Presented at the Neonatal Society 2014 Spring Meeting.
Hutchinson E, Matthews R (joing first authors), Santhakumaran S, Hyde MJ, Boyle R, Modi N
Section of Neonatal Medicine, Imperial College London, 369 Fulham Road, London, UK
Background: It has been hypothesised that the increase in caesarean section (CS) rate from 9% to 25% over the last three decades, is causally associated with the concomitant increase in food allergy. Plausible mechanisms include differences in gut colonisation (1) and lack of labour-induced stress response (2). An increased risk of other immunological conditions in infants following birth by CS, when compared to vaginal delivery (VD), including asthma (3) and diabetes mellitus (4), have also been reported. We aimed to conduct a systematic review and meta-analysis to evaluate the association between mode of delivery and allergic sensitisation to food in offspring aged <4 years.
Methods: A literature search was conducted in PubMed and Web of Science using pre-defined search terms, following a registered protocol (5). For inclusion, a study should have included data on mode of delivery and allergic sensitisation to cow’s milk, egg or peanut (the most common food allergies at this age (6), using either specific immunoglobulin E blood testing (sIgE) or skin prick testing (SPT), in infants <4 years of age. Reference lists were hand-searched for further potential studies. Authors were contacted to obtain data from studies in which it was clear that data on mode of delivery and food allergic sensitisation had been recorded, but the association not reported. A meta-analysis of all studies with relevant data was carried out in RevMan5 using the Mantel-Haenszel method. Random effects models were used throughout as the assumption of a common fixed effect is not appropriate when using observational data. Data for confounders was generally not available and therefore all data included are unadjusted. Forest plots are used to illustrate results and funnel plots used to investigate publication bias. Results are presented as Odds Ratios (OR) and 95% confidence interval. Heterogeneity is represented using I2, the proportion of variation in study results due to between-study differences.
Results: Nine studies, comprising 10203 infants (CS = 2536) were included in the meta-analysis. Primary Outcome: CS was not associated with an increased risk in allergic sensitisation to any food, when compared to VD (unadjusted OR 1.08 [0.91, 1.28]; p=0.39; I2=8%; 6 studies; 8483 subjects: 6218 VD, 2265 CS). Secondary outcomes: No association was found between mode of delivery and allergic sensitisation to cow’s milk, egg and peanut: cow’s milk (unadjusted OR 1.19 [0.94, 1.50]; p=0.15; I2=0%; 9 studies; 7766 subjects: 6055 VD, 1711 CS); egg (unadjusted OR 1.18 [0.97, 1.45]; p=0.11; I2=14%; 9 studies; 10167 subjects: 7647 VD, 2520 CS); and peanut (unadjusted OR 1.29 [0.79, 2.11]; p=0.30; I2=59%; 7 studies; 9168 subjects: 6784 VD, 2384 CS). There was no evidence that type of CS (in-labour or pre-labour), family history of allergy, or method of measuring allergic sensitisation (sIgE or SPT) altered the association with mode of delivery.
Conclusion: Our findings contrast with other published literature and cast doubt on the ‘microbiota’ hypothesis of the pathogenesis of food allergy. This suggests that early changes in intestinal microbiota, such as those following birth by CS, lead to abnormal development of gut-associated lymphoid tissue and consequent food sensitisation. We were not, however, able to adjust the analyses for confounders, such as breastfeeding, which may influence immunological outcomes in offspring.
Corresponding author: matthew.hyde02@imperial.ac.uk
References
1. Gronlund et al. JPGN. 1999; 28:19-25.
2. Zanardo et al. IJOG. 2006; 95:52-3.
3. Thavagnanam et al. Clin Exp Allergy. 2008; 38:629-33.
4. Cardwell et al. Diabetologia. 2008; 51:726-35.
5. Hyde et al. PROSPERO. 2013; CRD42013004158
6. Rona et al. J Allergy Clin Immunol. 2007; 120:638-46.