Published online Feb 15, 2024.
https://doi.org/10.4168/aair.2024.16.2.154
Postpartum Maternal Anxiety Affects the Development of Food Allergy Through Dietary and Gut Microbial Diversity During Early Infancy
,1,†
Jiwon Kim,2,†
Sukyung Kim,3
Hye-In Jeong,4
Mijeong Kwon,2
Hyun Mi Kim,2
Ji Sun Shim,5
Kyunga Kim,5
Jihyun Baek,6
Yechan Kyung,7
Suk-Joo Choi,8
Soo-Young Oh,8
Jaewoong Bae,9
Hong-Hee Won,1,10
Jihyun Kim,2,11
and Kangmo Ahn2,11
Abstract
Purpose
We aimed to investigate the mediating factors between maternal anxiety and the development of food allergy (FA) in children until 2 years from birth.
Methods
In this longitudinal cohort of 122 mother-child dyads from pregnancy to 24 months of age, we regularly surveyed maternal psychological states, infant feeding data, and allergic symptoms and collected stool samples at 6 months of age for microbiome analysis. Considering the temporal order of data collection, we investigated serial mediating effects and indirect effects among maternal anxiety, dietary diversity (DD), gut microbial diversity, and FA using structural equation modeling.
Results
Among the 122 infants, 15 (12.3%) were diagnosed with FA. Increased maternal anxiety between 3 and 6 months after delivery was associated with a lower DD score. Infants with low DD at 4 months showed low gut microbial richness, which was associated with FA development. When the infants were grouped into 4 subtypes, using consensus clustering of 13 gut bacteria significantly associated with maternal anxiety and DD, Prevotella, Eubacterium, Clostridiales and Lachnospiraceae were more abundant in the group with lower FA occurrence.
Conclusions
Postpartum maternal anxiety, mediated by reduced DD and gut microbial diversity, may be a risk factor for the development of FA in infants during the first 2 years of life.
INTRODUCTION
Food allergy (FA) in early childhood and subsequent dietary restrictions impair the quality of life and place a considerable psychological burden on both patients and their caregivers.1, 2, 3 FA occurs due to a combination of epicutaneous sensitization and the failure of oral tolerance induction during a critical window of opportunity in the first year of life.4 Recent research has shown that gut microbiota and dietary diversity (DD) during early life play a role in the induction of oral immune tolerance. Gut microbiota interacts with innate and adaptive immune systems within the gut mucosa to induce tolerance to food proteins.5, 6 In a birth cohort study in the United Kingdom, increased infant DD at 6 and 9 months significantly reduced the likelihood of FA development.7 Our previous study of a Korean birth cohort also revealed that higher DD within the first 6 months of life reduced the onset of FA by 12 months of age in high-risk infants, suggesting the beneficial roles of exposure to various foods in the gastrointestinal tract on oral immune tolerance.8 Primary prevention of FA is crucial to reducing its incidence and the progression to the atopic march. Although there is no conclusive evidence, early introduction of allergenic foods, an increase in DD, supplementation of prebiotics or probiotics, and appropriate use of emollients during infancy have been suggested to prevent the development of FA.7, 8, 9
Maternal stress and burnout due to physical and emotional exhaustion can hinder parents’ abilities to care for their children and provide appropriate responses.10 Psychological and emotional problems, including anxiety and depression, have been reported by approximately 20% of women within the first year after giving birth.11 Maternal anxiety during prenatal and postnatal periods can also affect their children’s emotional state.12 Additionally, prenatal maternal distress, mediated by oxidative stress, can influence the development of atopic dermatitis (AD) in infants.13 Maternal psychiatric symptoms increase the risk of developing eczema and inhalant allergy in children.8 However, little is known about the link between maternal anxiety and FA development.
Structural equation modeling (SEM) has been widely used to investigate causal relationships between observed variables.14 To comprehensively understand the relationship between variables, an SEM model converts multiple variables to latent factors that contain composite information. In a path analysis, SEM constructs relationships among variables and estimates associations.15 In the Canadian Healthy Infant Longitudinal Development cohort, for example, mediation approaches demonstrated that maternal overweight was linked to childhood overweight, mediated by cesarean section and infant gut microbiota.16 SEM path analysis in the same cohort showed that a cesarean section and a high ratio of Enterobacteriaceae/Bacteroidaceae at 3 months of age were causally associated with overweight and atopy at 1 year.17
In this study, we hypothesized that maternal psychological factors, specifically postpartum maternal anxiety, can affect the interplay between DD, gut microbial diversity, and the development of FA during the first 2 years of life. We used SEM path analysis to evaluate the relationship between prenatal and postpartum maternal anxiety, DD, gut microbial diversity, and FA development in infants 24 months after birth.
MATERIALS AND METHODS
Study population and design
In this birth cohort, we screened 244 pregnant women at 28 weeks of gestation or later and finally enrolled 161 infants. At enrollment, parents completed a questionnaire to supply basic demographic information and underwent a skin prick test (SPT) for common inhalant allergens. A family history of allergic diseases was established when 1 of 2 criteria was met: at least one parent had both a positive skin test response and a history of asthma or allergic rhinitis, or at least one parent or sibling had physician-diagnosed AD.
All infants were regularly followed up at 2, 6, 12, and 24 months of age for AD and FA. The diagnosis of AD was based on Hanifin and Rajka’s criteria.12 FA was confirmed by a positive oral food challenge test or a convincing history of adverse reactions within 2 hours of food ingestion plus positive serum specific IgE (≥ 0.35 kU/L) determined by the Immuno-CAP system (Thermo Fisher Scientific, Waltham, MA, USA). When the parents reported any suspicious allergic symptoms, even if they were not on a regular visit schedule, the infants were brought to the outpatient clinic and examined by pediatric allergy specialists. Immediate reactions included urticaria, eyelid edema, lip edema, cough, wheezing, vomiting, diarrhea, abdominal pain, hypotension, and altered mentality. This study was approved by the Institutional Review Board of Samsung Medical Center (IRB No. SMC-2016-12-111), and written informed consent was obtained from all the parents before participation in this study.
Additionally, gut microbiota samples were collected from all children at 6 months after birth and sequencing data were processed using the amplicon sequence variant-based pipeline (Supplementary Table S1). The details about SPTs, bacterial 16S rRNA sequencing of fecal samples, and statistical analyses are provided in Supplementary Data S1 and Supplementary Fig. S1.
Evaluation of maternal psychological states and DD
Maternal anxiety level was measured before birth and at 2, 6, and 12 months after birth using the Korean-translated version of the State-Trait Anxiety Inventory (STAI).13 It includes the state anxiety inventory and the trait anxiety inventory, each of which is composed of 20 items, scored from 1 (least likely) to 4 (most likely). An STAI score ≥ 40 was considered as the existence of anxiety symptoms.18 To evaluate the effects of maternal anxiety on DD and FA, we defined maternal anxiety as (1) the level of maternal anxiety measured at 2 months after birth (STAI2-mo), (2) the average level of maternal anxiety at 2 months and 6 months (STAIavg), and (3) the change in maternal anxiety between 2 and 6 months (STAIdelta).
Infant feeding data were collected using a standardized questionnaire at ages 2, 6, and 12 months, as described previously.5 The infants were surveyed about when they first started eating 11 different food groups, such as grains, vegetables, fruits, meat, fish, egg yolks, egg whites, dairy, wheat, peanuts, and legumes/nuts. Detailed dietary diversity (DDD) was calculated by summing the number of food groups introduced to each child’s diet from 3 to 12 months of age at intervals of 1 month. Therefore, the maximum score of DDD was 11 at each month.
RESULTS
Clinical characteristics of the study population
A total of 161 babies participated in this cohort, and 122 children (69 boys and 53 girls) and their parents completed the study (Fig. 1). Overall, 15 children (12.3%) were reported to have FA. Of these 15, 10 were diagnosed between 6 and 12 months, and 5 were diagnosed between 12 and 24 months after birth. Maternal anxiety level and gut microbiome data were obtained before FA diagnosis. Hen’s eggs were the most frequent food allergen (13/15), followed by peanuts (5/15), walnuts (3/15), cow’s milk (2/15), and wheat (1/15) (Table). Ten children (8.2%) were allergic to only one food, while 3 (2.5%) and 2 (1.6%) were allergic to 2 and ≥ 3 foods, respectively. Eleven children (9.0%) showed cutaneous manifestations, while 4 (3.3%) experienced anaphylaxis. No differences were found in birth weight, gestational age, delivery mode, sex, family history of allergic diseases, duration of breastfeeding, perinatal use of antibiotics, maternal education levels, and presence of AD before 12 months between participants with FA and those without (Table). Thirty children (24.6%) breastfed until 4 months of age. Those who were breastfed for longer than 4 months had decreased DD scores between 4 and 6 months (Supplementary Table S2). No differences were seen in maternal anxiety levels and DD scores at 3–8 months of age between the FA group and the healthy control. Maternal anxiety (STAI scores ≥ 40) was observed in 96 of 117 women (82.1%), 98 of 122 (80.3%), 94 of 122 (77.0%), and 84 of 110 (94.4%) in the prenatal period, and when the children were 2, 6, and 12 months of age, respectively. Maternal anxiety at 6 months increased in 51 cases (41.8%) and decreased in 62 (50.8%) compared to at 2 months.
Fig. 1
CONSORT flow diagram (A) and study design (B) for this study.
FA, food allergy; DDD, detailed dietary diversity; STAI, State-Trait Anxiety Inventory.
Table
Demographic and clinical characteristics of participants (n = 122)
Serial mediation analysis
We performed a mediation pathway analysis to investigate the relationship between STAI as an exposure and FA as an outcome, considering the temporal order of variable collection. Our model used 2 serial mediators (DD and gut microbial diversity) to examine the direct and indirect effects of maternal anxiety on the other variables. Increased maternal STAIdelta was significantly associated with reduced scores of DDD4-mo (P = 0.004) and DDD5-mo (P < 0.001) (Fig. 2A). Furthermore, DDD4-mo was significantly associated with gut microbial richness at 6 months of age (P < 0.001) (Fig. 2A). Additionally, gut microbial diversity at 6 months showed a negative association with the development of FA by 2 years of age, when DDD scores at 3, 4, 5, and 6 months were analyzed independently (P = 0.002, 0.004, 0.001, and 0.001, respectively) (Fig. 2A). We observed a significant indirect effect of an increase in maternal anxiety between 2 and 6 months on the increased risk of FA via the hypothesized path (indirect effect, P = 0.042) (Fig. 2A). We also investigated whether there was an effect from another path other than our proposed serial mediation effect. Maternal anxiety had neither an impact on change in Chao1 (P = 0.561) nor a direct effect on FA development (P = 0.678) (Fig. 2B).
Fig. 2
Statistical significance is denoted by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001.
Serial mediation estimates from maternal anxiety to food allergy development mediated sequentially through diet diversity and gut microbial richness (Chao1). (A) Forest plots of the results of an association analysis for each pathway in serial mediation. Monthly DDD scores at ages 3, 4, 5, and 6 months are analyzed independently. The plots display the average effect, standard error, and 95% CIs, with P values denoted in bold for statistically significant effects (P < 0.05). (B) A structural equation model diagram for 4-month DDD. The model includes all possible paths as covariates, with coefficients and P values for each path.
CI, confidence interval; DDD, detailed dietary diversity; STAI, State-Trait Anxiety Inventory.
No significant associations were found between prenatal maternal anxiety and DDD scores at 3, 4, 5, and 6 months (P = 0.986, 0.579, 0.591, and 0.786, respectively). Similarly, no significant associations were found between STAI2-mo and DDD scores at 3, 4, 5, and 6 months ( P = 0.568, 0.187, 0.164, and 0.738, respectively). Additionally, there were neither significant direct effects (all P > 0.05) nor indirect effects of an increase in STAI before birth and at 2 months after birth on the development of FA through the hypothesized path (all P > 0.05).
Effects of DD on gut microbial diversity and composition during infancy
Given the evidence for gut microbial involvement in the serial path between DD and FA development, we assessed whether DD and other factors contributed to distinct intestinal colonization. We conducted a multivariable regression analysis with potential covariates such as birth weight, sex, birth mode, and AD that could influence gut microbiome composition. Monthly collected DD scores across 3 to 6 months, average maternal anxiety during early infancy, and breastfeeding variables were significantly associated with gut microbial composition (false discovery ratio-adjusted P < 0.05) (Fig. 3A). Specifically, DD before 5 months accounted for a greater amount of variance than did breastfeeding. Fitted vectors of associated variables, including Chao1, were projected onto principal component analysis ordination axes (Fig. 3B). These vectors represent the directions of increased variables in the ordination space. An inverse relationship between breastfeeding and DD was observed. STAIavg and Chao1 also appeared to trend in the opposite direction, with high levels of maternal anxiety between 2 and 6 months associated with low gut microbial richness. In the univariable PERMANOVA test, antibiotics use, birth mode, and breastfeeding had significant effects on gut microbiota (P = 0.042, P = 0.033, and P < 0.001, respectively) (Supplementary Fig. S2).
Fig. 3
*Statistically significant variables are displayed with an asterisk (false discovery ratio-adjusted P < 0.05).
The relationship of clinical characteristics with gut microbial composition based on beta diversity. (A) The amount of variance is explained by environmental variables. (B) A biplot of statistically significant variables is found in (A). Each point represents a sample with colored FA cases in orange and healthy controls in light green.
FA, food allergy; DDD, detailed dietary diversity; STAI, State-Trait Anxiety Inventory.
We then conducted association tests to identify which genera of bacteria were significantly influenced by maternal anxiety and the monthly number of food types. Using multivariable analysis, we identified several taxa associated considerably with STAIdelta and monthly DDD at 3, 4, 5, and 6 months (Supplementary Table S3).
We then selected genus-level taxa that were associated with multiple factors, and the UpSet plot presented the co-association sets with bacterial names (Fig. 4). Prevotella was positively associated with DDD at 4, 5, and 6 months and negatively associated with STAIdelta but not with DDD at 3 months (Supplementary Fig. S3). When we focused specifically on associations between STAIdelta and DDD4-mo, we detected 6 taxa, including Prevotella, Neisseria, Haemophilus, and unclassified Clostridiales, that were negatively associated with STAIdelta and positively associated with DDD4-mo. Actinomyces and unclassified Enterobacteriaceae had a positive association with STAIdelta.
Fig. 4
UpSet plot visualizing taxa associated with more than one monthly dietary diversity and maternal anxiety. Taxa associated with a single variable were not included. The number to the right of each bar represents the number of the taxa associated with variables having black dots on the left side. Associated taxa are shown together and colored according to the directions of all significant associations. Green bars represent taxa concordantly associated with dietary diversity at 4 or 5 months and maternal anxiety and used in consensus clustering analysis.
DDD, detailed dietary diversity; STAI, State-Trait Anxiety Inventory.
Early gut microbiome marker to indicate healthy individuals without FA
To determine the effect of bacteria associated with maternal anxiety and DD on the development of FA, we conducted a consensus clustering analysis based on gut microbial compositions. For this analysis, we used 13 relevant bacteria that were concordantly associated with both STAIdelta and with DDD at either 4 or 5 months in the multivariable analysis, as presented as green-colored bars in Fig. 4. A total of 122 children were clustered into 4 subtypes using Euclidean distance of CLR-normalized microbial abundance (Fig. 5A). None of the infants in subtype C4 developed FA until 2 years of age, as seen in Fig. 5A and B (Fisher’s exact test comparing FA incidence between C4 and other subtypes, P = 0.021). We further identified specific bacterial groups that were enriched in each subtype using nonparametric Wilcoxon rank sum tests (Supplementary Table S4, Fig. 5C). Prevotella and Eubacterium were specifically abundant in subtype C4, which also showed significant differences in the Clostridiales order and Lachnospiraceae family compared with other subtypes (Fig. 5C and D). Additionally, the C4 subtype had a significantly lower abundance of Hemophilus and Clostiridium. In contrast, other subtypes (C1, C2, and C3) had Clostridium, Hemophilus, and Actinomyces, respectively, as representative markers.
Fig. 5
Four subtypes of gut microbial features. (A) Consensus similarity matrix. The optimal 4 consensus clusters were identified using K-means clustering on Euclidean distance. (B) CLR abundance heatmap of individual samples. Each row represents a gut microbial taxon, and each column represents a sample. Values were standardized by row to show relative changes. (C) Pairwise Wilcoxon rank sum test results of subtypes (consensus cluster) and taxa. Asterisks represent statistical significance (false discovery ratio-adjusted P < 0.05 and absolute estimate value > 1.0). (D) Boxplots of CLR abundance of cluster-enriched bacteria.
FA, food allergy; CLR, centered log ratio.
DISCUSSION
To the best of our knowledge, this is the first study to investigate the influence of maternal psychological factors during the early postpartum period on infant diet, gut microbial diversity, and subsequent development of FA. We demonstrated that increased maternal anxiety between 2 and 6 months after birth was associated with reduced DD in infants, increasing the likelihood of developing FA during the first 2 years of life. The effect of maternal anxiety on FA development was mediated by reduced gut microbial diversity. It has been suggested that DD and gut microbiota during the first year of life have an impact on FA development. Our previous study revealed that the introduction of solid foods within 6 months of birth increased gut microbial diversity and reduced the risk of developing hen’s egg allergy until 12 months of age, highlighting the importance of DD and the gut microbial environment during early infancy.8 Similarly, Azad et al. found that lower microbiota richness at 3 months of age increased the risk for food sensitization by 12 months of age, while no association was observed between richness at 12 months and food sensitization.19 Taken together, gut microbial richness around the age of 6 months plays a pivotal role in preventing FA in infants. The critical window period for dietary interventions that establish a beneficial gut microbial environment as a potential intermediate bridge may be between 3 and 6 months.
In the current study, children with greater DD scores and lower maternal anxiety levels before 6 months of age were associated with increased proportions of Prevotella, Neisseria, Haemophilus, and unclassified Clostridiales in their gut microbiome. Using consensus clustering, we classified the children’s gut microbiome into 4 subtypes, each with distinct intestinal microenvironments that suggest different functional roles and interactions with the host. Of these subtypes, a specific group with food tolerance showed higher levels of beneficial bacteria such as Prevotella, Lachnospiraceae, and Clostridium. These findings are consistent with previous studies showing that patients with allergic diseases had a less diverse gut microbiome characterized by a high proportion of Enterobacteriaceae and a reduction in the abundance of Streptococcus, Bacteroides, Prevotella, and Coprococcus.19, 20, 21, 22, 23, 24 Another study also showed that Clostridiales populations were lower in infants with FA and that fecal microbial transplantation containing these Clostridiales species prevented the development of FA in mice through the induction of ROR-γt+ regulatory T cells in a MyD88-dependent manner.25
Maternal psychological factors during the perinatal period have been shown to influence gut immunity and the development of allergic diseases in childhood.26, 27, 28 In particular, prenatal maternal depression is associated with low fecal secretory IgA concentrations and an increased risk of AD and asthma in offspring.26, 27 This mechanism is supported by fetal programming through activating the hypothalamic-pituitary axis and reducing the glutathione-to-glutathione disulfide ratio associated with oxidative stress in response to maternal stress.27, 29 High levels of cortisol and oxidative stress induce T helper cell type 2 skewing and increase the risk of allergic diseases.30 In contrast, our study suggests that exposure to maternal stress early in life was a more significant risk factor for FA development in the offspring than stress before childbirth. Notably, our study demonstrated that STAIdelta significantly impacted infant DD and later development of FA, whereas STAI at specific time points did not.28, 31 This suggests that the growing anxiety of mothers during a critical window period greatly influences parenting attitudes and infant feeding, finally contributing to FA development.
Maternal depression early in life, which has been of interest in previous studies, was not statistically associated with children’s FA in our present study. However, the potential effects of breastfeeding on DD and allergies in infants should be considered in future studies. Lower DD was observed in children who continued breastfeeding at 4 months, and breastfeeding and DDD affected compositional changes in the opposite direction (Fig. 3B). Some mothers who want to breastfeed for longer than 4 months tend to hesitate to introduce diverse types of foods, for fear of undermining the balanced beneficial effect of breastmilk and DD. The complex interplay between maternal anxiety, breastfeeding, DD, and FA in infants highlights the need for a holistic approach to understanding the development of gut microbiota and immune systems in early life.
In the present study, we employed serial mediation analysis using an SEM as a causal inference model with variables arranged in temporal order.32 Measuring DD variables before 6 months, collecting gut microbiome data at 6 months, and diagnosing FA after 6 months of age suggest that the SEM can establish causal relationships between these variables and the development of FA in infants. Unlike multiple regression models, a serial mediation SEM allows for controlling variables in previous paths when testing later paths, enhancing the accuracy of the partial mediation effect from maternal anxiety to FA.33 Furthermore, we adjusted for other paths, such as maternal anxiety to microbial diversity and maternal anxiety to FA, when we evaluated the hypothesized indirect effect (a-b-c). Overall, using SEM for serial mediation analysis allowed us to examine the causal relationships between variables in the context of FA development in infants.
Our study has some limitations, mainly stemming from a small sample size. Additionally, our study did not investigate the specific factors responsible for the increase in maternal anxiety between 2 months and 4 months after birth, which may contribute to postpartum stress. The postpartum period represents a challenging transition to parenthood, marked by a multitude of stressors, doubts about parenting skills, sleep disturbances, and unceasing demands.34 These stressors can lower parental self-efficacy and self-esteem, leading to symptoms of depression, anxiety, and stress.35 Furthermore, psychological issues in parents can have adverse effects on infant outcomes, including breastfeeding and nutrition.36 High levels of postpartum maternal anxiety, as observed in our study, may have influenced infant DD, potentially due to concerns about allergic symptoms and the added burden of preparing weaning diets.36 Future research is necessary to investigate the precise causes of postnatal psychological stress and explore the mechanisms through which this stress may impact DD during early infancy. Understanding these factors can provide insights into developing targeted interventions to reduce maternal stress and promote healthier infant feeding practices, potentially mitigating the risk of FA.
In conclusion, postpartum maternal anxiety, mediated by reduced DD and gut microbial diversity, may be a risk factor for the development of FA in infants during the first 2 years of life. The abundance of beneficial gut bacteria such as Prevotella, Eubacterium, Clostridiales and Lachnospiraceae at 6 months of age is associated with a lower likelihood of developing FA. These findings have important implications for developing personalized interventions, including maternal psychological management and dietary strategies to improve gut microbial diversity and prevent the occurrence of FA during early childhood.
SUPPLEMENTARY MATERIALS
MethodsSupplementary Data S1
Deblur quality statistics on gut microbial samplesSupplementary Table S1
Nonparametric tests for the difference in dietary diversity between children breastfed for longer than 4 months and those who were notSupplementary Table S2
Significant associations with variables from multivariable association testsSupplementary Table S3
Association of gut microbiota subtypes with taxa abundancesSupplementary Table S4
Rarefaction curve and sample depths of fecal microbiome samples collected at 6 months of age. A vertical line in the rarefaction curve (left) and the lowest horizontal line in the sample depth plot (right) represent the minimum sequencing depth of samples (x = 6,866).Supplementary Fig. S1
The scatter plot of gut microbial composition with clinical variables based on beta diversity. Statistical significances in the distribution of samples between infants who received antibiotics until 2 months of life (A), those born by cesarean section (B), and those breastfed until 4 months of age (C) are shown. Ellipses indicate 95% confidence intervals. P values were determined by PERMANOVA.Supplementary Fig. S2
The heatmap of effect directions from multivariable association tests on taxa abundance.Supplementary Fig. S3
Disclosure:There are no financial or other issues that might lead to a conflict of interest.
ACKNOWLEDGMENTS
This study was supported by Samsung Medical Center Grant #SMO1230141 and by the Korea Environment Industry and Technology Institute through Environmental Health Action Program funded by Korea Ministry of Environment (grant No. 2017001360006).
References
-
Fleischer DM, Chan ES, Venter C, Spergel JM, Abrams EM, Stukus D, et al. A consensus approach to the primary prevention of food allergy through nutrition: guidance from the American Academy of Allergy, Asthma, and Immunology; American College of Allergy, Asthma, and Immunology; and the Canadian Society for Allergy and Clinical Immunology. J Allergy Clin Immunol Pract 2021;9:22–43.e4.
-
-
Findling Y, Barnoy S, Itzhaki M. Burden of treatment, emotion work and parental burnout of mothers to children with or without special needs: a pilot study. Curr Psychol 2023;42:19273–19285.
-
-
Fan Y, Chen J, Shirkey G, John R, Wu SR, Park H, et al. Applications of structural equation modeling (SEM) in ecological studies: an updated review. Ecol Process 2016;5:1–12.
-
-
Simonyté Sjödin K, Hammarström ML, Rydén P, Sjödin A, Hernell O, Engstrand L, et al. Temporal and long-term gut microbiota variation in allergic disease: a prospective study from infancy to school age. Allergy 2019;74:176–185.
-
-
Bullock HE, Harlow LL, Mulaik SA. Causation issues in structural equation modeling research. Struct Equ Modeling 1994;1:253–267.
-
Cited by
MeSH Terms
Figures
1 / 5
Tables
1 / 1
ORCID IDs
Funding Information
-
Samsung Medical Center
SMO1230141
-
Ministry of Environment
2017001360006
