Mode of delivery at birth and the metabolic syndrome in midlife: the role of the birth environment in a prospective birth cohort study.

OBJECTIVES: The aim of this study is to examine the hypothesis that mode of delivery at birth may be associated with metabolic disorders in adult midlife.
SETTING: Population cohort study PARTICIPANTS: The National Child Development Study consists of individuals born during 1 week in 1958 in Great Britain. Respondents with biomedical data on the metabolic syndrome at age 45 were included. OUTCOME MEASURE: The metabolic syndrome was defined based on the National Cholesterol Education Program Adult Treatment Panel (NCEP-ATP) III classification.
RESULTS: 7156 were born naturally; among the caesarean births, 106 were non-elective and 85 were elective caesareans. The metabolic syndrome is present in 37.7% of those born by non-elective caesareans, 25.9% of those born by elective caesarean and 27.5% of those born by vaginal delivery. In a multivariate logistic regression model adjusted for antenatal factors, birth history, mother's characteristics and the socioeconomic environment at birth, only birth by non-elective caesarean remained associated with the metabolic syndrome in adulthood compared with vaginal delivery (OR 1.51, 95% CI 1.00 to 2.30). Mother's obesity (OR 1.61, 95% CI 1.12 to 2.34) and low maternal education level (OR 1.47, 95% CI 1.30 to 1.67) were also independently associated with midlife metabolic syndrome.
CONCLUSIONS: Birth by non-elective caesarean in 1958 may be associated with metabolic syndrome in adulthood after adjusting for prior confounding factors. We suggest that the birth context of emergency caesareans in 1958 is suggestive of a 'fetal stress' mechanism affecting health across the lifecourse. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://group.bmj.com/group/rights-licensing/permissions.

Being born by emergency caesarean in 1958 may be associated with the metabolic syndrome in midlife.

Mode of delivery may be a proxy for the birth environment and contextually variable clinical practices.

Given the possible context of emergency caesareans in 1958, a ‘fetal stress’ hypothesis is suggested for the subsequent association with the metabolic syndrome.

It is possible that an unknown confounding factor during early life was omitted from the analyses, which might explain the differences observed in MS outcome between the mode of delivery groups.

Introduction

In recent years, a number of studies have suggested that the mode of delivery at birth may be associated with obesity and metabolic disruption across the lifecourse. This stems from epidemiological research showing associations between birth by caesarean section and obesity in childhood.1 2 Such associations deserve further investigation given, on the one hand, the dramatic increases in caesarean sections in recent decades, from 21% in 1996 to 32% in 2007,3 and on the other hand, the burden of morbidity due to metabolic diseases.4 The hypothesised mechanism for this association involves the colonisation of the gut microbiota.5

Animal models have shown that modifications to rat gut microbiota have led to metabolic disruptions and ultimately obesity in affected animals.6 The gut microbiota is a potential source of inflammatory molecules that may contribute to metabolic diseases.7 8 This possible link between gut microbiota and metabolic disruptions is relevant to mode of delivery at birth due to the colonisation of the gut flora that occurs when the baby ingests maternal vaginal flora as she/he passes along the birth canal. If a caesarean section is carried out to deliver the baby, this phase of birth is skipped, and the baby is not exposed to the vaginal flora. The colonisation of their digestive tract therefore occurs differently to a baby which was delivered naturally.9–11 Recent reports have linked differences in infant gut microbiota with subsequent obesity.12

To explore the hypothesis that mode of delivery may be associated with metabolic disruptions, it is important to consider the context surrounding the pregnancy, birth and, where possible, variations in the mode of delivery. Caesarean sections have become part of routine practice in maternity wards, often planned well in advance in the case of at-risk pregnancies.13 However, the practice of caesareans was not so commonplace up to three decades ago, and individuals born under rather different practices and clinical conditions are now in their 40s and 50s. In this paper, we use a prospective birth cohort study of individuals born in 1958 to explore the possible association between mode of delivery at birth and the occurrence of the metabolic syndrome (MS) in midlife (45 years) under different contextual circumstances surrounding birth. Study participants delivered naturally, born via planned caesarean and via unplanned caesarean will be compared in terms of their MS profile at the age of 45 using available biomedical data from a birth cohort study.

Methods

Sample and participants

This study used data from the 1958 National Child Development Study (NCDS) which included all births during 1 week in 1958 (n=18 558) in Great Britain. Subsequent data collections were carried out on cohort members aged 7, 11, 16, 23, 33, 42, 46 and 50. The NCDS has been described in detail elsewhere.14 A biomedical survey (9377 cohort members participating) was conducted when participants were aged 44–46 years (figure 1).

Figure 1

Flow chart showing the sample selection.

Ethics

Written informed consent was obtained from the cohort member's parents for childhood measurements.. NCDS data are open access datasets available to non-profit research organisations.

Outcome measure

MS was defined using NCEP-ATP III (National Cholesterol Education Program Adult Treatment Panel III) clinical criteria, except for plasma glucose, which was not recorded and replaced by glycated haemoglobin (HbA1c) ≥6.5%.15

Exposure variable

Mode of delivery was categorised into three groups: emergency caesarean, elective caesarean or vaginal delivery.

Covariates

The variables taken into account covered four areas:

Mother's socioeconomic and health characteristics before the current pregnancy: mother's educational level (left school before/after minimum leaving age); household overcrowding (people per room); mother's self-reported pre-pregnancy weight and her height measured after the birth were used to construct the mother's pre-pregnancy body mass index (BMI) (weight in kg/(height in m)2). Since some mothers were younger than 18 years of age, age-specific BMI cut-offs were used in order to categorise BMI into four groups: thinness, normal, overweight and obese (corresponding to the cut-offs of <18.5, 18.5–24.9, 25–29.9 and ≥30 kg/m2 for adults, respectively). Mother's parity in 1958, including miscarriages after 28 weeks, was also extracted.

Previous pregnancy complications: previous pregnancy problems (yes/no), constructed based on whether the mother had previously had: an abortion or ectopic pregnancy, previous stillbirths, a previous neonatal death or other previous pregnancy complications.

Information on the current pregnancy: maternal age at birth; whether the mother smoked during pregnancy beyond the fourth month (yes/ no); abnormality during pregnancy (none/at least one abnormality including: antepartum haemorrhage, placenta praevia, vaginal bleeding and other abnormalities); hypertensive pathology (none/hypertension/toxaemia/proteinuria/eclampsia) and total number of antenatal visits (<5, 5–9 and >9 visits).

Details of the labour and birth: time elapsed since rupture of membranes (≥12 h before delivery ie, premature rupture of the membranes (PROM)/<12 h before delivery): whether labour was induced (yes/no); birthweight for gestation (<10th centile, 10–90th centile, >90th centile); gestational age was calculated as the duration between the first day of the mother's last menstrual period and childbirth, and categorised into groups (<38, 38, 39–41 and >41 weeks).

Statistical analyses

We first determined the prevalence of MS, and used the χ2 test to assess whether this prevalence differed by mode of delivery. The covariates were summarised as frequencies and percentages for categorical variables, means and SDs for continuous variables. χ2 or Fisher's exact tests were performed in order to compare the sample characteristics according to the exposure or the outcome. Comparisons of means by mode of delivery category were computed using analysis of variance, whereas the comparisons of means by MS status were carried out using the Student t test, after validating assumptions of normality and homoscedasticity.

Unadjusted and adjusted logistic regression models were carried out to explore the relationship between MS and mode of delivery. Complete case and multiple imputation analyses were conducted.

To control for possible bias due to missing data, we imputed data for covariates with missing data using the multiple imputation program ICE in STATA V.11.16 For more details, see the online supplementary data.

Results

Among 7347 observations, the prevalence of MS was 27.6% (36.6% for men, 18.6% for women, p<0.001). In total, 191 cohort members (2.6% of the sample) were delivered by caesarean section (106 emergency, 85 elective caesarean sections). The prevalence of MS in the emergency and in the elective caesarean were 37.7% (95% CI 28.5% to 47.0%) and 25.9% (95% CI 16.5% to 35.3%), respectively. The estimated prevalence of MS was 27.5% (95% CI 27.2% to 27.8%) within the vaginal delivery group, 37.7% (95% CI 34.9% to 40.5%) within the emergency caesarean group and 25.9% (95%CI 23.1% to 28.7%) within the elective caesarean group (global χ2 test p=0.061).

Sample characteristics according to mode of delivery are reported in table 1. Several maternal characteristics, parity, problems during previous pregnancies, abnormalities during the current pregnancy, induced labour, PROM and gestational age were highly associated with the cohort member's mode of delivery at birth (p<0.001). Specifically, older maternal age at birth, nulliparous mothers, induced labour, PROM and overdue birth (>41 weeks) were more frequent in the emergency caesarean section group. On the other hand, problems during previous pregnancies (past stillbirth and neonatal deaths and past complications of pregnancy), abnormality during pregnancy and premature birth (<38 weeks) were more frequent in the elective caesarean delivery group. Previous caesarean was also a strong predictor of elective caesarean delivery (data not shown).

Table 1

Description of covariates in terms of mode of delivery

	Vaginal delivery	Emergency caesarean	Elective caesarean	p Value
	(N=7156)	(N=106)	(N=85)	χ2
Variable	n (%)	n (%)	n (%)
Cohort member's gender				0.156
Male	3580 (50.0)	61 (57.6)	48 (56.5)
Female	3576 (50.0)	45 (42.4)	37 (43.5)
Maternal age, years				<0.001 F
<18	69 (1.0)	0 (0.0)	0 (0.0)
18–35	6414 (89.7)	87 (82.1)	63 (74.1)
>35	668 (9.3)	19 (17.9)	22 (25.9)
Mother's educational level				0.045
Low	5224 (73.2)	69 (65.7)	54 (64.3)
High	1913 (26.8)	36 (34.2)	30 (35.7)
Overcrowding				0.049
1–1.5 people/room	6140 (88.1)	96 (96.0)	73 (89.0)
≥1.5 people/room	832 (11.9)	4 (4.0)	9 (11.0)
Smoking after the fourth month of pregnancy				0.589
No	4766 (67.4)	74 (71.8)	55 (65.5)
Yes	2303 (32.6)	29 (28.2)	29 (34.5)
Maternal BMI before pregnancy				0.075 F
Underweight	305 (4.5)	3 (3.1)	4 (5.3)
Normal weight	4915 (72.6)	69 (71.1)	48 (63.2)
Overweight	1293 (19.1)	18 (18.6)	16 (21.0)
Obesity	262 (3.8)	7 (7.2)	8 (10.5)
Parity				<0.001
Nulliparous	2636 (36.9)	67 (63.2)	20 (23.5)
1 previous pregnancy	2269 (31.7)	20 (18.9)	32 (37.7)
≥2 previous pregnancies	2250 (31.4)	19 (17.9)	33 (38.8)
Problems with previous pregnancies				<0.001
No	5638 (78.9)	73 (68.9)	36 (42.4)
Yes	1504 (21.1)	33 (31.1)	49 (57.6)
Hypertensive pathology during pregnancy				0.002
No	4654 (68.3)	59 (62.1)	41 (51.3)
Yes	2164 (31.7)	36 (37.9)	39 (48.7)
Abnormality during pregnancy				<0.001
No	5358 (74.9)	50 (47.2)	23 (27.1)
Yes	1794 (25.1)	56 (52.8)	62 (72.9)
Total number of antenatal visits				0.240 F
<5	295 (4.2)	2 (1.9)	4 (4.8)
5–9 visits	1912 (27.1)	25 (24.3)	30 (36.1)
>9	4859 (68.7)	76 (73.8)	49 (59.2)
Birthweight				0.392
<10th centile	525 (8.4)	4 (4.8)	6 (8.5)
10–90th centile	5094 (81.7)	67 (79.8)	59 (83.1)
>90th centile	619 (9.9)	13 (15.4)	6 (8.4)
Gestational age, weeks				<0.001
<38	770 (9.8)	14 (12.1)	25 (25.5)
38	610 (7.8)	15 (12.9)	23 (23.5)
39–41	5504 (70.4)	59 (50.9)	45 (45.9)
>41	935 (12.0)	28 (24.1)	5 (5.1)
Whether labour induced				<0.001
No	6242 (87.2)	68 (64.2)	81 (95.3)
Yes	914 (12.8)	38 (35.8)	4 (4.7)
PROM >12h				<0.001
No	5599 (85.0)	55 (55.6)	83 (100.0)
Yes	985 (15.0)	44 (44.4)	0 (0.0)

Per cent may not add up exactly to 100 due to rounding.

BMI, body mass index; F, Fisher's exact test; PROM, premature rupture of the membranes.

Table 2 shows the relationships between mode of delivery, the covariates and MS. A low maternal level of education, smoking after the fourth month of pregnancy and maternal obesity were associated with a higher prevalence of MS. We also found significant links between MS and the following: hypertensive pathology, induced labour and PROM.

Table 2

Characteristics of mothers and cohort members before pregnancy, during pregnancy and labour, and at birth in terms of the metabolic syndrome at age 44–46 years (n=7347)

	Metabolic syndrome
	No	Yes	p Value
	(N=5317)	(N=2030)	χ2
Variable	n (%)	n (%)
Mode of delivery			0.061
Vaginal	5188 (97.6)	1968 (97.0)
Emergency caesarean	66 (1.2)	40 (2.0)
Elective caesarean	63 (1.2)	22 (1.1)
Cohort member's gender			<0.001
Male	2339 (44.0)	1350 (66.5)
Female	2978 (56.0)	680 (33.5)
Maternal age, years			0.166
<18	44 (0.8)	25 (1.2)
18–35	4743 (89.3)	1821 (89.7)
>35	525 (9.9)	184 (9.1)
Mother's educational level			<0.001
Low	3751 (70.8)	1596 (78.8)
High	1549 (29.2)	430 (21.2)
Overcrowding			0.330
1–1.5 people/room	4573 (88.4)	1736 (87.6)
≥1.5 people/room	599 (11.6)	246 (12.4)
Smoking after the fourth month of pregnancy			<0.001
No	3606 (3606)	1289 (64.2)
Yes	1643 (31.3)	718 (35.8)
Maternal BMI before pregnancy			<0.001
Underweight	230 (4.6)	82 (4.3)
Normal weight	3738 (74.2)	1294 (67.9)
Overweight	902 (17.9)	425 (22.3)
Obesity	171 (3.4)	106 (5.6)
Parity			0.089
Nulliparous	1938 (36.5)	785 (38.7)
1 previous pregnancy	1716 (32.3)	605 (29.8)
≥2 previous pregnancies	1662 (31.3)	640 (31.5)
Problems with previous pregnancies			0.771
No	4163 (78.5)	1584 (78.2)
Yes	1143 (21.5)	443 (21.8)
Hypertensive pathology during pregnancy			<0.001
No	3498 (69.2)	1256 (64.7)
Yes	1555 (30.8)	684 (35.3)
Abnormality during pregnancy			0.554
No	3941 (74.2)	1490 (73.5)
Yes	1374 (25.9)	538 (26.5)
Total number of antenatal visits			0.163
<5	230 (4.4)	71 (3.5)
5–9 visits	1436 (27.4)	531 (26.5)
>9	3579 (68.2)	1405 (70.0)
Birthweight			0.252
<10th centile	374 (8.0)	161 (9.3)
10–90th centile	3807 (81.8)	1413 (81.2)
>90th centile	472 (10.4)	166 (9.5)
Gestational age, weeks			0.268
<38	463 (9.6)	196 (10.8)
38	383 (7.9)	157 (8.6)
39–41	3414 (70.7)	1246 (68.4)
>41	567 (11.8)	224 (12.3)
Whether labour induced			0.002
No	4666 (87.8)	1725 (85.0)
Yes	651 (12.2)	305 (15.0)
PROM> 12 h			0.003
No	4203 (85.6)	1534 (82.7)
Yes	708 (14.4)	321 (17.3)

Per cent may not add up exactly to 100 due to rounding.

BMI, body mass index; F, Fisher's exact test; PROM, premature rupture of the membranes.

In table 3, we report unadjusted (model 1) and adjusted (model 2) ORs resulting from the final logistic regression model. Model 2 is adjusted for the effects of the cohort member's gender, maternal age, mother's educational level, smoking habits during pregnancy, BMI, parity, problems with previous pregnancies, hypertensive pathology during pregnancy, birthweight, gestational age, induction of labour and PROM (table 3—model 2). The results show that emergency caesarean delivery was associated with an increased proportion of MS compared with vaginal delivery (OR=1.51, p=0.05 results from imputed data).

Table 3

Unadjusted and adjusted logistic regression models showing relationship between mode of delivery and the metabolic syndrome in midlife: complete case analyses and analyses using multiply imputed data

	Complete case	Multiple imputations
	OR (95% CI)	OR (95% CI)
Model 1: unadjusted
Mode of delivery
Vaginal (ref)
Emergency caesarean	1.60 (1.42 to 1.80)***	1.60 (1.08 to 2.37)*
Elective caesarean	0.92 (0.74 to 1.07)	0.92 (0.57 to 1.50)
Model 2: adjusted†
Mode of delivery
Vaginal (ref)
Emergency caesarean	2.18 (1.28 to 3.71)**	1.51 (1.00 to 2.30)*
Elective caesarean	1.00 (0.53 to 1.91)	0.93 (0.56 to 1.56)
Cohort member's gender
Female (ref)
Male	2.48 (2.18 to 2.82)***	2.58 (2.31 to 2.87)***
Maternal age
Years	0.98 (0.97 to 0.99)**	0.98 (0.97 to 0.99)***
Mother's educational level
High (ref)
Low	1.48 (1.27 to 1.73)***	1.47 (1.30 to 1.67)***
Smoking after the fourth month of pregnancy
No (ref)
Yes	1.28 (1.11 to 1.46)**	1.24 (1.10 to 1.39)***
Maternal BMI before pregnancy
Normal weight (ref)
Underweight	0.87 (0.64 to 1.20)	0.97 (0.74 to 1.27)
Overweight	1.19 (0.85 to 1.67)	1.33 (0.99 to 1.78)
Obesity	1.45 (0.94 to 2.24)	1.61 (1.12 to 2.34)**
Hypertensive pathology during pregnancy
No (ref)
Yes	1.24 (1.08 to 1.43)**	1.18 (1.05 to 1.33)**
Birthweight
10–90th centile (ref)
<10th centile	1.05 (0.83 to 1.34)	1.08 (0.88 to 1.32)
>90th centile	0.92 (0.73 to 1.13)	0.96 (0.78 to 1.12)

*p≤0.05, **p≤0.01, ***p≤0.001.

†Also adjusted for: overcrowding, parity, previous pregnancy problems, total number of antenatal visits, gestational age, induced labour, PROM.

BMI, body mass index; PROM, premature rupture of the membranes.

An increased probability of having MS in midlife was also associated with: being male, a lower maternal education level, maternal smoking in pregnancy, maternal pre-pregnancy obesity and maternal hypertensive pathology during pregnancy. Respondents whose mothers were younger at the time of their birth were less likely to have MS in midadulthood.

Discussion

These analyses show that differences in the prevalence of MS at 45 years may be associated with the mode of delivery at birth, after controlling for possible confounders. However, the association observed is not for respondents born by caesarean section overall versus those born vaginally. Rather, emergency caesarean delivery remained associated with an increased risk of having MS in midlife compared with individuals born vaginally. These findings differ from studies carried out on cohorts of individuals born recently, showing that caesarean section per se is a risk for metabolic disorders via the gut microbiota hypothesis.1 11 The results from our study using data from births in 1958 suggest that mode of delivery may be a proxy variable for qualities in the birth environment that may have had long-term implications for cohort members’ health. We suggest that in 1958, caesarean sections were a rare phenomenon (2.6% prevalence), carried out electively in the case of high-risk pregnancies. When caesareans were emergency, we hypothesise that the birth context was most likely stressful, resulting in an emergency caesarean section. The stressful nature of the birth may play an important part in the observed association between emergency caesarean births and MS in midlife.

Factors occurring prior to and at the time of birth may contribute to the association between emergency caesareans and the subsequent development of MS. Based on the information from the NCDS, the women who had non-elective caesareans in 1958 were more likely to be overdue, to have had an induced labour and to have experienced rupture of the membranes more than 12 h before the birth. Such births seem to have gradually become emergency situations presumably after a long labour, with 97% of babies born thus described as having experienced ‘fetal distress’ by the duty midwife (data not shown). Abnormalities occurring at birth as indicated by an induced labour or a late delivery were more frequent in the emergency caesarean section group. Historically, the main reasons given by clinicians for carrying out caesareans, other than having previously had a caesarean, are the relatively undefined concepts of ‘fetal distress’, ‘failure to progress’ during labour and breech presentations.17 Given the rare occurrence of caesareans in the late 1950s, we can only speculate that the conditions surrounding a labour ending in an emergency caesarean were likely to have been fraught and stressful for those involved, not least for the baby. We put forward a fetal stress hypothesis, whereby babies born by emergency caesarean were subject to physiological stress and possibly their mother's psychological stress and its consequences in the postnatal period. Such a context of stress may have affected the baby's physiological and psychological stress responses thereafter. Early life stress has been associated with physiological alterations leading individuals along negative health trajectories.18 19 Furthermore, in later life, a stressful environment, such as job stress, has been associated with a greater prevalence of MS.20

The colonisation of the gut microbiota during vaginal births, and the lack thereof during caesarean births, has been put forward as a hypothesis for links observed in previous studies on caesarean delivery and metabolic disorders in childhood or adulthood.1 Some authors have reported that mode of delivery is associated with a differential colonisation of the gut flora. Higher proportions of bacteria from the Firmicutes group, and a lower frequency of members of the Bacteroidetes group, have been observed in children delivered by caesarean section compared with those born vaginally.11 Moreover, infants born by caesarean delivery were significantly less often colonised with bacteria of the Bacteroides fragilis group than vaginally delivered infants, and these subgroups represent the majority of the microbiota found in the adult gut.5 21

To support this hypothesis, we would have observed differences in the prevalence of MS between caesareans per se and vaginal birth; however, this was not the case. No association was observed between overall caesarean section in 1958 and MS in midlife. Differences between the gut microbiota of individuals born by emergency caesarean section versus those delivered vaginally but not in those born by elective caesarean section have been reported. A lower frequency of Escherichia–Shigella has been observed in other studies for those born by emergency caesarean.22 We cannot exclude that babies born by emergency caesarean were more likely to experience a prolonged exposure to vaginal bacteria and possibly to infectious pathogens, due to PROM (44% of emergency caesareans exposed to PROM vs 15% vaginal delivery and 0% elective caesareans). Mode of delivery has previously been associated with childhood obesity at 3 years of age, and the authors postulated that a longer exposure to bacterial flora could be a mechanism involved in childhood obesity.1

We reported that maternal pre-pregnancy obesity was associated with an increased probability of having MS at the modal age of 45 years. Previous work has shown that children exposed to maternal obesity in early life had a twofold increased risk of developing MS, with a trend towards a higher incidence of insulin resistance.23 Offspring exposed to maternal hyperglycaemia during their intrauterine development were also more prone to metabolic disorders in young adulthood leading to insulin resistance.24 25 Different paths of childhood growth with smaller gains in BMI during infancy could precede the development of MS or hypertension.26 27

There are a number of limitations to our study. The definition of MS proposed by different organisations has varied over the past decade. The prevalence of MS is lower when using definitions other than ATPIII; however, the risks of cardiovascular events, diabetes mellitus and hypertension are similar for ATPIII and American Heart Association (AHA) or International Diabetes Federation (IDF) definitions.28 Glycaemia was not recorded in the cohort study biomedical survey; therefore, we used the HbA1c value with a cut-off above 6.5% to define hyperglycaemia. HbA1c has been defined as a marker to identify diabetes status.29 The reliability of glucose measurements varies widely across laboratories and may result in misclassification of >12% of patients.30 By contrast, HbA1c values are relatively stable after collection.31 The NCDS cohort provides a rare opportunity to study conditions and characteristics at birth and in early life collected prospectively, in relation to good quality biological data sampled in midlife. Unfortunately, no information was collected at the time about gestational diabetes; however, we include birthweight and variables on other pregnancy complications which may capture the effect of insulin resistance during pregnancy. It was possible to include a large number of potential confounding variables in the statistical models; however, it is possible that a key unknown confounding factor during early life was omitted, which might explain the differences observed in MS outcome between the mode of delivery groups.

Conclusion

These findings suggest that mode of delivery at birth may be an important variable to take into account to understand the aetiology of metabolic disorders. It is likely to represent factors occurring in the environment proximal to the birth which may have an impact on the baby's health across the lifecourse. Our findings show that in 1958, emergency caesarean sections may be associated with an increased prevalence of MS in midlife. We suggest that given the maternity practices of the time, physiological stress experienced by the baby during delivery may be an important mechanism in the subsequent development of metabolic disorders.