Methods
Study population
We used data from the French National Perinatal Survey, which was conducted in 2010. The National Perinatal Surveys are routine surveys designed to monitor perinatal health, medical care, and health behavior. They include all births (live and stillbirths) during a 1-week period in all maternity units in France (n = 535 maternity units in 2010). Women are interviewed in the maternity unit 2 or 3 days after delivery about their sociodemographic characteristics, health behaviors, birth planning, and fertility treatments. Data on maternal health are collected from medical records. Comparisons for several perinatal indicators (eg, maternal age, gestational age) that are available in birth certificate statistics and hospital discharge statistics show that the sample was representative of all births in 2010. The National Council on Statistical Information (Comité du Label) and the French Commission on Information Technology and Liberties approved this survey (registration number 909003).
The survey sample included 14,681 women. For this analysis, we excluded 221 multiple births and 129 stillbirths because their obstetric management was different. We also excluded women with missing prepregnancy BMI (6.4%; n = 920). These women were significantly older, more often born outside of France, multiparous, and had lower levels of education. Finally, we excluded underweight women (BMI, <18.5 kg/m 2 ; n = 1114 women); this group faces increased risks of adverse perinatal outcomes that have been well-documented elsewhere. Our final analysis included 12,297 women with live born singleton pregnancies.
Our primary outcome measure was cesarean delivery that was differentiated by timing (prelabor or intrapartum). The perinatal survey recorded information about mode of delivery (spontaneous vaginal, instrumental, cesarean delivery) and mode of onset of labor (spontaneous, induced, prelabor cesarean delivery). Prelabor cesarean delivery included planned and emergency cesarean delivery that was performed before the onset of labor.
Our principal independent variable was maternal prepregnancy BMI (weight [kg]/height [m 2 ]) calculated with prepregnancy weight and height that had been reported by each woman and grouped according to World Health Organization recommendations : normal weight (18.5–24.9 kg/m 2 ), overweight (25–29.9 kg/m 2 ), and obese (≥30 kg/m 2 ).
Covariables
We selected variables that were shown in previous studies to be related to maternal BMI and mode of delivery and that included maternal social and demographic characteristics (age, parity, educational level, country of birth), preexisting medical conditions (chronic hypertension, preexisting diabetes mellitus), previous cesarean delivery, complications of the current pregnancy (preeclampsia, gestational diabetes mellitus that required insulin therapy, induction of labor, and LGA and SGA infants) and gestational age. LGA was defined as neonatal weight >90th percentile; SGA was defined as <10th percentile, based on French norms. We used actual weight at birth and not prenatal diagnosis of SGA or LGA because only approximately 20% of SGA and LGA babies are detected before birth. However, we tested our models using both definitions (SGA and LGA at birth and prenatally suspected SGA or LGA) because the latter may be more important in determining obstetricians’ decisions. We also reran our models using an absolute birthweight threshold (≥4000 g). Because gestational age is not linear, we modeled its effect using a fractional polynomial of degree 2 with powers (3,3).
To understand the results of our multivariable analyses better, we also investigated the indications for labor induction. Investigators could select up to 2 indications for labor inductions using preestablished categories: postterm/premature rupture of membranes (PROM)/fetal distress/suspected macrosomia/placenta previa/maternal condition/without medical indication/other reason (to be specified). Based on these responses, we created a mutually exclusive hierarchic classification: fetal indications/PROM at ≥37 weeks of gestation/diabetes mellitus/preeclampsia or gestational hypertension/prolonged pregnancy/suspected LGA/other indications/no medical indication. Fetal indications included situations with nonreassuring fetal status: intrauterine growth restriction with abnormal umbilical Doppler scan results, abnormal fetal cardiac rhythm or fetal malformation, and PROM at <37 weeks of gestation. Bleeding, previous intrauterine fetal death, and other maternal medical conditions were assigned to the other indications group. The no medical indication group included elective induction on patient’s demand. Because obese women had a higher frequency of labor induction, we also compared gestational age at delivery for all women and after induction for prolonged pregnancy.
Analysis strategy
We first compared maternal characteristics by prepregnancy BMI and parity and then described overall, prelabor, and intrapartum cesarean delivery rates in each of the BMI groups for primiparous and multiparous women and by previous cesarean delivery and induction of labor.
We then modeled risk ratios (RRs) and risk differences (RDs) associated with prepregnancy BMI for prelabor and intrapartum cesarean deliveries. RRs express the strength of the association and, along with odds ratios (ORs), are used most often in the literature to measure the risk of cesarean delivery by maternal weight. However, RDs provide an indication of the number of potentially avoidable cesarean deliveries and the impact of preventive action. For example, an RR of 2 denotes a doubling of risk in women with the characteristic under study but does not provide information about the absolute increase. An RR of 2 when the absolute increase is 1% (RD, 0.01) is not likely to be as clinically important as an RR of 2 when the absolute increase is 10% (RD, 0.10). We sought to assess both these dimensions.
We tested for interactions between BMI and the covariates in our models. These tests confirmed the existence of significant interactions with parity and previous cesarean delivery, but there were no significant interactions with other covariates. All multivariable models were stratified by parity and previous cesarean delivery.
We built our multivariable models progressively adjusting first for preexisting maternal characteristics, then for the obstetric complications of the current pregnancy, and finally for gestational age at delivery. We used modified Poisson regression with robust variance to estimate adjusted RRs. RD were generated from these models with the use of predictive margins (with the postestimation command lncom in Stata software; version 12.1; StataCorp, College Station, TX). To check the robustness of our results, we also ran our models in a lower risk population, which we defined as women without any of the risk factors for cesarean delivery who were included in our analyses, except BMI. The aim was to see whether RRs and RDs for obese women were similar to those in our fully adjusted models. A probability value of .05 was considered significant. These analyses were carried out with Stata software.
Results
Table 1 shows the mothers’ social, demographic, medical, and obstetric characteristics by prepregnancy BMI and parity. Both primiparous and multiparous obese women had less education and more preexisting medical conditions (hypertension, diabetes mellitus) and pregnancy complications (gestational diabetes mellitus, preeclampsia) than normal-weight women ( P < .05 for all these variables). Multiparous obese women were more likely to have had a previous cesarean delivery (26.4% compared with 17.9% for normal-weight women; P < .001). Obese women also had labor induced more often, especially among primiparous women (42.6% vs 23.8% for normal-weight women; P < .001). Trends for overweight women were similar to those of obese women.
Variable | Normal weight | Overweight | Obese | P value |
---|---|---|---|---|
Primiparous women, n | 3997 | 868 | 444 | — |
Social and demographic characteristics, % | ||||
Age ≥35 y | 10.2 | 8.8 | 9.9 | .44 |
Born outside of France | 10.8 | 11.1 | 7.4 | .08 |
High school education or less | 38.2 | 46.9 | 55.9 | < .001 |
Preexisting medical/obstetric factors, % | ||||
Hypertension | 0.5 | 0.9 | 5.2 | < .001 |
Type 1 or 2 diabetes mellitus | 0.2 | 0.6 | 1.1 | .004 |
Obstetric factors in current pregnancy, % | ||||
Preeclampsia | 2.2 | 3.6 | 7.0 | < .001 |
Gestational diabetes mellitus treated with insulin a | 1.2 | 2.1 | 7.2 | .005 |
Large-for-gestational-age b | 5.2 | 9.0 | 9.3 | < .001 |
Small-for-gestational-age c | 11.2 | 8.6 | 8.2 | .02 |
Breech and abnormal lies | 5.2 | 3.4 | 2.9 | .02 |
Delivery at >41 +0 wk of gestation | 20.7 | 22.0 | 26.1 | .02 d |
Induction of labor | 23.8 | 31.0 | 42.6 | < .001 |
Multiparous women, n | 4624 | 1435 | 872 | |
Social and demographic characteristics, % | ||||
Age ≥35 y | 26.2 | 28.4 | 26.3 | .24 |
Born outside of France | 12.3 | 18.3 | 14.5 | < .001 |
High school education or less | 45.8 | 58.2 | 69.5 | < .001 |
Preexisting medical/obstetric factors, % | ||||
Hypertension | 1.8 | 3.6 | 8.8 | < .001 |
Type 1 or 2 diabetes mellitus | 0.3 | 0.9 | 1.7 | < .001 |
Previous cesarean delivery | ||||
1 | 15.1 | 17.9 | 19.6 | < .001 |
>1 | 2.8 | 4.6 | 6.8 | |
Obstetric factors in current pregnancy, % | ||||
Preeclampsia | 0.9 | 2.2 | 4.4 | < .001 |
Gestational diabetes mellitus treated with insulin a | 1.2 | 3.0 | 7.0 | < .001 |
Large-for-gestational-age b | 9.8 | 13.8 | 18.8 | < .001 |
Small-for-gestational-age c | 6.6 | 5.5 | 4.1 | .01 |
Breech and abnormal lies | 3.5 | 3.5 | 3.8 | .87 |
Delivery at >41 +0 wk of gestation | 16.4 | 19.8 | 19.2 | .01 d |
Induction of labor | 18.6 | 22.7 | 23.6 | < .001 |
a Preexisting or gestational diabetes mellitus that requires insulin during pregnancy
b Birthweight >90th percentile, French postnatal curves Audipog
c Birthweight <10th percentile, French postnatal curves Audipog
The overall cesarean delivery rate was higher for obese than for normal-weight women: 34.6% vs 20.9% for primiparous women ( P < .001) and 29.5% vs 15.7% for multiparous women ( P < .001; Table 2 ). The rate of prelabor cesarean delivery was not associated with maternal BMI in primiparous women ( P = .68) but was in multiparous women ( P < .001). Differences by BMI group especially were marked for multiparous women with previous cesarean deliveries; 65.8% of obese women with previous cesarean deliveries had repeat prelabor cesarean deliveries vs 47.6% for normal-weight women with previous cesarean deliveries ( P < .001). Prelabor cesarean deliveries were also more frequent for multiparous obese women without a previous cesarean delivery: 7.2% vs 3.2%, respectively ( P < .001). Intrapartum cesarean delivery rates were higher in both obese primiparous and multiparous women compared with their normal-weight counterparts, with larger absolute differences for primiparous women, especially after induction of labor: 38.8% vs 26.1% ( P < .001).
Variable | Normal weight, n a (%) | Overweight, n a (%) | Obese, n a (%) | P value |
---|---|---|---|---|
Primiparous women | ||||
Overall cesarean delivery | 3965 (20.9) | 862 (25.9) | 442 (34.6) | < .001 |
Prelabor cesarean delivery | 3965 (7.5) | 962 (7.3) | 442 (8.6) | .68 |
Intrapartum cesarean delivery b | 3661 (14.4) | 796 (19.9) | 404 (28.5) | < .001 |
After spontaneous labor b | 2723(10.4) | 530 (15.3) | 216 (19.4) | < .001 |
After induction of labor b | 938 (26.1) | 266 (29.0) | 188 (38.8) | .002 |
Multiparous women | ||||
Overall cesarean delivery | 4575 (15.7) | 1422 (20.7) | 864 (29.5) | < .001 |
Prelabor cesarean delivery | 4575 (11.0) | 1422 (14.1) | 864 (22.6) | < .001 |
Intrapartum cesarean delivery b | 4063 (5.2) | 1219 (7.6) | 668 (9.0) | < .001 |
No previous cesarean delivery | 3664 (6.3) | 1078 (6.9) | 619 (13.6) | < .001 |
Prelabor cesarean delivery | 3770 (3.2) | 1108 (3.0) | 639 (7.2) | < .001 |
Intrapartum cesarean delivery b | 3541 (3.1) | 1045 (4.0) | 573 (6.8) | < .001 |
After spontaneous labor b | 2771 (2.2) | 758 (2.8) | 391 (4.1) | .06 |
After induction of labor b | 770 (6.6) | 287 (7.3) | 182 (12.6) | .02 |
Previous cesarean delivery | 805 (59.9) | 314 (69.1) | 225 (73.8) | < .001 |
Prelabor cesarean delivery | 805 (47.6) | 314 (52.9) | 225 (65.8) | < .001 |
Intrapartum cesarean delivery b | 418 (23.2) | 147 (34.0) | 77 (23.4) | .03 |
After spontaneous labor b | 355 (22.3) | 122 (34.4) | 63 (19.1) | .02 |
After induction of labor b | 63 (28.6) | 25 (32.0) | 14 (42.9) | .06 |
a Indicates the total number of women at risk for cesarean delivery in each body mass index group
Table 3 presents the RRs and RDs for prelabor cesarean delivery for primiparous women and multiparous women with and without previous cesarean deliveries. These risks did not significantly differ between normal-weight and obese primiparous women. For multiparous women without previous cesarean deliveries, prepregnancy obesity was associated significantly with a risk of prelabor cesarean delivery after adjustment for social, demographic, medical, and obstetric characteristics: RR, 1.82 (95% confidence interval [CI], 1.25–2.64) with a corresponding adjusted RD of 0.03 (95% CI, 0.01–0.05). For obese multiparous women with a previous cesarean delivery, the risk of a prelabor cesarean delivery was also significantly higher than for similar normal-weight women; the corresponding adjusted RD was higher than for women without a previous cesarean delivery: 0.14 (95% CI, 0.05–0.23).
Variable | n | Crude risk ratio | 95% confidence interval | Adjusted risk ratio a | 95% confidence interval | Adjusted risk difference a | 95% confidence interval |
---|---|---|---|---|---|---|---|
Primiparous women | |||||||
Normal weight | 3965 | Reference | Reference | Reference | |||
Overweight | 862 | 0.97 | 0.75–1.27 | 1.01 | 0.77–1.31 | 0.00 | −0.02 to 0.02 |
Obese | 442 | 1.15 | 0.83–1.58 | 1.00 | 0.70–1.42 | 0.00 | −0.03 to 0.03 |
Constant | 0.05 | ||||||
Multiparous women without previous cesarean delivery | |||||||
Normal weight | 3704 | Reference | Reference | Reference | |||
Overweight | 1085 | 0.91 | 0.62–1.34 | 0.70 | 0.45–1.10 | 0.00 | −0.02 to 0.00 |
Obese | 626 | 2.22 | 1.59–3.09 | 1.82 | 1.25–2.64 | 0.03 | 0.01–0.05 |
Constant | 0.03 | ||||||
Multiparous women with previous cesarean delivery | |||||||
Normal weight | 805 | Reference | Reference | Reference | |||
Overweight | 314 | 1.1 | 0.93–1.31 | 1.11 | 0.93–1.33 | 0.04 | −0.03 to 0.12 |
Obese | 225 | 1.41 | 1.19–1.66 | 1.36 | 1.13–1.63 | 0.14 | 0.05–0.23 |
Constant | 0.33 |
a Adjusted for maternal age (<30/30-35/>35 years), country of birth (France/other European country/Northern Africa/Africa/other), educational level (≤high school education), preexisting hypertension, preexisting diabetes mellitus, preeclampsia, gestational diabetes mellitus treated with insulin, large for gestational age (birthweight >90th percentile), small for gestational age (birthweight <10th percentile), and gestational age at delivery.
Table 4 presents estimates of crude and adjusted RRs and RDs of intrapartum cesarean delivery and includes an additional model to capture the effect of adjustment for induction of labor. Obese primiparous women had a significantly higher risk of intrapartum cesarean delivery than primiparous women with normal weight; after adjustment for all factors that included induction of labor, the adjusted RR was 1.64 (95% CI, 1.36–1.98), and the RD was 0.10 (95% CI, 0.05–0.14).
Variable | n | Crude risk ratio | 95% confidence interval | Adjusted risk ratio a | 95% confidence interval | Adjusted risk ratio b | 95% confidence interval | Adjusted risk difference b | 95% confidence interval |
---|---|---|---|---|---|---|---|---|---|
Primiparous women | |||||||||
Normal weight | 3661 | Reference | Reference | Reference | Reference | ||||
Overweight | 796 | 1.38 | 1.17–1.62 | 1.27 | 1.08–1.51 | 1.22 | 1.04–1.44 | 0.03 | 0.00–0.06 |
Obese | 404 | 1.98 | 1.66–2.35 | 1.83 | 1.52–2.24 | 1.64 | 1.36–1.98 | 0.10 | 0.05–0.14 |
Constant | 0.08 | ||||||||
Multiparous women without previous cesarean delivery | |||||||||
Normal weight | 3541 | Reference | Reference | Reference | Reference | ||||
Overweight | 1045 | 1.28 | 0.91–1.82 | 1.04 | 0.72–1.50 | 0.97 | 0.67–1.40 | 0.00 | 0.00–0.01 |
Obese | 573 | 2.17 | 1.52–3.09 | 1.78 | 1.22–2.60 | 1.66 | 1.15–2.39 | 0.02 | 0.00–0.04 |
Constant | 0.02 | ||||||||
Multiparous women with previous cesarean delivery | |||||||||
Normal weight | 418 | Reference | Reference | Reference | Reference | ||||
Overweight | 147 | 1.48 | 1.10–2.00 | 1.41 | 1.03–1.94 | 1.41 | 1.02–1.94 | 0.10 | 0.00–018 |
Obese | 77 | 1.03 | 0.65–1.62 | 1.07 | 0.64–1.70 | 1.08 | 0.68–1.71 | 0.02 | −0.09 to 0.13 |
Constant | 0.18 |
a Adjusted for maternal age (<30/30-35/>35 years), country of birth (France/other European country/Northern Africa/Africa/other), educational level (≥high school education), preexisting hypertension, preexisting diabetes mellitus, preeclampsia, gestational diabetes mellitus treated with insulin, large for gestational age (birthweight >90th percentile), small for gestational age (birthweight <10th percentile), and gestational age at delivery
The RR for intrapartum cesarean delivery for multiparous obese women without a previous cesarean delivery was also significantly higher than for women of normal weight after adjustment: 1.66 (95% CI, 1.15–2.39), although the corresponding RD was lower: 0.02 (95% CI, 0.00–0.04). For obese multiparous women with a previous cesarean delivery, the risk of intrapartum cesarean delivery did not significantly differ from that of normal-weight women. However, because trial of labor after a previous cesarean delivery was rare among obese women, our analysis included only a small group of these women (n = 77).
Results of the adjusted models in the low-risk population were very similar to those in the overall study sample (data not shown). Results were also similar when prenatally suspected SGA and LGA were used instead of SGA and LGA at birth and when an absolute birthweight threshold (>4000 g) was used (data not shown).
In view of the large differences in rates of labor induction, especially among primiparous women, and its association with risks of intrapartum cesarean delivery, we report the indications for induction by prepregnancy BMI group. Table 5 shows that labor was induced more frequently in obese, than in normal-weight, women for prolonged pregnancy, diabetes mellitus, preeclampsia, and suspected LGA, but these rates did not differ for inductions for fetal indications, PROM, other indications, or no medical indication. Despite the higher frequency of induction in obese women, especially for prolonged pregnancy, gestational age at birth did not differ between obese and normal-weight women.