Indoor endotoxin, proximity to a major roadway, and severe asthma exacerbations among children in Puerto Rico

Abstract

Background:

Few studies have examined concurrent exposure to household endotoxin and traffic-related air pollution (TRAP) in relation to childhood asthma, yet both factors are associated with asthma outcomes.

Objective :

We examined whether proximity to a major roadway (a TRAP proxy) modifies the estimated effects of indoor endotoxin on asthma outcomes in children.

Methods:

Cross-sectional study of 200 children with asthma (ages 6-14 years) living in Puerto Rico. Residential distance to a major roadway was calculated as the distance from the participant’s residential US Census Block centroid to the nearest major road. The outcomes of interest were severe asthma exacerbations, missed school days for asthma, atopy, lung function, and bronchodilator response. Logistic, linear, or negative binomial regression was used for the multivariable analysis.

Results:

In the multivariable analysis, there was an interaction between indoor endotoxin and residential distance to a roadway on severe asthma exacerbations (P=0.02) and bronchodilator response (P=0.07). In an analysis stratified by distance to a roadway, each log10-unit increase in endotoxin was associated with 4.21 times increased odds of severe asthma exacerbations among children living within 499 meters (the lower three quartiles of residential distance) to a road (95% CI= 1.5-12.0). Among subjects living further than 499 meters away from a roadway, each log-10 unit increase in endotoxin was associated with reduced odds of severe asthma exacerbations (OR=0.03, 95% CI= 0.001-0.67). Similar but non-significant findings were observed for BDR.

Conclusion:

Our findings suggest that residential proximity to a major road modifies the estimated effect of endotoxin on severe asthma exacerbations in children.

INTRODUCTION

Asthma is the most common chronic respiratory disease of childhood, affecting more than 6 million children in the U.S.¹. In this country, asthma disproportionately burdens minority and economically disadvantaged children².

Severe asthma exacerbations are a major cause of school absences, caregiver missed work days, and healthcare costs for asthma. Children with persistent asthma and at least one severe exacerbation in the prior year have a twofold increased risk of subsequent severe exacerbations despite use of controller medications⁴, a finding that suggests individual susceptibility due to poorly understood risk factors including environmental exposures.

Proximity to a major roadway (a proxy for traffic-related air pollution)^5–8 and household endotoxin (a cell-wall component of gram negative bacteria⁹)^10–13 are both common in urban environments and independently associated with worsening of asthma -including emergency department (ED) visits- in children.

Endotoxin is present not only in household dust but also in particulate matter¹⁴. Although animal and in vitro studies suggest that co-exposure to air pollution and endotoxin¹⁴ increases airway inflammation, few human studies have examined the relation between concurrent exposure to air pollution and endotoxin and childhood asthma^15–17.

A recent study of children and adults living in the contiguous U.S. demonstrated synergistic effects between two outdoor air pollutants (nitrogen dioxide (NO₂) and particulate matter <2μm (PM_2.5)) and household dust endotoxin on ED visits for asthma in children¹⁶. Proximity to a major roadway, which was not assessed in that study, differs from EPA assessment of criteria pollutants in several ways. First, proximity to a major roadway encompasses pollutants such as diesel-exhaust particles and ultra-fine particulate matter^18,19, which may not be captured by central monitoring stations due to decremental decay from roadways. Ultrafine particulate matter is small (<0.1μm) are more likely to reach the pulmonary alveoli than larger PM²⁰ and has been associated with asthma exacerbations^21,22. Second, particulate matter collected near roadways has been shown to cause worse inflammation than background (ambient) particle pollution¹⁴. Third, residential proximity to a major roadway can be easily calculated and thus may have clinical value.

In the U.S., Puerto Ricans have a greater burden of asthma than non-Hispanic whites or members of other Hispanic subgroups²³. The reasons for this are insufficiently understood but likely include exposure to environmental risk factors such as pollutants²³, which could affect expression of asthma-susceptibility genes through gene-by-environment interactions or epigenetic mechanisms^24,25. Given the ubiquity of household endotoxin and recent findings¹⁶, we hypothesized that concurrent exposure to household endotoxin and living near a major roadway would have joint detrimental effects on asthma among Puerto Rican children living in the island of Puerto Rico. We examined this hypothesis in a cross-sectional study of children with asthma in Puerto Rico.

METHODS

Study population

Subject recruitment and study procedures have been previously described in detail³. In brief, from March 2009 to June 2010, 351 children with asthma (defined as physician-diagnosed asthma and wheeze in the previous year) were recruited from randomly selected households in the metropolitan area of San Juan (Puerto Rico), using multistage probability sampling. All study participants were aged 6 to 14 years, had four Puerto Rican grandparents, and had lived in the same household for at least one year. Of the 351 study participants, 200 had endotoxin levels measured in house dust and were thus included in this analysis.

Study procedures

At the same study visit, participants completed a protocol including questionnaires, spirometry, and collection of blood and house dust samples. The child’s parents (usually [93%] the mother) completed a questionnaire slightly modified from one used in the Collaborative Study on the Genetics of Asthma²⁶ and a separate questionnaire on household characteristics. These questionnaires were used to obtain information about the child’s general and respiratory health, sociodemographic characteristics, family history, exposure to second-hand smoke, pet ownership, and respiratory health.

Spirometry was conducted with the EasyOne spirometer (NDD Medical Technologies, Andover, MA). Participants were free of respiratory infections for at least 4 weeks prior to testing and were instructed (when possible) to avoid use of inhaled short- and long-acting bronchodilators for 4 and 12 or more hours before testing, respectively. Expiratory maneuvers were judged acceptable if they met or exceeded American Thoracic Society criteria for children²⁷. The best FVC and FEV₁ values were selected for analysis. Percent predicted values were then calculated utilizing Global Lung Initiative (GLI2012) prediction equations²⁸. After completing spirometry, subjects were given 200 mg (two puffs) of an albuterol metered-dose inhaler using a spacer, and spirometry was repeated after 15 minutes.

Dust samples were collected from three different locations in a participant’s household: the one in which the child usually slept, the living room/television room, and the kitchen. The dust samples were combined and then sifted through a 50-mesh metal sieve. The fine dust was reweighed, extracted, and aliquoted for analysis of allergens using monoclonal antibody Multiplex array assays that used the same reagents as in the established ELISA (dust mite [Der p 1], cockroach [Bla g 1], and mouse urinary protein [Mus m 1]).

After the household dust was used for analysis of allergens, the remaining dust samples were weighed and aliquoted in a gravimetrics laboratory with a filtered air supply and precise temperature and humidity control. Samples were weighed to the nearest microgram using a dual chamber microbalance (Model XP26; Mettler, Columbus, OH). Endotoxin was measured via kinetic chromogenic Limulus amebocyte lysate assay, using multiple dilutions and a 12-point standard curve based on control standard endotoxin from Escherichia coli ranging from 0.0244 to 50.0 EU/mL²⁹. Non-detectable levels were assigned a constant value (half the lowest detectable level). Afterwards, endotoxin levels were converted to the log₁₀ scale for data analysis. Given that dust endotoxin levels can vary by season³⁰, we also recorded the month and season of collection ((1) Dry- December thru April, and (2) Wet- May thru November) of dust collection in Puerto Rico.

Serum IgE to each of five common allergens in Puerto Rico (dust mite [Der p 1], cockroach [Bla g 2], cat dander [Fel d 1], dog dander [Can f 1], and mouse urinary protein [Mus m 1]) were determined by using the UniCAP 100 system (Pharmacia & Upjohn, Kalamazoo, Mich). For each allergen, an IgE level of 0.35 IU/mL or greater was considered positive.

Proximity to a major roadway was determined for each participant as previously described⁶. In brief, home residential addresses were geocoded to US Census block (smallest census unit) at the University of Puerto Rico, and euclidean distance in meters from block centroid to nearest major roadway was obtained in ArcMap (ArcGIS 10.1; Esri, Redlands CA). A major roadway was defined by Esri 2012 Data and Maps major road layer as any interstate, inter-metropolitan area and intra-state highways, and other major roads.

Written parental consent was obtained for participating children, from whom written assent was also obtained. The study was approved by Institutional Review Boards of the University of Puerto Rico (San Juan, Puerto Rico), Brigham and Women’s Hospital (Boston, Massachusetts), and the University of Pittsburgh (Pittsburgh, Pennsylvania).

Statistical Analysis

Our outcomes of interest were at least one severe asthma exacerbation (defined as a hospitalization or visit to the ED or urgent care requiring treatment with systemic corticosteroids, or having received a course of systemic steroids for asthma³¹) in the previous year, average number of missed school days per month due to asthma in the previous year, atopy (defined as at least one positive IgE to the allergens tested), FEV₁ and FEV₁/FVC (percent predicted or <−1.64 z score based on GLI2012 prediction equations²⁸), and bronchodilator response (defined as an increment in FEV₁ equal or greater than 12% after administration of inhaled albuterol). Bivariate analyses were conducted using Wilcoxon rank-sum and chi squared tests for categorical variables, and the Spearman’s correlation test for continuous variables. Logistic (e.g., for at least one severe asthma exacerbation and atopy), linear (e.g., for FEV₁ and FEV₁/FVC), and negative binomial (for average missed school days per month) regression were used for the multivariable analyses of the outcomes of interest. All models included residential distance to a roadway (in meters), endotoxin levels, and an interaction term (residential distance to a roadway*endotoxin). We decided a priori to adjust for household income (< vs. ≥$15,000/year, the median annual household income in Puerto Rico in 2008-2009)³², current exposure to second-hand smoke (SHS), and (with the exception of models for percent predicted FEV₁ and FEV₁/FVC) age and sex. In separate models, we further adjusted for covariates if they were differentially associated with either exposure at p< 0.05. Interaction terms were considered statistically significant at P<0.10¹⁰. In the case of a significant interaction term, the analysis was stratified by highest quartile of residential distance to a road vs. others, as previously described¹⁶. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

eTable 1 shows a comparison of the main characteristics of children with asthma who were (n=200) and were not (n=151) included in the current analysis. Compared to children excluded from this analysis due to lack of available endotoxin measurements, those included were less likely to be exposed to SHS but more likely to have an annual household income ≥$15,000 per year, and to have higher levels of dust mite and cockroach allergens in house dust.

Table 1 summarizes the main characteristics of the 200 study participants, whose households had a geometric mean dust endotoxin level of 36.9 EU/mg. Approximately half of all participants reported SHS exposure and having had at least one severe asthma exacerbation in the previous year, but only 1% had carpets or rugs in their bedroom or living room. Nearly 80% of participants had electric stoves and 65% were atopic, but relatively few had a reduced FEV1 (21%) or FEV1/FVC (32%). Whereas nearly two thirds of participants owned a dog, less than a fifth owned a cat.

Table 1:

Main characteristics of study participants (n=200)

Variable	Median (IQR) or n (%)*
Age, years	10 (8, 12)
Female sex	82 (41%)
Annual household income >$15,000 per year	46 (23%)
Endotoxin (EU/mg)
Geometric mean (standard error)	36.9 (2.6)
Median (interquartile range)	33.4 (20.9, 58.8)
5th-99th percentile	8.9-189.1
Residential distance to a major roadway, meters
Geometric mean (standard error)	243 (2.5)
Median (interquartile range)	210 (172, 499)
5th-99th percentile	50-1007
Second-hand tobacco smoke exposure, current	105 (53%)
Rugs/carpet in child’s bedroom &/or living room	2 (1%)
Current dog ownershipa	126 (63%)
Current cat ownershipa	28 (14%)
Log10 mouse urinary protein (Mus m 1, ng/g) in house dustb	0.78 (0.30, 1.62)
Log10 dust mite allergen (Der p 1, μg/g) in house dustb	0.72 (0.43, 1.03)
Log10 German cockroach allergen (Bla g 2, U/g) in house dustb	0.18 (−0.10, 1.10)
Gas stove (vs. electric)	43 (22%)
Wet season of dust collectionc	119 (60%)
At least one severe asthma exacerbation in prior year	102 (51%)
Bronchodilator response, ≥12%	30 (19%)
Atopyd	125 (65%)
FEV1 % predicted	92.5 (82.2, 102.5)
Reduced (z-score ≤−1.64)	36 (21%)
FEV1/FVC ratio % predicted (GLI2012)	92.8 (85.9, 97.3)
Reduced (z-score ≤−1.64)	55 (32%)
Average school days missed per month due to asthma	2 (0, 5)

^*Values displayed as median (IQR) or n (%) unless otherwise specified.

^aCurrent ownership represents current or within the last 6 months reported ownership

^bHousehold dust was collected from bedroom, living/television room, and kitchen and combined

^cDry season Dec-April; Wet season May-Nov

^dAtopy defined as at least 1 positive specific IgE to one of 5 aeroallergens (cat, dog, dust mite, cockroach, mouse)

Table 2 displays a comparison of mean endotoxin or mean residential distance to a major road by selected variables. Endotoxin was significantly higher in homes reporting current cat ownership than in those reporting no cat ownership (57.7 EU/mg vs. 34.2 EU/mg, P <0.005). An annual household income ≥$15,000 per year, no exposure to SHS, having an electric stove, and having had the dust sample collected during the wet season were each significantly associated with greater residential distance to a major roadway (P <0.05 in all instances). Children who did not report a severe asthma exacerbation in the previous year or who had fewer reported missed school days lived further away from a major roadway, but these associations were not statistically significant (P=0.08-0.11). Endotoxin was weakly and positively correlated with mouse urinary protein allergen (P=0.04), and weakly and negatively correlated with dust mite allergen (eTable 2). Residential distance to a major roadway was weakly and negatively correlated with German cockroach allergen (P <0.01).

Table 2:

Geometric mean (95% CI) endotoxin level and proximity to a major roadway, by characteristics of study DarticiDants (n=200)

	Endotoxin (EU/mg)		Proximity to a major roadway (m)
	GM (95% CI)	P	GM (95% CI)	P
Age, years
6-10	33.9 (29.0, 39.6)	0.13	237 (200, 280)	0.46
11-15	41.8 (33.1, 52.8)		252 (204, 312)
Sex
Female	38.5 (30.8, 48.1)	0.52	238 (191,296)	0.80
Male	35.9 (30.4, 42.3)		247 (210, 291)
Annual Household income ≥$15K
Yes	40.2 (30.4, 53.1)		299 (240, 373)	0.04
No	36.2 (31.0, 42.3)	0.61	233 (199, 272)
Second-hand tobacco smoke exposure, current
Yes	38.1 (32.2, 45.0)	0.69	217 (181,260)	0.01
No	35.8 (28.9, 44.4)		274 (227, 332)
Current dog ownership
Yes (currently or in last 6 months)	36.5 (30.8, 43.3)	0.83	261 (223, 306)	0.17
No (never or not in last 6 months)	37.6 (30.3, 46.7)		215 (170, 271)
Current cat ownership
Yes (currently or in last 6 months)	57.7 (41.2, 80.9)	0.005	199 (137, 290)	0.09
No (never or not in last 6 months)	34.2 (29.7, 39.5)		253 (220, 291)
Stove type
Gas	44.6 (32.8, 60.7)	0.33	302 (233, 393)	0.04
Electric	35.1 (30.3, 40.6)		229 (197, 266)
Season of dust collection
Dry	35.9 (29.4, 43.8)	0.83	188 (156, 227)	0.0003
Wet	37.7 (31.5, 45.0)		289 (243, 344)
Severe asthma exacerbation in last year
Yes	37.8 (31.1, 46.0)	0.55	213 (177, 256)	0.11
No	36.0 (30.1, 43.1)		279 (232, 335)
Bronchodilator response, ≥12%
Yes	42.6 (29.8, 60.8)	0.94	218 (156, 303)	0.55
No	41.2 (34.8, 48.8)		236 (200, 278)
Atopy
Yes	38.7 (32.3, 46.4)	0.41	255 (216, 301)	0.14
No	34.9 (28.5, 42.9)		213 (171,265)
FEV1 z-score
≤−1.64 (reduced)	34.0 (25.9, 44.7)	0.16	232 (173, 310)	0.68
>−1.64 (normal)	42.2 (36.0, 49.5)		239 (204, 280)
FEV1/FVC z-score
≤−1.64 (reduced)	39.1 (30.9, 49.3)	0.93	259 (203, 330)	0.48
>−1.64 (normal)	41.0 (34.5, 48.8)		228 (192, 270)
Average asthma school days missed per month
0-3	34.0 (28.8, 40.3)	0.11	265 (224, 314)	0.08
≥4	41.8 (33.5, 52.1)		202 (163, 250)

P values obtained from Wilcoxon rank sum test

Bold= p < 0.05

We found an interaction (at P <0.10) between residential distance to a major roadway and endotoxin level on severe asthma exacerbations (P=0.02) and bronchodilator response (P=0.07). No such interactions were detected for any of the other outcomes of interest (eTable 3). Figure 1 shows the results of the analysis of endotoxin level and at least one severe asthma exacerbation, after stratification by residential distance to a major roadway (as upper quartile or >499 meters vs. the others, for ease of exposition). Among children living closer to (or within 499 meters of) a major roadway, each log-unit increment in endotoxin level was significantly associated with 2.33 increased odds of at least one severe asthma exacerbation in the previous year, after adjustment for age, sex, annual household income and current exposure to SHS (Model 1). After additional adjustment for cat ownership, indoor allergen levels, type of gas stove in the home, and season of dust collection, each log-unit increment in endotoxin level was significantly associated with 4.21 increased odds of at least one severe asthma exacerbation in the previous year (Model 2). Among children living >499 meters away from a major roadway, endotoxin level was significantly associated with 0.03 to 0.07 times decreased odds of at least one severe asthma exacerbation, after adjustment for relevant covariates (models 1 and 2). Similar but non-statistically significant associations between endotoxin level and bronchodilator response were observed in the analysis stratified by residential distance to a major roadway.

Figure 1.

Results of the multivariable analysis of the relation between house dust endotoxin and severe asthma exacerbations, stratified by residential distance to a major road. Model 1 was adjusted for age, sex, annual household income and current exposure to second-hand smoke. Model 2 was additionally adjusted for cat ownership, indoor allergen levels, type of gas stove in the home, and season of dust collection.

Table 3 shows the predicted probabilities of having had at least one severe asthma exacerbation in the previous year, by residential distance to a roadway and endotoxin level. High endotoxin levels had differential effects on predicted probability of severe asthma exacerbations, depending on proximity to a major roadway (close vs. further away). No such effect of endotoxin on probabilities for BDR were appreciated in an analysis stratified by residential distance to a roadway.

Table 3-.

Predicted probabilities by differing exposures to endotoxin & proximity to a major roadway

Endotoxin level/Residential distance to a roadway	N	Predicted Probability %	95% CI
At least one severe asthma exacerbation in the prior year
Low/Close	112	51	(35 - 67)
Low/Far	32	50	(29 - 71)
High/Close	33	70	(47 - 86)
High/Far	17	18	(5 - 46)
Bronchodilator response, ≥12%
Low/Close	112	18	(8 - 36)
Low/Far	32	23	(7 - 54)
High/Close	33	25	(9 - 52)
High/Far	17	14	(3 - 45)

For endotoxin, low is ≤58.8EU/mg & high is >58.8EU/mg (75^th percentile); for proximity to a major roadway= close is ≤499m & far is >499m (75^th percentile). Predicted probabilities obtained from models adjusting for age, sex, current ETS exposure, and annual household income.

Figure 2 demonstrates the predicted probability plots for the analysis using models with continuous household endotoxin*proximity to a roadway for at least one severe asthma exacerbation in the previous year and bronchodilator response. For children living closer to a major roadway, increasing endotoxin concentrations were associated with higher predicted probability of a severe asthma exacerbation and positive BDR. However, when living further away from a major roadway, increasing endotoxin levels are associated with decreased predicted probability of an asthma exacerbation or BDR.

Figure 2.

Contour plot of ≥1 severe asthma exacerbation and bronchodilator response, according to the interaction between house dust endotoxin and proximity to a major road.

All models were adjusted for age, sex, current exposure to second-hand smoke, annual household income, cat ownership, levels of indoor (dust mite, cockroach, and mouse urinary protein) allergens, type of stove, and season of dust collection. Predicted probabilities all fit at median or largest proportion for cohort (age 10 years, male sex, no exposure to second-hand smoke, annual household income <$15,000 per year, no cat ownership, median log-10 level for each allergen, electric stove, and wet season for dust collection).

DISCUSSION

We show that living closer to (<499 meters from) a major roadway modifies the estimated effect of indoor endotoxin on severe asthma exacerbations in the previous year. Among children living closer to a major roadway, increasing endotoxin levels were associated with increased odds of severe asthma exacerbations. On the other hand, increasing endotoxin levels were associated with reduced odds of severe asthma exacerbations among children living further away (>499 meters) from a major roadway. Similar but non-statistically significant results were observed for BDR. Our findings add to those of a few studies reporting an interaction between air pollutants and endotoxin on asthma outcomes in children^15,16.

The average indoor house dust endotoxin level in our study, conducted in the San Juan metropolitan area of Puerto Rico, was nearly twice as high as that previously reported in studies conducted in the U.S. and Washington, D.C. (36.9 EU/mg vs. 15.3 EU/mg¹⁶ and 16.2 EU/mg respectively³³), lower than those reported for homes of children with asthma in Saskatchewan, Canada (mild asthma= 58.7 EU/mg and moderate/severe asthma=49.2 EU/mg)³⁴, and similar to those reported for the homes of farmers and their spouses living in Iowa and North Carolina (38.4 EU/mg)³⁵. Several factors can affect house dust endotoxin levels, such as home age, presence of children, farm or rural living, flooring materials (the presence of carpet), presence of pets, poverty, geography, and relative humidity^33,36,37. Our study is limited to homes with children, which have been shown to have increased house endotoxin levels. The reasons for the negative correlation between dust mite and endotoxin in this study are unclear. Dust mite and endotoxin levels have been weakly correlated in some but not all studies³⁷.

Previous studies have reported positive associations between endotoxin and pet ownership or mouse urinary protein³⁸. Consistent with those findings, current cat ownership and mouse urinary protein were positively correlated with household endotoxin in the current study. In contrast to a U.S. study³⁸, endotoxin levels in our study were not different between homes who did or did not own dogs. Notably, a prior study reported that the association between household endotoxin levels and dog ownership varied across U.S. climate regions, perhaps due to variability in the time that dogs spend outdoors³³. We are unable to determine if the dog(s) was/were indoor, outdoor, or both in the current study.

Our findings are in partial agreement with those of a recent U.S. study that reported that higher levels of outdoor air pollutants (PM_2.5 & NO₂) and endotoxin synergistically increase odds of ED visits for children with asthma (OR for PM_2.5 = 3.75 95% CI 1.27 to 11.07, and OR for NO₂ = 3.45 95%CI 1.65 to 7.18)¹⁶. Our study expands on those results, as we show modification of the estimated effect of endotoxin on severe asthma exacerbations by residential distance to a major roadway.

Endotoxin has previously been reported to interact with air pollutant co-exposures in both in vitro²⁹ and clinical studies^15,16,39. A study of urban children with asthma found that increasing household airborne endotoxin was significantly associated with 0.38 times reduced odds of acute visits for asthma in children exposed to no/lower airborne nicotine, but with 1.15 times significantly increased odds of acute visits for asthma in children exposed to higher nicotine levels¹⁵. As such, those results suggest that tobacco smoke exposure potentiates the impact of endotoxin on toll-like receptor 4 (TLR4)¹⁵. This same study did not find an association with airborne particulate matter (PM) and endotoxin on asthma outcomes. Non-human studies have demonstrated synergism between diesel exhaust particles and endotoxin on proinflammatory cytokines, yet have also demonstrated reduced levels of the pro-inflammatory cytokine IL-13 with co-exposure (lipopolysaccharide (LPS) + diesel exhaust particles)⁴⁰.

Our finding of a negative association between endotoxin and severe asthma exacerbations among children living further away from a major roadway may be due to endotoxin compositional differences or unmeasured correlated factors. Puerto Rico experiences annual African Dust Events (ADE) characterized by transatlantic Saharan dust deposition. A study in Puerto Rico found higher levels of endotoxin in particulate matter (PM) from ADE compared to non-ADE PM; in vitro ADE-PM exposed cells demonstrated higher pro-inflammatory chemokines IL-6 & IL-8 via toll-like receptors to a greater degree than non-ADE PM; IL-6 and IL-8 decreased with addition of an endotoxin inhibitor¹⁷. Further, the authors reported that endotoxin extracts from both ADE & non-ADE PM caused cellular cytotoxicity, thus concluding that quantity is likely less important than type of endotoxin (e.g. P. aeruginosa vs. E. coli). A study of Amish vs. Hutterite homes supports compositional importance and/or temporality of exposure, given almost 7-fold higher levels of endotoxin in Amish homes, a population with low asthma and atopic prevalence, as compared to Hutterites homes, a population with higher asthma and atopy prevalence³⁹. A greater abundance of Proteobacteria was found in Amish samples compared to increased abundance of Firmicutes and Bacteroidetes in Hutterite samples³⁹. A study of farmers and spouses from two US states demonstrated bacterial microbiome differences by state, farming type, home condition, and season thus supporting potential differences in endotoxin composition³⁵. In further support of geographic and compositional differences, a study in the U.S. mainland reported that effects of home endotoxin on respiratory health varied by climate, perhaps due to differences in the number of acyl chains attached to lipid A (as penta-acylated and hexa-acylated forms may have different effects on inflammation)³³.

We cannot assess temporal relationships in this cross-sectional study. Although adjustment of our analysis for season of dust collection or other covariates did not alter our results, we lack endotoxin speciation/type. Moreover, proximity to a major roadway may correlate with unmeasured factors such as polyaromatic hydrocarbons or NO₂. Indoor NO₂ has previously been reported to interact with airborne endotoxin on asthma outcomes (increased acute asthma visits with low NO₂, protective with high NO₂)¹⁵. Although we do not have data regarding indoor air pollutants, adjustment for gas stove-a primary source of indoor NO₂¹¹, did not alter our results. We lack information on additional sources of exposure, such as school proximity to a major roadway, yet findings from a recent study with data on both school and residential distance to a a major road suggest that childhood asthma symptoms are predominately driven by residential proximity to a major roadway⁵. Notably, ≥62% of children who participated in that study were likely to attend schools or to live in homes located within 100 m of a major roadway. Thus, in future studies of children living in close proximity to a major roadway, dust endotoxin compositional analysis in the homes may be more important than that in the schools.

We did not find any interaction with household endotoxin and proximity to a major roadway on outcomes other than severe asthma exacerbations, consistent with prior findings of no association between outdoor air pollution and endotoxin on inhalant sensitization¹⁶, FEV₁⁴¹, or FEV₁/FVC¹⁵. However, we had limited statistical power due to relatively small sample size and thus our results for BDR may have achieved statistical significance in a larger cohort. For example, a prior study of 146 children and adolescents showed suggestive but non-statistically significant evidence that increased airborne endotoxin exposure was associated with increased BDR in homes with high air nicotine but not in those with low air nicotine (P for interaction=0.16)¹⁵.

In conclusion, proximity to a major roadway modifies the estimated effect of indoor endotoxin on severe asthma exacerbations among Puerto Rican children with asthma in Puerto Rico. Large longitudinal studies assessing the composition of dust endotoxin in homes near and farther away from major roadways are needed to confirm and expand on our findings, yet this should not preclude continued efforts to implement health policies that ensure clean air for all.

Abbreviations:

ED

emergency department

PM_2.5

particulate matter <2.5 μm

NO₂

nitrogen dioxide

IgE

Immunoglobulin E

BDR

Bronchodilator response

ADE

African Dust Event