Background
Atopic dermatitis (AD), a hypersensitive skin disease characterized by inflamed and scaly skin lesions, pruritus and skin rash is one of the most common chronic diseases to affect children in Western society [
1]. Symptoms often appear by 6 months of age, but 65% of affected children will have symptoms by 18 months of age and in more than 50% of children, symptoms will persist through 7 years of age [
2]. AD symptoms often lead to depression, sleep deprivation, feelings of embarrassment, stigma, social isolation, and restricted ability to own pets or play sports [
2,
3]. Consequently, AD leads to a diminished quality of life for children [
3] and its medical care burdens children, parents and the health care system [
4,
5]. Substantial attention has been paid to understanding the risk factors for AD and their implications in the atopic march toward allergic rhinitis and ultimately asthma [
2,
6‐
11]. Given that AD is a potent early warning sign for later atopic diseases, ascertaining qualities of the early environment that may predispose children to AD is warranted.
A recent systematic review identified early environmental risk factors for childhood AD to be maternal prenatal psychological distress, including depression, anxiety and stress [
12]. A study of 1264 mother–infant pairs revealed that maternal report of prenatal mental status including vitality, vigor, happiness, anxiety, discouragement, nervousness, tiredness, exhaustion and work stress predicted AD in 2 year olds [
13]. A large (n = 23,791) cross-sectional study revealed that parental history of diagnosed prenatal and postnatal depression predicted AD in 6–13 year olds [
14]. Maternal depressive symptoms in the postnatal period were associated with AD in 3–12 month old infants as was postnatal anxiety, particularly about childrearing (n = 40) [
15]. In several other studies, maternal prenatal stressful life events such as divorce, mourning the death of a loved one, job loss or financial problems were associated with AD in 3–14 year olds, independent of sociodemographic factors or known atopy risk factors, like allergen exposure [
16‐
18]. Stressful life events during the second half of gestation were found to increase the odds of AD in 14 year olds [
17]. To our knowledge studies have not examined the differential impact of prenatal or postnatal depression, anxiety and stress on the development of childhood AD.
Although prenatal and postnatal depression, anxiety and stress are often co-morbid [
19‐
21], they likely exert unique contributions to AD development. Postnatally depressed mothers often fail to respond appropriately to infant cues, alleviate infant stress and overall, are less sensitive and responsive towards their infants [
22‐
24]. This propensity is widely held responsible for the poor developmental outcomes often observed in children of depressed mothers [
25,
26]. Prenatal psychological distress may also negatively affect the priming of the maternal brain for sensitive, responsive interactions with infants [
27]. Maternal
1 sensitivity is defined as a pattern of behaviour that pleases the infant and increases the infant’s comfort and attentiveness and reduces its distress and disengagement [
28]. Consistently alleviating infant distress by responding to behavioral signals, such as fussiness due to hunger or fatigue, in a timely fashion is an indicator of high maternal sensitivity, while failing to do so is indicative of low responsiveness [
29]. Moreover, high maternal sensitivity and responsiveness are contrary to maternal behaviours that are overtly or covertly hostile or attempt to overly control the infants’ behaviours in everyday interactions [
30]. As a result, healthy maternal–infant relationships are typically characterized as sensitive and responsive interactions that are attentive to the infants’ needs while remaining non-intrusive and non-controlling, and contribute to the regulation of the infant’s response to stress before the infant is physiologically able to self-regulate [
31‐
33]. In other words, infants develop their ability to regulate their stress response through their relationship with their maternal caregivers, especially during the first 3 years of life [
34,
35]. Caregivers who do not provide sensitive, responsive, non-controlling care promote development of precocious self-regulation, which may have physiological costs, including suppressed immune development [
36,
37].
As early as 1999, maternal–infant relationships were theorized to predict and/or compound childhood AD [
38]; however, to date, only two studies have examined this theorized association [
15,
39]. In contrast, many studies have identified associations between maternal–infant relationships and childhood asthma [
40‐
45]. One study compared children with (n = 20) and without (n = 20) AD, they found that self-reported maternal overprotectiveness and control were associated with AD in 3–12 month olds [
15]. Another study compared self-reported relationship characteristics of mothers of children less than 6 years of age with (n = 102) and without (n = 131) AD and found that mothers of children with AD were less affectionate and that mothers of children with more severe symptoms were more rejecting and less encouraging of children’s autonomy [
39]. They also examined maternal sensitivity via parental self-report and failed to find any difference between AD and non-AD groups [
39]. However, self-report measures of maternal–infant relationship quality are poorly correlated with observational measures [
46] and thus observations of are much preferred [
47,
48], yet much ignored in the research on AD or other atopic diseases. Given the role of AD in the atopic march and the consistent observation of associations between maternal–infant relationships and childhood asthma, and emerging evidence of associations with AD, ascertaining the influence of qualities of the maternal–child relationship (sensitivity, responsiveness, and control) using observational measures, on AD would be useful.
Furthermore, existing research has not yet examined protective factors in AD development. Social support has been described as a buffer to stressful life events, enhancing the mother’s self-esteem and self-efficacy and aiding in the transition to motherhood, consequently promoting healthy child development [
49,
50]. Social support refers to emotional, affirmational, informational and instrumental assistance, from social relationships including partners [
51]. Without appropriate social support during the transition to motherhood, this transition can be difficult and distressing, adding to maternal psychological distress and affecting the mother’s ability to care for her infant [
49]. Lack of social support has been strongly associated with maternal depressive symptoms [
51,
52]. Research has affirmed the protective role of social support, particularly from fathers, for normative and depressed mothers [
53‐
56]. To our knowledge, existing research has not yet examined maternal social support as a protective factor in children’s AD. Other identified risk factors include parental history of atopic disease [
57,
58], especially asthma, low maternal education, being a boy [
59], high birth weight and day care attendance in the first year of life [
60]. Finally, breastfeeding has been shown to predict both increased maternal sensitivity [
53,
61] and changes in risk of AD [
62].
To address the aforementioned gaps in the literature, we sought to determine the association between maternal–infant relationship qualities (sensitivity, responsiveness and control) and child AD by 2 years of age, considering risk (i.e. maternal prenatal and postnatal depression, anxiety and stress) and protective (i.e. social support) factors. Given the prominent role of maternal sensitivity, we also sought to identity predictors of maternal sensitivity. We hypothesized that lower maternal sensitivity, higher unresponsiveness and control would increase the risk for AD in young children. We also hypothesized that greater symptoms of depression and anxiety and reported stress would predict lower maternal sensitivity and increased risk of childhood AD. Finally, we hypothesized that greater reported postnatal social support would reduce the likelihood of AD.
Discussion
In a subsample of 242 women and their infants enrolled in the APrON longitudinal study, high maternal sensitivity was found to significantly protect against AD development at 18 months of age (adjusted OR = 0.73; 95% CI 0.56, 0.93, p = 0.012), independent of maternal asthma status. The maternal sensitivity and child AD association was also independent of maternal prenatal and postnatal anxiety, and social support. A mother’s assessment of the quality of support provided by her partner reduced the onset of this early childhood atopic disorder by 0.96 for every unit increase of perceived support. Maternal anxiety was a risk factor for AD in offspring. Further, positive associations with child AD were observed for maternal controlling and unresponsive behaviors (adjusted OR = 1.33, 95% CI 1.03, 1.71, p = 0.028 and adjusted OR = 1.35, 95% CI 1.05, 1.73, p = 0.020 respectively). Maternal sensitivity was predicted by maternal age above 30 and, in contrast to expectations, higher postnatal depressive symptoms.
Our results suggest that poor maternal–infant relationship quality (low sensitive, high unresponsive and high controlling behavior during interactions with infants) at 6 months of age increases the risk of child AD at 18 months of age. Our findings are consistent with evidence demonstrating the link between maternal–infant relationship quality and child atopic disease, although current understanding is largely limited to the etiology of asthma [
40‐
43,
45]. For example, identified difficulties in the parent–child relationship independently predict asthma onset at 3 and 6 years of age [
41], while the onset of asthma in early life has been associated with poor quality relationships with parents that persist into adulthood [
87].
Infants whose caregivers are less sensitive have higher levels of cortisol [
33,
88] and excess cortisol exposure during early infancy may influence the developing immune response via epigenetics or neuroendocrine dysregulation, potentially leading to AD [
87,
89,
90]. It is also possible that more sensitive mothers take greater care of their infants’ skin, by for example, use of moisturizer and regular bathing; however, how this may contribute to the atopic march toward asthma is unclear. To our knowledge, no evidence has been published demonstrating associations among maternal sensitivity, infant skin care, and the atopic march, but these associations are worth exploration as possible rival hypotheses. A recent review summarizing the increasing literature on links between maternal–infant relationships and asthma development, noted that only correlational, not causal links have been established. The review called for randomized controlled trials of social interventions focused on maternal–infant relationships with examination of atopic disease outcomes [
44].
Predictors of maternal sensitivity typically include low postpartum depression, anxiety, and stress, high paternal support and adult (as opposed to adolescent) motherhood [
91‐
93]. Depression is known to deplete maternal ability to express positive affect towards her infant, and could possibly be exacerbated by lack of paternal support [
91]. Our finding that older adult mothers are more sensitive to their infants is in keeping with research suggesting that younger mothers engage less frequently than older adult mothers in behaviors that promote secure attachment, such as positive verbal feedback [
94]. But, contrary to much evidence [
61,
62,
90,
92,
93,
95], neither maternal stress, anxiety nor breastfeeding were significantly associated with maternal sensitivity. We also observed an unexpected association between higher postnatal depressive symptoms and maternal sensitivity, which may have been due to small sample size. Women reporting excessive depression, anxiety or stress are typically less positively reinforcing and interactive with their infant, thought to lead to poor child development outcomes [
22‐
24,
96]. However, new data from the Avon Longitudinal Study of Parents and Children (n = 704) revealed that perinatal depressive symptoms predicted little variance in maternal–infant relationship qualities, but rather maternal interpersonal sensitivity in pregnancy is the most significant contributor to forming healthy maternal–infant relationships in the context of perinatal depressive symptoms [
97]. In other words, other variables that are correlated with both depressive symptoms and qualities of the maternal–child relationship may underpin observed associations between depression and child development and deserve exploration. Moreover, depressive and anxiety symptoms in pregnancy are typically worse than in the postpartum period [
98,
99], so significant differences would be expected; however, how these differences may influence maternal–child relationships or onset of children’s AD is unknown. Studies have shown that high levels of perceived support reported by depressed or anxious mothers may buffer the association with child atopic disease, suggesting accessible support may improve the quality of the mother–child relationship [
100‐
102]. In this study, although greater reported postnatal social support was shown to reduce the likelihood of AD, consistent with various studies [
34,
53,
54], social support was not found to be a predictor of sensitive maternal behavior.
Prenatal pregnancy-specific anxiety (adjusted OR = 2.74; 95% CI 1.04, 7.19, p = 0.041) and postnatal anxiety (adjusted OR = 1.16, 95% CI 1.01, 1.33, p = 0.034) predicted AD independent of paternal support and maternal sensitivity. A trend for an inverse association between postnatal depression and AD was also observed. These findings are compatible with several studies on childhood atopic disease associations with maternal prenatal or postnatal stressors [
13,
16‐
18,
59,
103], prenatal anxiety but not depression, as suggested above [
88,
104‐
107]. Mutual adjustment for anxiety and depression may have uncovered the negative influence of postnatal depression on the health care utilization for infant care and thus, physician diagnosis of AD [
108]. Consistent with our results, some studies found postnatal anxiety about childrearing to be associated with AD in 3–12 month old infants [
15]. As maternal anxiety becomes more prevalent [
109], it may gain a more prominent influence on the maternal–infant relationship [
97]. Given that low maternal sensitivity and anxiety are interrelated, the association with childhood AD may be similarly mediated by epigenetics or neuroendocrine dysregulation [
88,
104,
110], and altered infant cytokine profiles [
111]. Infant neuroendocrine and immunologic maturation is highly plastic in the face of environmental stressors [
112‐
115], especially during the early postnatal period. For those who are already genetically susceptible to the development of atopy [
116], as observed in our sample of children whose mothers reported a history of asthma, maternal sensitivity may make the difference between the development of AD or not.
The pathogenesis of AD is known to involve abnormal levels of specific cytokines released by T-helper 2 cells detected in cord blood or infant peripheral blood [
117‐
120]. The altered differentiation of these cytokines may be induced through excessive glucocorticoid exposure at an early age, driven by maternal perinatal psychosocial distress [
121,
122]. Future research could focus on understanding the associations among: (1) excessive, long-term fetal/infant glucocorticoid exposure during critical developmental periods, (2) Th2 specific cytokine levels in infant cord or peripheral blood, and (3) fetal programming, driven by prenatal distress (anxiety, depression and stress), and (4) poor maternal–infant relationship qualities, linked to postnatal distress. Moreover, genetic risk, suggested by the heritability of atopic disease suggest a need for genome-wide association studies [
13].
This study has many strengths, including the outcome of childhood AD assessed via maternal report of a physician diagnosis of AD and observational assessment of maternal–infant relationship quality, but there are also limitations. First, physician corroboration of maternal report of physician diagnosis of AD was not possible. Second, only “maternal” caregiving was assessed; rather than seeking to reinforce gender stereotypes, we recognize that primary caregivers may be mothers, fathers or others. However, we also recognize that the vast majority of infants’ primary caregivers are mothers in Canada [
123] and in our study [
64]. Third, only a small number of mothers reported depressive symptoms in the postpartum period. This may have contributed to the potentially spurious findings of inverse association between postpartum depression and outcomes. Additionally, not all participants reported perceptions of the quality of their partners’ support, as only 88% of participants reported on their child’s father. Nonetheless, the variability in primary support provider may also reflect the reality of modern families. Furthermore, while we collected data on maternal asthma, we did not collect additional data on mother’s atopy status or family history of atopy. Finally, we acknowledge that this is a relatively high SES sample with low sociodemographic risk that may limit generalizability.
To our knowledge, this is the first study to uncover the association between maternal–infant relationship qualities (sensitivity, control, unresponsiveness) and childhood AD after accounting for risk (maternal depression, anxiety and stress) and protective (social support) factors and well-known covariates. We found low maternal sensitivity to be a significant risk factor for childhood AD, in the presence and absence of perinatal anxiety and low social support. Caregivers who are emotionally engaged and supportive of their children during infancy may be well equipped to prevent poor infant immunological development, reducing the likelihood of childhood AD and potentially asthma [
34]. Limited work has begun to establish that interventions focused on the quality of maternal child-relationship may influence asthma symptoms [
45]. These results suggest interventions that improve maternal–infant relationship quality, anxiety and support could reduce odds of childhood AD. Existing parenting intervention programs, including the Nurse Family Partnership [
124] and Keys to Caregiving [
125], and support services may already have effects on AD and the atopic march, that have not to date been measured. In addition, strengthening of relationships between caregivers serves as a means of controlling maternal anxiety [
95] and may therefore be effective in helping prevent childhood AD. Additional intervention studies ought to focus on improving maternal child relationship quality and reducing odds of AD, as a means to forestall the atopic march to allergic rhinitis and ultimately asthma.
Authors’ contributions
NLL, ALK, and NC drafted the manuscript. HNN conducted statistical analysis and compiled all tables and figures. LA and MJH coded and logged participant data. NLL, ALK, HNN, TSC and GFG all contributed to the formulation of study design and data analysis and interpretation. All authors read and revised the manuscript as appropriate and agreed upon this final version. All authors read and approved the final manuscript