Risk of Atopic Dermatitis and the atopic march paradigm in children of mothers with atopic illnesses: a birth cohort study from the United Kingdom

Abstract

Background:

Atopic Dermatitis (AD) is thought to precede the onset of other allergic illness (OAI) in a temporal progression (i.e., atopic march), yet the timing and progression has been questioned. It is also unclear how parental allergic illness impacts the development of these illnesses in offspring.

Objective:

(1) explore risk of incident AD and (2) timing of allergic disease onset in children of mothers with AD compared to mothers without AD from the United Kingdom.

Methods:

We created a birth-cohort of mother-child pairs using IQVIA Medical Research Data database and developed cox proportional models to examine the above associations (Hazard Ratio [95% Confidence Interval]).

Results:

Among 1,224,243 child-mother pairs, mean child (standard deviation) follow-up time was 10.8 (8.3) years and 50.1% were males (N=600,905). Children were 59% (HR=1.59 [1.57,1.60]) more likely to have AD if their mothers had AD compared to no AD with mean age of first AD diagnosis at 3.3 (4.8) years. Most children with any diagnosis of AD present with AD first (91.0%), however, in those with asthma, only 67.8% developed AD first.

Conclusion:

Children born to mothers with AD are more prone to develop AD and some develop OAI first, suggesting that not all follow the same sequential pathway.

Capsule summary:

• The atopic march describes the temporal association of initial onset of atopic dermatitis at an early age followed by subsequent development of other atopic illness.

• We observed that while this may be true for some, not all patients with allergic diseases including atopic dermatitis, will follow the same “sequential” march.

Introduction

Atopic dermatitis (AD) is a common, pruritic, inflammatory skin disease characterized by periods of acute disease flares.^1,2 AD is a potentially lifelong disease that begins in early childhood. ^1,2 AD has been associated with other allergic illness (OAI) such as asthma, seasonal allergies, and food allergies.^3–5 AD and these OAI may share a common “atopic” pathophysiology and/or environmental triggers that lead to disease exacerbation and while genetic risk factors and heritability may play a role, there is insufficient data demonstrating a common genetic etiology.^6–9 Understanding the pathophysiology and potential triggers that lead to onset of AD is critical as early childhood AD and OAI have been associated in a stepwise process called the atopic march. ^10,11 In this model, allergen presentation occurs via a defective skin barrier which leads to systemic sensitization and immune activation with resultant airway and allergic nasal hyperresponsiveness as seen in asthma and allergic rhinitis.^12,13

Studies using birth cohorts can also provide insight into the relationship between the presence of atopic illness in a parent and subsequent development of these same illnesses in their offspring. At least 25 birth-cohorts have been established to examine the “progressive” appearance of these illnesses.¹⁴ However, in most, the primary focus has been asthma and not AD and, because of their selective nature, most lack generalizability and may be prone to misclassification as they rely on caregiver-reported questionnaires.^14,15 The goal of this study was to examine the (1) risk of incident AD and (2) timing of AD and OAI onset in children of mothers with AD compared to mothers without AD using a large, population-based, database from the United Kingdom (UK).

Materials and methods

Study design:

a retrospective longitudinal cohort study using data from the IQVIA Medical Research Data (IMRD) dataset, which incorporates data from The Health Information Network (THIN). This large, primary care database collects anonymized, longitudinal, data from a patient’s clinical and prescribing record. IMRD includes data from 10% of the population of the UK and is believed to be representative of the general population.^16,17 IMRD contains information from 19,508,644 patients across 832 general practices which is collected using a series of READ codes, a numerical classification system to record health-care related symptoms and diagnoses.^16–18 Diagnostic codes for AD have been previously validated in this database.^19,20 Study data collection began in 2004, as this was the year that all practices that contributed data to IMRD and ended in 2021.

Definition of study population:

All children were registered to their general provider’s (GP) practice within 60 days of their delivery date as recorded in IMRD and had at least two office visits. To properly match mother and child pairs, all pairs used the same GP and resided in the same household, a method consistent with prior work.^19,21 We limited our study to mother-child pairs as there is no reliable method linking the father’s medical history with the child.

Follow-up time:

Follow-up time began on the child’s delivery date. End of follow-up occurred at the earliest of outcome diagnosis (i.e., AD or OAI in child); patient transfer out of practice, withdrawal of practice from IMRD, death or end of database data collection.

Exposure of interest:

Maternal history of AD and OAI before or during pregnancy

Outcome of interest:

(1) incident AD or (2) OAI in the children.

Covariates of interest:

included age of mother at delivery, Townsend deprivation index (a UK measure of economic inequality dichotomized to the top two deciles of deprivation), and a measure of ethnicity (likelihood of White ethnicity in region of residence).

Statistical analysis:

Descriptive statistics were used to summarize characteristics of the mother-child cohorts. Continuous and categorical variables were summarized using means (standard deviation [SD]) and proportions (percentages [%]) and tested using t-tests or Mann-Whitney tests or chi-square tests. Cox proportional hazards models (Hazards ratio, HR [95% confidence interval, 95%CI]) were used to examine the association between exposures in the mother identified a priori and onset of AD and OAI in the child. We performed sensitivity analysis to account for clustering within the mother as mothers may have had more than one child. Individuals with missing data were dropped from analysis as the prevalence of missing data across variables was never greater than 0.001%.

Protection of study subjects:

The study was approved by IMRD, Scientific Review Committee for UK Ethics as protocol number 22SRC042, and University of Pennsylvania Institutional Review Board. It was developed in accordance to RECORD guidelines for acceptable reporting of cohort studies.

Results

We analyzed 1,224,243 child-mother pairs. The characteristics of the mothers are listed in Table I. The average age at the time the child was born was 28.6 years (28.5, 28.6). 15.28% (15.21,15.34) of the mothers had AD. Mothers with AD also had a higher prevalence of OAI compared to mothers without AD (Table I).

Table I.

Characteristics of mothers in the overall cohort, mothers without atopic dermatitis and mothers with atopic dermatitis.

	Overall Cohort	Mothers without ADa	Mothers with AD
N (%)	1,224,243 (100)	913,693 (74.63)	310,550 (25.37)
Age (median, IQR)	28.6 (28.5, 28.6)	28.6 (28.6, 28.6)	28.4 (28.4, 28.5)
Race (White)b	74.5 (74.4,74.6)	75.2 (75.2,75.3)	70.5 (70.3,70.7)
Townsend Indexb,c	29.0 (28.9,29.1)	28.7 (28.6,28.8)	30.5 (30.3,30.8)
Prevalence of Atopic Dermatitisc	15.28(15.21,15.34)	0.00(0.00,0.00)	100.00(100.00,100.00)
Prevalence of OAI d
Seasonal Allergiesb	14.19(14.13,14.25)	12.17(12.11,12.23)	25.39(25.20,25.59)
Asthmab	16.88(16.81,16.95)	14.97(14.90,15.03)	27.49(27.29,27.70)
Food Allergyb	0.30(0.29,0.31)	0.23(0.22,0.24)	0.72(0.68,0.76)
Prevalence of any OAI e,f
One AIf	21.36(21.28,21.43)	19.64(19.57,19.72)	30.87(30.66,31.08)
Two AI	4.92(4.88,4.96)	3.82(3.78,3.85)	11.04(1090,11.18)
Three AI	0.06(0.06,0.06)	0.03(0.03,0.03)	0.22(0.20,0.24

^aAtopic Dermatitis

^bPrevalence measured as percentages (95% Confidence Interval)

^cTownsend Deprivation Index (40% most deprived)

^dOAI=Other Allergic Illness (seasonal allergies, asthma, or food allergies)

^eAI=any Allergic illness (AI) including AD

The characteristics of the children are summarized in Table II. Mean (SD) follow-up time for children was 10.8 (sd: 8.3) years with more than 13 million person-years of follow-up across the cohort. Children of mothers with AD had a slightly higher follow-up time compared to children of mothers without AD (11.6 [7.9]) and 10.7 [8.4]). The overall cohort included slightly more males (N=600,905 [50.1%]) and was predominately White (73.7%). AD was diagnosed in 25.37% (25.29, 25.44) of children during follow-up. The prevalence of AD was higher among children of mothers with AD compared to children of mothers without AD (37.17 [36.95, 37.39]) and 23.24 [23.16, 23.32]). A similar trend was observed for OAI. Children with AD were also more likely to have asthma (2.30 [2.28, 2.33]), seasonal allergies (2.65 [2.61, 2.68]) and food allergies (4.69 [4.51, 4.87]) compared to children without AD. However, no difference between sex, ethnicity, and Townsend deprivation score, and risk of AD was seen. In addition, children born to mothers with AD were more likely to have AD as well as other OAI (p-value <0.00001). Mean (95%CI) age of first recorded AD diagnosis in children was 3.32 (3.30, 3.34) years and 5.52 (5.49, 5.54), 9.14 (9.11,9.18), and 4.48 (4.39,4.58) for asthma, seasonal allergies and food allergies. The average age of first recorded diagnosis of AD was earlier in children of mothers with AD (2.98 years [2.95,3.02]) compared to children of mothers without AD (3.42 [3.40,3.44]) and was earlier if the mom had a history of all four atopic illnesses (2.09 [1.57,2.62]) (Table III).

Table II.

Characteristics of children in overall cohort and if the mother did have or did not have atopic dermatitis. Prevalence measured as percentage with 95% CI.

	Overall Cohort	Children of Mothers without ADa	Children of Mothers with AD
N (%)	1,224,243 (100)	913,693 (74.63)	310,550 (25.37)
Follow-up time, years (SD)b	10.8 (8.3)	10.7 (8.4)	11.6 (7.9)
Sex (male)c	50.1 (50.0,50.2)	49.8 (49.7,49.96)	50.9 (50.8,51.1)
Food allergyc	0.96 (0.99,0.98)	0.49 (0.52,0.50)	2.32 (2.42,2.37)
Race (White)c	73.7 (73.6,73.8)	74.3 (74.2,74.4)	71.8(71.7,72.0)
Townsend Indexc,d	29.7 (29.6,29.8)	29.7 (29.6,29.8)	29.8 (29.5,30.0)
Prevalence of Atopic	25.37(25.29,25.44)	23.24(23.16,23.32)	37.17(36.95,37.39)
Dermatitisc
Prevalence of OAI e
Seasonal Allergyc	8.14(8.09,8.19)	7.47(7.42,7.52)	11.86(11.72,12.01)
Asthmac	13.45(13.39,13.51)	12.72(12.65,12.78)	17.48(17.31,17.65)
Food allergyc	0.98(0.96,0.99)	0.90(0.88,0.92)	1.40(1.34,1.45)
Prevalence of any AI f
One AI	32.68(32.60,32.65)	31.47(31.38,31.56)	39.42(39.19,39.64)
Two AI	9.37(9.32,9.42)	7.77(7.72,7.82)	18.23(18.06,18.40)
Three AI	1.74(1.71,1.76)	1.18 (1.16,1.20)	4.82(4.72,4.92)
Four AI	0.03(0.03,0.03)	0.01(0.01,0.01)	0.01(0.01,0.12)

^aAtopic Dermatitis

^bAge of the children at end of follow-up or follow-up time since birth

^cPercentages (95% Confidence Interval)

^dTownsend Deprivation Index (40% most deprived)

^eOAI=Other Allergic Illness

^fAI=any OAI including AD

Table III.

Age (mean, years [95%Confidence Interval, CI]) of first recorded diagnosis of Atopic Dermatitis (AD) and other allergic illness (OAI) and timing of onset of AD relative to OAI in the overall cohort, children of mothers with AD and children of mothers without AD

	Overall Cohort	Children of Mothers without AD	Children of Mothers with AD
N (%)	1,224,243 (100)	913,693 (74.63)	310,550 (25.37)
Age first recorded diagnosis Atopic Dermatitis
years (95%CI)	3.32 (3.30, 3.34)	3.42 (3.40, 3.44)	2.98 (2.95, 3.02)
Age first recorded diagnosis Asthma
years (95%CI)	5.52 (5.49, 5.54)	5.60(5.57,5.63)	5.18(5.13,5.24)
Age first AD recorded diagnosis seasonal allergies
years (95%CI)	9.14(9.11,9.18)	9.33(9.29,9.37)	8.49(8.42,8.56)
Age first AD recorded diagnosis food allergies
years (95%CI)	4.48 (4.39,4.58)	4.51(4.40,4.61)	4.40(4.21,4.59)
Age at 1st recorded diagnosis of AD, years (range) if mother had OAI a
No OAI	3.46(3.44,3.48)	3.54(3.52,3.57)	3.07(3.03,3.12)
One OAI	3.03(3.00,3.06)	3.09(3.05,3.12)	2.90(2.84,2.96)
Two OAI	2.93(2.87,2.99)	3.02(2.94,3.10)	2.81(2.72,2.89)
Three OAI	2.38(2.00,2.77)	2.56(2.03,3.09)	2.09(1.57,2.62)
Age at first AD recorded diagnosis relative to OAI
AD preceding asthma	2.92(2.91,2.94)	3.01(3.00,3.04)	2.60(2.57,2.63)
AD preceding seasonal allergies	3.05(3.04,3.07)	3.15(3.13,3.17)	2.70(2.67,2.73)
AD preceding food allergies	3.36(3.25,3.28)	3.678(3.45,3.39)	2.91(2.88,2.94)
AD occurring after asthma	8.40(8.31,8.49)	8.64(8.54,8.75)	7.61(7.44,7.78)
AD occurring after seasonal allergies	10.99(10.84,10.15)	11.30(11.11,11.48)	10.109.83,10.37)
AD occurring after food allergies	6.78(6.51,7.05)	6.59(6.27,6.90)	7.36(6.83,7.89)

^amaternal history of asthma, seasonal allergy and/or food allergy

In fully adjusted Cox proportional hazards models, children were 59% more likely to have a recorded diagnosis of AD if their mothers also had a diagnosis of AD compared to mothers without AD (HR=1.59 [1.57, 1.60]) (Table IV). Children of mothers with asthma, seasonal allergies or food allergies also had a 7% (HR=1.07 [1.06, 1.08]), 30% (HR=1.30 [1.29, 1.32]), and 29% (HR=1.29 [1.22, 1.36]) increased risk of developing AD compared to children born to mothers without the corresponding OAI (Table IV). The risk of a child having AD increased as the number of the other atopic illnesses increased in the mother. For example, if the mother had asthma and seasonal allergies, the risk of AD was 1.25 (1.23,1.26) and increased to 1.43 (1.30,1.58) if the mother also had food allergies.

Table IV.

Hazard ratio (95% confidence intervals) for unadjusted and adjusted models examining the association between characteristics of the mother and risk of atopic dermatitis in their children^a

	Risk of Atopic Dermatitis in the child
	Unadjusted HR (95% CI)	Adjustedb HR (95% CI)
Characteristics of the mother
Mother’s race (White)	0.90 (0.90,0.91)	0.91 (0.90,0.92)
Mother’s age at birth	1.01 (1.01,1.02)	1.01 (1.01,1.01)
Maternal history of Atopic Dermatitis	1.70 (1.69,1.71)	1.59 (1.57,1.60)
Maternal history of Seasonal Allergies	1.45 (1.44,1.46)	1.30 (1.29,1.32)
Maternal history of Asthma	1.18 (1.17,1.19)	1.07 (1.06,1.08)
Maternal history of Food allergy	1.57 (1.49,1.65)	1.29 (1.22,1.36)

^aIn all cases p<0.00001

^bEach model was adjusted for all covariates in the table

We observed that among children with AD, 91.0% had a recoded diagnosis of AD first, whereas 6.47% and 2.53% had a recorded diagnosis of asthma or seasonal allergies first. Among children with any recorded diagnosis of asthma, 66.0% had a recorded diagnosis of asthma first, 27.8% will have a recorded diagnosis of AD first and 6.30% will have a recorded diagnosis of seasonal allergies first. Lastly, in children with a recorded diagnosis of seasonal allergies, 38.8% had a recorded diagnosis of seasonal allergies first, 36.8% had a recorded diagnosis of AD first and 24.35% had a recorded diagnosis of asthma first (figure 1). When AD diagnosis preceded the diagnosis of asthma or seasonal allergies, the age of onset of AD was earlier (asthma: 2.92 [2.91, 2.94], and seasonal allergies: 3.05 [3.04, 3.07]) and, later if AD occurred after asthma (8.40 [8.31,8.49]), seasonal allergies (10.99 [10.83,11.15]), or after food allergies (6.78 [6.51,7.03]) (Table III). Lastly, we conducted sensitivity analyses accounting for clustering within mother, as mothers may have had more than one child, with nearly identical results.

Figure 1.

Proportion of children with any recorded diagnosis of atopic dermatitis, asthma and seasonal allergies and timing of onset of other allergic illness.

Discussion:

The purpose of this study was to examine the association between maternal history of atopic illnesses and the risk of AD and other atopic illnesses of their offspring. Using the IMRD dataset, we developed a cohort of 1.2 million children and their mothers. This represents one of the largest cohort studies on this topic. We observed that children born to mothers with AD had almost twice the risk of developing AD earlier in life compared to children born to mothers without AD. These findings are consistent with previous smaller studies and a meta-analysis of these studies.²² Also consistent with previous reports, the children in our study were at an increased risk of developing other co-occurring atopic illnesses such as asthma, seasonal allergies, and food allergies.^22,23 We also observed that most children born to moms with AD who develop AD, develop it before age 3 and children born to mothers with AD, asthma and seasonal allergies are almost three times more likely to have AD. AD tended to occur earlier in life than asthma and seasonal allergies as children with AD who also have asthma, developed AD first 91% of the time. However, in our study, the diagnosis of AD can follow asthma and does so approximately 30% of the time. We also observed that maternal history of asthma or seasonal allergies may also be an independent risk factor for development of AD the child, a finding consistent with previous studies.²²

AD is thought to precede the development of other atopic illnesses, a pattern referred to as the atopic march.^23–25 As part of this concept, children who develop AD are likely to then develop asthma and/or seasonal or allergic rhinitis and are susceptible to developing other potentially atopic illnesses like food allergies among others.²³ The breakdown of the skin barrier in AD is thought to be the first step in a sequential pathway that emphasizes the notion that AD contributes to systemic allergen sensitization due to skin breakdown with subsequent environmental exposures that then leads to immunologic dysregulation and development of OAI.^1,2,23 Nevertheless, this hypothesis has limitations and few studies have looked at these atopic illnesses as risk factors for AD or that these illnesses might occur before the diagnosis of AD in childhood..^23,24,26 For example, in our study, the incidence of AD was most common in children less than 3 years of age, whereas the onset of asthma and seasonal allergies occurred later. However, diverging from the classical atopic march, we observed that some children developed asthma or seasonal allergies prior to developing AD and mothers without AD, but who had asthma or seasonal allergies, were also more likely to have children with AD than mothers without asthma or seasonal allergies, findings that are consistent with other studies. ^{23,24,26–30}

Another inconsistency with the classical atopic march hypothesis is the lack of discovery of genetic variation common to all three allergic illnesses, a fact that would seem important in interpreting a study on the association between mothers and their children. This hypothesis is likely not consistent with the most common AD genetic association, Filaggrin loss of function (FLG LOF) variation which is associated with earlier onset AD.³¹ However, FLG LOF is inconsistently associated with asthma and seasonal allergies. ^32–36 Overall, more consistent with the evidence to date might be the notion that some children have global and not disease-specific immune dysregulation, and that the clinical manifestation of their immune dysregulation is dependent on exposures as well as other genetic susceptibilities. This hypothesis is likely consistent with recent demonstrations that blocking the IL-4 pathway seems to be associated with clinical improvement in AD, asthma, seasonal allergies as well as other atopic illnesses. Despite this, not all allergic illnesses respond similarly to cytokine blockade suggesting differences in pathophysiology.^37,38 In addition, it is not clear if restoring skin barrier function early-on has an effect at preventing the progression of the atopic march as results of recent clinical that have examined this approach have been inconsistent although additional trials are ongoing.^39–43

An alternative hypothesis is that children with AD may be more prone to OAI rather than AD being part of the causal pathway that results in the development of other allergic illnesses. A study of 218,485 children (<18 years of age) from the United States who were observed for a period of 5 years with the purpose of examining pediatric allergy patterns found that 13.4% of children had 2 or more allergic conditions and that some (i.e., asthma and allergic rhinitis) are more often comorbid with each other.⁴⁴ This study did not examine the timing of onset of these comorbidities relative to one another and it is possible that some diagnosis may have been missed if, for example, patients were seen in a different care setting (i.e., hospital-based visit). Nevertheless, when taken together, our results support the notion of a potential shift in the current paradigm of the atopic march from a sequential trajectory to one of co-occurring disease.

One limitation of our study is that we used recorded diagnosis to determine the presence or absence of our exposure and outcomes of interest (i.e., AD and OAI). However, we do not expect such misclassification to be different between groups. We also did not have detailed information about AD disease severity. Exposure to systemic medications is often used as proxy but could lead to misclassification, however, in the setting of children mostly under 10 years of age the use of newer systemic agents per NICE criteria is uncommon. Studies have also shown that more severe and persistent AD outcomes are associated with FLG LOF.⁶ However we do not have access to FLG LOF data.⁴⁵ While we adjusted for confounding variables, these atopic illnesses may be on the same causal pathway and hence should not be considered confounders. Unmeasured confounders could have impacted the strength of our associations. It is also important to realize that AD and OAI are potentially heterogenous diseases. It is possible that the time of disease onset and disease associations are related to this heterogeneity. As the pathophysiology of these diseases becomes more complete new categories of disease may be discovered thereby altering the relationships reported in this study. Lastly, our results may not be generalizable to other populations. Strengths include its large sample size, use of a validated method for determining if a child had AD, the longitudinal nature, and, not having to depend on patient recall to assist with data acquisition.

In conclusion, children born to mothers with AD are more likely to be diagnosed with AD, as well as OAI. Our study suggests that children born to mothers with AD not only had a higher risk of acquiring AD at an earlier age but are also susceptible to developing OAI in childhood with AD occurring first, but not always, in many cases. The notion of a sequential atopic march that sees children with AD progressively developing OAI may not be fully accurate. An alternative hypothesis that children with AD are more prone to OAI rather than AD being part of the causal pathway should be considered. Future studies in other cohorts are needed to explore the timing of onset of these allergic diseases with respect to each other.