Background

Atopic dermatitis (AD) is a common allergic inflammatory skin disorder in children. Its etiology is not fully understood, but it is presumed to be multifactorial with interactions among genetic and environmental factors [1]. In terms of the genetic factors involved, the risk of developing AD is known to be higher if one parent has an allergic history than if neither parent such allergic history [2, 3], and mothers with AD may more readily transmit AD to their offspring compared with fathers [4, 5]. The sex of the child is also recognized to affect the development of allergic diseases. For example, boys are reported to be more susceptible to childhood asthma [6,7,8]. Nevertheless, the effects of sex on AD development, with and without maternal history of allergic disease, remain unclear.

In terms of the environmental factors involved, the severity of AD is known to change in relation to the seasons, and many people with AD in Japan experience exacerbations in winter [9]. A link has also been found between the prevalence of early childhood eczema and climate factors [10], with associations found with outdoor temperature and humidity [11]. Low indoor or outdoor humidity exacerbated subclinical eczema, thereby increasing the disease prevalence, through damage to the barrier function of the skin [10, 11]. Moreover, because rapid adaptations are required in the first few months of life after transitioning from the in utero environment to the very different ex utero environment, interest is growing in how prenatal as well as postnatal environmental exposures might contribute to AD risk in early childhood. Season of birth or month of birth can be used as a surrogate factor for potential environmental exposures during the prenatal and early postnatal periods. Factors that exhibit seasonal variation include temperature, humidity, dryness, exposure to sunlight, and outdoor physical activity [12, 13]. Studies of the relationship between season of birth and allergy have suggested that early infancy is of particular importance for the later development of allergic diseases [14, 15]. Many studies have reported that AD is more prevalent in children born in autumn than in spring in the northern hemisphere [14,15,16,17,18,19,20,21,22].

The International Study of Asthma and Allergies in Childhood (ISAAC) has developed a questionnaire for large-scale screening of AD and other allergic diseases in the general population [23, 24], but no widely used questionnaires have been developed to determine prevalence in infancy. Therefore, little information is available on the prevalence and development of skin lesions in the early postnatal period. Most previous studies assessed AD prevalence among children 6 years and older, and there are scant data on the prevalence of eczema and development of AD in younger children, especially those under 1 year of age. Up to 1 year of age, the diagnosis of AD is not common because the typical skin lesions are not obvious at that age. Moreover, physicians may not see children with eczema because over-the-counter medications are readily available to treat the symptoms. Yet, early eczema is regarded as the beginning of the atopic march [5, 25]. Indeed, atopic eczema is one of the earliest clinical manifestations of allergic disease [26] and occurs most commonly in the first months of life [27, 28]. It is important, therefore, to clarify how the season or month of birth affects the trajectory of the skin lesions in infants up to 1 year of age.

In this study, using data from a large Japanese birth cohort, we sought to clarify the effects of season of birth and maternal history of allergic disease on the development of eczema in infants up to 1 year of age. We wanted to determine how early in the postnatal period the association between the season of birth and eczema can be observed. Moreover, we examined the influence of sex of the infant and maternal history of allergic disease on the development of eczema and the diagnosis of AD in early infancy.

Methods

Study design and participants

We analyzed data obtained in the Japan Environment and Children’s Study (JECS), which is investigating the effects of environmental factors on children’s health. JECS is an ongoing nationwide government-funded prospective birth cohort study that was started in January 2011 [29, 30]. It is funded directly by the Ministry of the Environment, Japan, and involves collaboration among the JECS Programme Office (National Institute for Environmental Studies), the Medical Research Center for the JECS (National Center for Child Health and Development), and 15 regional centers [29, 30]. Pregnant women are contacted through cooperating health care providers and/or local government offices that issue the Maternal and Child Health Handbooks and those consenting to participate are registered. The population characteristics of the JECS cohort are highly representative of the Japanese population according to vital statistics published by the government [30].

Ethics statement

JECS participants were recruited during their pregnancy by research coordinators after receiving face-to-face explanations of the overall aims of the study, and informed consent was obtained from all participants. JECS is conducted in accordance with the Declaration of Helsinki and applicable national regulations and guidelines. All procedures involving human subjects for the JECS protocol were reviewed and approved by the Ministry of the Environment’s Institutional Review Board on Epidemiological Studies (100,910,001) and the ethics committees of all participating institutions.

Study data

This study used the jecs-qa-20210401 dataset (jecs-ta-20190930), which contains information on demographic factors, medical history, obstetric history, physical and mental health, lifestyle, occupation, housing situation, and socioeconomic status, obtained from participant responses to self-administered questionnaires or transcribed from the medical records by physicians, midwives, nurses, and/or research coordinators. The dataset contains the records of 103,057 pregnancies with information on mother–infant pairs. Of these, we excluded 5,647 multiple registrations (second or third registration of the same mother), 948 multiple births (twins or more), and 3,521 miscarriages or stillbirths. Among the remaining 92,941 singleton live births from unique mothers, an additional 11,352 were excluded due to unreturned questionnaires or missing response to items on the infant’s skin status or physician-diagnosed AD. Finally, we analyzed the data for 81,615 infants (Fig. 1).

Fig. 1
figure 1

Participant flow diagram

Full size image

Exposure classification

First, we considered the month of birth as an exposure factor in our analysis. Then, we used season of birth for analysis. In a previous study, we found the highest incidence of AD in 3-year-olds born in October–December and the lowest incidence in those born in April–June [20]. Therefore, in this study, we classified seasons as spring (April–June), summer (July–September), autumn (October–December), and winter (January–March).

Outcomes

Information on dermatitis was collected by distributing questionnaires to the caregivers (mainly mothers) of the infants at 1 month, 6 months, and 1 year after delivery. Although various definitions of AD and eczema have been reported [31], the present study refers to “eczema” as judged by the parents based on the symptoms they observed and to “AD” as diagnosed by a physician. We analyzed four outcomes reported on: eczema at 1 month of age, eczema at 6 months of age, eczema at 1 year of age, and physician-diagnosed AD reported at 1 year of age. Eczema at 1 month of age was defined as an affirmative response to the question, “Has your baby ever had a rash on the face, head, or neck or around the ear during the past month after birth?” Eczema at 6 months and eczema at 1 year of age were both defined as an affirmative response to the question, “Has your child ever had an itchy rash that was coming and going for at least 2 months?”, which was based on an item from the ISAAC questionnaire targeting 6–7-year-olds, the Japanese translation of which has been validated [23, 24]. Physician-diagnosed AD reported at 1 year of age was defined as an affirmative response to the question, “Has your child ever been diagnosed by a physician as having AD?”.

Statistical analysis

To identify the association between season of birth and the four outcomes, we performed logistic regression analysis [32] and determined the 95% confidence intervals (CIs). Multiple testing correction was performed using the Benjamini–Hochberg method [33], and the false discovery rate was set at less than 0.05. Then, in multivariable logistic regression analysis, we adjusted for the following potential modifiers: maternal age, annual household income, maternal education level, maternal history of allergic disease (AD, bronchial asthma, allergic rhinitis, pollinosis, allergic conjunctivitis, and food allergy), maternal intake of vitamin D during pregnancy, maternal active smoking, maternal passive smoking, infant sex, gestational weeks, feeding method during the first month after birth, presence of older siblings, pet ownership, and registered regional center. These covariates were selected based on a previous study [22]. Vitamin D intake was obtained from a food frequency questionnaire [34] and was adjusted for energy intake using the residual method. Multicollinearity assessed using generalized variance-inflation factors greater than 10 was not detected.

To examine whether the relationship between season of birth and prevalence of eczema or AD differed by infant sex or maternal allergic history, parent-judged eczema and physician-diagnosed AD were analyzed separately after stratification of the cohort by sex and maternal history of allergic disease. SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA) was used for all statistical analyses.

Results

A total of 81,615 infants were analyzed in this study. Distribution by month of birth ranged from 7.0% to 10.2% (Table 1). Most infants were born in the summer, followed by the autumn, spring, and winter (29.0%, 25.1%, 23.5%, and 22.5%, respectively; Table 1). The overall prevalence of parent-judged eczema was 61.0% at 1 month, 33.0% at 6 months, and 18.7% at 1 year and that for physician-diagnosed AD was 4.3% up to 1 year of age.

Table 1 Demographic and obstetric characteristics of the infants
Full size table

Logistic regression analysis to examine the association between month of birth and each outcome revealed that, compared with infants born in May as a reference, infants born in July had the highest risk of eczema at 1 month (adjusted odds ratio [aOR], 1.25; 95% CI, 1.16–1.35; Fig. 2a), whereas infants born in November had the highest risk at eczema at 6 months (aOR, 2.51; 95% CI, 2.32–2.73; Fig. 2b) and those born in October had the highest risk at 1 year (aOR, 1.17; 95% CI, 1.07–1.27; Fig. 2c). Infants born in October also had the highest risk of physician-diagnosed AD (aOR, 1.40; 95% CI, 1.18–1.66; Fig. 2d).

Fig. 2
figure 2

Odds ratios (95% CI) for outcomes in early infancy by birth month (N = 81,615). Adjusted for maternal age, annual household income, maternal educational level, maternal history of allergic disease, maternal intake of vitamin D during pregnancy, maternal active smoking during pregnancy, maternal passive smoking during pregnancy, infant sex, gestational weeks, feeding method, presence of older siblings, pet ownership, and registered regional center. Bold letters indicate statistically significant differences calculated using the Benjamini–Hochberg method.Abbreviations: AD, atopic dermatitis; aOR, adjusted odds ratio; cOR, crude odds ratio;95% CI, 95% confidence interval

Full size image

When the infants were classified by season of birth, with those born in spring as a reference, infants born in summer had the highest risk of eczema at 1 month (aOR, 1.19; 95% CI, 1.14–1.24; Table 2a), whereas those born in autumn had the highest risks of eczema at 6 months (aOR, 2.19; 95% CI, 2.10–2.30; Table 2b) and at 1 year (aOR, 1.08; 95% CI, 1.02–1.14; Table 2c) and of physician-diagnosed AD up to 1 year of age (aOR, 1.33; 95% CI, 1.20–1.47; Table 2d).

Table 2 Odds ratios (95% CIs) for outcomes in early infancy by birth season (N = 81,615)
Full size table

Figure 3 shows the relationship between season of birth and prevalence of eczema at 1 month, 6 months, and 1 year of age. At 6 months old, for infants born in spring, information was collected in autumn. For infants born in autumn, information was collected in spring. Although eczema at 6 months and at 1 year of age are opposite with regards to the season of observation, the highest frequency was identified for infants born in autumn at both ages. Between 1 and 6 months of age as well as between 6 months and 1 year of age, transitions in parent-judged eczema according to season of birth were categorized as no symptoms at either time point (symptom-free), symptoms improving from the first to second time point (improvement), symptoms occurring between the first and second time points (onset), and symptoms of the same severity at both time points (persistent). The results showed that infants with persistent eczema at either time point examined were more likely to be born in autumn (Supplementary Table 1).

Fig. 3
figure 3

Prevalence of eczema by season of birth. This figure shows the season of information collection for eczema in infants of the same age in months. At 6 months of age, for infants born in spring, information was collected in the autumn (October–December), while for infants born in autumn, information was collected in the spring (April–June). Thus, the seasons of information collection at 6 months and 1 year of age were opposite

Full size image

Finally, we stratified the results by sex and maternal history of allergic disease. Prevalence of the four outcomes in both sexes with a maternal history of allergic disease was higher than in those without such history: eczema at 1 month of age, 63.5% vs 58.5%; eczema at 6 months of age, 36.2% vs 29.7%; eczema at 1 year of age, 22.0% vs 15.5%; and physician-diagnosed AD, 5.6% vs 3.0%. When the groups with and without maternal history allergic disease were stratified by sex, prevalence was higher in boys than in girls for all four outcomes (Table 3).

Table 3 Prevalence of outcomes stratified by maternal history of allergic disease and infant sex
Full size table

To determine the association between season of birth and physician-diagnosed AD up to 1 year of age, subgroup analysis stratified by infant sex and maternal history of allergic disease revealed that, in girls, the aOR was not significant among those without maternal history of allergic disease, but for girls with such maternal history, those born in autumn showed a significant aOR compared with those born in spring (Table 4). Intriguingly, for boys with maternal history of allergic disease, those born in summer or autumn had a higher risk of AD (summer: aOR, 1.36; 95% CI, 1.15–1.60; autumn: aOR, 1.37; 95% CI, 1.16–1.62).

Table 4 Odds ratios for physician-diagnosed AD by birth season stratified by maternal allergic history and sex
Full size table

Discussion

In this large prospective birth cohort, environmental factors indicated by month of birth were found to influence the prevalence of eczema and AD diagnosis in infants younger than 1 year of age. In particular, the prevalence of eczema was found to be higher among those born in the autumn. Moreover, the importance of male sex and maternal history of allergic disease on the development of atopic eczema was evident. Because eczema is one of the earliest clinical manifestations of allergic disease and negatively affects quality of life [26, 28], our results may be significant from a preventive standpoint. Improved knowledge about the impact of season of birth on the development of eczema could inform future preventive measures to prevent skin barrier disruption, such as appropriate skin care from early infancy, and thereby reduce the risk of allergic diseases developing, especially in male infants with a maternal history of allergic disease.

It is well established that events during our early lives affect later life course trajectories [35, 36]. Previous studies have suggested that the initiation of antigen-specific responses can occur in utero, mediated by immunoglobulin E. [37]. It is difficult to examine in utero factors directly, however. Our results here do suggest the important involvement of early postnatal life in the development of eczema. Dry skin and the resulting skin barrier disruption are believed to be one of the non-allergic etiological factors of AD [38] because they lead to immunological dysregulation, which in turn may lead to allergens being sensitized epicutaneously, thereby inducing allergic inflammation of the skin. Our results suggest that the higher prevalence of eczema observed under certain climate conditions could be due to environmental triggering of subclinical disease in predisposed infants.

In Japan, children born in spring (the reference season in this study) spend their first few months in warm, humid weather, which results in moist skin, whereas those born in autumn spend their first few months in cold, dry weather, which can result in dry skin. They are also mainly at home in conditions that dry the skin. Winter in Japan, with its low indoor humidity and cold outside temperatures, is associated with a double negative influence on the skin barrier. Thus, changes in skin condition in the first few months of life could play an important role in triggering allergic skin inflammation and ultimately affect the prevalence of AD. Previous studies showed that the sex of the child was associated with the development of allergic diseases in childhood [6,7,8]. Moreover, mothers with AD may transmit it to their children [4, 5]. In the present study, the prevalence of physician-diagnosed AD was higher in boys with a maternal allergic history irrespective of season of birth (Table 3).

Infants born in the autumn in Japan tend to spend early infancy in a season with little sunshine. A previous study found higher levels of 25-hydroxy vitamin D and immunosuppressive cytokine IL-10 in the blood of infants born in spring than in those born in winter [39]. In line with this, low exposure to ultraviolet light and subsequently lower levels of vitamin D and/or IL-10 may explain the association between season of birth and development of AD. Vitamin D suppresses allergic sensitization by promoting Fop3+ regulatory T cells [40]. Indeed, reduced vitamin D and low fish intake by mothers during pregnancy have been correlated with increased incidence of AD in their children [41, 42]. Several trials have shown that vitamin D supplementation improves the clinical symptoms of AD [43,44,45]. In our analysis, we adjusted the data for the amount of dietary vitamin D intake during pregnancy. Therefore, we speculate that the influence of month of birth on the development of eczema in early infancy may be dependent on the vitamin D produced through exposure to sunshine in the children.

It is well recognized that the prevalence of AD is higher in infants born in autumn than in those born in spring [14, 15, 17,18,19,20]. A recent systematic review and meta-analysis also indicated that AD was significantly associated with autumn and winter births compared with spring births in the northern hemisphere [46]. However, our results show that eczema or rash was more prevalent at 1 month of age in infants born in the summer compared with those born in other seasons. Thus, the eczema seen at this age may be much more influenced by the season of observation. In Japan, hot and humid conditions in summer are associated with the prevalence of eczema in 1 month olds [47,48,49]. In our study, the prevalence was particularly high in 1 month olds (61.0%) compared with older infants, probably because eczema at this age shows a high degree of clinical heterogeneity, similar to other forms of eczema such as seborrheic dermatitis, intertrigo, and diaper dermatitis. In addition, about 60% of infants who had symptoms of eczema at 6 months no longer showed such symptoms at 1 year of age, which is a similar findings to that of previous Japanese studies where atopic dermatitis observed in infants younger than 6 months old remitted with age [50]. Although there are differences in prevalence between 6 months and 1 year of age, both time points show a higher prevalence among children born in autumn.

This study has several strengths. First, the prospective design of JECS, with the regular return of questionnaires, minimizes recall bias concerning the child’s AD. Also, JECS has a well-powered sample size, which enabled adjustment for various confounders in logistic regression analysis. The dataset covers about 45% of live births within the study areas, and the characteristics of the participants are similar to those of other women according to the Japanese vital statistics data, so the cohort is considered representative of the Japanese population [30]. We also studied the effect of possible confounders.

Nonetheless, this study has some limitations. Presence of eczema was determined from responses on a self-reported parental questionnaire. The rather broad question used might result in overestimates of eczema prevalence through the inclusion of other entities such as allergic contact dermatitis, and differences in the wording of survey questions at 1 month and at 6 months and 1 year of age might have an effect as well. We identified AD based on caregiver-reported physician diagnosis at 1 year of age, and the prevalence might differ if the outcome assessment were to be conducted using data collected from the medical records. Also, the reported diagnosis was not necessarily made by an allergy specialist. Furthermore, genetic polymorphisms known to be associated with the development of AD, such as filaggrin [51], could not be considered in this study.

A combination of multiple indicators, such as use of therapeutic agents and eczema characteristics, is considered to be the most confirmatory for the diagnosis of AD [31]. Because diagnosing AD in infants under 1 year of age is difficult, it is necessary to clarify those infants who are more susceptible to seasonal exposure, taking into account genetic polymorphisms and other factors. It is also necessary to clarify whether our results observed in Japan are generalizable to children in other parts of the world.

Conclusions

Our results indicate that eczema and diagnosis of AD are more prevalent at age 6 months and 1 year in infants born in autumn than in those born in spring. However, at 1 month of age, eczema is more prevalent in infants born in summer. Thus, eczema in 1-month-olds may be influenced by the season of observation. The climate in early infancy seems to affect skin condition and ultimately influence the development of AD. Our results suggest that caregivers of male infants with a maternal history of allergic disease, and particularly those born in autumn, should pay careful attention to the infant’s skin.