Household mold, pesticide use, and childhood asthma: A nationwide study in the U.S.

Abstract

Objectives:

To investigate the associations of household mold and pesticide use with risk of childhood asthma and examine the potential effect modification by child’s sex at a national level in the U.S.

Methods:

Nationally representative data were drawn from the cross-sectional 2017 and 2018 National Surveys of Children’s Health. Household mold and pesticide exposures during the past 12 months and physician-diagnosed childhood asthma were assessed by standard questionnaires administered to primary caregivers. Multivariable logistic regression models were used to calculate adjusted odds ratios (aOR) for current asthma, adjusting for child, caregiver, and household covariates. We also examined potential effect modification by child’s sex. Sampling weights accounted for the complex survey design.

Results:

Among 41,423 U.S. children in 2017–2018, the weighted prevalence of current asthma was 10.8% in household mold-exposed children, compared with 7.2% in non-exposed children (P <0.001). After adjusting for covariates including child’s obesity, children with household mold exposure compared to those with no household mold exposure had a 1.41-fold (95% CI: 1.07, 1.87) higher odds of current asthma. Associations between household mold and current asthma were pronounced among boys (aOR 1.57; 95% CI: 1.03–2.38) but not girls (aOR 1.28; 0.90–1.83; P for interaction <0.001). No significant associations were observed between household pesticide use and current asthma, after adjusting for covariates.

Conclusions:

Our findings suggest that household mold is associated with current asthma among children, independent of other major risk factors including child’s obesity status. Our findings may inform strategies targeting mitigation of household mold as an important indoor environment factor to address childhood asthma.

Introduction

Asthma is one of the most common childhood chronic diseases in the industrialized world, affecting 7.5–8.4% of U.S. children in 2017–2018 (Centers for Disease Control and Prevention, 2018, 2020). Asthma predisposes children to a myriad of long-term sequalae, including impaired overall health status, compromised social and school performance, and greater disability risk (Beuther et al., 2006; Fletcher et al., 2010; Newacheck and Halfon, 2000). Thus, it is imperative to identify potentially modifiable risk factors of asthma which may inform public health prevention or intervention strategies. Notably, children spend over 90% of their time indoors (Breysse et al., 2010). Given that people may have a better ability to modify their indoor than outdoor environment, modification of indoor environmental exposures has been suggested as a promising target for childhood asthma prevention and control.

Dampness and mold are common household exposures, present in 18% to 50% of buildings (Mendell et al., 2011). Previous reviews and meta-analyses have suggested that visible mold or residential fungal exposure was associated with childhood asthma in the U.S and worldwide (Caillaud et al., 2018; Fisk et al., 2007; Quansah et al., 2012). In particular, two population-based cross-sectional studies of 762 and 2849 children in the U.S have linked indoor fungal exposures to atopic disease, highlighting the importance of residential mold exposures in asthma (Salo et al., 2006; Sharpe et al., 2015). However, these data were predominantly Caucasian based among restricted age groups (mostly 6–17 years or with no data on specific age groups). On the other hand, longitudinal cohort studies in the U.S. have linked both qualitative and quantitative measures of household mold and its components to asthma or other atopic disease among children mostly in infancy (Behbod et al., 2015; Belanger et al., 2003; Gent et al., 2002; McConnell et al., 2002; Reponen et al., 2011; Rosenbaum et al., 2010; Schroer et al., 2009). These data were also restricted to certain racial/ethnic groups at city or state levels and suffered from varied degrees of residual confounding due to limited data on covariates. Collectively, contemporary, population-based data among children of diverse age, geographical, and racial/ethnic groups with comprehensive data on potential confounders in the U.S. at a national level are lacking.

Emerging data have also linked household chemical use to persistent wheezing among preschoolers and residential pesticide exposure to asthma in adolescents (Bukalasa et al., 2018; Sherriff et al., 2005; Xu et al., 2012). Nonetheless, data on household pesticide use and childhood asthma in the U.S. are scant. Children present a particularly vulnerable population to indoor pesticides, as they are more likely to be exposed to household than occupational pesticides (Silvers et al., 1994) and to more pesticide per unit of body weight than adults (Council, 1993), while their immune systems and lung functions are not fully developed (Almqvist et al., 2008).

Notably, substantial racial/ethnic disparities exist in the prevalence and health care use of asthma among U.S. children, with a higher prevalence in children of certain racial/ethnic minority groups (Non-Hispanic Black and certain Hispanic subgroups, such as Puerto Rican) vs. non-Hispanic White children (Centers for Disease Control and Prevention, 2020; Fitzpatrick et al., 2019). Given the potential racial/ethnic variation in housing conditions, examining the contribution of household mold and pesticide use to racial/ethnic disparities in childhood asthma is warranted. Moreover, given the established sex and age differences in asthma development and management (Carey et al., 2007; Postma, 2007; Shah and Newcomb, 2018; Skobeloff et al., 1992) and that asthma morbidity as well as residential mold and pesticide exposures tend to differ by socioeconomic status (Guha et al., 2013), examination of corresponding potential effect modification by each of these factors may inform potential risk-based preventive or intervention strategies.

To address these important knowledge gaps, in a nationally representative sample of children aged 3–17 years in the U.S., we investigated the associations of household mold and pesticide use with current childhood asthma and examined potential effect modification by child’s age, sex, and race/ethnicity and household poverty status. The ongoing Coronavirus Disease 2019 (COVID-19) pandemic has the potential of increasing indoor time and resulting in limited access to adequate medical care among children more than ever; investigation of household mold and pesticide use as important indoor environmental factors in relation to childhood asthma is timely and warranted.

Methods

Study sample and design

Data were drawn from the 2017 and 2018 National Surveys of Children’s Health (NSCH), a nationally representative caregiver-completed survey available both in English and Spanish regarding U.S. children 0–17 years of age (Data Resrouce Center for Child & Adolescent Health, 2018). Detailed methodology of the survey was previously described elsewhere (United States Census Bureau, 2016, 2018). In brief, households were selected based on child-presence flags provided by the Census Master Address File, with 60% addresses from Stratum 1 (flagged as households with children) and 40% from Stratum 2a (not flagged but has a higher probability of child presence than Stratum 2b), in order to improve sampling efficiency. During data collection, a screener by mail was first used to identify households with at least one child under 18 years old and enumerate the children in those households. After completing the screener, one child was randomly selected per household for parental/caregiver’s response to the survey. The household adult who knew best about the child’s health status (i.e., the child’s primary caregiver) completed the survey in English or Spanish. Notably, assessment of household mold and pesticide use was first implemented in the NSCH in 2017. Thus, we used data of 52,129 responses from the NSCH in 2017–2018. The overall weighted response rates were 37.4% and 43.1% for 2017 and 2018, respectively. The National Center for Health Statistics Research Ethics Review Board approved all data collection procedures for the survey. Written informed consent was obtained from all study participants.

Exclusion criteria

Among 52,129 children, we excluded those younger than 3 years old (n=6,900) and those without any current health insurance (n=2,646) to reduce the potential impact of misclassification or detection bias in diagnosis of asthma. We also excluded children with invalid or missing answers to questions about mold (n=570), pesticides (n=156), or current asthma (n=434), leaving a sample of 41,423 (Figure 1). Characteristics of participants in the analytical sample did not differ from characteristics of those excluded due to missing data on the outcome or exposure of interest.

Figure 1.

Study flowchart, 2017–2018 National Survey of Children’s Health

Outcome measure

The respondents were asked, “Has a doctor or other health care provider EVER told you that this child has Asthma?” and “If yes, does this child CURRENTLY have the condition.” The child was considered to have current asthma if the respondent reported “Yes” to both the first and the second questions and have no current asthma if the respondent reported “No” to the second question. To evaluate the robustness of our findings, we have also conducted a sensitivity analysis by comparing children with current asthma to those who never had a diagnosis of asthma in relation to household mold or pesticide use.

Exposure assessment

The respondents were asked, “DURING THE PAST 12 MONTHS, other than in a shower or bathtub, have you seen any mold, mildew or other signs of water damage on walls or other surfaces inside your home?” The child was considered to be exposed to household mold if the respondent answered “Yes” to the question.

The respondents were asked, “DURING THE PAST 12 MONTHS, how often were pesticides used inside your residence to control for insects?” This question contains multiple options which we collapsed into 2 categories based on distribution of survey responses: ever (more than once a week, once a week, once a month, once every 2–5 months, once every 6 months, once during the past 12 months) and never.

Covariates

Potential covariates were considered based on biological plausibility, prior knowledge, and statistical considerations, including: age (3–5, 6–11, 12–17 years), sex (boy, girl), race/ethnicity (Non-Hispanic White, Hispanic, Non-Hispanic Black, Other), insurance status (any public, private only), household highest education (less than high school, high school, some college credit but no degree or associate degree, college degree or higher), household poverty (0–199, 200–399, ≥400 percent of the Federal Poverty Level, calculated as the ratio of total family income to the national family poverty threshold), family structure (two parents, married; two parents, unmarried; single parent or other), household smoker (yes, no), primary caregiver’s mental and emotional health status (fair or poor, good, very good or excellent). Further, given that obesity is an important risk factor for childhood asthma and its severity (Peters et al., 2018), we additionally adjusted for child’s obesity status calculated using parent-reported weight and height in a secondary analysis. Notably, data on body mass index were only available among children aged 10–17 years in the NSCH given the reasonably high classification rate (97.5%) of obesity status based on parent-reported weight and height among older school-aged children but not preschool and elementary school-aged children (Akinbami and Ogden, 2009; Goodman et al., 2000). Inclusion or exclusion of potential covariates in final regression models was finally evaluated by comparing the odds ratios associated with exposures of interest (i.e., household mold or pesticide use), adjusted and unadjusted for the potential covariate, with inclusion if the coefficient of exposure of interest changes by 10% or more. The change-in-estimate approach to confounder selection performs well with regard to power, bias, mean squared error and confidence interval coverage for the primary exposure of interest (Mickey and Greenland, 1989).

Statistical analysis

To generate nationally representative results, the NSCH sampling weights were applied for all analyses, which accounted for potential non-response biases and non-coverage of non-telephone households (Luke and Blumberg, 2006). The sampling weights were developed from base sampling weights which were the inverse of probability of the selection of the address, and further adjusted for non-response, within-household subsampling factors, and a post-stratification raking that matched sample demographic estimates to various demographic controls as described in detail previously (United States Census Bureau, 2019).

Unweighted and weighted numbers and unweighted and weighted prevalence of current asthma by participant characteristics were reported. The Rao-Scott Chi-squared tests were used to calculate p-values for group comparisons of categorical variables from complex survey data (Lewis, 2013). Multivariable logistic regression analyses were conducted to assess the associations of household mold and pesticide use (with both exposures of interest being mutually adjusted for in each model) with parent-reported physician diagnosis of current asthma, adjusting for the aforementioned covariates. To evaluate the potential impact of applying the NSCH sampling weights, we also conducted sensitivity analysis without applying the sampling weight.

We tested for an interaction between household mold and child’s sex, age, race/ethnicity, and household poverty status by including a cross product, respectively. P for interaction was calculated using the likelihood ratio test. Based on the interaction results, we conducted stratified analysis and examined potential effect modification. Further, given the relatively low prevalence and potential misclassification of asthma in children aged 3–5 years, we conducted a sensitivity analysis restricted to children aged 6–17 years. We also conducted sensitivity analyses by including children without any current health insurance (n = 2,646) or children who had been diagnosed with asthma but did not have current asthma (n = 2,093) to evaluate the robustness of study findings. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC) and the statistically significant level was set at a two-tailed P-value <0.05.

Results

Among 41,423 children in the 2017–2018 NSCH study, the overall weighted prevalence of current asthma was 8.6% (Table 1). The weighted prevalence of current asthma was 11.9% in mold-exposed children, compared to 8.2% in non-exposed children (P < 0.001). The weighted prevalence of current asthma was 11.9% in children exposed to household pesticide use once a month or more, significantly higher than the prevalence among children with lower frequency of pesticide use exposure (range: 8.3–8.4%). Additionally, the weighted prevalence of current asthma varied significantly by participant characteristics, including child’s age, sex, race/ethnicity, insurance type, and obesity status; household education, poverty status, family structure, and household smoker; and respondent’s mental and emotional health status (all P-values <0.05).

Table 1.

Unweighted numbers and weighted prevalence of asthma by household mold, household pesticide use, and participant characteristics, 2017–2018 National Survey of Children’s Health

	Unweighted N of all	Unweighted N of asthma	Unweighted % of asthma	Weighted N of asthma	Weighted % (95% CI) of asthma*	P-value†
Overall	41423	3629	8.8	4679292	8.6 (8.0, 8.9)
Household mold						<0.001
No	37500	3160	8.4	4014659	8.2 (7.6, 8.8)
Yes	3923	469	12.0	664633	11.9 (9.9,14.0)
Household pesticide use						0.30
Never	23420	1967	8.4	2396309	8.3 (7.5, 9.1)
Ever	18003	1662	9.2	2282983	8.9 (8.0, 9.8)
Child’s age, year						<0.001
3–5	7285	369	5.1	613957	5.7 (4.5, 6.9)
6–11	14607	1284	8.8	1918055	8.7 (7.7, 9.7)
12–17	19531	1976	10.1	2147280	9.9 (9.0,10.8)
Child’s sex						0.004
Female	19979	1592	8.0	2060574	7.7 (7.0, 8.5)
Male	21444	2037	9.5	2618718	9.4 (8.5,10.4)
Child’s race/ethnicity						<0.001
Non-Hispanic White	29001	2293	7.9	2081085	7.3 (6.7, 7.9)
Hispanic	4616	457	9.9	1048250	8.1 (6.4, 9.7)
Non-Hispanic Black	2588	444	17.2	1135016	15.4 (13.1,17.6)
Other‡	5218	435	8.3	414941	7.3 (6.0, 8.6)
Child’s insurance type						<0.001
Private only	31138	2381	7.6	2443914	7.2 (6.6, 7.8)
Any public	10285	1248	12.1	2235378	10.9 (9.7,12.2)
Child’s obesity status (10–17 years)						0.013
Underweight	1551	132	8.5	163708	8.9 (6.2,11.5)
Normal	15885	1442	9.1	1596000	9.1 (8.1,10.2)
Overweight	3444	382	11.1	368123	8.7 (6.8,10.6)
Obesity	3133	459	14.7	529767	12.9 (10.6,15.1)
Household education						0.002
Less than high school	841	92	10.9	452359	10.0 (6.4,13.6)
High school or equivalent	5254	564	10.7	1024089	9.8 (8.2,11.5)
Some college or associate degree	9796	997	10.2	1222585	10.1 (8.8,11.4)
College degree or higher	25532	1976	7.7	1980260	7.2 (6.6, 7.8)
Poverty status, % of federal poverty level						<0.001
0–199	10659	1207	11.3	2319186	10.7 (9.4,11.9)
200–399	14498	1213	8.4	1278268	7.7 (6.8, 8.5)
≥400	16266	1209	7.4	1081839	6.7 (6.0, 7.4)
Family structure						<0.001
2 Parents, married	29413	2286	7.8	2515176	7.1 (6.4, 7.7)
2 Parents, unmarried	2554	274	10.7	546966	12.7 (9.5,16.0)
Single parent or other	9132	1038	11.4	1545165	11.1 (9.8,12.4)
Region						0.61
Midwest	9838	845	8.6	1088643	9.3 (8.3,10.3)
Northeast	7436	659	8.9	757673	8.5 (7.2, 9.8)
South	13895	1260	9.1	1764571	8.5 (7.5, 9.5)
West	10254	865	8.4	1068406	8.1 (6.7, 9.6)
Household smoker						<0.001
Yes	6139	673	11.0	923787	11.0 (9.4,12.6)
No	35173	2950	8.4	3752947	8.2 (7.5, 8.8)
Primary caregiver’s mental and emotional health status						<0.001
Fair or Poor	1771	259	14.6	406900	15.9 (12.2,19.5)
Good	7321	780	10.7	1023943	10.5 (9.1,12.0)
Very good or excellent	31837	2554	8.0	3170171	7.7 (7.0, 8.3)

^*All percentages were calculated with weighted data to reflect the national representative prevalence in the target U.S. population.

^†Obtained by Rao-Scott Chi-squared tests.

^‡Other: American Indian or Alaska Native, Asian, Native Hawaiian and other Pacific Islander, and multiracial/other.

Compared to children with no household mold exposure, children with household mold exposure had a 1.52-fold (95% CI: 1.23–1.87) higher odds of current asthma (crude model in Table 2). Compared to children never exposed to household pesticide use, those exposed to pesticides at least once per month had a 1.49-fold (1.13, 1.98) higher odds of current asthma (crude model). After adjusting for covariates (Model 1), the effect size of adjusted odds ratios (aOR) was attenuated but remained significant for household mold, yes vs. no (aOR 1.25; 95% CI: 1.00–1.57), and attenuated to non-significant for household pesticide use, once a month or more vs. never (aOR 1.07; 95% CI: 0.80–1.42). In Model 2, restricted to children aged 10–17 years with further adjustment for child’s obesity status, associations slightly strengthened for household mold (aOR 1.41; 95% CI: 1.07–1.87) and remained non-significant for household pesticide use. In a sensitivity analysis without applying the sampling weight, results remained similar with slightly greater effect sizes (Table S1).

Table 2.

Unadjusted and adjusted odds ratios of current asthma among children aged 3–17 years, 2017–2018 National Survey of Children’s Health

	Crude	Model 1*	Model 2†
Household mold	1 (reference)	1 (reference)	1 (reference)
No	1.52 (1.23, 1.87)	1.25 (1.00, 1.57)	1.42 (1.07, 1.87)
Yes
Household pesticide use	1 (reference)	1 (reference)	1 (reference)
Never	1.08 (0.93, 1.26)	1.00 (0.86, 1.16)	0.91 (0.76, 1.08)
Ever
Child’s age, year
3–5	0.55 (0.44, 0.70)	0.56 (0.44, 0.71)
6–11	0.87 (0.74, 1.02)	0.86 (0.73, 1.02)	1.02 (0.98, 1.06)
12–17	1 (reference)	1 (reference)
Child’s sex
Male	1 (reference)	1 (reference)	1 (reference)
Female	0.80 (0.69, 0.93)	0.80 (0.69, 0.93)	0.94 (0.78, 1.12)
Child’s race/ethnicity
Non-Hispanic White	1 (reference)	1 (reference)	1 (reference)
Hispanic	1.11 (0.88, 1.41)	1.03 (0.82, 1.29)	1.03 (0.77, 1.38)
Non-Hispanic Black	2.30 (1.91, 2.77)	1.91 (1.55, 2.36)	1.80 (1.38, 2.33)
Other	1.00 (0.81, 1.23)	0.99 (0.80, 1.23)	1.04 (0.79, 1.37)
Child’s insurance type
Any public	1 (reference)	1 (reference)	1 (reference)
Private only	0.63 (0.54, 0.73)	0.86 (0.69, 1.08)	0.82 (0.62, 1.08)
Household education
Less than high school	1 (reference)	1 (reference)	1 (reference)
High school or equivalent	0.98 (0.63, 1.50)	1.11 (0.70, 1.73)	1.29 (0.77, 2.15)
Some college credit (no degree) or associate degree	1.01 (0.66, 1.52)	1.22 (0.78, 1.91)	1.61 (0.95, 2.75)
College degree or higher	0.70 (0.47, 1.04)	1.10 (0.70, 1.73)	1.61 (0.96, 2.70)
Poverty status, % of federal poverty level
0–199	1 (reference)	1 (reference)	1 (reference)
200–399	0.69 (0.58, 0.82)	0.88 (0.72, 1.07)	0.81 (0.64, 1.03)
≥400	0.60 (0.51, 0.71)	0.89 (0.71, 1.12)	0.95 (0.73, 1.24)
Family structure
2 Parents, married	1 (reference)	1 (reference)	1 (reference)
2 Parents, unmarried	1.92 (1.41, 2.60)	1.48 (1.08, 2.02)	1.32 (0.84, 2.09)
Single parent or other	1.63 (1.39, 1.92)	1.16 (0.96, 1.40)	1.16 (0.93, 1.45)
Household smoker
Yes	1 (reference)	1 (reference)	1 (reference)
No	0.72 (0.60, 0.87)	0.85 (0.70, 1.04)	0.76 (0.59, 0.97)
Primary caregiver’s mental and emotional health status
Fair or poor	1 (reference)	1 (reference)	1 (reference)
Good	0.62 (0.46, 0.85)	0.75 (0.54, 1.02)	0.78 (0.53, 1.15)
Very good or excellent	0.44 (0.33, 0.58)	0.59 (0.44, 0.80)	0.58 (0.40, 0.85)
Obesity status (10–17 years)
Underweight	1 (reference)		1 (reference)
Normal	0.97 (0.68, 1.37)		0.97 (0.69, 1.37)
Overweight	0.95 (0.73, 1.24)		0.89 (0.67, 1.18)
Obesity	1.47 (1.17, 1.86)		1.28 (0.99, 1.65)

^*Model 1: Adjusted for all variables listed in the table for Model 1. Both exposures of interest (i.e. household mold and pesticide use) were mutually adjusted for in each model.

^†Model 2: Restricted to children aged 10–17 years and adjusted for covariates in Model 1 and obesity status. Both exposures of interest (i.e. household mold and pesticide use) were mutually adjusted for in each model.

^‡Due to age restriction (i.e., 10–17 years) for data on childhood body mass index, categorization is not applicable here. Age was included as a continuous variable in Model 2.

In stratified analysis, household mold was significantly associated with current asthma among boys (aOR 1.57; 95% CI: 1.03–2.38) but not girls (aOR 1.28; 95% CI: 0.90–1.83; P-for-interaction <0.001; Table 3) and among non-Hispanic White children (aOR 1.71; 95% CI: 1.28–2.28) but not racial/ethnic minority children (P for interaction <0.001; Table 4), after adjusting for covariates. No significant effect modification by child’s age (3–5, 6–11, 12–17 years) or household poverty status (0–199, 200–399, ≥400% of federal poverty level) was observed (all P for interaction >0.15; data not shown). In a sensitivity analysis restricted to children aged 6–17 years, results were materially unchanged for the association between household mold and childhood current asthma (aOR 1.41; 95% CI: 1.07–1.87; Table S2). In a sensitivity analysis by additionally including children without any current health insurance (n=2,646), results were slightly attenuated but remained robust for the association between household mold and childhood current asthma (aOR 1.34; 95% CI: 1.03–1.76; Table S3). In a sensitivity analysis additionally excluding children who ever had been diagnosed with asthma but did not have current asthma (n = 2,093) out of the denominator, results remained robust with slightly greater effect size for the association between household mold and childhood asthma (aOR 1.46; 95% CI: 1.10–1.93; Table S4).

Table 3.

Unadjusted and adjusted odds ratios of current asthma in association with household mold and pesticide use stratified by child’s sex among children aged 3–17 years, 2017–2018 National Survey of Children’s Health

	Boys		Girls
	Crude*	Adjusted†	Crude*	Adjusted†
N (asthma)	2,037	1,321	1,592	1,156
N (all)	21,444	12,709	19,979	11,768
Household mold
No	1 (reference)	1 (reference)	1 (reference)	1 (reference)
Yes	1.58 (1.17, 2.14)	1.57 (1.03, 2.39)	1.43 (1.08, 1.90)	1.28 (0.90, 1.84)
Household pesticide use
Never	1 (reference)	1 (reference)	1 (reference)	1 (reference)
Ever	1.13 (0.92, 1.40)	0.88 (0.69, 1.12)	1.04 (0.84, 1.27)	0.99 (0.77, 1.26)

^*P for interaction between household mold and child’s sex <0.10. P for interaction between household pesticide use and child’s sex >0.15.

^†Adjusted for child’s age, sex, race/ethnicity, insurance type, household highest education, poverty status, family structure, household smoker, primary caregiver’s self-reported mental and emotional health status, and mutually adjusted for household mold and pesticide use as exposures of interest. P for interaction between household mold and child’s sex <0.001. P for interaction between household pesticide use and child’s sex >0.15.

Table 4.

Unadjusted and adjusted odds ratios of current asthma in association with household mold and pesticide use stratified by child’s race/ethnicity among children aged 3–17 years, 2017–2018 National Survey of Children’s Health

	Non-Hispanic White		Hispanic		Non-Hispanic Black		Other
	Crude*	Adjusted†	Crude*	Adjusted†	Crude*	Adjusted†	Crude*	Adjusted†
N (asthma)	2,293	2,275	457	450	444	435	435	427
N (all)	29,001	28,686	4,616	4,531	2,588	2,534	5,218	5,132
Household mold
No	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)
Yes	1.85 (1.39, 2.47)	1.71 (1.28, 2.28)	0.97 (0.57, 1.65)	0.88 (0.50, 1.52)	1.38 (0.88, 2.14)	1.15 (0.71, 1.85)	1.06 (0.60, 1.87)	0.95 (0.53, 1.69)
Household pesticide use
Never	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)	1 (reference)
Ever	1.05 (0.88, 1.24)	1.02 (0.87, 1.20)	0.71 (0.46, 1.11)	0.70 (0.45, 1.10)	1.49 (1.06, 2.10)	1.39 (0.99, 1.95)	1.01 (0.69, 1.47)	1.00 (0.71, 1.42)

^*P for interaction between household mold and child’s sex <0.001. P for interaction between household pesticide use and child’s sex >0.15.

^†Adjusted for child’s age, sex, race/ethnicity, insurance type, household highest education, poverty status, family structure, household smoker, primary caregiver’s self-reported mental and emotional health status, and mutually adjusted for household mold and pesticide use as exposures of interest. P for interaction between household mold and child’s sex <0.001. P for interaction between household pesticide use and child’s sex >0.15.

Discussion

In a nationally representative sample of children aged 3–17 years of diverse geographical and sociodemographic backgrounds in the U.S. in 2017–2018, children exposed to household mold were significantly more likely to have current asthma, whereas household pesticide use was not associated with current asthma, after adjusting for major covariates including child’s obesity status. We also observed significant sex differences: household mold exposure was only significantly associated with current asthma among boys not girls. The positive and significant household mold-asthma association was more pronounced among non-Hispanic White versus racial/ethnic minority children, which however, could be partially attributed to the smaller sample sizes in racial/ethnic minority groups. No significant effect modification was observed by child’s age or household poverty status. Findings from our nationwide study reinforced the potential of mitigating household mold to address childhood asthma across various age and socioeconomic groups and suggest that boys and non-Hispanic White children may be at higher risk of childhood asthma in association with household mold compared to their counterparts. Future large-scale population-based prospective studies are needed to confirm our findings.

Household mold and childhood asthma

Our findings were similar to some but not all previous studies. Specifically, cross-sectional and case-control studies conducted in China found that children with household mold exposure had increased odds of asthma (Chen et al., 2011; Wang et al., 2013; Zheng et al., 2002). One prospective cohort study in Finland also confirmed that the presence of dampness or mold in the home was an independent risk factor for childhood asthma (Jaakkola et al., 2005). In the U.S., studies have been inconsistent, with some studies reporting positive associations (Iossifova et al., 2009; Jones et al., 2011; Reponen et al., 2012; Reponen et al., 2011; Vesper et al., 2008; Vesper et al., 2006) and a few showing null associations (McConnell et al., 2002) between household mold and childhood asthma. However, inferences from these studies were largely limited by small sample sizes ranging from 19 to 1233 children, restricted geographical and racial/ethnic variation, and possible residual confounding due to unavailability of data on important covariables, including child’s insurance type and obesity status (Caillaud et al., 2018). Our findings extend the literature by providing contemporary, population-based, and racially/ethnically diverse data on household mold and parent-reported physician diagnosis of asthma among children in the U.S. at a national level.

The exact mechanism underlying the household mold-childhood asthma association remains to be elucidated. However, previous data suggest its biological plausibility. Microbes and mycotoxins can produce volatile and irritative organic compounds which may induce immunoglobulin E-mediated hypersensitivity of the respiratory tract (Chen et al., 2011) and subsequently an increased risk of asthma. Further, there could be a high degree of variance in the amounts and types of molds, as observed previously (Baxi et al., 2013). Future investigations assessing specific type and amount of mold in relation to childhood asthma are warranted.

Household pesticide use and childhood asthma

In our study, frequency of household pesticide use was not associated with current asthma status after adjusting for covariates. This is consistent with some, but not all, previous studies. One study suggested that pesticide spraying was not associated with the number of asthma medical visits (Karpati et al., 2004). In contrast, others reported that para-occupational pesticide exposure was strongly, cross-sectionally associated with childhood asthma (OR 4.61; 95% CI 2.06–10.29) among Lebanese children, whereas residential and domestic pesticide exposures were only marginally associated with childhood asthma with 1.01 and 1.00 lower bounds of the 95% CIs, respectively (Salameh et al., 2003). In addition, a birth cohort study reported that early-life exposure to organophosphate pesticides was associated with respiratory symptoms consistent with possible asthma diagnosis at 5 and 7 years of age in an agricultural community in California (Raanan et al., 2015). Indeed, children could be exposed to pesticides through multiple pathways, including prenatal exposures, living or studying near areas with pesticide use, attending agricultural areas, having parents working in agriculture, and eating and drinking contaminated food and water (Buralli et al., 2020). Differences in exposure type, time window of measurements, and study populations might have contributed to the inconsistent findings. Many studies on children’s pesticides exposure focused on rural areas where children are more likely to be exposed to high-level para-occupational pesticides compared to relatively lower-level urban household exposures. It is noteworthy that in our study, almost half (43.5%) of participants reported any use of household pesticide use, and high frequency use (i.e., once a month or more) was rare (4.9%). Moreover, unlike mold exposure which may affect children continuously, household pesticide exposure may be sporadic, present for a relatively short time, and likely to degrade with time. Similarly, compared to (para-)occupational pesticide exposures, household pesticide exposure may have lower exposure doses and durations. Additionally, pest antigens are known to be associated with risk of asthma (Sheehan et al., 2010; Wu and Takaro, 2007). Therefore, lower levels of or even removal of pest antigens due to the use of pesticides could be a competing risk to the potential adverse impact of pesticide exposure on childhood asthma. Collectively, these differences in exposure nature and dosage and risk competition may have contributed to the null association of household pesticide use and childhood asthma that we observed.

Effect modification by sex and race/ethnicity

We also observed significant sex differences, with significant associations between household mold and asthma being observed among boys but not girls. We observed a significantly higher weighted prevalence of asthma in boys than in girls (9.4% vs. 7.7%; P=0.004), which is in line with previous data (Chen et al., 2003; Schatz and Camargo Jr, 2003). Previous studies have suggested that sex differences in lung development may be associated with sex differences in the onset of childhood asthma (Leon Hsu et al., 2015). Boys have smaller central airways than girls, which would increase boys’ susceptibility to asthma (Chen et al., 2003). Further, studies have illustrated sex differences in the development of immune responses during early childhood. The median of polyhydroxyalkanoate-induced Interferon-γ responses of boys was 1.6 times the level of girls at age 3 years (Uekert et al., 2006). It is plausible that a certain level of mold exposure may disproportionally disturb immunological processes and increase the risk of asthma among boys as compared to girls. Collectively, these sex differences might contribute to sex-specific associations of household mold exposure and childhood asthma.

When stratified by race/ethnicity, associations between household mold and asthma were only significant among Non-Hispanic Whites but not racial/ethnic minority children, suggesting that Non-Hispanic White children may be more susceptible to asthma risk attributable to household mold exposure compared to other potential asthma risk factors. Additionally, we cannot rule out the possibilities of insufficient statistical power due to smaller sample sizes of racial/ethnic minority children and more important roles of other measured or unmeasured factors in asthma among racial/ethnic minority children.

Strengths and limitations

Our study has some notable strengths. To our knowledge, it is the first large-scale investigation of current asthma among children in relation to two emerging but still largely understudied indoor exposures (i.e., household mold and pesticide use), using a contemporary nationally representative sample of U.S. children aged 3–17 years. The NSCH study design allows us to derive nationally representative estimates of childhood health and wellbeing, while accounting for potential biases due to non-response and non-coverage of non-telephone households (Luke and Blumberg, 2006). Further, our study includes various information on physical, mental, and social factors of the child, caregiver, and household, which allows us to minimize potential residual confounding.

Some limitations of this study merit note. The determination of asthma status was based on parent-reported physician diagnosis, which may be subject to potential recall and reporting bias. However, the weighted prevalence of current asthma (8.6%) among children aged 3–17 years in this study in 2017–2018 was similar to the national prevalence among children aged <18 years in 2017 (8.4%) (Centers for Disease Control and Prevention, 2018), supporting the potential validity and accuracy of the outcome assessment. The exposures of household mold and pesticide use were also parent-reported and not objectively measured. However, previous studies demonstrated that the agreement between parent-reported mold and field investigation by health inspectors was reasonably high (kappa statistic up to 0.70), with no indication of differential reporting by status of childhood respiratory symptoms (Andrae et al., 1988; Verhoeff et al., 1995). Despite that we were not able to assess seasonable variability in household mold exposure, the NSCH survey collected data on household mold during the past 12 months to integrate seasonal variability into the data. Information on the amounts and types of mold and pesticides was lacking. Hence, we could not investigate exposure-response relationships or ascertain the specific types of mold driving the observed associations. Nonetheless, our findings of a robust and strong association between household mold and childhood asthma may stimulate further investigations with objective measurements of mold types and amounts. We examined the potential effect modification by child’s race/ethnicity; however, we could not rule out the possibility of insufficient statistical power among the minority groups. Given that the prevalence of asthma may differ even within certain racial/ethnic subgroups (e.g., among Hispanics, higher in Puerto Ricans and lower in Mexicans compared to non-Hispanic Whites) (Centers for Disease Control and Prevention, 2020), future investigations to examine the potential effect modification across racial/ethnic subgroups are warranted. Further, while we focused on household mold and pesticide use as two major indoor environmental factors, there were no data collected on other indoor environmental risk factors such as indoor allergens or emissions from gas and wood stoves or building materials; however, household mold has been found to be an independent predictor of allergen load (Salo et al., 2006). Future studies with more comprehensive, objective assessment of indoor environment factors and potential interactions amongst these factors are needed. Finally, due to the cross-sectional design of the NSCH surveys, temporality is an issue we cannot ignore. We cannot rule out the possibility that some of the null findings we observed were due to households with asthmatic children that had intentionally improved their indoor environments and reduced allergen exposure. Our findings may stimulate future prospective cohort studies to examine the associations between household environment and risk of childhood asthma.

Conclusion

This study suggests that household mold exposure is associated with childhood current asthma, independent of other major risk factors, including obesity status. Given that children may spend up to 90% of their time indoors and that indoor environments are potentially modifiable, our findings may inform strategies targeting improvement of household environment to address childhood asthma. Future prospective investigations are needed to confirm our findings. Further research is warranted amid the COVID-19 pandemic, as children may be more vulnerable than ever to extended time of exposure indoors and limited access to adequate medical care.