Association between Atopic Diseases and Neurodevelopmental Disabilities in a Longitudinal Birth Cohort

Abstract

Reports on the association between the prevalence of atopic diseases and neurodevelopmental disabilities (NDs) have been inconsistent in the literature. We investigated whether autism spectrum disorder (ASD), attention deficit-hyperactivity disorders (ADHD), and other NDs are more prevalent in children with asthma, atopic dermatitis (AD) and allergic rhinitis (AR) compared to those without specific atopic conditions. A total of 2580 children enrolled at birth were followed prospectively, of which 119 have ASD, 423 have ADHD, 765 have other NDs, and 1273 have no NDs. Atopic diseases and NDs were defined based on physician diagnoses in electronic medical records. Logistic regressions adjusting for maternal and child characteristics estimated the associations between NDs (i.e., ASD, ADHD, and other NDs) and asthma, AD and AR, respectively. Children with asthma, AD or AR had a greater likelihood of having ADHD or other NDs compared with children without specific atopic conditions. The association between ASD and asthma diminished after adjusting for maternal and child factors. Either mothers or children having atopic conditions and both mothers and children with atopic conditions were associated with a higher prevalence of ADHD in children, compared with neither mothers nor children having atopic conditions. Children diagnosed with multiple atopic diseases were more likely to have NDs compared with those without or with only one type of atopic disease. In conclusion, in this US urban birth cohort, children with atopic diseases had a higher co-morbidity of NDs. These findings have implications for clinical care and future research on shared early life antecedents.

Lay summary

This US birth cohort study found a significant increase in the co-morbidity of ADHD, ASD and other NDs among children with atopic diseases (AD, AR, and asthma), especially, those with multiple atopies and with maternal atopic disease history. The findings have implications for etiologic research that searches for common early life antecedents of NDs and atopic conditions. Findings from this study also should raise awareness among health care providers and parents about the possible co-occurrence of both NDs and atopic conditions, which calls for coordinated efforts to screen, prevent and manage NDs and atopic conditions.

Introduction

Neurodevelopmental disabilities (ND) are a heterogeneous group of conditions that emerge early in life and often persist throughout the lifespan. NDs have a long-term negative impact on education, employment, health and well-being (Howlin & Magiati, 2017) (Arnold et al., 2020). Common ND diagnoses include autism spectrum disorder (ASD), attention deficit-hyperactivity disorders (ADHD), intellectual disabilities and other disorders in communication, learning disabilities and motor problems. The overall prevalence of NDs in the US among children aged 3–17 years rose from 16.2% to 17.8% between 2009 and 2017 (Zablotsky et al., 2019). NDs are highly comorbid conditions (Levy et al., 2010) (Muskens et al., 2017). The presence of psychiatric and medical comorbidities further impedes long-term positive outcomes for children with NDs, warranting improved assessment of and interventions for NDs.

Atopic diseases are a group of chronic diseases with shared underlying immune system dysfunction, including atopic dermatitis (AD), allergic rhinitis (AR), and asthma (Moreno, 2016). These three atopic diseases are quite prevalent in childhood. In the US, the reported prevalence of clinically diagnosed, pediatric eczema and asthma was 9.5% and 12.8%, respectively; and a quarter of children and adolescents aged 6 to 19 had a history of rhinitis (Drury et al., 2016) (Shargorodsky et al., 2015). These chronic conditions influence quality of life among children and their caregivers (Haanpää et al., 2018) (Hwang et al., 2019), and often persist into adolescence and adulthood (Mortz et al., 2015) (To et al., 2020). Since these conditions share common causes, atopic diseases are often comorbid among children (Hong et al., 2012).

Several studies have reported a higher prevalence of atopic conditions among children with NDs compared with their typically developing peers (Levy et al., 2010) (Muskens et al., 2017). The health impacts of atopic diseases may be particularly profound among children with NDs because it is difficult to diagnose these diseases among a population with limited social and communication skills. If left untreated, atopic diseases could actually worsen the symptoms of NDs (Lu et al., 2014). Based on findings from available studies, the association between ASD and atopic diseases remains inconclusive (Kotey et al., 2014) (Zheng et al., 2016) (Jónsdóttir & Lang, 2017) (Billeci et al., 2015). As for ADHD, two systematic reviews have concluded that children with ADHD are more likely to have asthma, atopic eczema, and allergic rhinitis (Schans et al., 2017) (Miyazaki et al., 2017). For other types of NDs, like intellectual disabilities and learning disorders, there is a paucity of research outside of a recent paper showing a positive association between atopic dermatitis and learning disability independent of sociodemographic characteristics and comorbid neurodevelopment illness (Wan et al., 2020).

Three major limitations inhibit causal knowledge about the association between NDs and atopic disease. A primary limitation is that prior research rarely considered perinatal and early life factors as confounders or effect modifiers despite the fact that many perinatal and early life factors, like maternal obesity during pregnancy (Li et al., 2016) (Harpsøe et al., 2013), low birthweight (Franz et al., 2018) (Tedner et al., 2012) and preterm birth (Agrawal et al., 2018) (Trønnes et al., 2013), are associated with both NDs and atopic diseases. Second, atopic diseases are usually identified based on retrospective self-report or parent-report, which might introduce information and recall biases (Xu et al., 2018) (Yaghmaie et al., 2013). Third and finally, maternal atopic conditions may be an important etiologic component in the underlying mechanism of co-occurrence of child atopic condition and NDs. While several studies have assessed the association of maternal atopic diseases with various NDs, including ADHD, ASD and other ND, the findings have been inconsistent (Croen et al., 2005) (Croen et al., 2019) (Cowell et al., 2019).

To further the field, the primary goal of this study was to examine the co-occurrence of NDs and child atopic conditions. Additionally, we explored whether these associations varied by prenatal and perinatal risk factors, including maternal preconception obesity and pregestational and gestational diabetes, child birthweight and gestational age at birth. A secondary aim was to examine the joint association of maternal and child atopic conditions with NDs. This study fills an important gap in the literature by overcoming several major limitations in the literature by employing a prospective birth cohort that leverages electronic medical records to objectively assess the associations of NDs with atopic conditions, while taking into account important prenatal and perinatal factors.

Methods

Participants

The analytic sample used for our analysis was comprised of participants from the Boston Birth Cohort (BBC). Since 1998, the BBC has recruited mother/infant pairs from Boston Medical Center (BMC) on a rolling basis. Most of the women enrolled in the BBC self-identify as a low-income, urban, minority population with a high proportion of preterm birth infants. Study enrollment was conducted within 24–72 hours after delivery of a singleton live birth. Written informed consent was required for enrollment. Children with major birth defects were excluded. Information on demographic characteristics, socio-economic status, life-style, and clinical data were collected through a standardized questionnaire. Maternal venous blood samples were collected within 24–72 hours of birth. Clinical diagnoses for children based on International Classification of Diseases (ICD)-9 (before October 1, 2015) and ICD-10 (after October 1, 2015) for every clinical visit were collected by an electronic medical records (EMR) system since 2003. Children who were aged 2–18 years old at the last follow-up visit and had information about NDs and atopic diseases were included in the current study. Ethical approval for data collection and analysis was obtained from the Institutional Review Boards of Boston Medical Center and Johns Hopkins Bloomberg School of Public Health.

Identification of children with NDs

According to their EMR, children in the follow-up study who were ever diagnosed with ICD-9 codes 299.XX, or ICD-10 codes F84.0, F84.8, or F84.9 were classified as having ASD. Children who were ever diagnosed with ICD-9 codes 314.XX, or ICD-10 codes F90.X, were categorized as having ADHD. The other ND group was defined by a child ever having had ICD-9 codes 315.XX or ICD-10 codes F81.X outside of ASD and ADHD in the EMR. In order to ensure a relatively “clean” reference group, children without any diagnosis of ASD, ADHD, other NDs and/or any other mental, behavioral or neurodevelopmental disorder codes (i.e., ICD-9 mental disorders codes 290–319 and ICD 10 mental, behavioral and neurodevelopmental disorders codes F01-F99) constituted the no NDs group. A diagnostic hierarchy was created for this study in which ASD was coded first, ADHD second, other NDs third, and no NDs fourth. Thus, children with ASD could also have ADHD, but not vice versa; the same was applied for ADHD vs. other NDs.

Identification of children with atopic diseases

Child atopic diseases were also extracted from the EMR (asthma: ICD-9 codes 430.XX or ICD-10 codes J450.XX; atopic dermatitis: ICD-9 codes 691.XX or ICD-10 codes L20.XX; allergic rhinitis: ICD-9 codes 477.XX or ICD-10 codes J30.XX). Asthma, AD and AR were coded as binary outcomes (ever had been diagnosed with the disease (i.e., had a given diagnosis code on at least 1 visit) versus never had been diagnosed with the disease) based on retrospective physician diagnosis. For the analysis, a 3-categorical variable was used to summarize the number of atopic diseases: 0= never had any of the above atopic diseases; 1= had been diagnosed with a single type of the above atopic diseases; and 2= had been diagnosed with two or three types of the above atopic diseases. Self-reported maternal atopic conditions were collected in interviews with mothers about prior history of physician-diagnosed food allergy, AD, AR and asthma, and were coded as a dichotomized variable: 0=never had any diagnosed atopic conditions; 1=ever had been diagnosed with any atopic conditions.

Covariates

Maternal potential confounding factors considered in the analysis were: maternal race/ethnicity (self-reported Black, non-Black), maternal highest education level (below high school, high school, college and above), maternal age at time of delivery (aged <=35 years vs >35 years), maternal pre-pregnancy obesity and diabetes, maternal smoking status during pregnancy (never vs. ever). A previous study reported that the joint effect of maternal obesity and diabetes on ASD or intellectual disability was greater than the effect of either obesity or diabetes alone (Li et al., 2016).Thus, we combined maternal obesity and diabetes as a composite variable to reflect either a body mass index (at pre-pregnancy weight) >30 or pregestational/gestational diabetes. Child confounding factors included child sex (female, male), child birthweight (<2500g vs >=2500g), gestational age at birth (<37 weeks vs >=37 weeks) and child age at last visit (aged <6 years; 6–12 years; >12 years).

Statistical Analysis

Descriptive analysis was performed to describe the distribution of maternal and child factors and atopic diseases across different groups of NDs. Chi-square tests were applied to 1) assess the difference between any ND group and the no ND group, and 2) assess the difference across the three ND groups (ASD, ADHD and the other ND group). As for the dichotomized variable “any ND” (any ND group versus no ND group), binary logistic regression models were employed to 1) test the association between any ND and each atopic condition (asthma, AD or AR), separately; 2) test the association between any ND and the number of atopic diseases (0, 1, 2 and more). Multinomial logistic models examined the three ND groups (ASD, ADHD and other NDs) as the dependent variable, and each atopic disease and number of atopic diseases as the independent variable, respectively. Adjusted binary and multinomial logistic models included child sex, child birthweight, gestational age at birth, child age at last visit, maternal age at delivery, maternal highest education level, maternal race/ethnicity, maternal smoking during pregnancy, and maternal obesity and diabetes as covariates. Among the covariates, maternal smoking status during pregnancy, maternal diabetes and obesity, and maternal education level had missing values, all of which were missing less than 10%. Missing values were imputed using multiple imputation by chained equations (MICE). We further tested for effect modification in the adjusted models using the interaction terms of any NDs with each of the maternal and child characteristics, and then conducted stratified analyses by the maternal and child characteristics.

Next, we examined the joint effect of maternal and child atopic diseases on NDs. Since the proportion of missing values was 26.6% for maternal atopic conditions, we conducted MICE for maternal atopic conditions and the three covariates with missing values mentioned above (maternal smoking status during pregnancy, maternal diabetes and obesity and maternal education level) with 10 iterations. After imputation, we created a new variable with three levels to combine maternal and child atopic diseases (neither mother nor child had atopic disease, either mother or child had atopic diseases, and both mother and child had atopic diseases). We then assessed the joint association of maternal and child atopic diseases with NDs using binary logistic models (any ND versus no ND) and multinomial logistic models (three ND group versus no ND group). We applied this approach, rather than including maternal atopic condition as a covariate, since we considered maternal atopic condition as an effect modifier in the relationship between child atopic conditions and NDs. Moreover, we performed the sensitivity analysis using complete case data (only the participants with complete information on child and maternal atopic conditions were included; missing data for other covariates were imputed by MICE, as mentioned above) (Table S1). Odds ratios (ORs) and 95% confidence intervals (95% C.I.) were reported for binary logistic regression coefficients, and Relative Risk Ratios (RRRs) were reported for multinomial logistic regression coefficients. The significance level was set at p<0.05. All statistical analyses were conducted in STATA 14.0 (College Station, TX).

Results

Table 1 presents the comparison of demographic characteristics across children with different types of NDs and no NDs. A total of 2580 children were included in the final analytic dataset: 1339 boys (51.9%) and 1543 Black individuals (59.8%); mean (SD) age at last follow-up visit was 9.0 (3.9) years. More than a quarter of the children were born prematurely (29.2%) or had low birthweight (27.7%). More than half of the mothers had a high school or higher education level. About a quarter of the mothers had smoked (24.4%) or had diabetes or obesity during pregnancy (28.8%). One in five mothers had atopic conditions (19.8%), while the proportion of missing values was 26.6%. The proportions of children with ASD, ADHD (and without comorbid ASD) and other NDs (after removing children with co-diagnosis of ASD or ADHD) were 4.6%, 16.4% and 29.6% in this sample. In total, there were 1307 children (50.7%) who were diagnosed with at least one type of ND.

Table 1.

Population characteristics for participating mother-child pairs, stratified by type of neurodevelopmental disabilities (NDs)

	No ND (n=1273)	Any ND (n=1307)	ASD (n=119)	ADHD (n=423)	Other NDs (n=765)
Maternal characteristics
Maternal race/ethnicity, n(%)
Black	745(58.5)	798(61.1)	63(52.9)	257(60.8)	478(62.5)
White	68(5.3)	78(6.0)	6(5.0)	29(6.9)	43(5.6)
Hispanic	291(22.9)	294(22.5)	36(30.2)	98(23.2)	160(20.9)
Other	169(13.3)	137(10.5)	14(11.8)	39(9.2)	84(11.0)
Maternal education, n(%)b
Below high school	352(27.6)	363(27.8)	25(21.0)	122(28.8)	216(28.2)
High school	461(36.2)	485(37.1)	42(35.3)	168(39.7)	275(36.0)
College degree or above	454(35.7)	453(34.7)	50(42.0)	130(30.7)	273(35.7)
Missing	6(0.5)	6(0.5)	2(1.7)	3(0.7)	1(0.1)
Maternal age at delivery ≥ 35 years, n(%)	217(17.0)	247(18.9)	26(21.8)	69(16.3)	152(19.9)
Maternal smoking during pregnancy, n(%)ab
Never	996(78.2)	939(71,8)	89(74.8)	276(65.2)	574(75.0)
Ever	267(21.0)	362(27.7)	29(24.4)	146(34.5)	187(24.4)
Missing	10(0.8)	6(0.5)	1(0.8)	1(0.2)	4(0.5)
Maternal diabetes and obesity, n(%)a
None	864(67.9)	822(62.9)	64(53.8)	275(65.0)	483(63.1)
Either diabetes or obesity or both	334(26.2)	410(31.4)	45(37.8)	124(29.3)	241(31.5)
Missing	75(5.9)	75(5.7)	10(8.4)	24(5.7)	41(5.4)
Maternal Atopy Conditions*, n(%)b
None	692(54.4)	689(52.7)	56(47.1)	205(48.5)	428(56.0)
Ever	231(18.2)	281(21.5)	24(20.2)	129(30.5)	128(16.7)
Missing	350(27.5)	337(25.8)	39(32.8)	89(21.0)	209(27.3)
Child Characteristics
Female, n(%)ab	741(58.2)	500(38.3)	31(26.0)	119(28.1)	350(45.8)
Child age at last visit (years), n(%)ab
2–6 (<6)	389(30.6)	268(20.5)	26(21.8)	23(5.4)	219(28.6)
6–12 (<12)	643(50.5)	691(52.9)	65(54.6)	241(57.0)	385(50.3)
12–18(<18)	241(18.9)	348(26.6)	28(23.5)	159(37.6)	161(21.0)
Preterm birth (<37 weeks), n(%)a	314(24.7)	443(33.9)	46(38.7)	126(29.8)	271(35.4)
Low birthweight (<2500 g), n(%)a	293(23.0)	422(32.3)	42(35.3)	125(29.6)	255(33.3)
Asthma, n(%)ab	292(22.9)	439(33.6)	37(31.1)	169(40.0)	233(30.5)
Atopic Dermatitis, n(%)a	360(28.3)	480(36.7)	41(34.4)	166(39.2)	273(35.7)
Allergic Rhinitis, n(%)ab	222(17.4)	351(26.9)	29(24.4)	139(32.9)	183(23.9)
Number of Atopic Diseases, n(%)ab
0	662(52.0)	522(39.9)	56(47.1)	142(33.6)	324(42.4)
1	395(31.0)	421(32.2)	27(22.7)	142(33.6)	252(32.9)
≥2	216(17.0)	364(27.8)	36(30.2)	139(32.9)	189(24.7)

^aThe p value for the chi-square test between the any ND and no ND group was less than 0.05

^bThe p value for the chi-square test among the three ND groups (ASD, ADHD and Other NDs) was less than 0.05.

^*Maternal atopy conditions include asthma, atopic dermatitis, and allergic rhinitis.

Children with NDs were, on average, more likely to be boys, older in age at the last follow-up visit, and were more likely to have preterm birth and low birthweight compared to their peers without any NDs (Table 1). The mothers of children with any NDs were more likely to have ever smoked and have diabetes or obesity during pregnancy compared to the mothers of children with No-NDs. Across the NDs groups (ASD, ADHD and other NDs), children with ASD and ADHD were more likely to be boys; children with ADHD were older in age at the last follow-up visit; and the mothers of children with ADHD were more likely to have ever smoked during the pregnancy and have atopic conditions. As for atopic diseases, 731 (28.3%) of 2580 children had a diagnosis of asthma, 840 (32.6%) had a diagnosis of AD, and 573 (22.1%) had a diagnosis of AR. The proportion of children diagnosed with atopic diseases was significantly higher in children with any NDs compared with children with no NDs. Children with ADHD had the highest proportion of those diagnosed as having asthma, AD and AR compared with children with ASD and other NDs.

Table 2 shows the results for both the crude and adjusted associations of NDs with atopic diseases. Children with a diagnosis of asthma, AD and AR were 38% (adjusted OR (aOR)= 1.38 (95%CI:1.15–1.66), p=0.001), 52% (aOR=1.52(1.28–1.81), p<0.001) and 42% (aOR=1.42(1.17–1.74), p=0.001) more likely to have a diagnosis for any type of ND, when compared to children without specific diagnosed atopic conditions, respectively. When each specific ND was analyzed, the risks of ADHD and other NDs were higher in those children with asthma, AD and AR in both the crude and adjusted models. However, having a diagnosis of ASD was not associated with any of the atopic conditions. The association of asthma with ASD was identified in the crude model, however it did not maintain significance in the adjusted model (adjusted RRR (aRRR) =1.16 (0.76–1.79), p=0.490).

Table 2.

The association of child atopic diseases with neurodevelopmental disabilities (NDs)

		Asthma
	Count	Crude Model		Adjusted Modela
		OR/RRR# (95%CI)	P value	OR/RRR# (95%CI)	P value
Any NDs	439	1.70(1.43–2.02)	<0.001	1.38(1.15–1.66)	0.001
ASD	37	1.52(1.01–2.28)	0.046	1.16(0.76–1.79)	0.490
ADHD	169	2.24(1.77–2.82)	<0.001	1.69(1.31–2.18)	<0.001
Other NDs	233	1.47(1.20–1.80)	<0.001	1.26(1.02–1.56)	0.028
No NDs	292	Ref.	-	Ref.	-

		Atopic Dermatitis
	Count	Crude Model		Adjusted Modela
		OR/RRR# (95%CI)	P value	OR/RRR# (95%CI)	P value
Any NDs	480	1.47(1.25–1.74)	<0.001	1.52(1.28–1.81)	<0.001
ASD	41	1.33(0.90–1.98)	0.156	1.38(0.92–2.08)	0.121
ADHD	166	1.64(1.30–2.06)	<0.001	1.71(1.33–2.19)	<0.001
Other NDs	273	1.41(1.16–1.70)	<0.001	1.45(1.19–1.77)	<0.001
No NDs	360	Ref.	-	Ref.	-

		Allergic Rhinitis
	Count	Crude Model		Adjusted Modela
		OR/RRR# (95%CI)	P value	OR/RRR# (95%CI)	P value
Any NDs	351	1.74(1.44–2.10)	<0.001	1.42(1.17–1.74)	0.001
ASD	29	1.52(0.98–2.38)	0.062	1.22(0.77–1.94)	0.396
ADHD	139	2.32(1.80–2.97)	<0.001	1.61(1.23–2.11)	<0.001
Other NDs	183	1.49(1.19–1.86)	<0.001	1.35(1.07–1.70)	0.010
No NDs	222	Ref.	-	Ref.	-

^aAdjusted for child sex, child age at last visit, child birthweight, gestational age, maternal race/ethnicity, maternal education, maternal age at delivery, maternal smoking during pregnancy, maternal diabetes and obesity.

Note. For outcome, the reference group is no NDs; for each exposure, the reference group is without the specific atopic condition.

^#Odds ratios (ORs) were reported for logistic regression coefficients of binary outcome (Any NDs, yes, no), and Relative Risk Ratios (RRR) were reported for multinomial logistic regression coefficients (ASD, ADHD, and compared other NDs compared to No NDs).

When atopic diseases were assessed as a summary variable (number of atopic diseases; see Table 3), having a diagnosis of two or three atopic diseases was associated with ASD (but not having only one atopic disease), whereas having at least one type of atopic disease was associated with any NDs, ADHD or other NDs. Additionally, having two or three atopic diseases had a larger effect on the risk of having any NDs, ADHD or other NDs, compared with having only one atopic disease. We further tested if the associations of having any NDs and asthma, AD or AR varied across subgroups defined by maternal and child characteristics (see Figures S1–S3). There were no significant interactions between having any NDs and the maternal and child characteristics for the outcome of atopic diseases (all p >0.05). There were some differences in point estimates of the associations for preterm birth and low birthweight, but the 95% confidence intervals overlapped.

Table 3.

The association of the number of atopic disease diagnoses for a child with the diagnosis of neurodevelopmental disabilities (NDs)

	Number of atopic diseases =1			Number of atopic diseases ≥2
	N	Crude Model	Adjusted Modela	N	Crude Model	Adjusted Modela
Any NDs	421	1.35(1.13–1.62)**	1.25(1.04–1.50)*	364	2.14(1.74–2.62)***	1.78(1.43–2.20)***
ASD	27	0.81(0.50–1.30)	0.71(0.44–1.16)	36	1.97(1.26–3.08)**	1.59(1.00–2.53)*
ADHD	142	1.68(1.29–2.18)***	1.49(1.12–1.96)**	139	3.00(2.27–3.97)***	2.20(1.63–2.98)***
Other NDs	252	1.30(1.06–1.60)*	1.24(1.00–1.53)*	189	1.79(1.41–2.26)***	1.61(1.26–2.06)***

^aAdjusted for child sex, child age at last visit, child birthweight, gestational age, maternal race/ethnicity, maternal education, maternal age at delivery, maternal smoking during pregnancy, maternal diabetes and obesity.

^*p <0.05

^**p<0.01

^***p<0.001

Note. For the outcome, the reference group is no NDs; for the exposure, the reference group is “Number of atopic diseases=0”

Next, we explored the joint effect of maternal and child atopic conditions. According to our descriptive analysis using only complete cases (only participants with complete information for child and maternal atopic conditions), the prevalence of each ND group increased with an increasing number of child atopic diseases with the exception of ASD among complete cases (N=1893), whether the mother had atopic conditions or not (see Figure 1). The prevalence of ADHD was higher for children with maternal atopic conditions across different numbers of child atopic conditions, whereas the prevalence of other NDs showed the opposite pattern. We then assessed the combined effect of maternal and child atopic conditions on NDs using imputed data in the regression models (see Table 4). Either mother or child having atopic diseases and both mother and child having atopic conditions was associated with a higher likelihood of having any NDs or ADHD. The effect size of the associations was larger for both mother and child having atopic conditions (for any NDs, aOR=1.56(1.18–2.06); for ADHD, aRRR=2.65(1.84–3.82)) compared with either mother or child having atopic diseases (for any NDs, aOR=1.26(1.04–1.54); for ADHD, aRRR=1.54(1.12–2.10)). The regression models, using complete case data (see Table S1), generated similar results as the imputed data analysis (Table 4).

Figure 1.

Prevalence of neurodevelopmental disabilities (NDs) stratified by maternal and child atopic conditions

Note: Group 1: no maternal atopic condition and no child atopic condition; Group 2: no maternal atopic condition and 1 type child atopic condition; Group 3: no maternal atopic condition and 2+ types child atopic condition; Group 4: with maternal atopic condition and no child atopic condition; Group 5: with maternal atopic condition and 1 type child atopic condition; Group 6: with maternal atopic condition and 2+ types child atopic condition

Table 4.

Joint association of maternal and child diagnosed atopic diseases with the diagnosis of neurodevelopmental disabilities (NDs)

	Either mom or child with atopic diseases			Both mom and child with atopic diseases
	N	Crude Model	Adjusted Modela	N	Crude Model	Adjusted Modela
Any NDs	466	1.40(1.16–1.69)***	1.26(1.04–1.54)*	200	1.88(1.45–2.44)***	1.56(1.18–2.06)**
ASD	44	1.48(0.89–2.44)	1.29(0.78–2.15)	11	1.59(0.86–2.93)	1.29(0.67–2.49)
ADHD	159	1.80(1.34–2.42)***	1.54(1.12–2.10)**	101	3.32(2.37–4.66)***	2.65(1.84–3.82)***
Other NDs	263	1.24(1.01–1.53)*	1.17(0.94–1.45)	88	1.38(1.00–1.89)*	1.22(0.88–1.70)

^aAdjusted for child sex, child age at last visit, child birthweight, gestational age, maternal race/ethnicity, maternal education, maternal age at delivery, maternal smoking during pregnancy, maternal diabetes and obesity.

^*p <0.05

^**p<0.01

^***p<0.001

Note. For the outcome, the reference group is no NDs; for the exposure, the reference group is “neither mother nor child with atopic disease”.

Discussion

Our findings provide evidence of a higher prevalence of atopic diseases among children with NDs (as compared to those without NDs) in this US urban population. We found that children diagnosed with more than one type of atopic disease were more likely to have a diagnosed ND. Diagnoses of ADHD and other NDs were associated with an increased likelihood of having asthma, AD and AR. ASD was not associated with any of the three types of atopic diseases in the adjusted models. Our findings are consistent with previous studies demonstrating significant associations between ADHD and atopic diseases (Schans et al., 2017) (Miyazaki et al., 2017). In our study, the strength of the associations between ADHD and the three atopic diseases was similar, while the previous meta-analysis by Miyazaki et al. reported that the association between ADHD and asthma was stronger than the associations between ADHD with AD and AR (Miyazaki et al., 2017). Findings for the relationship between ASD and specific atopic diseases remained inconclusive. Our finding of an insignificant association between ASD and AD is in contrast with the conclusion from a systematic review conducted by Billeci and colleagues (Billeci et al., 2015). As for asthma, we found no evidence of an association with ASD independent of other maternal and child covariates, which is in agreement with a recent meta-analysis of cross-sectional and case-control studies (Zheng et al., 2016) and provides more evidence from this prospective birth cohort. The relationship between ASD and AR has been less studied than ASD and asthma or AD, and among relevant studies there has been a lack of consistent conclusions (Akintunde et al., 2015) (Zerbo et al., 2015).

Interestingly, we found a consistent association between other NDs and the three types of atopic diseases, which suggests the need for more attention to the diagnosis of atopic diseases among children with other types of NDs besides ASD and ADHD. It is important to note that the other NDs group consists of heterogeneous diagnoses in developmental delay, including developmental speech or language disorders, developmental learning difficulties, developmental coordination disorders and other specific delays in development. While evidence has shown that childhood AD conditions have been associated with developmental delays (Jackson-Cowan et al., 2021) and, particularly, learning disability (Wan et al., 2020), less is known about the link between other atopic conditions and specific delays in development. Future studies with larger sample sizes are needed to more definitively answer these questions. Moreover, our study found that children with a cluster of atopic diseases (i.e., more than two types of atopic diseases) were more likely to have NDs, which indicates that the comorbidity pattern of atopic diseases is not only common in the general child population but also should arise particular concern among those who care for children with NDs. A previous study reported that a greater number of atopic comorbidities before age three years was associated with a greater risk of ADHD and ASD in later life (Chen, M. H. et al., 2014). Our findings further confirmed that having other NDs (other than ADHD and ASD) was also associated with having a greater number of atopic comorbidities and extends on the above reported association identified in early childhood to include all of childhood.

Several studies have found associations between increased risk for NDs and atopic conditions in the mother (Croen et al., 2019) (Liu et al., 2019), as well as links between maternal and child atopic conditions (Paaso et al., 2014) (Wang et al., 2012). On the one hand, maternal atopic conditions may reflect inherited vulnerability to atopic diseases in their offspring. On the other hand, immune system disturbances during pregnancy from maternal atopy might affect the development of fetal immune responsiveness and be associated with the etiology of both child atopic conditions and NDs (Matson et al., 2007) (Ravaccia & Ghafourian, 2020). Our findings indicated that the combined effect of maternal and child atopic conditions on ADHD was substantially larger than their respective effects, which might support the hypothesis that environmental factors act as a second hit in individuals with high genetic susceptibility and thus require further investigation (Ravaccia & Ghafourian, 2020).

Our findings support the hypothesis that atopic diseases and NDs may share a common pathogenic pathway. Atopic disease is a group of diseases that share a common underlying immune system dysfunction (Moreno, 2016). Furthermore, evidence has identified immune system dysfunction and inflammatory process as potential etiological mechanisms for NDs. Several immune-related abnormalities and changes were found in individuals diagnosed with NDs (Bilbo, Block, Bolton, Hanamsagar, & Tran, 2018). For example, increased levels of cytokines have been reported in children with ASD, and a correlation between serum levels of inflammatory cytokines and ADHD symptom severity also has been observed (Hu et al., 2018) (Molloy et al., 2006) (Cortese et al., 2019). The evidence linking ASD and ADHD with immune system dysfunction also includes observed changes in the pattern of immune cells, the presence of autoantibodies, and increased levels of immunoglobulin (Hughes et al., 2018) (Verlaet et al., 2014). A further hypothesis about the link between immune system dysfunction and NDs is that maternal inflammatory events induced by environmental factors during pregnancy could disrupt fetal immune and nervous system development, and thus increase the risk of NDs in offspring; this hypothesis is supported by findings from various laboratory experiments (Bilbo et al., 2018). Moreover, immune system dysfunction found in family members of individuals with NDs suggests potential shared genetic liabilities (Atladóttir et al., 2009) (Croen et al., 2019). The complex genetic associations between NDs and immune system dysfunction have been explored in various types of studies. For example, a case report provided evidence that 22q13 deletion syndrome, which has autistic behaviors and global developmental delays as clinical features, may be associated with immune system dysfunction (Chen, C. P. et al., 2010). A population-based study reported the association of the C4B null allele with both autism and family history of autoimmune diseases (Mostafa & Shehab, 2010). Some recent studies identified a significant genetic correlation between ADHD and various immune-related phenotypes at a genome-wide level, including asthma, psoriasis and serum C-reactive protein levels (Zhu et al., 2019) (Tylee et al., 2018). Given the high occurrence of atopic diseases among children diagnosed with NDs, specific atopic disease management guidelines for children with NDs to help guide clinicians and parents is essential. Although there are effective treatments for atopic diseases, it is often difficult for children with NDs to adhere to treatment plans designed to address these chronic conditions. For example, children with typical development might more easily master the knowledge and skills required to perform self-care for asthma, as compared with children with NDs, which could improve the effectiveness of treatment and decrease the burden for parents.

Our stratified analysis, which explored the influence of maternal and child risk factors on the association between NDs and atopic diseases, aimed to identify the target populations at high risk and thus support the diagnosis of atopic diseases in children with NDs. The findings from our study indicated that the associations between NDs and atopic diseases were stronger among children with preterm birth and low birthweight, though the differences were not significant. Since preterm birth and low birthweight are two known risk factors for asthma and NDs, separately (Matheson et al., 2017) (Schieve et al., 2016), further research with larger sample sizes might explore the potential biological mechanisms underlying how preterm birth and low birthweight influence the association between NDs and atopic diseases.

Some limitations of our study should be taken into consideration when interpreting the findings. First, the diagnoses of NDs and atopic diseases were extracted from the EMR data based on ICD codes, instead of standardized diagnostic assessments; as such, misclassification bias may exist in the EMR data for NDs and atopic diseases. All the identified ASD cases in the BBC have been verified by a comprehensive medical chart review (including all inpatient and outpatient records and clinical evaluations for ASD), as diagnosed by a developmental or psychiatric specialist along with corresponding ICD-9 and/or ICD-10 codes. This validation was conducted by our field team consisting of a pediatric developmental specialist and highly trained research staff at the BMC. For ADHD and atopic disorders, all cases were identified through a systematic EMR query using all the relevant ICD-9 and 10 codes with corresponding physician diagnoses. In addition, a sensitivity analysis was conducted in which we required cases to have 2 or more clinical visits with diagnostic codes for the disease (see Supplementary Tables S2–S5). The results of the sensitivity analyses were consistent with the original analysis. While we cannot eliminate the possibility of case misclassification, we expect that misclassification, if present, is likely to be random, and likely to bias the association towards the null. Second, the number of NDs cases is relatively small in this cohort, especially ASD cases. Therefore, our results are likely to be influenced by random errors and have wide confidence intervals with decreased precision, especially for the results from the stratified analyses. Third, it is possible that children with ND diagnoses are followed more closely by health care providers, which can potentially increase the chances of identifying atopic diseases in this group as compared to children with typical development, and vice versa. However, this difference is difficult to tease out, because while atopic conditions and NDs tend to be diagnosed in early childhood, ADHD is typically diagnosed at school age. To explore the cumulative incidence of ASD and ADHD across the postnatal follow-up period, we performed additional analyses stratified by atopic status. As shown in Supplementary Figure S4, for ASD, we observed a similar age of onset and level of incidence regardless of atopic status. For ADHD, we observed a higher level of incidence among atopic children, but the group difference only became obvious after age 6 years, corresponding to the typical age when ADHD begins to be recognized and diagnosed. Although we cannot eliminate possible detection bias, the higher incidence of ADHD among children with atopic conditions is more likely attributable to co-morbidity. Future studies are needed to further delineate temporal relationships between atopic diseases and NDs and the biological mechanisms underlying their co-morbidity. Lastly, data came from children and mothers living in one urban site in the US and are therefore not generalizable to all children in the US. As presented in Table 1, the BBC is a high-risk birth cohort, which has a high proportion of preterm birth, maternal obesity and diabetes, all of which are well-known risk factors of ASD and ADHD. In addition, this is also a cohort at high risk of atopic conditions largely due to residence in an urban, low-income setting with poor housing conditions (presence of mice, cockroaches, etc.), which have been associated with increased risk of asthma or atopic diseases in general. However, as a result of our work in this enriched cohort, this study had greater statistical power and as such has contributed to a better understanding of the link between atopic diseases and NDs, within a traditionally understudied and underrepresented population. (Abuabara et al., 2017)

This is the first study of the association between atopic diseases and neurodevelopmental disabilities in a US urban, low income, predominantly minority birth cohort. We found a significant positive association between ASD, ADHD and other NDs and atopic diseases in general. Because NDs and atopic diseases are cared for by different pediatric subspecialties, findings from this study, if further confirmed, could lead to increased awareness among health care providers and parents about the possible co-occurrence of both NDs and atopic conditions in a given child, which could result in more coordinated efforts to screen, detect, prevent, and treat atopic diseases among children with NDs, and vice versa. These findings also have implications for the need of future etiology research to search for common early life antecedents of NDs and atopic conditions.

Abbreviations:

ND

neurodevelopmental disability

ASD

autism spectrum disorder

ADHD

attention deficit hyperactivity disorders

AD

atopic dermatitis

AR

allergic rhinitis

BBC

Boston Birth Cohort

BMC

Boston Medical Center

ICD

International Classification of Diseases

EMR

Electronic Medical Records

BMI

body mass index