Does prenatal anxiety or depression symptoms are associated with asthma or atopic diseases throughout the offspring’s childhood? An updated systematic review and meta-analysis

Asthma is the most common respiratory disease among children, while atopic diseases such as atopic dermatitis affect about 20% of infants under 2 years old. Recently, studies suggest that they may all be related to prenatal depression or anxiety, but further research and analysis are needed. This study aimed to explore the association between prenatal mental disorders and childhood asthma or atopic disease in a systematic review and meta-analysis of nine studies. Methods PubMed, Embase, and Cochrane Library were searched up to May 2020. The primary outcome was childhood asthma and childhood atopic dermatitis. Random-effects models were used because of high heterogeneity indicated by I 2 >50% and Q-test P<0.10. Results A total of 598 studies were initially identied, but nine studies met the inclusion criteria. Prenatal mental disorder was associated with childhood asthma (n=6 studies; ES=1.146, 95%CI: 1.054-1.245, P=0.001; I 2 =93.5%, P heterogeneity <0.001) whereas no signicant association was found for childhood atopic dermatitis (n=4 studies; ES=1.211, 95%CI: 0.982-1.494, P=0.073; I 2 =78.5%, P heterogeneity <0.001). Childhood asthma seems to be related more to depression (n=1 study; ES=1.170, 95%CI: 1.061-1.291, P=0.002) and anxiety/depression (n=4 studies; ES=1.157, 95%CI: 1.050-1.275, P=0.073; I 2 =95.3%, P heterogeneity <0.001). This meta-analysis demonstrated that prenatal mental disorders increase the risk of childhood asthma. We limited the included samples to pregnant women to investigate the association between prenatal psychological factors and offspring’s physical health. Future studies should include large high-quality cohort studies to investigate the behavioral, environmental, and genetic causes for this association.


Background
In terms of the developmental origins of health and disease, there is evidence that adverse early-life exposure associated with maternal psychiatric diseases can alter the immune system and exacerbate the risk of asthma and atopic diseases such as atopic dermatitis (AD) [1][2][3][4]. Asthma is one of the most common respiratory diseases that seriously affects children's physical and mental health, as recognized by the World Health Organization [5,6]. Recent investigations have investigated the nature and mechanism of disease at the earliest stage in the rst few years of life [7][8][9]. Psychological and emotional factors are considered to trigger the exacerbation of asthma, and emotional stimulation has been proved to lead to the increase of respiratory resistance in asthma [10]. AD, also called atopic eczema, affects many children, especially infants under 2 years old. Some studies showed that the prevalence of AD in children under 2 years old is as high as 20% and reported a twofold to threefold increase in prevalence in the last 30 years [11]. There is evidence for a causal link between maternal mental disorder and AD, and a shared genetic pathway contributes to this familial liability [12].
A meta-analysis of 41 studies carried out in 2019 revealed a signi cant in uence of parental mental diseases and childhood asthma but no association with AD [13]. In the previous analysis, multiple childhood physical disorders were investigated, and asthma and AD were only two of them, and not all included studies examined those two diseases. Besides, the meta-analysis examined parental mental disorders, not exclusively the mothers', which could make more sense biologically because of possible transplacental in uence from the mother's hormones [14,15]. Besides, a new study was published in 2019 and reported that maternal mental disorder is not associated with offspring asthma signi cantly, whereas low job control might be a more relevant risk factor [16]. Although the association was signi cant, the odds of environmental effects on the periods between delivery and diagnosis of asthma or AD still cannot be ignored. Therefore, it is still unclear whether childhood asthma or AD is related to maternal mental diseases. We sought to undertake an updated meta-analysis and systematic review of all studies investigating the prenatal impact of anxiety and depression on childhood asthma and AD.

Methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were implemented for this systematic review and metaanalysis [17]. We searched for relevant articles using the PICO principle, followed by screening based on the inclusion and exclusion criteria.

Eligibility criteria
We included observational studies that examined the associations between prenatal mental diseases (depression and anxiety) and asthma and AD, with language limited to English. The participants were children and their biological mothers who undertook an intentional study with medical records access. The analysis was based on mothers with prenatal anxiety or depression (exposure) vs. mothers without prenatal anxiety or depression (non-exposure).

Search strategy
For this systematic review and meta-analysis, we searched PubMed, Embase, and Cochrane Library for studies published up to May 2020 with language and article type restriction. Supplementary Table 2 presents the search strategies.

Data extraction
The information was extracted independently by two investigators. The study characteristics (authors, year of publication, country, study design, and sample size), treatment parameters (exposure of the mother during pregnancy, questionnaire for the diagnosis of exposure, age of children when the study was taken, the outcome measurement, the covariates and the effect size were extracted from the included studies. Disagreements were resolved by discussion.

Data synthesis
In this meta-analysis, we extracted the adjusted odds ratios (aOR) from each included study, unless we determined any additional variables in the causal relationship between exposure and outcome. If aOR were not available, the crude ORs were extracted. When studies showed multiple exposures with multiple impact sizes, we reported only those exposures that the investigator considered the most severe and chronic. When a study reported multiple results separately, we extracted two results for different analyses. When a study reported only analyses of different asthma types, such as early or late-onset transient asthma, we treated the effect size as two independent outcomes. When a single study was available, the effect sizes of that study were presented.

Quality of the evidence
We assessed the cohort or case-control studies using the Newcastle-Ottawa Scale (NOS) [18]. Quality assessment was evaluated independently by two reviewers. The discrepancies in the assessment were resolved through discussion until a consensus was reached.
Statistical analysis STATA SE 14.0 (Stata Corp., College Station, Texas, USA) was used for all analyses. The studies were grouped by types of outcomes. The comparison between results was employed effect and corresponding 95% con dence interval (CI) for each group. We used Cochran's Q test and I 2 index to calculate statistical heterogeneity, for which high heterogeneity was de ned as I 2 >50% and P<0.10 in the Q-test [19]. A random-effect model was chosen for high heterogeneity. Pvalues <0.05 were considered statistically signi cant. We planned to conduct two sensitivity analyses, including removing the poor-quality studies ranked by determining the meta-analysis's robustness, including the estimated comparison between the type of exposure, study design, and sample collection continent. We did not estimate the potential publication bias with funnel plots for sensitivity analysis because the number of studies included in the meta-analysis was less than 10. For outcomes with less than 10 studies, the funnel plots and Egger's test could yield misleading results and were not recommended [20]. Figure 1 shows the study selection procedure, and the reasons for the excluded studies are also shown. A total of 598 studies were retrieved from PubMed (n=302), Embase (n=265), and Cochrane (n=31). After removing the duplicates, 432 articles were examined. Ten articles were excluded because of being notes/reports, 109 articles were excluded because they were conference abstracts, 25 articles were excluded because of being reviews, and 15 articles were excluded because of the language. Then, 273 articles were left for full-text screening and 264 were excluded because of being not accessible (n=10), study aim/design (n=54), population (n=100), outcome (n=28), intervention (n=63), and animal (n=9). No additional records were identi ed through other sources.
Characteristics of the included studies Table 1 presents the included studies [10,12,16,[21][22][23][24][25][26]. There were four prospective cohort studies [7,[18][19][20], four retrospective cohort studies [12,16,24,25] and one case-control study [26]. Most included studies were from Europe [10,12,16,21,23,26], and the others were from America [22,24,25]. These studies included 982,942 patients. The included studies investigated different exposures, including depression, anxiety, depression or anxiety, mental health service use, prenatal distress, and negative life events. Studies were grouped into the categories according to their outcomes: ve for asthma [10,12,16,21,24], three for AD [22,23,26], and one for asthma and AD [25]. To ensure the reliability and quality control of this meta-analysis, we scored each study using the NOS criteria, and studies with more than 5 stars were included in the meta-analysis. After evaluation, six studies scored 9 stars [10,12,21,23,25,26], and three studies scored 8 stars [16,22,24] (Supplementary Table 1a). Letourneau et al. [22] was examined infants of only 18 months of age, and Liu et al. [16] examined children when they were 0-6 years of age; both studies examined their population before the usual onset age for asthma and AD, possibly explaining their lower quality, at least in part. Radhakrishnan et al. de ned exposure as any use of mental health service, which may include a wide variety of mental illnesses besides anxiety and depression [24].

Effect of prenatal depression on childhood AD
Four studies (eight datasets) [22,23,25,26] could be included for the meta-analysis of prenatal depression on childhood AD. The result indicated that there was no statistically signi cant difference between the control and treatment groups, which means that prenatal depression may not in uence childhood AD (ES=1.211, 95%CI: 0.982-1.494, P=0.073; I 2 =78.5%, P heterogeneity <0.001) ( Figure 2B and Table 2b).

Sensitivity analyses
The sensitivity analyses indicated that publication bias was not signi cant since there was no individual study that affected the observed result for childhood asthma (Supplementary Figure 1A) and childhood AD (Supplementary Figure 1B).

Discussion
This study aimed to explore the association between prenatal psychiatric disorders and childhood asthma or atopic disease in a systematic review and metaanalysis of nine studies. The results indicate that prenatal mental disorders increase the risk of childhood asthma. We limited the included samples to pregnant women to investigate the association between prenatal psychological factors and offspring's physical health. Future studies should include large high-quality cohort studies to investigate the behavioral, environmental, and genetic causes for this association.
The strengths of this meta-analysis include the large number of patients included. Besides, only maternal exposure was considered. Finally, only studies on anxiety or depression and asthma or AD were included. The focus of the previous meta-analysis was to explore the effects of parental mental illnesses on children's physical health, revealing that prenatal mental disorder contributes to poor fetal growth and further suggesting the impact of maternal mental diseases during pregnancy on children [13]. Compared with the previous study, more attention was paid to the impact of maternal mental health during pregnancy on childhood systemic autoimmune diseases by the present meta-analysis. Even though the number of included studies was different from the previous meta-analysis [13], the included sample size that entered the nal analysis was similar to each other and could explain the results' consistency.
Nevertheless, this meta-analysis has limitations, and the results must be weighed against these limitations. All identi ed studies were observational and are therefore subject to confounding bias. Moreover, there were several retrospective studies. Factors such as selection bias, recall bias, and information bias were inevitably inherent in our analysis. Although we extracted the adjusted effect sizes for analysis, the covariates of each model were different. Some studies may be subject to over-adjustment, where analyses adjust for variables on the causal pathway between the exposure and the outcome. Although we actively tried to include unpublished research, all the identi ed studies were from the published literature. Therefore, it might well be that some positive ndings are the result of publication bias. Furthermore, some caveats should be considered while interpreting the ndings of subgroup analysis of childhood asthma since only Cookson et al. [10] reported an association between prenatal maternal anxiety and asthma, and another study by Magnus et al. [21] examined an association between prenatal maternal depression and childhood asthma.
Some studies have reported that anxiety/stress during pregnancy can aggravate the activation of the hypothalamic-pituitary-adrenal (HPA) axis, leading to the release of cortisol, which cannot be entirely metabolized by the placenta, and may even promote the release of placental glucocorticoids and crosses to the fetus, in which they can in uence the fetal brain development and may result in airway in ammation and hyperresponsiveness [27,28]. Maternal stressinduced changes in cortisol levels might affect fetal immune regulation and TH2 lymphocyte dominance by directly affecting cytokine production [14]. In human subjects, prenatal mental diseases were associated with changes in the inherent and adaptive immune responses in infants' umbilical cord blood at high risk for atopic disease [15]. β 2 -adrenoreceptors [29] expressed throughout the body were stimulated by the stress hormone adrenaline [30,31]. The investigators identi ed that maternal mental diseases during pregnancy might affect fetal growth, especially low-birth-weight infants with smaller lungs and airways, leading to a high asthma risk [32][33][34]. Given this wealth of possible mechanisms, focusing on maternal mental disorder's association during pregnancy and childhood asthma is meaningful.
The present study is a systematic review and meta-analysis (which has not been previously prospectively registered) on prenatal anxiety or depression symptoms and childhood asthma or AD. In the nine studies' nal analysis, the relationship between prenatal mental disorder and childhood asthma was statistically signi cant compared with childhood AD. The present study results are consistent with the previous meta-analysis [13], further con rming that prenatal mental disorders are associated with childhood asthma.
The association between prenatal maternal depression and childhood asthma seems more signi cant but only included one study that met the criteria [21]. In theory, prospective studies are more reliable than retrospective ones. Albeit not statistically signi cant in prospective studies [10,21], our ndings pointed out that statistical signi cance might be observed if new studies are conducted in the future, and the sample size is expanded. Therefore, this result is generally reliable, and the concerns could be overlooked. The results showed that asthma was signi cantly associated with maternal preconception mental status in Europe [10,12,16,21]. Although the combined overall aOR value was statistically signi cant (P=0.001), the association was not statistically signi cant for the North American studies (P=0.059) [24,25]. This difference might be due to the large con dence intervals in some studies and the conservative estimates of the North American subgroups by the random-effects model.
Although the result of this updated meta-analysis is consistent with the previous one, we limited the included studies to pregnant women to investigate the association between prenatal psychological factors and offspring's physical health. Future research might be able to deepen our understanding of when and how these vulnerable children are at risk of preventable illnesses. Those results also highlight the need for investments in the development of prevention programs during pregnancy, targeting maternal mental health promotion.

Limitations
All identi ed studies come from observational settings and are therefore subject to confounding bias. Moreover, there were several retrospective studies in our analysis. Factors such as selection bias, recall bias, and information bias were inevitably inherent in our analysis. Despite we extracted the adjusted effect sizes for analysis, the covariates of each model were different. Some studies may be subject to over-adjustment, where analyses adjust for variables on the causal pathway between the exposure and the outcome. Although we actively tried to include unpublished research, all the identi ed studies were from the published literature. Therefore, it might well be that some positive ndings are the result of publication bias.

Conclusions
In conclusion, this meta-analysis supports that prenatal mental disorders increase childhood asthma risk, whereas no signi cant association was found for childhood AD.

List Of Abbreviations
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA); the Newcastle-Ottawa Scale (NOS); con dence interval (CI) Declarations Ethics approval and consent to participate: Not applicable.
Consent to publish: Not applicable.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.  Figure 1 Flow chart of the study selection.

Figure 2
A) Forest plot of asthma. Children who were born to mothers with prenatal mental illness had increased odds of developing asthma. B). Forest plot of AD. The impact of prenatal mental illness was not signi cant for childhood AD