Cesarean Section and Risk of Severe Childhood Asthma: A Population-Based Cohort Study

Methods 

Established in 1967, the Medical Birth Registry of Norway (MBRN) is based on compulsory notification of all live births and stillbirths after 16 weeks of gestation.18 The MBRN contains data on mode of delivery (spontaneous vaginal, instrumental vaginal [forceps or vacuum extraction], or CS), year of birth, maternal age and marital status, birth order, maternal diagnoses, and infant gestational age, sex, and medical condition. The MBRN is routinely linked to the Central Population Registry, to obtain data on death dates.

From 1988 onward, data on type of CS (planned or emergency) were extracted from the birth notification form. All 4 of the following criteria had to be met for a CS to classify as planned: (1) an indication for CS identifiable before delivery (eg, narrow pelvis, previous CS), (2) onset of delivery marked as “provoked” opposed to “spontaneous,” (3) delivery recorded as CS, and (4) CS performed during regular working hours (ie, 7:00 AM to 5:00 PM). Consequently, the vast majority of attempted vaginal deliveries diverted to CS were classified as emergency CS.

Data on social security cash benefits received for asthma were obtained from the registry of the Norwegian National Insurance Scheme (NIS), which provides a basic benefit to any child in Norway with a chronic disease or disability involving significant long-term expenses, along with an attendance benefit if the child needs extra attendance or nursing. To qualify for the attendance benefit, the asthmatic condition must be moderate or severe, even after appropriate treatment has been initiated. The benefits are provided on the basis of a physician's diagnosis, coded in the NIS registry in accordance with International Classification of Diseases codes (ICD9 code 493, and for new cases from December 1998 onward, ICD10 code J45), irrespective of family income. Data on maternal educational level were obtained from Statistics Norway Register of Education. The data were last updated in December 2002.

We initially had data on 1,869,380 children born between 1967 and 1998. We excluded children who died before age 1 year (37,438), children registered with any birth defect except congenital dislocation of the hip (42,386), and children in a multiple pregnancy (43,241), leaving records of 1,756,700 children for analysis.

We assessed the association between mode of delivery and asthma using Cox proportional hazard models in SPSS version 14.0 (SPSS Inc, Chicago, IL). Children were censored when they reached age 18 years, or in the year 2002. We tested the proportional hazards assumption for each covariate using graphical methods and found no marked deviations. We checked for interactions between the exposure and each of the covariates and again found none. Two-sided P values <.05 were considered statistically significant.

We used models in which mode of delivery was the exposure and any benefit received for asthma (ie, basic benefit, attendance benefit, or both) was considered the outcome. The first model compared CS and instrumental vaginal delivery with spontaneous vaginal delivery for children born between 1967 and 1998; the second model compared emergency CS, planned CS, and instrumental vaginal delivery with spontaneous vaginal delivery for children born between 1988 and 1998. Using background information on possible confounders, we adjusted the analyses for the categorical variables maternal age (<25, 25 to 34, and 35+ years), birth order (1, 2+), record of maternal asthma (yes/no), maternal educational level (0 to 9 years, 10 to 12 years, and 13+ years), the child's sex, and, in the first model, year of birth (1967 to 1971, 1972 to 1976, 1977 to 1981, 1982 to 1986, 1987 to 1992, and 1993 to 1998).

In an attempt to further reduce confounding, we separately analyzed the subset of children delivered by presumably healthy mothers (ie, with no record of any maternal diagnoses before or during pregnancy), age 20 to 39 years and married or cohabiting, a total of 1,033,187 children. Also, analyses were stratified according to gestational age (<37 weeks, 37 to 41 weeks, and 42+ weeks). In addition, a subgroup of CS deliveries, classified as planned, with a record of cephalopelvic disproportion and no record of premature rupture of membranes, was examined separately.

Results 

The rate of CS gradually increased from 1.7% in 1967 to a plateau of 11% to 12% between 1985 and 1998 (Figure). The cumulative incidence of asthma by year of birth increased until 1992. The subsequent observed decrease likely was due to the fact that many children born thereafter had not yet developed asthma or applied for benefits by 2002. The overall cumulative incidence of asthma (by age 18 or year 2002) was 4.0/1000.

View Large Image Download to PowerPoint

Figure. Incidence rates of CS (per 100) from the MBRN (solid line) and cumulative incidence rates of benefits for asthma (per 1000) from the NIS (dashed line) at age 18 years or in the year 2002, by year of birth, 1967 to 1998.

The prevalence of asthma was 2.3% in the women who delivered by CS, 1.9% in those who had instrumental vaginal delivery, and 1.4% in those who had spontaneous vaginal delivery (Table I; available at www.jpeds.com).

Table I. Characteristics of births of singletons without birth defects by mode of delivery. Norway, 1967 to 1998

		Year of birth 1967 to 1998			Year of birth 1988 to 1998
		Spontaneous vaginal delivery	Instrumental vaginal delivery	CS	Spontaneous vaginal delivery	Instrumental vaginal delivery	Planned CS	Emergency CS
	Number of births	1,520,088 (86.5%)	99,877 (5.7%)	136,735 (7.8%)	505,150 (81.3%)	44,102 (7.1%)	26,016 (4.2%)	44,097 (7.1%)
	Maternal age, %
	<25 years	37.8	38.5	26.7	26.5	29.1	14.5	24.3
	25 to 34 years	54.3	54.4	58.3	63.6	61.8	63.9	61.8
	35+ years	7.9	7.0	15.0	10.0	9.1	21.6	13.9
	Birth order, %
	1	38.1	80.0	48.6	37.5	78.3	30.6	56.2
	2+	61.5	19.5	50.9	61.8	21.0	68.9	43.1
	Unknown	0.4	0.6	0.5	0.7	0.8	0.5	0.7
	Maternal education, %
	<10 years	17.1	11.1	12.5	8.2	6.4	8.6	8.8
	10 to 12 years	53.5	52.7	53.3	54.6	53.1	53.0	55.2
	13+ years	25.7	32.3	30.1	33.6	36.8	35.1	32.2
	Unknown	3.8	3.9	4.0	3.6	3.7	3.2	3.7
	Married/cohabiting, %	89.6	86.6	88.7	91.0	89.7	93.0	90.3
	Maternal asthma, %	1.4	1.9	2.3	2.3	2.6	3.0	3.0
	Gestational age, %
	<37 months	3.8	2.6	12.4	3.7	2.6	8.8	15.8
	37 to 41 months	76.5	71.3	66.4	75.0	70.6	76.6	58.4
	42+ months	12.7	18.7	11.9	11.8	17.6	4.8	14.6
	Unknown	7.0	7.4	9.3	9.5	9.2	9.8	11.2
	Male sex, %	50.5	57.9	53.4	50.3	57.6	51.5	54.7

Between 1967 and 1998, children born by CS had a 52% increased risk of asthma (hazard ratio [HR] = 1.52; 95% confidence interval [CI] = 1.42 to 1.62), and children born by instrumental vaginal delivery had a 19% increased risk (HR = 1.19; 95% CI = 1.08 to 1.31) compared with children born through spontaneous vaginal delivery (Table II). Year of birth was the only important confounder; adjusting also for maternal age, birth order, maternal education, maternal asthma, and the child's sex did not substantially alter the HR. The risks were similar in the subset of children delivered by presumably healthy mothers.

Table II. Cumulative incidence and risk of asthma by mode of delivery: Singletons without birth defects born in Norway between 1967 and 1998, with follow-up until age 18 or December 2002

	Mode of delivery	Cumulative incidence of asthma		HR for asthma (95% CI)
		n/N	Per 1000	HR crude	HR adjusted⁎	HR adjusted†
	Spontaneous vaginal	5537/1,520,088	3.6	1	1	1
	Instrumental vaginal	493/99,877	4.9	1.39(1.26to1.52)	1.46(1.33to1.61)	1.19(1.08to1.31)
	CS	1055/136,735	7.7	2.20(2.06to2.35)	2.17(2.03to2.32)	1.52(1.42to1.62)

⁎ Adjusted for maternal age, birth order, maternal education, maternal asthma, and sex. † Adjusted for maternal age, birth order, maternal education, maternal asthma, sex, and year of birth.

Between 1988 and 1998, delivery by emergency CS was associated with a 59% increased risk of asthma (HR = 1.59; 95% CI = 1.44 to 1.75); delivery by planned CS, with a 42% increased risk of asthma (HR = 1.42; 95% CI = 1.25 to 1.61); and instrumental vaginal delivery, with a 14% increased risk of asthma (HR = 1.14; 95% CI = 1.01 to 1.28) (Table III). Excluding from the analyses all children with a record of rupture of membranes 6 hours or more before delivery did not changes these results. Moreover, the risks were similar in the subset of children delivered by presumably healthy mothers (HR = 1.62; 95% CI = 1.37 to 1.92 for emergency CS and HR = 1.17; 95% CI = 0.89 to 1.54 for planned CS).

Table III. Cumulative incidence and risk of asthma by mode of delivery: Singletons without birth defects born in Norway between 1988 and 1998⁎ with follow-up through December 2002

	Mode of delivery	Cumulative incidence of asthma		HR of asthma (95% CI)
		n/N	Per 1000	HR crude	HR adjusted†
	Spontaneous vaginal	3696/505,150	7.3	1	1
	Instrumental vaginal	306/44,102	6.9	0.95(0.85to1.07)	1.14(1.01to1.28)
	Planned CS	264/26,016	10.1	1.39(1.23to1.58)	1.42(1.25to1.61)
	Emergency CS	475/44,097	10.8	1.48(1.34to1.63)	1.59(1.44to1.75)

⁎ Data on type of CS (planned vs emergency) were not available before 1988. † Adjusted for maternal age, birth order, maternal education, maternal asthma, and sex.

Children with a record of cephalopelvic disproportion delivered by planned CS had a 23% increased risk of asthma compared with children born by spontaneous vaginal delivery (HR = 1.23; 95% CI = 0.99 to 1.54).

Discussion 

Between 1967 and 1998, children born by CS had a moderately increased risk and children born by instrumental vaginal delivery had a slightly increased risk of developing asthma. Between 1988 and 1998, this association apparently was stronger for emergency CS than for planned CS. The association also was consistently stronger for preterm children than for term children.

Important strengths of our study include the large, population-based cohort of children and the long follow-up period. The ability to study planned and emergency CS separately was lacking in most earlier studies. Several MBRN variables have been validated and found to be satisfactory,19, 20 as have NIS variables,21 indicating reliable data on both exposure and outcome.

One weakness of our study is the lack of data on other possible risk factors for asthma. Less breast-feeding has been observed after CS,22 but a recent review found no convincing evidence that breast-feeding provides primary prevention of childhood asthma.23 Thus, breast-feeding is not likely to be an important confounder.

Parental smoking is an established risk factor for asthma, and a recent Norwegian study found a tendency toward an increased risk of emergency CS in mothers who smoke.24 Some studies in which smoking habits were adjusted for found no association between CS and asthma,15, 16, 17 whereas others in which smoking habits were not adjusted for did find an association.8, 11 This is not a consistent finding, however; some studies that adjusted for smoking habits still found an association.10 In any case, smoking habits may confound our results.

We cannot rule out underreporting of maternal asthma to the MBRN. The prevalence among mothers increased from 0.2% in 1967 to just under 3% in the 1990s. This figure is low, and underreporting could contribute to residual confounding, because women with asthma have a greater risk of CS delivery25 and their offspring have a higher risk of developing asthma.

The cumulative incidence of childhood asthma in the NIS registry for children born between 1967 and 1998, was 4.0/1000. By year of birth, it increased from 0.1/1000 for children born in 1967 to a peak of 10.9/1000 for those born in 1992. These figures are much lower than those reported in prevalence studies in Norway based on questionnaires and self-reporting (ranging from 25/1000 in the 1970s to 100/1000 in the 1990s).26 Because only a small percentage of children suffering from asthma received cash benefits, the question of how they were selected arises. We believe that our cases are true cases of asthma of a certain degree of severity, because (1) all subjects in the NIS registry are physician-diagnosed, and (2) according to the criteria of the NIS, the asthmatic condition must be of a certain degree of severity for a family to qualify for a cash benefit. Some studies have implied that along with degree of severity, both socioeconomic27 and geographic28 factors may influence the rate of application for cash benefits. Although the associations we found between CS and asthma were generally the same in all geographic regions and within all educational groups, we cannot exclude the possibility that these factors may have biased our results.

Several recent studies have attempted to evaluate the association between mode of delivery and asthma. Some, which did not adjust for smoking habits, found odds ratios (ORs) in the range of 1.2 to 1.3;8, 11 others that adjusted for maternal smoking habits found no significant associations.15, 16, 17 One study excluded any child whose mother had a record of previous or perinatal morbidity;9 their OR of 1.3 is similar to our finding in this group. We did not reproduce the finding of Renz-Polster et al10 of an association between CS and asthma only in girls; unlike Debly et al,12 neither did we find this association only in children born preterm, although the association was stronger at lower gestational age.

We found that CS was associated with a moderately increased risk of asthma, consistent with many previous studies.7, 8, 9, 10, 11 The association was apparently stronger for emergency CS than for planned CS, both overall in all strata of gestational age and particularly in the subgroup of “healthy mothers.” According to our definition of planned CS versus emergency CS, it is likely that some planned CSs were performed after rupture of membranes, when the child had been exposed to the mother's microbial flora; however, a larger proportion of children born by emergency CS would have experienced this. Thus, if CS caused asthma through the child's lack of exposure to maternal microbial flora (the hygiene hypothesis), then we would expect planned CS to have a stronger effect than emergency CS. Alternatively, if CS caused asthma through mechanisms associated with respiratory morbidity in newborns, then planned CS should still have a stronger effect than emergency CS, because the process of labor reduces the risk of neonatal respiratory problems.29 Either way, the effect of planned CS should be stronger than the effect of emergency CS—the opposite of what we observed.

In an attempt to examine children with a minimal risk of having been exposed to maternal microbial flora, we separately evaluated those children with a record of cephalopelvic disproportion and no record of premature rupture of membranes delivered by planned CS. Their point estimate of a 23% increased risk of asthma is lower than that of all children born by planned CS (42%) and those born by emergency CS (59%), further supporting the unexpected finding that emergency CS seemed to have a stronger effect than planned CS.

Instrumental vaginal delivery was associated with a slightly increased risk of asthma. Although these children were exposed to the maternal microbial flora, they could be at increased risk for respiratory distress due to asphyxia and its treatment. Thus, this finding supports the respiratory distress hypothesis, but does not exclude the hygiene hypothesis.

In conclusion, in our large population-based cohort study of 1.7 million children, we confirmed a moderate association between CS and severe asthma, consistent with many previous studies. Because asthma constitutes an important and increasing disease burden in children today, and the rate of CS continues to rise, further exploration of the reasons for this association, as well as the possible different effects of planned CS and emergency CS, is important.