Long-term Air Pollution Exposure, Plant-based Diet and Asthma Exacerbations in the Nurses’ Health Study II

Abstract

Rationale:

Short-term ambient air pollution exposure may worsen asthma health. Effects of longer-term air pollution exposures on asthma exacerbations and risk mitigation by dietary factors are unknown.

Objective:

To examine associations between 48-month air pollution exposure and asthma exacerbations and whether a plant-based diet modifies these relationships.

Methods:

Women with asthma in the Nurses’ Health Study II were followed from 1997 to 2014. We estimated 48-month time-varying average residential ambient fine particulate matter (PM_2.5), nitrogen dioxide (NO₂) and ozone (O₃) exposures using nationwide spatiotemporal models. Plant-diet index (PDI) scores were calculated based on food frequency questionnaires administered every 4 years. Air pollution and diet assessments were repeated measures within-individuals, while asthma exacerbations in the past year were captured in 1998 and 2014. Average air pollutant exposure was assessed in the 48-months prior to each outcome assessment year. Single and multi-pollutant logistic regression models with generalized estimating equations to account for repeated measures within participants were used to assess the effects of each air pollutant on asthma exacerbation risk. We also evaluated effect measure modification by PDI scores on the effects of each air pollutant on asthma exacerbation risk using two-way interaction terms.

Results:

Of 4326 participants, median 48-month PM_2.5, NO₂ and O₃ concentrations were 13.7 μg/m³, 12.0 ppb and 25.5 ppb, respectively, from July 1993 to June 1997 and 8.9 μg/m³, 6.6 ppb and 27.8 ppb, respectively, from July 2009 to June 2013. In adjusted single pollutant models, greater exposures to both PM_2.5 and NO₂ were associated with higher odds of asthma exacerbation (OR 1.43; 95% CI 1.14–1.80, and OR 1.25; 95% CI 1.12–1.38, respectively). In multi-pollutant models, greater exposure to NO₂ was associated with higher odds of asthma exacerbation (OR 1.23; 95% CI 1.06–1.42). There were no statistically significant interactions between pollutants and PDI score on asthma exacerbations.

Conclusions:

Long-term exposure to ambient NO₂ and PM_2.5 even at low levels, may increase asthma exacerbation risk in women, but is not attenuated by a plant-based diet as measured herein. Further research is needed on long-term effects of inhaled pollutants on asthma health and personal, modifiable strategies to reduce risk.

INTRODUCTION

Recent ecological disasters, including record-breaking global heat waves and the unprecedented intensity of the 2023 Canadian wildfires, have amplified public awareness of ambient air pollution and the transcontinental impact on air quality, health and socioeconomics.^1,2 Climate-driven weather changes may worsen air quality, with key pollutants such as fine particulate matter (PM_2.5), nitrogen dioxide (NO₂) and ozone (O₃) posing major respiratory risks.³ PM_2.5 from sources including wildfire smoke, windblown dust from droughts, vehicle and industrial emissions,⁴ has been linked to premature death and worsening cardiorespiratory health.^5,6 NO₂ is a gas primarily from fuel combustion and industrial activities, concentrations of which may increase due to extreme weather-driven fossil fuel demand.⁷ Ground-level O₃ forms through reactions between nitrogen oxides and volatile organic compounds in heat and sunlight, peaking during heatwaves.⁸ Both NO₂ and O₃ are associated with increased respiratory mortality.^9–11

Populations with pre-existing lung disease are particularly susceptible to poor air quality. Asthma is a major public health concern affecting over 300 million people worldwide.¹² Numerous case-crossover studies have demonstrated how high pollutant exposure over hours to days is temporally associated with worse asthma control and exacerbations.^13–15 However, exposure over years may be more representative of the cumulative burden of inhaled pollutants and their longitudinal effects on respiratory health. For example, long-term exposure over a decade to criteria ambient air pollutants, such as PM_2.5, NO₂ and O₃, have been linked to developing adult-onset asthma,^16–18 but their effects on exacerbation risk in adults with established asthma remain understudied. Notably, adult women, may have heightened susceptibility to inhaled environmental exposures, contributing to greater asthma prevalence and severity, including more frequent asthma exacerbations and higher death rates, compared to men.¹⁹ Thus, the long-term effects of ambient air pollution on asthma health merit dedicated investigation in this population.

Beyond public health policies addressing cleaner air, there is growing effort to identify ways to mitigate the impact of ambient air pollution exposure. Individual modifiable factors, such as the intake of certain nutrients, have been considered, building on potential relationships between diet and asthma health.²⁰ Specifically, anti-oxidants, omega-3 polyunsaturated fatty acids and vitamin D may alter the asthmatic response to inhaled pollutants.^21–23 However, the sum of different nutrients may demonstrate greater impact on health than individual nutrients alone, emphasizing the importance of considering dietary patterns.²⁴ While individual plant-derived nutrients, including vitamin C, vitamin E and fiber, have been suggested to improve asthma health, with some capacity to dampen the negative effects of inhaled pollutants,^25–28 few studies have examined a plant-based dietary pattern and asthma health. The Plant-Based Diet Index (PDI) score is a commonly used metric that assesses consumption of plant-based foods and has been widely applied across various disease states.^29–31 One French cohort study found that consumption of a plant-based diet reduced asthma symptoms in women.³² However, the potential role of a plant-centered diet in modifying the impact of air pollution exposure on asthma exacerbations is unknown.

To address these gaps, we investigated the relationships between long-term (48-month) residential exposure to several key ambient air pollutants (PM_2.5, NO₂ and O₃) and asthma exacerbations, and potential effect modification by a plant-based diet, in a national cohort of women from the Nurses’ Health Study II (NHS II). Some of these results have been previously reported in an abstract.

METHODS

Study design

We analyzed data collected in the NHS II cohort, which enrolled 116,430 female registered nurses between 25 and 42 years old in 1989 across 14 states in the United States (U.S.) (Figure 1).³³ Participants received follow-up questionnaires on their health and lifestyle characteristics every other year with response rates of 85–90%.³³ Return of the questionnaires has been determined to imply consent. The study was approved by the institutional review board of Mass General Brigham (Boston, MA).

Figure 1.

Flow diagram of participant selection in the NHS II cohort. Missing data resulted from participant failure to respond. NHS II: Nurses’ Health Study II. PM_2.5: Fine particulate matter; NO₂: Nitrogen dioxide; O₃: Ozone; PDI: Plant-diet Index.

Population

This analysis included women who self-reported a physician-diagnosis of asthma and used asthma medications in the past year from 1997–1998 and 2013–2014. This strategy of identifying asthma was previously validated in this cohort using medical records.³⁴ Participants with missing diet, air pollution or asthma exacerbation data were excluded (Figure 1). All clinical and sociodemographic data were collected through questionnaires. Baseline median household income and home values were calculated using the 2000 U.S. Census tract data. Asthma severity data were available in 1998 and 2014 and similar to previous analyses,³⁵ categorized as intermittent, mild, moderate or severe persistent based on the frequency of reported symptoms in the past four weeks, symptoms between exacerbations, medications used in the past 12 months, and work or usual activity days missed due to asthma.

Ambient air pollution exposure assessment

We estimated time-varying average ambient PM_2.5, NO₂ and O₃ exposure at each participant’s geocoded residential address using validated nationwide spatiotemporal models.^36–38 Prediction models for monthly PM_2.5 from 1988 onward incorporated data from several pollution monitor networks across the U.S. and geospatial predictors. NO₂ and O₃ predictions were estimated using regionalized weekly spatiotemporal models for the contiguous U.S. from 1990 onward.^37,38 Additional methodology on air pollution exposure assessment may be found in the Online Supplement and elsewhere.^36–38 Moving was defined as relocating >5 km away from the previous residence during each survey period (1991 to 2013) and was assumed to occur on June 1 of each biennial questionnaire cycle. This may have resulted in minor exposure misclassifications; hence, a sensitivity analysis was conducted limited to non-movers.

Ambient pollutant exposure was averaged over the 48 months before each outcome assessment (July 1993 to June 1997 for the 1998 outcome period; July 2009 to June 2013 for the 2014 outcome period) to ensure exposures preceded outcomes. A 48-month window was chosen to best reflect long-term exposure without being too distant from the outcome (asthma exacerbation), which is consistent with previous analyses in the NHS cohort,³⁹ and other studies showing health effects of PM_2.5 to be stronger with longer-term exposure up to 48-months compared to 3 months.^40–42 However, in sensitivity analyses, we additionally assessed 24- and 12-month exposure windows prior to each outcome year.

Plant-based diet assessment

Diet data were collected using a validated semi-quantitative food-frequency questionnaire (FFQ) starting in 1991 and every four years thereafter.⁴³ Plant-based diet adherence was assessed using the previously derived PDI score,²⁹ which categorized intake into 18 groups (12 plant-based, 6 animal-based) (Table E1). Consumption of these foods in servings per day was ranked into quintiles and given positive or reverse scores, with plant-based foods receiving positive scores and animal foods reverse scores. Those in the highest quintile of a food group obtained a score of five while those in the lowest quintile obtained a score of one. Scores were summed to generate the PDI score (range: 18–90).²⁹

The healthy PDI (hPDI) and unhealthy PDI (uPDI) were also calculated based on food quality, such that nutritious plant-foods were given positive scores and less nutritious plant-foods given reverse scores for the hPDI, and the converse for the uPDI (Table E1). The PDI, hDPI and uPDI scores were then averaged from 1991 until the first outcome year in 1998 (consisting of data from FFQs administered in 1991 and 1995), and from 1999 (after the first outcome year) to the second outcome year in 2014 (consisting of data from FFQs administered in 1999, 2003, 2007 and 2011). Daily total caloric intake and fruit and vegetable servings were also averaged using the same time frames.

Asthma exacerbations

The main outcome of interest was a dichotomized self-reported asthma exacerbation status (yes or no). Asthma exacerbation was defined as at least one of the following in the past year (1997–1998 or 2013–2014) due to uncontrolled asthma: hospital admission, emergency department or urgent care visit, with two repeated outcome measures (1998 and 2014). Two separate time periods were used for outcome assessment to account for changes in ambient air pollution, dietary patterns and/or asthma management over time and examine if these relationships remain consistent across time in the same cohort.

Covariates and statistical analyses

Covariates were selected a priori and included age, race, physical activity, smoking status and pack year history, asthma severity, body mass index (BMI), total caloric intake, Census region, Census tract median household income and Census tract home value. To ensure that each outcome was modeled with predictors and covariates relevant to its specific time period, covariates were matched to the time points of the outcomes, such that for the 1998 outcome, covariates from 1997 were used and for the 2014 outcome, covariates from 2013 were used.

We described cohort characteristics by asthma exacerbation status and compared included and excluded participants using Wilcoxon two-sample test or Chi-square test for continuous and categorical variables, respectively. Spearman’s correlations were used to evaluate correlations between 1) PM_2.5, NO₂, and O₃ exposures, and 2) average PDI scores between each dietary follow-up year. We then used logistic regression models to assess the impact of each air pollutant on asthma exacerbation with generalized estimating equations to account for repeated measures within subjects. Single and multi-pollutant models were constructed to examine the independent effects of PM_2.5, NO₂ and O₃, both with and without covariate adjustment.

We also evaluated associations between PDI, hPDI, uPDI scores and asthma exacerbation, adjusted for the above covariates and each pollutant. Finally, to assess whether the PDI, hPDI and uPDI scores modified associations between air pollution and asthma exacerbation, we tested a two-way interaction term between PDI, hPDI and uPDI scores and each air pollutant with asthma exacerbation. Statistical significance was determined using a p-value threshold of <0.05. All statistical analyses were conducted using SAS version 9.4.

RESULTS

Population characteristics

Among a total of 4326 women, the median age was 43 years, the cohort was predominantly White (96.7%) and 2877, 66.6% were never smokers. Most participants had intermittent or mild asthma (2930, 67.7%). In 1998, 1312 (30.3%) had an asthma exacerbation, while 1069 (24.7%) had an asthma exacerbation in 2014. In 1998, women who reported an asthma exacerbation in the past year had more severe asthma, significantly lower Census tract median household home value, higher BMI, more weekly physical activity, greater total caloric intake and a greater proportion were current smokers (Table 1). Characteristics of the same participants in 2013 are presented in Table E2. Compared to included participants, more excluded individuals were non-White, reported greater physical activity, had less severe asthma and less exacerbations (Table E3).

Table 1.

Baseline characteristics of women with asthma in the NHS II by asthma exacerbation status in 1998 and 2014

Characteristics in 1997*	Overall (n = 4326)	Asthma exacerbation in 1998		p-value	Asthma exacerbation in 2014		p-value
		No (n = 3014)	Yes (n =1312)		No (n = 3257)	Yes (n = 1069)
Age (years)	43 (39 – 46)	43 (39–46)	43 (40–46)	0.22	43 (39–46)	43 (39–46)	0.53
Race, n (%)							0.97
White	4185 (96.7)	2926 (97.1)	1259 (96.0)		3151 (96.8)	1034 (96.7)
Non-white	141 (3.3)	88 (2.9)	53 (4.0)		106 (3.2)	35 (3.3)
Census region, n (%)
Northeast	784 (18.2)	552 (18.3)	232 (17.7)	0.79	608 (18.7)	176 (16.5)	0.2
Midwest	1337 (31.0)	918 (30.5)	419 (32.0)		986 (30.3)	351 (32.9)
West	677 (15.7)	471 (15.7)	206 (15.7)		502 (15.4)	175 (16.4)
South	1521 (35.2)	1068 (35.5)	453 (34.6)		1155 (35.5)	366 (34.3)
Physical activity (METhours/week)†
Quintile 1	1.3 (0.5–2.3)	1.4 (0.6–2.3)	1.2 (0.4–2.4)	0.01	1.4 (0.5–2.3)	1.2 (0.4–2.3)	0.7
Quintile 2	5.2 (4.2–6.5)	5.2 (4.2–6.5)	5.2 (4.0–6.5)		5.2 (4.2–6.5)	5.3 (4.0–6.5)
Quintile 3	10.9 (9.3–12.7)	10.9 (9.2–12.9)	10.7 (9.5–12.6)		10.9 (9.3–12.9)	10.8 (9.3–12.6)
Quintile 4	20.9 (17.7–24.6)	20.8 (17.7–24.5)	21.0 (17.9–24.7)		20.9 (17.7–24.4)	20.9 (17.8–24.7)
Quintile 5	42.7 (34.6–59.1)	41.8 (34.3–59.6)	43.8 (35.4–58.4)		42.6 (34.7–59.4)	42.7 (33.5–58.7)
Census tract household income ($k)	43.2 (34.7–53.8)	43.6 (34.7–54.2)	42.5 (34.6–52.7)	0.08	43.2 (34.7–54.0)	43.3 (34.5–53.1)	0.34
Census tract home value ($k)	95.1 (65.4–163.1)	96.8 (65.6–167.9)	92.3 (65.1–155.8)	0.04	97.2 (66.2–167.0)	91.3 (64.9–153.6)	0.01
Smoking status, n (%)				0.01			0.61
Never	2877 (66.6)	2011 (66.8)	866 (66.1)		2177 (66.9)	700 (65.5)
Former	1179 (27.3)	838 (27.8)	341 (26.0)		882 (27.1)	297 (27.8)
Current	265(6.1)	162 (5.4)	103 (7.9)		194 (6.0)	71 (6.7)
Pack years‡	10 (5–17)	10 (5–17)	10 (5–18)	0.08	10 (5–17)	10 (5–17)	0.66
BMI (kg/m2), n (%)				<0.01			<0.01
BMI <25	2091 (48.4)	1565 (51.9)	526 (40.2)		1640 (50.4)	451 (42.3)
25 ≤ BMI <29	1123 (25.9)	777 (25.8)	346 (26.4)		822 (25.3)	301 (28.2)
BMI ≥30	1109 (25.7)	672 (22.3)	437 (33.4)		794 (24.4)	315 (29.5)
Asthma severity§, n (%)				<0.01			<0.01
Intermittent	1133 (26.2)	1042 (34.6)	91 (6.9)		961 (29.5)	172 (16.1)
Mild persistent	1797 (41.5)	1240 (41.4)	557 (42.5)		1343 (41.2)	454 (42.5)
Moderate persistent	1120 (25.9)	655 (21.7)	465 (35.4)		785 (24.1)	335 (31.3)
Severe persistent	276 (6.4)	77 (2.6)	199 (15.2)		168 (5.2)	108 (10.1)
48-month PM2.5 (ug/m3)¶	13.7 (11.6–15.7)	13.6 (11.5–15.6)	13.9 (11.7–15.9)	0.01	13.6 (11.6–15.7)	13.8 (11.7–15.8)	0.47
48-month NO2 (ppb)¶	12 (8.2–16.2)	11.7 (8.2–16.0)	12.4 (8.3–16.6)	0.02	11.9 (8.1–16.1)	12.1 (8.5–16.2)	0.39
48-month O3 (ppb)¶	25.5 (23.6–27.4)	25.5 (23.6–27.5)	25.3 (23.5–27.3)	0.24	25.5 (23.6–27.5)	25.3 (23.5–27.3)	0.34
Total caloric intake	1804.5 (1474.0–2174.0)	1792.5 (1464.5–2156.5)	1826.3 (1505.0–2211.3)	0.03	1784.0 (1464.0–2155.0)	1848.5 (1514.0–2229.5)	<0.01
PDI score #				0.07			0.76
Quintile 1	47.0 (45.3–48.5)	47.0 (45.5–48.5)	47.0 (45.0–48.5)		47.0 (45.0–48.5)	47.0 (45.5–48.5)
Quintile 2	51.5 (51.0–52.5)	51.5 (50.5–52.5)	51.5 (51.0–52.5)		51.5 (50.5–52.5)	51.5 (51.0–52.0)
Quintile 3	55.0 (54.0–55.50)	55.0 (54.0–55.5)	55.0 (54.0–55.5)		55.0 (54.0–55.5)	55.0 (54.0–55.5)
Quintile 4	58.0 (57.0–58.50)	58.0 (57.0–58.5)	58.0 (57.0–58.5)		58.0 (57.0–58.5)	58.0 (57.5–59.0)
Quintile 5	62.5 (61.0–64.5)	62.5 (61.0–64.50)	62.0 (61.0–64.5)		62.5 (61.0–64.5)	62.5 (61.0–64.5)

^*Values expressed as median (interquartile range), unless otherwise specified. Missing data: n = 7 for geographic region; n = 10 for Census tract household income and home value; n = 5 for smoking status; n = 3 for BMI; n = 309 for total caloric intake.

^†Physical activity was captured by deriving a MET·hours/week score from the time spent per week performing various physical activities and multiplied by the typical energy expenditure requirements in METs, then categorized into quintiles.

^‡Calculated among both former and current smokers.

^§Derived from the asthma questionnaire in 1998.

^¶Values represent an average of 48 months (from July 1993-June 1997) prior to the 1998 asthma questionnaire.

^#PDI score represents the cumulative average from 1991–1997 (which includes two food frequency questionnaires in 1991 and 1995).

MET: Metabolic equivalent; BMI: Body mass index; PDI: Plant diet index.

Ambient air pollution and asthma exacerbation

Median (IQR) PM_2.5, NO₂ and O₃ concentrations in the 48-months before 1998 and 2014 were 13.7 (11.6–15.7) ug/m³, 12.0 (8.2–16.2) ppb and 25.5 (23.6–27.4) ppb, respectively, and 8.9 (7.7–10.2) ug/m³, 6.6 (4.8–8.8) ppb and 27.8 (26.3–29.5) ppb, respectively (Figure 2). Concentrations for NO₂ and O₃ were well below current U.S. Environmental Protection Agency (EPA) annual standards of 53 and 70 ppb, respectively, in both time periods. Relative to the current U.S. EPA annual standards of 9 ug/m³, PM_2.5 concentrations were higher in the 48-months before 1998, but lower in the 48-months before 2014 (Figure 2).^44–46 NO₂ and PM_2.5 concentrations, however, were still generally above the updated 2021 World Health Organization annual standards for NO₂ and PM_2.5 of 10 ppb and 5 ug/mg³, respectively.⁴⁷ Correlations within each pollutant between the two time periods and between PM_2.5 and NO₂ were positive and moderate to strong. PM_2.5 concentrations were weakly negatively correlated with O₃ in both time periods (Table E4), which may reflect higher PM_2.5 levels blocking sunlight needed for O₃ formation in colder weather or other geographic factors.⁴⁸

Figure 2.

Median (IQR) concentrations of ambient air pollutants averaged from July 1993 to June 1997 and July 2009 to June 2013, relative to the 2024 United States Environmental Protection Agency annual standards (red dashed line).

Higher 48-month average exposure to both ambient PM_2.5 and NO₂ was associated with significantly higher odds of having an asthma exacerbation in the past year in adjusted single pollutant models (OR 1.43; 95% CI 1.14–1.80, and OR 1.25; 95% CI 1.12–1.38, respectively), while there was no relationship between O₃ exposure and odds of asthma exacerbation (OR 0.85; 95% CI 0.70–1.04) (Table 2). In adjusted multipollutant models, higher exposure to NO₂ was associated with significantly greater risk of asthma exacerbation (OR 1.23; 95% CI 1.06–1.42), while there were no statistically significant relationships with PM_2.5 and O₃ (OR 1.15; 95% CI 0.88–1.51, and OR 1.07; 95% CI 0.85–1.36), respectively) (Table 2). In sensitivity analyses using 24- or 12-month time periods, these relationships remained robust (Table E5), likely as concentrations of pollutants did not vary much in the 12–48 months prior to each outcome (Table E6). After excluding women who moved at any point in this study (n = 2612, 60.4%), there were no significant changes in relationships between PM_2.5 and NO₂ and odds of asthma exacerbation in adjusted single-pollutant models, but associations between NO₂ and asthma exacerbations in the adjusted multi-pollutant model were no longer statistically significant, likely due to loss of power (Table E7).

Table 2.

Associations between 48-month average ambient air pollutants and odds of asthma exacerbation in the past year

Ambient air pollutant	Single pollutant models*				Multi-pollutant models†
	Unadjusted OR‡	95% CI	Adjusted OR‡	95% CI	Unadjusted OR‡	95% CI	Adjusted OR‡	95% CI
PM2.5	1.61	1.41–1.84	1.43	1.14–1.80	1.51	1.27–1.81	1.15	0.88–1.51
NO2	1.23	1.14–1.32	1.25	1.12–1.38	1.08	0.96–1.22	1.23	1.06–1.42
O3	0.78	0.67–0.91	0.85	0.70–1.04	1.07	0.88–1.31	1.07	0.85–1.36

^*Single pollutant models adjusted for age, race, geographic region, physical activity, Census tract median household income, Census tract median home value, asthma severity, smoking status, pack years, body mass index, total caloric intake and plant diet index score.

^†Multi-pollutant models adjusted for the above covariates and ambient air pollutants.

^‡OR represents changes in odds for every 10 μg/m³ increase in 48-month PM_2.5, or 10 ppb increase in 48-month NO₂ or O₃.

Plant-diet index score, air pollution and asthma exacerbation

Median PDI scores ranged from 47.0 in quintile one to 62.5 in quintile five in 1997 and did not substantially change for the same participants in 2013 (Table E2). Correlations in average PDI scores between years were moderate to strong (Table E8). The median daily servings of fruit and vegetables consumed among those in quintile one compared to five were 0.64 and 1.88, and 1.67 and 3.81, respectively. There were no significant associations between the PDI, hPDI and uPDI scores (quintile five compared to one) and odds of having an asthma exacerbation (OR 0.98; 95% CI 0.81–1.19; OR 1.03; 95% CI 0.85–1.24; OR 0.91; 95% CI 0.75–1.10, respectively) in adjusted models (Table E9). There were also no statistically significant interactions between each of PM_2.5, NO₂ and O_3, and PDI score with asthma exacerbation (p_int = 0.69, p_int = 0.29, and p_int = 0.60, respectively), or with hPDI or uPDI scores (data not shown).

DISCUSSION

In this secondary analysis of a multi-center longitudinal cohort study, we found that long-term exposures to even low levels of ambient air pollution over 48 months across two time periods were associated with increased odds of asthma exacerbations among adult women. Specifically, in single pollutant models, both PM_2.5 and NO₂ were significantly associated with greater asthma exacerbation risk, whereas in multi-pollutant models, only NO₂ was significantly associated with greater asthma exacerbation risk. This may be due to collinearity between PM_2.5 and NO₂ or exposure to NO₂ as the main driver of the increased risk of asthma exacerbations. Additionally, most women in our study had only mild asthma, highlighting air pollution as an important risk factor for asthma exacerbations even among those with mild disease. Greater adherence to a plant-based diet did not significantly attenuate these associations, and further study in larger populations are needed to identify if other modifiable, individual-level factors may be protective against long-term air pollution exposure. In the meantime, our findings may help to inform public policies centered on regulating ambient air pollution exposures and reinforce the ongoing need to champion clean air policies, including revisiting U.S. EPA annual standards, particularly for NO₂.

Recently, there has been growing attention to the number and intensity of wildfires in the U.S. and worldwide, and the impact of climate change on air quality. Wildfires can significantly elevate ambient PM_2.5 and NO₂ concentrations, among other pollutants, resulting in substantially worse air quality.⁴⁹ Short-term exposure to ambient air pollution may worsen asthma control in adults;^{13–15,49–52} in contrast, few studies have explored associations between long-term air pollution exposure over years – typically at much lower concentrations than short-term exposures over hours to days – and asthma exacerbation risk. One recent special report noted only three European studies linking long-term exposure to traffic-related NO₂ with greater asthma exacerbations, although average NO₂ levels were higher than in our cohort.¹⁸ In this study, our observation that higher 48-month average exposure to ambient NO₂ and PM_2.5 was associated with significantly greater odds of asthma exacerbation in the past year remained robust even when 24- and 12-month exposure windows were considered. Notably, average ambient concentrations of NO₂ and PM_2.5 were both relatively low in the 48-months prior to 1998 (median of 12.0 ppb and 13.7 ug/m³, respectively) and 2014 (6.6 ppb and 8.9 ug/m³, respectively), highlighting potentially negative consequences of long-term exposure to even low concentrations of ambient pollutants. However, we did not observe a relationship between low levels of 48-month O₃ exposure and asthma exacerbations in this cohort. Other studies have shown an association between O₃ and asthma exacerbation in children and adult-onset asthma incidence only among those with higher exposure levels.^53,54

While advocacy for improving air quality remains paramount, modifiable factors, such as nutrient and dietary patterns, represent attractive targets to improve respiratory health at an individual-level.⁵⁵ There is existing precedence for the role of diet as a potential modifier of the asthmatic response to the inhaled environment. One cohort study in children found that higher omega-3 fatty acid intake was protective against the effects of indoor PM_2.5 on asthma symptoms,²² while another showed that low vitamin D levels potentiated the adverse effects of indoor PM_2.5 on asthma symptoms in children.²¹ A healthy diet – typically involving more consumption of fruits and vegetables and less animal-based products – also appeared protective against wheezing in adolescents with substantial secondhand smoke exposure.^56,57 Putative mechanisms of benefit include systemic anti-inflammatory effects derived from fiber, carotenoids and other antioxidants,⁵⁵ and reduced exposure to pro-inflammatory mediators from animal-products and red and processed meats, which have been linked with increased asthma morbidity.⁵⁷ Despite this body of literature, in our cohort, consumption of a plant-based diet did not significantly modify the relationship between ambient air pollution and asthma exacerbation. This may be due to inadequate consumption of plant-based foods to offset the harms of air pollution exposure, even though participants with the highest intake of plant-based foods consumed median daily servings of fruits and vegetables of 1.67 and 3.81, respectively, which is in keeping with dietary guidelines.⁵⁸ Alternatively, it is possible that a plant-based diet may have differential effect modification by asthma severity, such that women with mild asthma (as predominantly represented in this cohort) may derive less benefits compared to those with severe asthma. Although we did not observe significant protective effects with intake of a plant-based diet, further work adopting an exposomic lens to account for the totality of environmental exposures, including diet, is warranted.

This study has several strengths. Availability of repeated exposure and outcome data over two distinct time periods allowed us to examine longitudinal effects of dietary patterns and ambient air pollution on asthma exacerbations and account for variations within person over time. This enhanced the robustness in our evaluation of relationships between PM_2.5 and NO₂ and asthma exacerbations, despite potential changes in concentrations of air pollution exposure, dietary habits and asthma management over time. We also investigated the influence of not only individual key air pollutants, but also aggregate effects in multi-pollutant models, further strengthening our findings.

Some limitations must be noted. First, this cohort enrolled only female nurses who were predominantly white, limiting generalizability to other women of color who may have different air pollution exposures and dietary habits. Second, over a third of women with asthma in the NHS II were excluded in this study due to missing data. A greater number of excluded individuals were non-white, more physically active and had less severe asthma, and it is unclear if the effects of ambient air pollution on asthma exacerbations extends to this group. Third, the study design of the NHS II relies predominantly on self-report, which carries inherent recall bias with using a FFQ to document dietary habits and introduces potential misclassification bias for identifying participants with asthma (although this approach has been previously validated in this cohort).³⁴ Fourth, the PDI score cannot be used to distinguish a whole-foods plant-based diet from a plant-based diet high in ultra-processed foods, which may account for the lack of significant interaction with air pollutants and asthma exacerbation. Fifth, most women reported moving residences, which may have caused some inaccuracies in capturing air pollution exposures. However, in sensitivity analyses excluding those who moved, results were not substantially different, although statistical significance was lost likely due to reduced power. Finally, ambient air pollution estimates are based on participants’ residential address and does not account for personal, indoor or ambient inhaled pollutants in places away from home. Air pollution data was limited to NO₂, PM_2.5 and O₃ and did not include other key air pollutants such as sulfur dioxide or carbon monoxide. There was also lack of data on secondhand smoke exposure. Yet, our findings still provide an important basis for the understanding of the long-term impact of inhaled ambient pollutants on asthma health in adult women.

CONCLUSIONS

In this longitudinal cohort study of women with asthma, we observed that exposure to even relatively low concentrations of NO₂ and PM_2.5 was associated with greater asthma exacerbation risk, although the consumption of a plant-based diet did not attenuate these relationships. While further studies are needed, our results add to growing evidence that supports lowering acceptable standards for key air pollutants, and highlights the need to identify personal, modifiable strategies of increasing resilience among susceptible populations, including women with asthma.