Background The mode of delivery has recently gained attention as another potential perinatal risk factor for childhood obesity but results are conflicting.
Objective To examine whether caesarean section is independently associated with childhood obesity after adjusting for a broad range of confounding factors.
Methods The current study used a population-based survey in Grade 5 students linked to a provincial perinatal registry in the Canadian province of Nova Scotia. Associations between caesarean section and childhood overweight and obesity at age 10/11 years were examined using multiple logistic regression.
Results Of the 4298 students who participated in the 2003 Children's Lifestyle and School Performance Study (response rate 51.1%), 3426 (80%) could be linked with information in the Atlee Perinatal Database, and 2988 mother-child pairs (70%) had complete information on the exposure and outcome. Compared to vaginal delivery, caesarean section was associated with offspring obesity (OR) 1.49, 95% CI 1.10 to 2.00) in the univariate analysis. After adding maternal prepregnancy weight to the multiple regression model, the OR for obesity dropped from 1.48 to 1.20 (95% CI 0.87 to 1.65). When caesarean section with and without labour were considered separately, we found no statistically significant associations relative to the vaginal delivery group (OR 1.24, 95% CI 0.84 to 1.82 and OR 1.03, 95% CI 0.58 to 1.84).
Conclusion Our results do not support a causal association between caesarean section and childhood obesity. Maternal prepregnancy weight was an important confounder in the association between caesarean delivery and childhood obesity and needs to be considered in future studies.
- Cesarean section
Statistics from Altmetric.com
What is already known on the subject
A number of perinatal factors have been shown to be associated with obesity in the offspring.
Delivery by caesarean section has recently been reported as a risk factor for childhood obesity but the association may be due to confounding.
What this study adds
Findings do not support a causal association between caesarean section and childhood obesity.
Maternal prepregnancy weight is an important confounder in the association between caesarean delivery and childhood obesity and needs to be considered in future studies.
A number of perinatal factors have been shown to be associated with obesity in the offspring including high maternal prepregnancy weight, excessive gestational weight gain, and high birth weight.1–3 Recently, the mode of delivery gained attention as another potential perinatal risk factor for childhood obesity when a number of studies investigated the association between caesarean section and obesity in the offspring.4–10 The hypothesis underlying these studies is that the bacterial exposure of the newborn following caesarean section compared to vaginal delivery results in differences in the colonisation of the intestine,11 which in turn influences energy metabolism in the gut and may contribute to the later development of obesity.12 However, results from the studies are conflicting and most studies were not able to adjust for all potentially important confounders, the most important of which is maternal prepregnancy weight: Mothers who are obese are more likely to have obese offspring13–15 and require caesarean section more frequently than non-obese mothers.16 Other potential confounders comprise breast-feeding, socioeconomic status, maternal age, parity and smoking. Given the increase in caesarean section rates worldwide, clarification of the contribution of surgical delivery to the current obesity epidemic is of interest to policy makers and clinicians alike. The objective of the current study, therefore, was to examine whether caesarean section is independently associated with obesity in childhood after adjusting for a broad range of confounding factors.
The current study was a population-based retrospective cohort study in the Canadian province of Nova Scotia using data from the cross-sectional 2003 Children’s Lifestyle and School Performance Study (CLASS)17 that were linked with a province-wide perinatal database.
Children’s lifestyle and school performance study
The CLASS was a population-based survey of grade 5 students and their parents in Nova Scotia. The study consisted of a home questionnaire that was completed by the parents; a version of the Harvard Youth/Adolescent Food Frequency Questionnaire18 adapted for Canadian settings that was administered to the students in the schools by study assistants; and a measurement of the students’ height and weight. The home questionnaire collected information on sociodemographic factors, breast-feeding practice, the child’s physical activity and sedentary activity, and their physical and mental wellbeing. Standing height was measured by a research assistant to the nearest 0.1 cm after students had removed their shoes; body weight with the students wearing light indoor clothing was measured to the nearest 0.1 kg on calibrated digital scales (Tanita HD-314, Tokyo, Japan). In addition to the above information, participating parents were asked to provide their Nova Scotia Health Insurance number and informed consent to allow linkage with administrative health databases. More than 97% of students in Nova Scotia attend public schools. Of the 291 public schools in Nova Scotia with grade 5 classes, 282 (96.9%) participated in the study. The average rate of return of questionnaires and consent forms was 51.1% per school. One of the seven provincial school boards did not allow measurements of height and weight. A total of 4298 students participated in the study and had their height and weight measured.
The Nova Scotia Atlee Perinatal Database (NSAPD) has collected demographics, procedures, interventions, maternal and newborn diagnoses, and morbidity and mortality information for all pregnancies and births occurring in hospitals in the province since 1988.19 Data are collected from hospital charts by trained health records personnel using standardised data collection forms. As part of an ongoing data quality-assurance programme for the NSAPD, routine data checks and edits are made at the time of data collection, and reabstraction studies are preformed regularly.20 ,21 Linkage of the CLASS data with the NSAPD was carried out by the Reproductive Care Program of Nova Scotia that administers the database. A combination of deterministic and probabilistic matching was used to link the two datasets. Of the 4298 students in the CLASS study with measured height and weight data, 3426 (79.7%) could be linked with information in the NSAPD. The most common reason for an unsuccessful linkage was that children were born outside Nova Scotia (12.4%); for the remaining children, parents had provided an invalid or no health insurance number. We excluded 373 students whose mothers’ prepregnancy weight was not recorded and 65 students with missing data for birth weight or gestational age, leaving a final sample of 2988 children.
This study, including data collection, parental informed consent forms and data linkage with the NSAPD, was approved by the Health Sciences Human Research Ethics Board of Dalhousie University, the IWK Health Centre Research Ethics Board and the Joint Data Access Committee of the Reproductive Care Program of Nova Scotia.
The primary outcome was a child’s body mass index (BMI) at age 10/11 years. Children were categorised as normal weight, overweight or obese using the age-specific and gender-specific BMI cut-off points for children and youth established by the International Obesity Task Force22 based on health-related adult definitions of overweight (≥25 kg/m2) and obesity (≥30 kg/m2). For supplementary analyses of the primary outcome, BMI z-scores were calculated using the LMS method based on the International Obesity Task Force reference population.23
Main exposure of interest
The main exposure was mode of delivery, grouped as planned or unplanned caesarean delivery versus spontaneous or assisted vaginal delivery as recorded in the NSAPD. To assess the influence of whether or not labour had occurred before caesarean section in a supplementary analysis, the exposure was further disaggregated into: caesarean section, no labour; caesarean section, first/second stage labour; vaginal delivery.
From the CLASS study the following variables were derived: gender; physical activity (parental report; ≤2 times per week, >2–4 times per week, >4–7 times per week, >7 times per week); Diet Quality Index24 (range 0–1, with 1 indicating highest diet quality); breast feeding (parental report; <1 week or none, > 1 week to 3 months, >3 months to 6 months, >6 months); household income ($0 to $20 000, $20 001 to $40 000, $40 001 to $60 000, >$60 000); highest household educational attainment (secondary school or less, college, university); marital status (married/common-law, single/divorced); area of residence (urban, rural); average neighbourhood household income (from the 2001 census).
From the NSAPD the following variables were derived: Prepregnancy weight; gestational weight gain (according to the Institute of Medicine's recommendations25; within recommended range, inadequate, exceeding recommendations); smoking status at delivery (none, 1–12 cigarettes/day, or ≥13 cigarettes/day); parity (1, 2, ≥3); maternal age; pre-existing or gestational diabetes (yes, no); previous infant with a birth weight >4080 g (yes, no); gestational age; birth weight z-score; weight-for-gestational age (small (SGA; ≤10th percentile), appropriate (AGA) or large for gestational age (LGA; ≥90th percentile)); delivery hospital (eight categories); type of delivery hospital (tertiary care, regional, community).
As maternal height and BMI were not routinely collected for the NSAPD at the inception of the cohort, we predicted maternal BMI category (underweight (<18.5 kg/m2), normal weight (18.5–<25 kg/m2), overweight (25–<30 kg/m2), obese (≥30 kg/m2)) from maternal prepregnancy weight based on height/weight data from 54 821 mothers for whom height and weight were recorded in the NSAPD from 2002–2011; predicted BMIs were used to categorise gestational weight gain based on the Institute of Medicine's recommendations. Weight-for-gestational age categories and birth weight z-score were determined based on Canadian reference values.26
Descriptive statistics for maternal and child characteristics by mode of delivery were presented as means or relative frequencies as applicable. Data on delivery hospital and type of delivery hospital were suppressed to protect the anonymity of the hospitals. A series of multinomial logistic regression models was used to examine the association between caesarean section and weight status (normal weight, overweight, obese) at age 10/11 years. The models were incrementally adjusted for sex, parental education, area-level income, urban residence, maternal age (Model 1); prepregnancy weight (as a continuous variable) (Model 2); birth weight z-score (Model 3); breast feeding (Model 4). The remaining covariates were then added to the model individually and were retained in the final model (Model 5) if they changed the estimate of the main effect by ≥10% in either direction. Models 1–5 were also run using the children's BMI z-score23 as the outcome.
Model 1 was further stratified by prepregnancy weight (<67 kg vs ≥67 kg), and Models 1 and 2 were stratified by prepregnancy weight, gestational age (<37 weeks vs ≥37 weeks), and weight for gestational age (AGA, SGA, LGA) to examine whether the association between mode of delivery and obesity differs across the strata of these variables.
As CLASS participation rates in areas with lower estimates of household income were slightly lower than the average, inverse probability weighting was used to overcome potential non-response bias. On the basis of average household incomes according to postal code data from the 2001 census for participants and non-participants, response rates per decile of household incomes by postal code were calculated. Decile-specific response weights were then calculated as the inverse of the decile-specific response rate. All analyses in the present paper are weighted for non-response. Stata V.12 (Stata Corp, College Station, Texas, USA) was used to perform the statistical analyses.
The prevalence of overweight and obesity in the cohort was 23.1 and 9.8%, respectively. Nineteen per cent of children were born via caesarean section. A breakdown of the sociodemographic and clinical characteristics for the vaginal and caesarean delivery groups can be found in table 1. There were statistically significant differences for a number of maternal and child characteristics between the two exposure groups. Mothers in the caesarean section group were older, had a higher prepregnancy weight, were more likely to be primipara, to bottle feed, and to have a SGA or LGA infant or a previous delivery of a baby with a birth weight >4080 g. Offspring of mothers in the caesarean section group had lower birth weight z-score and gestational age compared with those of mothers in the vaginal delivery group.
Results from the regression results are shown in table 2. In the univariate regression analysis, caesarean section was associated with overweight (OR 1.42, 95% CI 1.15 to 1.77) and obesity (OR 1.49, 95% CI 1.10 to 2.00). After the multiple regression base model (Model 1) was adjusted for maternal prepregnancy weight (Model 2) the OR for obesity dropped from 1.48 to 1.20 and was no longer statistically significant; the OR in the fully adjusted model was 1.19 (95% CI 0.85 to 1.67). The OR for overweight remained statistically significant after full adjustment (OR 1.29, 95% CI 1.03 to 1.64). Running the six regression models with BMI z-score as the outcome yielded similar results: Children delivered by caesarean section had 0.17 SD (95% CI 0.06 to 0.27) (Model 1) and 0.08 SD (95%CI −0.24 to 0.18) (Model 2) higher BMI before and after adjustment for maternal prepregnancy weight compared with infants delivered vaginally.
Results from the stratified analyses are summarised in table 3. Children born by caesarean section had elevated odds for obesity in both strata of maternal prepregnancy weight (<67 kg vs ≥67 kg) but the associations were not statistically significant. Among term infants, there was a strong positive association between caesarean delivery and obesity in the multiple regression base model (Model 1) while there was no statistically significant association among preterm infants. After adjusting for prepregnancy weight (Model 2), neither stratum showed a significant association between caesarean section and obesity. In the three weight-for-gestational age strata (AGA, SGA, LGA), the strongest association between caesarean section and obesity was observed among AGA infants, but the association did not remain significant after controlling for maternal prepregnancy weight. In term infants and in AGA infants, there was a statistically significant association with overweight which persisted even after adjustment for maternal prepregnancy weight.
We recalculated Models 1 and 2 using a 3-level exposure (caesarean section, no labour; caesarean section, first/second stage labour; vaginal delivery). The odds of obesity were similar for caesarean section with (OR 1.46, 95% CI 1.01 to 2.12) and without (OR 1.37, 95% CI 0.81 to 2.32) labour for Model 1. After controlling for maternal prepregnancy weight (Model 2), there were no statistically significant associations in the ‘labour’ group (OR 1.24, 95% CI 0.84 to 1.82) and ‘no labour’ group (OR 1.03, 95% CI 0.58 to 1.84).
In this retrospective cohort study, we assessed the association between mode of delivery and childhood obesity at age 10/11 years. Caesarean delivery was associated with 49% increased odds of obesity in the offspring before adjusting for confounders but after adjustment the association was attenuated and not statistically significant. Maternal prepregnancy weight was the strongest confounder of the main effect estimate.
To date, seven studies have examined the association between caesarean section and childhood obesity.4–10 The association was first described in the English literature in 2011 as an incidental finding by Rooney et al.6 Most other studies specifically examined the relationship motivated by the hypothesis that differences in gut microbial composition as a result of caesarean delivery may increase energy harvesting in the intestine, which may contribute to obesity later in life.27–29 Our lack of finding a strong relationship between caesarean section and childhood obesity is in keeping with two population-based studies,9 ,10 while the other five studies reported an association between caesarean delivery and later childhood obesity4–8 with OR estimates between 1.47 and 5.23. Factors that may have contributed to the heterogeneity of results between studies are the adjustment for maternal BMI, the inclusion of preterm infants in the sample, the timing of the children's BMI measurement ranging from 3 years8 to 23 years,10 the proportion of births by caesarean section in the sample, the study design and the country. Given the heterogeneity of the published studies on the topic and the high potential for publication bias, we feel that a critical review of the existing evidence rather than a meta-analysis30 will help to determine whether there is a true association between caesarean section and later childhood obesity. The characteristics of the seven studies are shown in table 4.
Maternal prepregnancy weight or BMI must be considered a strong confounder in the association between surgical delivery and obesity in the offspring. Obese mothers are more likely to require a caesarean section owing to increased maternal soft tissue, fetal macrosomia, dysfunctional labour, and intrapartum complications.31–33 Increased maternal prepregnancy weight is also associated with childhood obesity.3 ,13–15 ,34 Despite its potential confounding effect some studies did not adjust for maternal prepregnancy weight or BMI4–7 and their findings may, at least in part, be due to confounding. The only study that adjusted for maternal prepregnancy BMI and found an association between caesarean section and later obesity was the study by Huh et al.8 Notable differences between the latter study and our study are the use of a convenience sample of affluent families attending an urban/suburban group practice versus a population-based sample, and an earlier age BMI assessment (3 years vs 10/11 years).
The gut flora in obese individuals has been shown to be high in the Firmicutes species relative to Bacterioidetes.12 ,35 ,36 The different gut flora composition possibly influences energy metabolism and endocrinological activity of the intestine.37 The infant's gut is sterile until birth when it is colonised with bacteria from the birth canal. The gut flora changes considerably during the 1st months of life in response to environmental influences such as feeding practices, diet, hospitalisation and antibiotic use.11 ,38 Some studies have reported similar flora in the infant and the mother, indicating an ‘inheritance’ of the microbial composition.12 ,39 The postnatal variability of the gut flora in response to the environment and the similarity with the maternal flora would rather speak against a strong role of caesarean delivery in the gut flora development, which is supported by the results of Salminen et al40 who found no difference in the microbial composition in children aged 7 years delivered vaginally or by caesarean section. Collado et al41 in their study of the gut microbiota of 42 mother-infant pairs found that faecal concentrations of Bacteroides and Staphylococcus were significantly higher in infants of overweight mothers, indicating that the infant gut flora may be primarily determined by maternal BMI rather than the mode of delivery. However, the findings from this study were limited by the small sample size and an imbalance between the groups of normal weight and overweight mothers with regard to the proportion of infants born vaginally.
Huh et al8 have suggested that caesarean section may be a proxy for antibiotic use and may explain the observed association between mode of delivery and childhood obesity. The administration of antibiotics to the mother or the infant may alter the gut microbial structure,11 thereby decreasing numbers of Bacteriodes with antiobesogenic properties.37 Unfortunately, we were not able to investigate perinatal or postnatal antibiotics use in the present study. Preterm infants and SGA infants are more likely to receive antibiotics perinatally or in the first months of life than term infants and as a result their gut flora composition differs from that of healthy term infants.38 One may speculate that the absence of an association between mode of delivery and later obesity in the two latter groups (see table 3) may be due to administration of antibiotics post partum or in early infancy, which removed any effect of intrapartum transmission of microbes. Only one previous study was able to investigate planned and unplanned caesarean section separately.8 In accordance with their findings, there was no difference in offspring obesity between women undergoing caesarean section with and without labour in the present study. Based on the gut flora hypothesis we would have expected to see a protective effect owing to an exposure to maternal vaginal flora in infants delivered by caesarean section after the onset of labour compared to those delivered by caesarean section without labour, but this hypothesis was not supported by our results.
The strengths of our study include the use of measured childhood BMI, and the linkage of a province-wide perinatal database with an information-rich survey, which allowed adjustment for a broad range of potentially confounding factors. Due to the universal health care system in Canada, access to and use of perinatal care likely had little influence on the health of mothers and children in our study. Limitations include the absence of a recorded maternal height in the database at the time this cohort was initiated which prevented the adjustment of the sample for maternal prepregnancy BMI. We were not able to assess childhood obesity before age 10 years due to the lack of data on early child growth recorded in the CLASS study. An a priori sample size calculation showed that our study was sufficiently powered to detect a 40% higher prevalence of obesity in children born by caesarean section relative to the 10% prevalence of obesity in the vaginally delivered infants. We can therefore not exclude that our study was underpowered to detect an effect below this threshold.
Taken together, results from our primary analyses as well as the subgroup analyses do not support a causal association between caesarean section and childhood obesity. Maternal prepregnancy weight was an important confounder in the association between caesarean delivery and childhood obesity and needs to be considered in future studies.
Contributors KF wrote the manuscript. CGW contributed to the writing and analysis. ACA and PJV provided the data, and contributed to the writing of the manuscript. SK conceived the study, conducted the statistical analyses, and led the writing of the manuscript.
Funding KF was supported by a Beth Rafuse Studentship from the Dalhousie Medical Research Foundation. PJV was supported by a Canada Research Chair in Population Health and the Alberta Innovates Health Scholarship.
Competing interests None.
Ethics approval IWK Health Centre Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.