0.97)	0.93 (0.90–0.96)		< 0.001	0.95 (0.93–0.97)

* adjusted for air pollution at same-day, Lag 1 and Lag 2, seasonality, day-of-week and COVID effect

IRRs from the bi-pollutant models were broadly consistent with the single-pollutant estimates, with similar effect directions, comparable magnitudes, and no substantial changes in statistical significance after adjusting for a second pollutant.

Stratification analysis

Stratification analyses was performed based on gender (male vs. female), ages (adult: 18–60 vs. older adult: >60 yrs), asthma management step (step 1/2/3 vs. step 4/5) and most recent blood eosinophil count (≥ 0.3 × 10⁹/L vs. < 0.3 × 10⁹/L). The estimated IRRs and 95% CIs for same-day exposure and average of one-week air pollution exposure and risk of prescribing OCS for the stratified models are presented in Figs. 2, 3, 4 and 5 for gender, age group, asthma management and blood eosinophilic count, respectively.

Fig. 2.

IRRs and 95% CIs for average of one-week lag and same-day exposure of NO₂, PM₁₀, PM_2.5 and O₃, stratified by gender

Fig. 3.

IRRs and 95% CIs for average of one-week lag and same-day exposure of NO₂, PM₁₀, PM_2.5 and O₃, stratified by age group

Fig. 4.

IRRs and 95% CIs for average of one-week lag and same-day exposure of NO₂, PM₁₀, PM_2.5 and O₃, stratified by asthma management step

Fig. 5.

IRRs and 95% CIs for average of one-week lag and same-day exposure of NO₂, PM₁₀, PM_2.5 and O₃, stratified by blood eosinophilic count ≥ 0.3 × 10⁹L and < 0.3 × 10⁹L

Both male and female individuals with asthma showed an increased risk of OCS prescription with elevated same day NO₂ ambient concentrations in the highest quartiles (Q3 and Q4) compared to lowest quartile (Q1), with IRRs higher in males. In contrast, IRRs with one-week average exposure showed females were more adversely affected by NO₂ and PM_2.5, than males, especially in the highest quartile (Q4). The IRRs of the highest quartile (Q4) compared to lower quartile (Q1) in the stratified models for females were 1.09 (95% CI, 1.05–1.13), 1.08 (95% CI, 1.05–1.11) and 1.05 (95% CI, 1.02–1.08) for NO₂, PM₁₀ and PM_2.5, respectively, while the corresponding values for males were 0.99 (95% CI, 0.94–1.04), 1.02 (95% CI, 0.97–1.06) and 1.01 (95% CI, 0.97–1.05).

Older adults showed increased IRRs for OCS prescription with elevated same day NO₂. However, IRRs for one-week average exposure, stratified by age group (Fig. 3) indicated that younger adults were more negatively affected by one-week prior NO₂, PM₁₀ and PM_2.5, exposures compared to older adults, again the effect particularly marked in the highest quartile (Q4). Estimated IRRs in the highest quartile (Q4) in the stratified models for younger adults were 1.07 (95% CI, 1.03–1.11), 1.06 (95% CI, 1.03–1.11) and 1.05 (95% CI, 1.01–1.07) for NO₂, PM₁₀ and PM_2.5, respectively, while the corresponding values for older adults were 1.04 (95% CI, 0.99–1.09), 1.05 (95% CI, 1.01–1.10) and 1.03 (95% CI, 0.99–1.07).

Marginally higher IRRs with elevated ambient pollutant exposure were evident in patients on step 4/5 asthma medications compared to those on step 1/2/3 medications with NO₂ and PM₁₀ (both same day and one-week average), though increased risks with PM_2.5 exposures were less evident. Estimated IRRs of the one-week highest quartile (Q4) in the stratified models for those on step 1/2/3 were 1.14 (95% CI, 1.01–1.28), 1.17 (95% CI, 1.06–1.28) and 1.14 (95% CI, 1.05–1.25) for NO₂, PM₁₀ and PM_2.5, respectively, while the corresponding values for those on step 4/5 were 1.15 (95% CI, 1.03–1.30), 1.17 (95% CI, 1.08–1.28) and 1.08 (95% CI, 0.99–1.18).

Blood eosinophil counts were only available for a subset of the patients. Eosinophilic patients (≥ 0.3 × 10⁹/L) had higher IRRs for OCS prescription with higher quartiles of one-week average NO₂ exposure than non-eosinophilic patients (< 0.3 × 10⁹/L), though risk was elevated in both groups. Similar increased risk was evident in eosinophilic patients with PM₁₀ exposure. Estimated IRRs of the one-week highest quartile (Q4) in the stratified models for eosinophilic patients were 1.32 (95% CI, 1.18–1.47), 1.17 (95% CI, 1.08–1.28) and 1.13 (95% CI, 1.04–1.22) for NO₂, PM₁₀ and PM_2.5, respectively, while the corresponding values for non-eosinophilic patients were 1.15 (95% CI, 1.06–1.26), 1.19 (95% CI, 1.11–1.28) and 1.08 (95% CI, 1.01–1.15).

Discussion

In this study, we hypothesised that despite the reductions in air pollution in London over the last decade, present levels of traffic-related air pollution still adversely affect people with asthma living within the city. To address this, we investigated the association between short-term air pollution exposure and asthma exacerbations treated in the primary care setting, measured by prescription of oral corticosteroid courses (OCS). We found that per IQR (µg/m³) increase in the average of one-week NO₂, PM₁₀ and PM_2.5 exposures, the risk of asthma exacerbation climbed by 5%, 2% and 2%, respectively, consistent with ambient pollutant exposure triggering late-phase asthmatic airway inflammation.

The majority of research examining the relationship between asthma and air pollution has focused on hospitalization or mortality due to asthma attacks [21–27]. In contrast, few studies have examined asthma exacerbations in the community as an outcome. Ashworth et al. [31] examined primary care data in Lambeth, South London (2009-13), and reported significant increases in respiratory consultations and inhaler prescriptions, with previous 1 or 2 days, or weekly average NO₂ and PM₁₀. A similar pattern was also seen with PM_2.5. These findings align well with our results, despite covering a period preceding the air pollution improvements seen in London following the introduction of London’s Ultra Low Emission Zone. Previously we investigated the association between daily NO_x (oxides of nitrogen) concentrations and risk of asthma exacerbation in a smaller primary health care cohort from east London and over a shorter time frame (February 2018–January 2020, corresponding to the pre-COVID period in the present analysis). In that study we observed that a 1 µg/m³ increase in NO_x in previous week was associated with increased risk of prescribing OCS by 2.7% [32]. In the present study we extended these observations to a larger cohort to consider the key criterion pollutants, and over a period of significant change in traffic pollution due to the COVID-19 travel restrictions. This allowed us to compare the changing impact of traffic-related pollutant exposures on OCS prescribing in the periods before and after the COVID-19 lockdowns, during which ambient pollution levels changed markedly.

In the stratification analysis, we found associations between air pollution and OCS prescription, as a surrogate of asthma exacerbations, in all subgroups of patients with asthma. However, the timing and magnitude of the observed risk varied between subgroups. For example, the risk associated with elevated ambient NO₂ was greater in males for same-day exposure, but greater in females when considering one-week average exposures. Similarly, older adults had increased risk for same-day exposure compared to younger adults where the increased risk was more evident with one-week average exposure. These differential patterns reflect likely differences in the initiation of immunological responses to inhaled ambient pollutants, and the relative contribution of early- and late-phase asthmatic airway responses to exacerbations. For example, Ding et al. (2020) have shown differences between men and women in the time courses of peripheral blood eosinophil responses following air pollution exposure [33]. Gender is known to have a major effect of asthma, with the majority of patients with severe asthma being female [34]. Increased risks in patients on step 4/5 medications (high dose inhaled corticosteroids generally with long-acting bronchodilators with/without maintenance oral corticosteroids or biologics), compared to those on lower medication steps, is also consistent with greater disease severity in those patients requiring more medication. Increased risk of exacerbation in eosinophilic patients with greater one-week prior pollutant exposure is consistent with our understanding of blood eosinophils as a biomarker of exacerbation risk [35] and is consistent with ambient pollutants acting via late phase asthma immune responses [36].

The COVID-19 pandemic significantly impacted air pollution levels, particularly primary emissions from traffic with measurements from 129 monitoring stations across the UK showing mean reductions in NO₂ off 38.3% and PM_2.5 by 16.5% [37]. Additionally, personal exposure to NO₂ and PM_2.5 declined by 52.3% and 15.7%, respectively during the pandemic [38], and there was a 50% reduction in road traffic overall [39]. If traffic related pollution was contributing to asthma exacerbations such reductions would be predicted to improved symptoms. Our findings indicated that risk of asthma exacerbation was only 2% lower during the COVID-19 period (compared to the pre-COVID period) after adjustment for air pollution exposure. This reduction likely resulted from a combination of factors, including reduced circulation of respiratory viruses, improved self-management driven by patient concerns during the pandemic, and government advice for high-risk individuals to follow stricter restrictions [40]. In addition, we found that the risk of asthma exacerbation was 28% higher in the post-COVID period, compared to the pre-COVID, consistent with the findings of Tydeman et al. [41].

The main limitation of this study was the lack of data on indoor air pollution exposure, particularly important during the Covid lockdown periods. Our study was entirely based on outdoor pollutant concentrations based on average measurements from twelve monitoring stations in East London to estimate the exposure for the entire cohort. Another limitation was due to incomplete clinical data for asthma registered patients, e.g. we didn’t have eosinophil counts for 51% and asthma management step for 74% of the cohort. Patients with complete data are more likely to have had recent clinical assessments or more active asthma management, which may not reflect the broader population with asthma. This introduces selection bias, particularly in the eosinophil analyses, as individuals with recorded eosinophil counts may represent a group with more severe, unstable, or closely monitored disease. As a result, subgroup estimates for eosinophilic status and medication step should be interpreted with caution, as they may not fully generalise to all adults with asthma in the dataset and may over represent patients at higher baseline risk. A further limitation concerns potential exposure misclassification arising from the limited number and uneven distribution of air pollution monitoring stations in East London. Because individual-level exposure was estimated using averaged concentrations from these sites, spatial variability in pollutant levels may not be fully captured.

Conclusion

In conclusion, our study sheds light on the intricate relationship between short-term air pollution exposure and mild asthma exacerbations treated in primary care settings. We found notable associations between increased levels of NO₂, PM₁₀, and PM_2.5 and heightened risks of asthma exacerbations, while observing a negative correlation with Ozone, reflecting its inverse relationship with NO₂. Furthermore, our stratified analyses revealed differences in risk between stratified subgroups of patients with asthma. Differences in timing of risk by age and gender require replication in other studies but potentially need to be considered in public health advice and in particular the duration following an air pollution episode that different patients should be advised to more closely monitor their asthma. Similarly, public health messaging should potentially be directed to those patients on higher medication steps. The increased risks in those patients with recent eosinophilia suggests that the risks associated with air pollution exposure could be reduced by biomarker-directed clinical management of patients where inhaled corticosteroids are increased until any blood eosinophilia is suppressed. We did not, however, test whether reducing eosinophilia alters pollution-related risk, so this should be viewed as a potential implication rather than a tested effect.

These findings underscore the urgency of implementing targeted interventions and policies to reduce air pollution, particularly traffic emission.

Authors’ contributions

**HH** : Conceptualization, Methodology, Formal Analysis, Writing – Original Draft**PEP**, **ISM** and **CJG** : Conceptualization, Methodology, Supervision, Writing – Review & Editing.

Funding

HH was funded by the NIHR, school of primary care research (SPCR) (grant reference number = 627). IM is part funded by the National Institute for Health and Care Research (NIHR) Health Protection Research Unit in Environmental Exposures and Health (NIHR 200922), a partnership between UK Health Security Agency and Imperial College London, and the MRC Centre for Environment and Health, which is funded by the Medical Research Council (MR/S0196669/1, 2019–2024). The views expressed are those of the author(s) and not necessarily those of the NIHR, UK Health Security Agency or the Department of Health and Social Care.

Data availability

The data that support the findings of this study are available from EMIS web but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of EMIS web.

Declarations

Ethics approval and consent to participate

This study involved secondary analysis of routinely collected, de-identified primary care electronic health record data obtained from general practices in East London. The research team did not have access to identifiable patient information at any stage, and all data analysed were anonymised and reported in aggregate form. The study received retrospective sponsorship approval from Queen Mary University of London (EDGE ID: 204982). The study did not undergo Health Research Authority (HRA) or NHS Research Ethics Committee (REC) review.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.