Mode of delivery and offspring atopic dermatitis in a Swedish nationwide study

Atopic dermatitis is a common chronic childhood disease associated with significant morbidity and healthcare costs. There is a known association between caesarean section and asthma, but the relationship between caesarean section and offspring atopic dermatitis remains uncertain.


| BACKG ROU N D
Allergic diseases and atopic conditions such as atopic dermatitis have risen globally, and yet the reason for this increase is not fully elucidated. 1 Atopic dermatitis is the most prevalent chronic noncommunicable disease of the skin globally. 2 It often presents as a relapsing condition characterized by pruritus, abnormally dry skin, and eczematous lesions. 3 The disease displays a high heterogeneity in its natural course and individual trajectories are unpredictable; however, it may have systemic effects that increase the likelihood of other comorbidities. Apart from its association with other atopic conditions such as food allergy and asthma, 4 children and adults with eczema are also more likely to experience more frequent bacterial and viral infections. 5,6 Other comorbidities include overweight/obesity 7 and also psychiatric conditions such as depression, anxiety, and autism. 8 The condition is often mild, but in some children and adults may be debilitating, require constant medication and significantly impair the quality of life. Further, Sweden ranks as having one of the highest disability-adjusted life-years for atopic dermatitis globally making it an important place in which to investigate the condition. 9 It has been suggested that the increased atopic dermatitis risk may be partially explained by the increase in caesarean sections across the globe. [10][11][12] In the last decade, caesarean sections have risen globally from 9% to 21%, with Europe reporting rates of circa 25.7% (23.4-28.0) in 2018. 13 Caesarean sections have been linked with decreased gut microbial diversity and a reduced ability of the infant body to develop adequate immune response mechanisms. 14 Whilst the indications for births by caesarean section are multifactorial, the mode of delivery may have clinical outcomes for children, including atopic dermatitis. 15 Previous studies have examined the association between birth by caesarean section and atopic dermatitis with inconsistent results. In various studies, including a prospective Australian birth cohort study (n = 10,383), 16 the United States National Surveys of Children's Health birth cohort study (n = 249,585), 17

and the British
West Midlands birth cohort study (n = 24,690) 18 there was no association found. [16][17][18] By contrast, a positive association between caesarean section and atopic dermatitis was observed in both a study conducted in Korean adolescents (n = 1302), 19 and in Ecuadorian children (n = 400). 20 As caesarean section may be associated with asthma, 11,21 and asthma and atopic dermatitis are genetically and phenotypically linked, the inconsistent results observed in different studies need to be explored further.
Some studies do not account for the temporal risk of early onset atopic dermatitis or the different types of methods of delivery. Birth by caesarean section has been associated with an altered gut microbiome, 22 but the differences between elective and emergency caesarean sections, or those between uncomplicated and complicated vaginal deliveries are not regularly accounted for. The use of different definitions of atopic dermatitis including differences based on physician or parent reporting, and an inability to adjust for different familial and socioeconomic confounders, including those shared by siblings, may also limit the generalizability of findings.
The aim of this nationwide register-based cohort study of Swedish mother and child pairs was to investigate the association between mode of delivery and atopic dermatitis in children aged ≤5 years. We additionally conducted a sibling analysis to adjust for unmeasured familial factors, thus further assessing causality. 23 2 | ME THODS

| Study population
This register-based cohort study included all singleton children born in Sweden between January 2006 and December 2018 (n = 1,400,713) who were identified in the Medical Birth Register. 24 This register includes ≥98% of all births in Sweden and reports on antenatal events, deliveries, and birth characteristics. By using a unique personal identity number, 25 information about mother and child was linked to other registers for health and socioeconomic factors. These include the longitudinal integrated database for health insurance and labour market studies (LISA) 26 29 We excluded children who had missing information on the mode of delivery (n = 1286) and children with missing information on maternal pregnancy (n = 21), Figure S1. Siblings were identified by shared maternity in the Medical Birth Register, and these were included in the sibling analysis (n = 853,884), of which 320,692 were discordant.

| Outcome
Atopic dermatitis was defined using a clinical algorithm established by Henriksen et al. 30 This is based on either a diagnosis of atopic

Key Message
Our study found that there is an increased risk for atopic dermatitis in children aged ≤5 years born by caesarean section.
Birth by instrumental vaginal delivery, emergency, and elective caesarean delivery was associated with atopic dermatitis.
In the sibling control analysis, birth by instrumental vaginal delivery, emergency, and elective caesarean delivery was associated with atopic dermatitis in children aged <1 year indicating some familial confounding. dermatitis in the National Patient Register or the dispense of drugs used in the treatment of atopic dermatitis as identified from the Swedish Prescribed Drug Register. A full description of the criteria is included in the Appendix S1.

| Exposure
Mode of delivery was extracted from the Medical Birth Register. 24 It was first defined as birth by vaginal delivery or caesarean section. Using information from the Medical Birth Register on the type of delivery, spontaneous onset of labour, caesarean section (elective or emergency caesarean section prior to the onset of labour, or emergency caesarean section after the onset of labour), and vaginal instrumental delivery were extracted. Based on these, the mode of delivery was categorized as: (1) vaginal, uncomplicated delivery, (2) vaginal, instrumental delivery, (3) emergency caesarean section (prior to or after start of labour) and (4) elective caesarean (before start of labour).

| Other variables
Potential confounders were identified based on prior knowledge and a directed acyclic graph, Figure S2 Register. 31 The LISA was used for information on the maternal level of education; compulsory (≤9 years), secondary (10-11 years), or tertiary (≥12 years) at the time of the child's birth.

| Statistical analyses
Time-to-event analysis using Cox proportional hazards regression models was performed with attained age as the analysis time scale to assess the association between mode of delivery and atopic dermatitis. The cluster robust sandwich estimator for standard errors was used to account for clustering of observations within families (siblings). The proportional hazards assumption was tested based on Schoenfeld residuals and found to be violated. We therefore allowed for time-varying effects, where follow-up time was divided into two intervals: age <1 year and age ≥1 year. The follow-up was restricted to age 5 as ≥80% of the childhood onset of atopic dermatitis occurs before age 5. 32 Observations were censored at age 5, emigration, death, or at the end of the study on December 2021 (whichever came first). Models were presented as (i) age-adjusted (as analysis time scale), (ii) age-and sex-adjusted, and (iii) adjusted for age, sex, birthweight, gestational age, maternal age, parity, maternal history of asthma, and maternal level of education.
Further, we conducted a sibling control analysis to account for unmeasured familial confounders, both environmental and genetic, shared by siblings using stratified Cox proportional hazards regression. Siblings were identified by shared maternity in the Medical Birth Register and restricted to siblings with the same mother.
Siblings discordant for exposure and outcome were informative for the estimation of the hazard ratios for the association of interest, although all siblings were included in the analyses as they would contribute to the estimation of covariate coefficients. The sibling models were adjusted for covariates not shared by siblings, that is, child's age, sex, gestational age, birthweight, parity, and maternal age. All results from Cox proportional hazards regression were reported as hazard ratios (HR) and 95% confidence intervals (CI).
Analysis was performed using STATA 16. 33

| Ethics statement
All data were pseudonymized prior to analysis. Study participants were not involved in the planning or writing of this research project. Ethical approval was obtained for this project from the Swedish Ethical Review Authority.

| RE SULTS
The study included 1,399,406 children after excluding 1307 children with missing information on the mode of delivery or with missing information on the mother's identity, Figure S1.  Table S1.

| Association of mode of delivery with atopic dermatitis
In this population of 1,399,406 children, 243,293 (17.4%) developed atopic dermatitis at a mean age of 2.72 years (SD 1.8). In total children, were followed up for 6,029,542 person-years. For children TA B L E 1 Baseline characteristics of mother and child pairs.  Table 3. An illustration of the survival estimates is shown in Figure S3.

| Sibling analysis
In sibling analyses, compared with vaginal delivery, in children aged

| DISCUSS ION
In this prospective population-based study investigating the asso-

| Comparison to other studies
To the best of our knowledge, only two previous studies have shown an association between mode of delivery and atopic dermatitis. 19,35 In the Korean study by Yu et al, (n = 1302; OR 1.80, 95% CI: 1.14-2.85) 19 there was higher odds of atopic dermatitis, but not asthma, in an adolescent population. In addition to variables we adjusted for, the researchers also adjusted for serum 25 attendance because we considered them mediators of the relationship between the mode of delivery and atopic dermatitis association rather than confounding structures. Although we could not account for ethnicity directly in our study, we adjusted for the maternal country of birth as a proxy and observed no difference in the overall estimates (results not shown). Further, by conducting a nationwide study, we were also able to include children of different ethnicities. There is a predominance of studies on individuals of European ancestry, and few nationwide studies that account for ethnic diversity. Atopic dermatitis, however, is more common and severe in non-Caucasian than Caucasian children. 36 48 The mechanism for the increased risk of atopy associated with forceps or instrumental vaginal delivery is unknown but may be associated with increased fetal stress that alters the programming of Th1 to a Th2 immune response. 18 Atopic dermatitis is associated with interactions between multiple genes and environmental factors. In our sibling analyses, in children aged under 1 year, an elevated risk of atopic dermatitis was observed in children born by caesarean section. Diepgen et al., in their study on familial aggregation, showed that a stronger correlation of atopic dermatitis exists between siblings than between siblings and parents, suggesting an important role of shared environmental factors. 49 We observed a higher risk in siblings born by caesarean section, which would suggest that it is an environmental confounder, which is not shared by siblings, but that the effect of the confounder attenuates as the children grow older as reflected by the lack of association in siblings aged ≥1.

| Strengths and limitations
This is a population-based nationwide study that utilized the extensive Swedish medical and administrative registers and is therefore larger than any previous study in children aged ≤5 years. 17 Our study has some limitations. Some important variables could not be controlled or accounted for and thus we cannot exclude residual confounding, 55 for example, paternal atopy. Neither could we explore possible mechanisms that may help to explain the association between caesarean section and atopic dermatitis such as breastfeeding. It is known that delivery by caesarean section reduces the likelihood of early initiation and increases the likelihood of early cessation of breastfeeding, and this may further delay the maturation of the gut microbiome. 55 This study does not represent children with mild disease who may have received over-the-counter medication or who had been seen by a general practitioner without being prescribed medication.

| CON CLUS ION
In this population-based cohort study, there was an association between birth by caesarean section and atopic dermatitis. This risk re-

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/pai.13904.