Study design and population

We conducted a retrospective cohort study, analyzing state-level maternal and infant hospital discharge data, linked to birth and death certificate data, for California, Missouri, and Pennsylvania from 2000-2009. The Institutional Review Board of the Departments of Health in California, Missouri, and Pennsylvania; the Children’s Hospital of Philadelphia approved this study. The data were input by the Department of Health for each respective state; then the data were cleaned and validated with the use of sources that included birth certificates and maternal and infant hospital data with strong concordance (eg, mode of delivery is >99.5% concordant). The records were created by linking birth certificate data with maternal hospital discharge records and newborn infant hospital discharge data records or death certificate data in the event of a fetal death. Records were linked with the use of previously described methods. With these techniques, >98% of birth and death certificates are matched to maternal and newborn infant hospital records. These data have been used extensively in our and others’ publications.

Live singleton births and in-hospital fetal deaths that occurred between 23 and 24 6/7 weeks of reported GA were included in the analysis. Because periviable births that are not resuscitated at the time of the delivery potentially may be classified as fetal deaths, it was important that fetal deaths not be excluded entirely from the analysis. We sought to distinguish these types of fetal deaths from fetal deaths that occurred out of the hospital or as intrauterine deaths. Such deaths were designated as “outpatient” or “intrauterine” fetal deaths with criteria described by Phibbs et al in previous work and excluded from the analysis ( Appendix ; Supplementary Material A ). Fetal anomalies were also excluded.

Variable selection and data analysis

The primary predictor of interest was cesarean delivery ( International Classification of Diseases , ninth revision, Clinical Modification [ICD-9-CM] code 669.7x and 74.x), which had to be documented in the maternal or the infant record. Ultimately, documentation from the maternal record is reflected in all but 28 of the 8157 cases (99.0%). Breech neonates were identified by the following ICD-9-CM codes: 652.2, 652.20, 652.21, 652.23, and 763.0. Death-related outcomes of interest included overall survival (defined as 6-month corrected age among intubated and nonintubated neonates) and survival to >24-hours, >1-week, and >6-month corrected age among neonates for whom intubation was performed or attempted (ICD-9-CM codes 96.01, 96.02, 96.03, 96.04, 96.05 and current procedural terminology code 31500). Morbidity outcomes included respiratory distress syndrome (RDS), bronchopulmonary dysplasia (BPD), grade III/IV intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), retinopathy of prematurity, and asphyxia. Also included were composite outcome measures of “major morbidity” designated as BPD, grade III/IV IVH, NEC, or asphyxia and a “composite” for death or asphyxia. ICD-9-CM codes used for specific diagnoses are listed in Appendix ( Supplementary Material B ). Maternal sociodemographic characteristics were also considered: age in 3 categories (<18, 18-35, >35 years); race or ethnicity designated in 4 categories (white, black, Hispanic, and other); parity in 4 categories (0, 1, 2, ≥3); education in 2 categories (<high school education or ≥high school education); median income by ZIP code (<$20,000, $20,000-40,000, $40,000-60,000, >$60,000) to approximate household income; and insurance payer (fee for service, Health Maintenance Organization, federally insured, uninsured, and other). In an effort to control for potential confounding factors, sociodemographic characteristics (insurance, race, and age) that were associated with mode of delivery were included as covariates in the final models. Likewise, maternal comorbidities, pregnancy complications, and delivery indications were also included in the full model, specifically: preexisting diabetes mellitus, gestational diabetes mellitus, chronic hypertension, pregnancy-induced hypertension (PIH), preterm labor, preterm premature rupture of membranes, placental abruption, repeat cesarean delivery, placenta previa, and chorioamnionitis. ICD-9-CM codes that were used for specific diagnoses are listed in Appendix ( Supplementary Material B ). Finally, year of delivery was included because the incidence of cesarean delivery increased over time in our cohort.

We conducted all analyses using SAS statistical software (version 9.2; SAS Institute Inc, Cary, NC). Descriptive statistics were calculated with χ ² tests and Fisher exact test, as appropriate. Logistic regression was performed for multivariable analyses, which included potential modifying factors such as sociodemographic factors, maternal comorbidities, pregnancy complications, and delivery indications in the model. Delivery hospital was also included as a fixed effect to account for potential clustering of outcomes at the level of the delivery hospital. We initially examined the relationship between mode of delivery and death and morbidity in the overall cohort. Separate analyses were conducted that excluded “emergent indications,” which were designated as fetal distress, PIH, previa, and abruption. We reasoned that these indications typically require immediate delivery and may also be associated with poorer outcomes, regardless of mode of delivery. We also examined survival over 3 time periods among the subset of neonates who were intubated. Finally, we constructed separate models to evaluate the potential interaction between cesarean delivery and birthweight in relationship to morbidity and death. Statistical tests were considered significant at α = .05, adjusted for multiple comparisons with the Bonferroni correction.

Study population

Our study population was comprised of 8157 maternal/infant observations. Among these, we identified 1854 breech deliveries (22.9%) that comprised our final study cohort. Table 1 provides overall baseline sociodemographic and clinical characteristics for the mother and infant pairs, stratified by GA and mode of delivery: 77.9% of the mothers were 18-35 years old; 21.6% of the women were black; 32.8% of the women were Hispanic, and 34.4% of the women were white. Approximately 27.6% of the women had less than a high school education, and 50.0% were federally insured. Overall, 45.5% and 76.6% of women who delivered 23- and 24-week breech neonates were delivered by cesarean delivery.

Women who were delivered by cesarean delivery differed significantly from women who were not delivered by cesarean across parity, insurer, and educational categories ( P < .01 for each). Mode of delivery also differed based on maternal comorbidities. Compared with women who delivered vaginally, women who delivered by cesarean were more likely to have PIH (7.89% vs 1.66%; P < .01) and a previous cesarean delivery (16.04% vs 8.90%; P < .01) and less likely to have a birthweight of ≤500 g (10.41% vs 22.51%; P < .01).

Absolute incidence and adjusted odds of overall survival by mode of delivery

Table 2 presents the absolute incidence and adjusted odds of overall survival (includes all intubated and nonintubated neonates) for 23- and 24-week breech neonates in our study population, stratified by mode of delivery. Among the entire cohort of neonates who were delivered by cesarean at 23 weeks, the incidence of overall survival for cesarean-born neonates was 52.2% (175/335) compared with 22.9% (94/402) for vaginally born neonates. Survival among this group was statistically significantly higher (adjusted odds ratio [AOR], 3.98; 95% confidence interval [CI], 2.24–7.06) for those delivered by cesarean as compared with those who had a vaginal delivery in multivariable analyses. At 24 weeks, the overall survival among neonates who were delivered by cesarean was 65.0% (556/856), compared with a survival of 41.0% (107/261) for vaginally delivered neonates. In multivariable analysis, at 24 weeks, breech neonates who were delivered by cesarean, when compared with vaginal delivery, experienced greater survival in the overall cohort (AOR, 2.91; 95% CI, 1.76–4.81). These associations remained statistically significant, even after excluding emergent indications.

Morbidity among survivors by mode of delivery and GA

Table 3 presents neonatal morbidity among intubated and nonintubated survivors, comparing cesarean and vaginal delivery by delivery indication and GA. Among intubated and nonintubated 23-week cesarean-born neonates who survived beyond 6-month corrected age, no differences were found in morbidity outcomes by mode of delivery. However, among those delivered for nonemergent indications, cesarean-born neonates were significantly more likely to experience each of the following events: sepsis (AOR, 3.62; 95% CI, 1.78–7.40), NEC (AOR, 7.90; 95% CI, 1.03–60.57), and RDS (AOR, 6.84; 95% CI, 3.20–14.63). Moreover, major morbidity , which was defined as BPD, IVH, NEC, or asphyxia, was nearly 3 times as high for neonates who were delivered by cesarean for nonemergent indications compared with those delivered vaginally (AOR, 2.83; 95% CI, 1.37–5.84).

Stay updated, free articles. Join our Telegram channel

Join