Short-term exposure to relative humidity and lung health in early adolescents

Abstract

Background:

Extremes in humidity can induce bronchoconstriction and trigger breathing symptoms in people with asthma. Less is known about how humidity influences measurements of lung health in children and adolescents. Our objective was to assess the extent to which short-term exposures to high and low relative humidity (RH) are associated with lung function and fractional exhaled nitric oxide (FeNO) in adolescents.

Methods:

We included adolescents (mean age 13.2 y, SD: 0.9) from a northeast US prospective prebirth cohort (n = 1019). We assigned daily RH levels to geocoded participant addresses. We defined low or high RH as ≤10^th or ≥90^th internal percentiles, respectively, of the cohort-specific RH distribution and the reference RH as the median. We evaluated the linearity of associations of RH in the 1–7 days before assessment with forced expiratory volume in 1 s (FEV₁), forced vital capacity (FVC), and FeNO using generalized additive models with penalized splines (df = 3). We log-transformed FeNO due to non-normality. For nonlinear relationships, we used distributed lag nonlinear models to explore the cumulative effects of lag 1–7 day RH on FEV₁, FVC, and FeNO.

Results:

Median RH was 65.6% (interquartile range [IQR] = 19.8%), 10th percentile 47.2%, 90th percentile 86.6%. Mean FeNO (SD) was 25.9ppb (26.9ppb). High (vs. median) RH was associated with 38.0% higher FeNO (95% CI = 10.3, 72.7). Exposure to low (vs. median) RH was associated with 186.2 ml lower FEV₁ (95% CI = −299.2, −73.3) and −130.2 ml lower FVC (95% CI = −251.9, −8.5).

Conclusion:

Short-term exposures to extremes of RH were associated with lower lung function and higher FeNO, a measure of airway inflammation, in adolescents.

What this study adds

Humidity is a commonly reported trigger for breathing symptoms in people with asthma, yet less is known about the effects of humidity on lung function and airway inflammation in the general population. We used a well-characterized cohort of early adolescents with high-resolution humidity and pollution data and found that short-term exposure to extremes in relative humidity is associated with lower lung function and higher airway inflammation. Our findings suggest that extremes in humidity may have implications for adolescent lung health.

Introduction

Humidity is a commonly reported trigger for breathing symptoms in people with asthma¹ and extremes of humidity have been shown to cause bronchoconstriction.^2,3 With climate change, extreme weather events, such as droughts and heavy precipitation events, are becoming more frequent and intense,⁴ causing extremes in humidity.⁵ Humidity may exert effects on the respiratory system through action as an irritant on vagal receptors in the airway and by altering the airway mucosa and the rheologic properties of bronchial mucus, which serve important roles in regulating airway inflammation.^2,6–8

Less is known about the effects of humidity on lung function and airway inflammation. Relative humidity is a measure of the amount of water vapor in an air-water mixture compared with the maximum amount possible for a given temperature. Among the limited number of prior studies on relative humidity and lung health, the findings are not conclusive.^3,6,9–11 In a German cohort of adolescents, short-term exposure to higher relative humidity showed trends toward lower airway inflammation, though did not reach significance.¹⁰ A controlled exposure study in healthy, nonsmoking young adults without respiratory disease found lower lung function in dry (25% relative humidity) versus ambient relative humidity (50% relative humidity), but not in humid (90% relative humidity) versus ambient air, holding temperature constant.⁹ Our aim was to assess the extent to which short-term exposure to high or low relative humidity is associated with lung function and eosinophilic airway inflammation (as assessed by fractional exhaled nitric oxide; FeNO), in early adolescents living in the northeast United States (US).

Methods

Population

We included early adolescents from Project Viva, a prebirth cohort. Mother-child pairs were enrolled during pregnancy between 1999 and 2002 at a large Massachusetts-based group medical practice (Atrius Harvard Vanguard Medical Associates) and participated in six study visits, including one during their early adolescent years. Detailed enrollment criteria have been described previously.¹² Of 2128 infants enrolled in Project Viva, we included 1019 participants seen at the early adolescent study visit between 2011 and 2016 with exposure data and any outcome data. We selected the early adolescent study visit because it is the first to measure both spirometry and FeNO. Study participants who moved outside of Massachusetts during the course of the study were followed, and their exposure data were updated based on their new address.

We obtained medical history and demographics from questionnaires and interviews with mothers. We asked if the adolescent had a physician diagnosis of asthma and defined current asthma as asthma diagnosis at the time of study visit in addition to wheezing or asthma medication use in the prior year based on parental reports on questionnaires. We defined household tobacco use as anyone currently smoking tobacco within the participant’s home. We categorized race and ethnicity based on maternal reports as non-Hispanic Asian, non-Hispanic Black, non-Hispanic White, Hispanic, or other, which includes responses by mothers identifying multiple races or ethnicities. We used annual household income (≤USD$70,000 vs. >USD$70,000) and maternal education collected at study enrollment (dichotomized to either having or not having a college education or higher) as markers of socioeconomic status. We measured height and weight at study visits.

Allergen extract-specific IgE antibodies were measured with ImmunoCap (Thermo Fisher Scientific/Phadia, Kalamazoo, Michigan) in a subset of participants at the early adolescent study visit.¹³ We defined aeroallergen sensitization as any IgE >0.35 IU/ml against common indoor and outdoor allergens in the northeast US—Aspergillus fumigatus (mold), Alternaria alternata (plant fungi), Betula pendula (silver birch tree), Quercus (oak tree), Lolium (ryegrass), and Ambrosia artemisiifolia (common ragweed), Dermatophagoides farina (dust mite), cat dander, or dog dander similar to prior studies.^14,15

All mothers provided written informed consent, and adolescents provided verbal assent. The study was approved by the institutional review boards of Beth Israel Deaconess Medical Center and Harvard Pilgrim Health Care.

Exposure assessment

Spatially and temporally resolved daily relative humidity, absolute humidity, and temperature measurements are based on the 800-m resolution Oregon State University Parameter-elevation Relationships on Independent Slopes Model (PRISM) Climate dataset.¹⁶ For analyses adjusting for fine particulate matter (PM_2.5) and ozone (O₃) in the same 7-day exposure window, we used modeled daily air pollution data generated using machine learning algorithms developed at the Harvard School of Public Health based on neural network, gradient boosting, and random forest with integrated satellite data, land-use data, and chemical transport model outputs with 1 by 1-km resolution.^17,18

We geocoded participant residential addresses at the early adolescent study visit using ArcGIS (ESRI, Redlands, California) and linked relative humidity, absolute humidity, temperature, and air pollution data to geocoded addresses and the study visit date, similar to prior studies.¹⁴

Outcome assessment

We measured FeNO with a hand-held electrochemical device (NIOX MINO) validated by the clinical standard chemiluminescence FeNO analyzer.¹⁹ Trained research assistants instructed participants to inhale through a nitric oxide (NO) scrubbing filter on two efforts to avoid ambient NO measurement and exhale into room air. On the third effort, participants exhaled into the FeNO analyzer. Only the last three seconds of exhalation were used to ensure lower airway NO measurements. Nose clips were not used. We defined FeNO as the mean of two efforts.

In accordance with the American Thoracic Society guidelines for acceptability and reproducibility, trained research assistants measured forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV₁) using the EasyOne Spirometer (NDD Medical Technologies, Andover, Massachusetts).²⁰ We required participants to produce at least three acceptable spirograms, two of which had to meet criteria for reproducibility defined as a ≤0.150 L difference between the two largest FEV₁ values and the two largest FVC values, for inclusion.

Statistical analysis

Potential confounders and predictors of outcome were selected a priori based on published and anticipated associations of relative humidity with lung function and FeNO.^10,21 We log-transformed FeNO due to non-normality, consistent with prior published work by our group.^14,15 We evaluated the linearity of associations of relative humidity with FEV₁, FVC, and FeNO using generalized additive models with penalized splines (df = 3), and found there were nonlinear exposure-response functions (eFigure 1; http://links.lww.com/EE/A326). Therefore, for our primary analyses, we used generalized additive models with distributed lag nonlinear models to evaluate the cumulative effects of relative humidity on FEV₁, FVC, and FeNO at lag 1–7 days (eFigure 2; http://links.lww.com/EE/A326). We defined low relative humidity as the 10th percentile and high relative humidity as the 90th percentile relative to the internal distribution of our study population. We report associations of the cumulative exposure of low or high relative humidity across the 7 days preceding the study visit (lag 1–7 days) with study outcomes compared to the median relative humidity level, defined as the median of the lag 1–7 day relative humidity. For log-transformed FeNO, effect estimates are presented as percent changes of the geometric outcome mean with 95% confidence intervals (CIs). For lung function, effect estimates are expressed as absolute differences in lung function with their 95% CIs. We adjusted for sex, age, height, weight, current asthma, current household tobacco use, household income, maternal education, study visit date, season as the sine and cosine of visit date, and mean temperature.

We tested for effect modification by current asthma diagnosis (yes or no), season (warm season defined as March to August and cool season defined as September to February), sex (male or female), mean temperature (defined as high and low based on median 7-day moving averages), PM_2.5 and O₃ (defined as high and low based on median 7-day moving averages). We also tested for effect modification by aeroallergen sensitization (yes or no) for associations of relative humidity with FeNO based on prior work by our group.¹⁵

We tested for correlation between relative humidity and air pollutants (O₃ and PM_2.5) and performed sensitivity analyses additionally adjusting for O₃ and PM_2.5, averaged across the same 7-day exposure time-window as relative humidity. In additional sensitivity analyses, we also evaluated associations using alternative definitions of “extreme,” with more extreme cutoffs of 2.5th and 5th percentiles for extreme low relative humidity, and 95th and 97.5th percentiles for extreme high relative humidity.

In secondary analyses, we also tested for associations of absolute humidity with lung function and FeNO using the same definitions we used for low and high relative humidity, specifically, the 10th percentile and 90th percentile, respectively, relative to the internal distribution of our study population. We then report associations of the cumulative exposure of low or high absolute humidity across the 7 days preceding the study visit (lag 1–7 days) with study outcomes compared to the median of the lag 1–7 day absolute humidity level.

All analyses were performed using R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). We used the dlnm and gam packages for analyses.

Results

The characteristics of the 1019 early adolescents included in this study are summarized in Table 1. Early adolescents (mean age 13.2 y, SD: 0.9) were evenly balanced by sex and were less than two thirds non-Hispanic White with a greater proportion of non-Hispanic Black participants (16.1%) relative to Massachusetts demographics. Around 84% of participants had home addresses in Massachusetts and 91.3% in the Northeast US, as defined by the US Census Bureau. The mean FeNO (SD) was 25.9 ppb (26.9 ppb), which, for reference, is below the upper limit of normal for FeNO (upper limit of normal is 35 ppb for ages under 12 and 50 ppb for ages 12 and older).²²

Table 1.

Characteristics of early adolescent study participants

Characteristics (n = 1019)	Mean ± SD or %
Female	49.7%
Age (years)	13.2 ± 0.9
Height (centimeters)	159.9 ± 9.0
Weight (kilograms)	54.1 ± 14.8
Race and ethnicity
Non-Hispanic Asian	2.8%
Non-Hispanic Black	16.1%
Hispanic	4.5%
Non-Hispanic White	64.4%
Other or >1 race or ethnicity	12.1%
Current asthma	13.1%
Any aeroallergen sensitizationa	36.3%
Any smoking in the home	11.3%
Maternal college degree	71.4%
Annual household income >USD$70,000	58.7%
Spirometry
FEV1 (L)	2.8 ± 0.6
FVC (L)	3.3 ± 0.7
FeNO (parts per billion)	25.9 ± 26.9

^aAeroallergen sensitization is defined as having IgE against common indoor and outdoor allergens using a cutoff of 0.35 IU/ml; 381 participants did not have aeroallergen data.

FeNO, fractional exhaled nitric oxide; FEV₁, forced expiratory volume in 1 second; FVC, forced vital capacity; PM_2.5, fine particulate matter with a mean aerodynamic diameter less than or equal to 2.5 microns.

The median (IQR) 7-day moving average relative humidity was 65.6% (19.8%), which was the reference relative humidity (Table 2). The 10th percentile for relative humidity for lag 1–7 days preceding the study visit was 47.2% and the 90th percentile was 86.6%. There was a weak negative correlation between relative humidity and air pollution (r = −0.06 for PM_2.5, r = −0.29 for O₃) (eFigure 3; http://links.lww.com/EE/A326).

Table 2.

Daily mean relative humidity and temperature and air pollution at lag 1–7 days for early adolescents

Exposure	Median (IQR)	10th Percentile	90th Percentile
Relative humidity (%)	65.6 (19.8)	47.2	86.6
Absolute humidity (g/m3)	7.5 (8.2)	2.3	16.1
Temperature (°C)	13.2 (17.4)	−3.1	24.5
Air pollutiona
PM2.5 (µ/m3)	6.7 (3.1)	4.5	10.6
Ozone (ppb)	39.5 (9.2)	32.0	47.5

^aAir pollution based on moving average of lag 1–7 days.

IQR, interquartile range; PM_2.5, fine particulate matter with a mean aerodynamic diameter less than or equal to 2.5 microns; ppb, parts per billion; O₃, ozone; µ/m³, microns per cubic meter.

Associations of low and high relative humidity with FeNO and lung function are shown in Figure 1. With an increase in relative humidity from the median to the 90th percentile relative humidity, FeNO was 38.0% higher (95% CI = 10.3, 72.7). With a decrease in relative humidity from the median to the 10th percentile, FeNO was 23.1% higher (95% CI = 0.7, 50.4). Short-term exposure to low relative humidity was associated with 186.2 ml lower FEV₁ (95% CI = −299.2, −73.3) and 130.2 ml lower FVC (95% CI = −251.9, −8.5), relative to the median. We did not find associations of high relative humidity with FEV₁ or FVC, which were −7.8 ml (95% CI = −136.1, 120.4) and 43.2 ml (95% CI = −95.0, 181.4), respectively.

Figure 1.

Percent difference (95% confidence intervals) in FeNO (panel A) and absolute differences in FEV₁ (panel B) and FVC (panel C) at high and low relative humidity (lag 1–7 days) compared to the reference median relative humidity. Models adjusted for sex, age, height, weight, current asthma, household tobacco use and income, maternal education, study visit date and season, and mean temperature in the same exposure window. FeNO, fractional exhaled nitric oxide; FVC, forced vital capacity; FEV₁, forced expiratory volume in 1s.

Effect modification analyses

Males showed more negative associations of high and low relative humidity with FVC and FEV₁, compared to females (P_interactions <0.05) (Figure 2 and eTable 3; http://links.lww.com/EE/A326). At high relative humidity, males had 172.7 ml lower FVC (95% CI = −388.7, 43.3), while females had 258.3 ml higher FVC (95% CI = 84.5, 432.2) (P_interaction <0.01).

Figure 2.

Absolute differences in lung function at low relative humidity (Panel A) and high relative humidity (Panel B) compared to the reference median relative humidity stratified by sex (P_interaction<0.05). Models adjusted for age, height, weight, current asthma, household tobacco use and income, maternal education, study visit date and season, and mean temperature in the same exposure window. Whiskers represent 95% confidence intervals. FVC, forced vital capacity; FEV₁, forced expiratory volume in 1s.

Season (warm vs cold) did not modify associations of relative humidity with FeNO or lung function (eTable 4; http://links.lww.com/EE/A326). Associations of low and high relative humidity with FeNO and lung function did not vary by air pollution (PM_2.5, O₃), temperature, or asthma diagnosis (eTables 5−8; http://links.lww.com/EE/A326). Similarly, associations of relative humidity with FeNO, a marker of allergic airway inflammation, did not differ by aeroallergen sensitivity (eTable 9; http://links.lww.com/EE/A326).

Sensitivity analysis

Results remained robust in sensitivity analyses using different definitions for “extreme” relative humidity, with more extreme relative humidity (2.5th, 5th, 95th, and 97.5th percentiles) associated with larger differences in FeNO and lung function. In sensitivity analyses additionally adjusting for PM_2.5 and O₃ air pollution exposure in the same exposure window, associations were similar in direction and magnitude (eTables 1 and 2; http://links.lww.com/EE/A326).

Secondary analysis

Similar to our findings that low relative humidity was associated with lower FEV₁, we found that low absolute humidity was associated with −13.7 ml lower FEV₁ (98% CI = −25.3, −2.1) relative to median (eTable 10; http://links.lww.com/EE/A326). Associations of absolute humidity with FVC were similar in direction to associations of relative humidity with FVC. Unlike relative humidity, we did not find associations of absolute humidity with FeNO.

Discussion

In an unselected cohort of early adolescents in the northeast US, exposure to high and low relative humidity over the preceding 7 days was associated with higher airway inflammation, and low relative humidity was associated with lower lung function. The differences in lung function are the equivalent of approximately 4–6 years of aging.²³ While short-term exposure to humidity is more often considered in people with asthma, we found differences in spirometry and FeNO regardless of asthma diagnosis. When using more “extreme” definitions for low and high relative humidity, we found similar direction and magnitude in our results.

There are few prior studies on relative humidity and lung function and airway inflammation. A recent study in a German cohort of adolescents (mean age 15.1 years) found a linear relationship between prior day exposure to relative humidity and FeNO, but overall, did not find significant associations between relative humidity and FeNO (0.01% lower FeNO per 5% increase in relative humidity at lag 1 day [95% CI = −0.02, 0.00]).¹⁰ In the normative aging cohort composed of elderly adults in the northeast US, higher relative humidity over the preceding 6 days was associated with lower lung function.²¹ Unlike the German cohort and the normative aging cohort studies, which found linear relationships between relative humidity and lung function and airway inflammation, we found U-shaped associations between lung function and airway inflammation. A similar U-shaped pattern was found for respiratory symptoms in prior studies—both low and high extremes in relative humidity are associated with increased respiratory symptoms.²⁴

FeNO and spirometry (FEV₁ and FVC) measure different aspects of respiratory physiology—FeNO is a marker of eosinophilic airway inflammation, while spirometry measures the volume of air exhaled as a function of time and can reveal obstructive and restrictive limitations. Evaluating both FeNO and lung function outcomes provides a more comprehensive look at the potential effects of humidity on lung health. Low and high relative humidity may influence lung function and airway inflammation through multiple pathways, including creating environmental conditions favorable to known asthma triggers (e.g., respiratory pathogens, aeroallergens) and decreasing host airway defenses (e.g., impaired mucociliary clearance).²⁵ Compared with human airway epithelial cells at 69% relative humidity, cells subjected to low relative humidity (20%) and high relative humidity (90%) demonstrate higher transepithelial resistance, a measure of functional intracellular tight junction formation and integrity, and lower cell viability.²⁶ This suggests that extremes of relative humidity can disrupt airway epithelial membrane and induce cell damage. In a chamber study exposing participants to different extremes of humidity for only 6 hours, young adults with asthma had decreases in FEV₁ at extremely low relative humidity (10–15%), regardless of whether the air was cold or warm.¹¹ This chamber study, and others,⁹ also supports our findings that relative humidity exerts an influence on lung function regardless of mean ambient temperature.

While our primary exposure was relative humidity, studies on absolute humidity exposure in children with exercise-induced bronchoconstriction (EIB) have found that low absolute humidity, but not relative humidity, is associated with increases in airway resistance measured by impulse oscillometry,²⁷ suggesting that absolute humidity may be a more relevant exposure in relation to airway resistance in EIB. While our study outcomes, spirometry, and FeNO, represent measurements of lung health distinct from impulse oscillometry, we similarly found that low absolute humidity was associated with lower lung function, suggesting an additional pathway by which absolute humidity may exert effects on the airways, even in adolescents without EIB. Prior studies, particularly in people with EIB, have shown that exposure to dry air can lead to obstruction on spirometry as the inspired air is humidified by water vapor from the bronchial epithelial lining, thereby drying the lining, which can cause cilia loss, acute inflammation, and vascular congestion.²⁸ Further, the inspired air is warmed to body temperature, which causes airway cooling followed by rapid airway warming.^27,29 On the other hand, breathing humidified air reduces the movement of water and heat from airway mucosa and thus limits airway obstruction.²⁹ While EIB is a distinct disease from asthma that shares common symptoms, we found that exposure to low relative humidity was associated with lower lung function regardless of asthma diagnosis. In a chamber study of adolescents aged 10.7 years (SD: 2.3) with asthma, exercising in low relative humidity (25%) was associated with lower lung function, while exercising in 90% relative humidity showed no association.² Cold dry air challenges had been used to test for bronchial hyperresponsiveness by FEV₁; however, these challenges typically use air cooled to −10°C with very low relative humidity, conditions that were not present in our study population.

The extent to which temperature and absolute humidity affect respiratory pathogens is unclear, but interestingly, many known respiratory pathogens (e.g., bacteria, fungi, viruses, and mites) exhibit an optimum zone of relative humidity, with a U-shape relationship similar to that of relative humidity and our study outcomes.^30,31 At relative humidity below 50%, aerosols may stay suspended in the air as their size decreases from rapid evaporation of water content, leading to an increase in aerosol abundance, some of which may be harmful to respiratory health, such as airborne pathogens.³⁰ Pollen concentrations also increase with lower relative humidity.³² Common respiratory viruses have a predilection for low or high relative humidity; adenoviruses prefer relative humidity above 70% while influenza prefers relative humidity below 50%.³⁰ High relative humidity, typically above 70–75%, is also required for most fungi and mite growth.³⁰ Indeed, high humidity may be the most important environmental factor influencing dust-mite growth.³³ Relative humidity also influences air chemistry; increases in the concentration of formaldehyde, a water-soluble gas, are directly proportional to increases in relative humidity.³⁰ Formaldehyde upregulates Th2 cytokines and induces airway eosinophil infiltration in mouse models³⁴ and, in human studies, has been associated with lower lung function.³⁵ Formaldehyde is generally considered an indoor air pollutant, though there is increasing emphasis on contributions to outdoor air concentrations.³⁶ The complex air chemistry dictating changes in concentrations of aerosols, gases, and aeroallergens in relation to changes in relative humidity points to the complexity of understanding mechanisms driving associations with FeNO and lung function. The overlapping U-shaped relationships of relative humidity with plausible mediators and our outcomes is intriguing and warrants further mechanistic studies.

We found that males had stronger negative associations of relative humidity with lung function compared with females at both extremes of relative humidity. In children aged 5–6 years, higher weekly indoor relative humidity was associated with a higher prevalence of respiratory symptoms (e.g., cough, chest colds, asthma exacerbations, or bronchitis) in males, while a similar association was not found in females.³⁷ In prior work in this cohort of adolescents, we similarly found differences by sex for associations of short-term rain exposure with lung function, with lower lung function in males compared with females.¹⁴ These sex differences may be related to changes in sex hormones, which can affect airway resistance and inflammation, as well as muscle-fat ratios, which together can alter lung function.^38,39 Other studies have suggested that time spent outside may differ between male and female children, with boys spending more time outside than girls.⁴⁰ If true, differences in exposure by sex may explain our findings with lung function. We did not find that sex-modified association of relative humidity with FeNO, similar to a German cohort study in adolescents.¹⁰

As extreme weather events, from flooding and heavy rainfall to droughts and heatwaves, become more common as a result of climate change and affect relative humidity levels, understanding the effects of high and low relative humidity on lung health is essential for informing public health recommendations for relative humidity and developing interventions to reduce exposure.

This study has multiple strengths. We utilize a well-characterized cohort study with high-resolution exposure estimates, including both weather and air pollution concentrations modeled to home addresses rather than derived from a central site monitor. Another strength is the use of a relatively broad pulmonary outcome measurement to characterize lung health. The use of time-varying distributed lag nonlinear models to evaluate the complex nonlinear associations is another strength. We define relative humidity percentiles based on the distribution within our study population, though an alternative approach is to define percentiles based on historical data specific to each grid cell and date. There are several limitations to our study related to the challenging nature of estimating personal environmental exposure. Adolescents are unlikely to spend all their time at their home address, to which we have linked exposure data. However, we would anticipate similar outdoor relative humidity at nearby locations, such as schools, where adolescents may spend time. Further, outdoor relative humidity, our study exposure, may not reflect the indoor environment where adolescents spend time. In one study, the indoor-to-outdoor correlation was weak (r = 0.55, β = 0.39) while the indoor-to-outdoor correlation for absolute humidity was much stronger (r = 0.96, β = 0.69).⁴¹ This study highlights the potential impact of outdoor relative humidity on lung health, rather than indoor relative humidity, where environmental controls, such as air conditioning and humidifier use may alter indoor RH levels. Some have argued that additional measurements of humidity should be considered in environmental epidemiology, such as absolute humidity for understanding the effects of atmospheric moisture or wet-bulb globe temperature for understanding human comfort and evaporative cooling in warm environments.⁴² Relative humidity remains the most common humidity variable in environmental epidemiology and is useful for understanding dampness.⁴²

Conclusion

Short-term exposures to high and low relative humidity were associated with higher FeNO, a measure of eosinophilic airway inflammation, and short-term exposures to low relative humidity were associated with lower lung function, in adolescents. These findings suggest future research evaluating the impact of climate variability on health should consider the respiratory health implications of humidity.

Conflicts of interest statement

N.J.N. reports receiving grants from the NIH and Eleanor and Miles Shore Faculty Development Award, honoraria from NIH, NIEHS, Stanford University, Harvard University, and Columbia University, and payments from Industrial Economics, Inc., unrelated to this work. D.R.G. and Oken report receiving grants from the NIH. M.B.R. reports receiving grants from the NIH, honoraria paid by USC, U. Utah, NYU, UNC, and UVM for serving as visiting professor and giving a lecture, and payments from the Conservation Law Foundation, unrelated to the submitted work. Other authors declare that they have no conflicts of interest with regard to the content of this report.

Acknowledgments

We thank the mothers, children, and staff of Project Viva. We would also like to thank the PRISM Climate Group at Oregon State University for their work on spatial climate analyses, which contributed to this work.