At a time when reducing cesarean delivery is a national health objective, cesarean delivery rates, on the other hand, have increased in preterm neonates. However, the most optimal route of delivery for preterm neonates continues to be hotly debated. Previous analyses have shown a lower mortality in cesarean births as compared with vaginal births in breech neonates with birth weights greater than 500 g.1–7 Other outcomes, such as neurologic sequelae and intraventricular hemorrhage, have also been found to benefit from cesarean delivery in breech preterm neonates.3–5,8 For neonates in vertex presentation, some analyses have also demonstrated a benefit from cesarean delivery, with lower mortality, higher 1-minute Apgar scores, and less intraventricular hemorrhage.9–11 In contrast, other studies have shown no improvement in mortality and other outcomes in vertex neonates born by cesarean delivery.12–17 Despite the uncertainty in benefit for preterm vertex neonates, cesarean delivery has increased in this group.18
Previous studies evaluating mode of delivery and outcomes in very low birth weight (VLBW) vertex neonates have had several limitations. They have been single-center or multicenter trials with relatively small sample size.9,10,13,15,16 Some have accumulated larger numbers by evaluating outcomes over a longer period, 10–15 years, in which practice patterns may have been evolving.10,17 Others were performed before the widespread use of antenatal steroids and surfactant therapy, which may influence the effect of delivery mode on outcome.11,13,14,16 Furthermore, little is known about the incidence, specific indications, or sociodemographic factors that might prompt cesarean delivery in VLBW vertex neonates.
We examined the births recorded in the 1999 to 2000 National Center for Health Statistics (NCHS) linked birth/death data set to examine the factors that were associated with cesarean delivery during this period. We also further investigated whether cesarean delivery was associated with a survival advantage.
MATERIALS AND METHODS
We used the U.S. birth cohort linked birth and infant death certificate data sets of 1999 to 2000 published by the NCHS.19 In this 2-year cohort population, 8,023,378 live births were recorded. Our analysis included neonates born only to U.S. residents (8,018,320) and who were singletons (7,770,437), born in the vertex position (7,426,979) and weighed between 500 and 4,000 g (6,659,088). We excluded those neonates born to mothers with a previous cesarean delivery or unknown method (704,538) and neonates whose deaths were associated with congenital anomalies (3,636). This resulted in an analytic cohort of 5,950,914 infants, of which 54,695 (0.92%) were VLBW (< 1,500 g). Sociodemographic characteristics of this population including maternal age, education, and race were characterized by percentage undergoing cesarean delivery. Cesarean rates according to group were compared using the χ2 statistic. This study was approved by the Stanford University Institutional Review Board.
To characterize what maternal characteristics may be associated with cesarean delivery in the VLBW neonate, we calculated cesarean rates for complications noted in the NCHS database. Recorded obstetric complications included the following 30 maternal medical risk factors and complications of pregnancy, labor, and delivery: anemia, cardiac disease, lung disease, diabetes, genital herpes, hydramnios or oligohydramnios, hemoglobinopathy, chronic hypertension, pregnancy-associated hypertension, eclampsia, incompetent cervix, previous infant more than 4,000 g, previous preterm or small for gestational age infant, renal disease, Rh sensitization, uterine bleeding, fever, moderate or heavy meconium, premature (> 12 hours) rupture of membranes, placental abruption, placenta previa, other excessive bleeding, seizure during labor, precipitous (< 3 hours) labor, prolonged (> 20 hours) labor, dysfunctional labor, cephalopelvic disproportion, cord prolapse, anesthetic complications, and nonreassuring fetal status. Maternal hypertension, preeclampsia, and eclampsia were combined into 1 category of hypertensive disorder. To identify possible differences in the aforementioned factors associated with cesarean delivery in VLBW and non-VLBW neonates, we compared the cesarean delivery rates for each of these 30 conditions, along with the fraction of cesarean delivery attributable to each condition, using the χ2 statistic. To facilitate presentation, data are only shown for those conditions that were present in more than 2.5% of cesarean deliveries in either the VLBW or non-VLBW group. Data across all 30 complications are available on request.
To attribute any survival advantages with the mode of delivery, birth weight–specific 28-day mortality rates per 1,000 live births, the relative risks (RRs) of mortality and their 95% confidence intervals (CIs) were estimated. As advised by the NCHS, we calculated mortality by assigning a weight to each neonate to account for death records that could not be linked to his corresponding birth certificate. This weight was dependent on the state of birth, with 1.4% of deaths in 2000 and 2.3% in 1999 without linked birth certificates.20,21 In 2000, Oklahoma had the lowest linkage at 93.2%, whereas 19 states had 100% linkage. In 1999, Ohio had the lowest linkage at 90.9%, whereas 12 states had 100%.
All states had uniform reporting requirements of demographic factors. Eleven states reported more detailed Asian subgroups than the others. Various states did not have specific items on their certificates. Of the variables relevant to our analysis, the only category that was not reported by a state was congenital anomalies by New Mexico. States also had varying levels of completeness of data. For example, in 1999 the mode of delivery was reported for 99% or more certificates in 46 states. However, 28.9% of certificates from Oklahoma lacked this information. Such records could not be included in our analysis. Birth weight was noted in 98.8% to 100% of records. For more detailed information regarding completeness of data by state, refer to the Technical Appendix from the Vital Statistics of the United States.22,23
To adjust for other risk factors, the complications associated with pregnancy were entered into a multivariate model to estimate the risk-adjusted survival advantage of mode of delivery. Sociodemographic factors were also included in the model, including maternal race (white, African American, or other), ethnicity (Hispanic or non-Hispanic), maternal age (reference 20–29 years), education (reference high school), and adequacy of prenatal care. Adequacy of care was measured by the Kessner index, which takes into account the month that prenatal care began, number of prenatal visits, and gestation.24 Adequacy was coded as a binary variable, with those mothers who had intermediate, low, or unknown care classified as having inadequate prenatal care.
We used 100-g intervals when determining cesarean delivery rates and calculated 95% confidence intervals. When comparing mortality rates by mode of delivery, we determined the optimal weight interval by performing sample-size calculations after preliminary analysis of the data. The infant mortality rate ranged from 2.3% to 5% for neonates born between 999 and 1,500 g. Using a significance level of 0.05 and power of 80%, and assuming a baseline mortality rate as low as 4%, with the ability to detect a 25% difference in groups, the sample size needed was approximately 10,600 per weight group.25 To obtain this degree of power, 200-g intervals were used to estimate the observed birth weight–specific mortality rates. In the multivariate analyses, neonates were also grouped into 200-g intervals. All analyses were computed using SAS 9.1 (SAS, Cary, NC).
The overall primary cesarean delivery rate in our vertex singleton cohort (500 to 4,000 g) was 11.6%, with a rate of 42.3% and 11.3% for VLBW and non-VLBW neonates, respectively (Table 1 and Fig. 1). At 500 to 599 g, 27.1 % of these neonates were delivered by cesarean. This rate increased to 43.7% for 700 to 800 g, peaked at 47.7 % at 1,100 to 1,200 g, and then declined to 40.6% at 1,400 to 1,500 g. Cesarean rates further declined for neonates more than 1,500 g and were less than 16.8% for neonates more than 2,500 g.
White VLBW neonates were more likely to undergo cesarean delivery than African Americans, whereas the opposite was true for non-VLBW neonates (Table 2). Higher education status and older age were both associated with increased cesarean delivery for VLBW neonates.
There were no recorded complications of pregnancy or delivery in 31.2% and 40.1% (P < .001) of VLBW and non-VLBW neonates, respectively, who were delivered by cesarean. Of the 30 medical conditions, each of 9 conditions accounted for greater than 2.5% of cesarean deliveries in VLBW neonates and each of 8 conditions in non-VLBW neonates. These conditions accounted for 95.3% of VLBW cesarean deliveries and 93.3% of non-VLBW cesarean deliveries. Five complications were frequently associated with cesarean delivery in both groups (Table 3). Hypertensive disorders had the strongest effect on the number of cesarean births in VLBW neonates. This was due to both the large proportion of VLBW neonates being born to mothers with this condition (18.1%) as well as a high cesarean rate in this group (78.7%). The largest medical factors associated with cesarean delivery in the non-VLBW neonates were nonreassuring fetal status, dysfunctional labor, and cephalopelvic disproportion. Cesarean delivery was more likely in the VLBW group given the presence of a medical condition frequently associated with cesarean, with the exception of cephalopelvic disproportion (Table 3, last column).
Table 4 shows the comparison of observed birth weight–specific neonatal mortality rate by mode of delivery. In the lowest birth weight group (500 to 699 g, Table 4 and Fig. 2), the RR of mortality associated with vaginal delivery was more than twice as high compared with cesarean delivery. The survival advantage associated with cesarean decreased with increasing birth weight and was not statistically significant above 1,300 g. We assigned a weight to each neonate to account for death records that could not be linked to their corresponding birth certificates, as advised by NCHS. When we performed the analysis without weighting, a similar survival advantage up to the 1,100–1,299 g group (RR 1.25, 95% CI 1.03–1.52) was observed for cesarean deliveries.
To control for possible differences in the incidence of risk factors, multivariate logistic models were constructed to assess the effect of mode of delivery in VLBW and non-VLBW cohorts using sociodemographic and medical risk factors that had a significant (P < .05) univariate association with neonatal mortality rate. Although maternal education, prenatal care, maternal fever, nonreassuring fetal status, placental abruption, and birth weight were associated with mortality in both cohorts, 11 of the maternal risk conditions served as risk factors in only 1 of the 2 cohorts (Table 5). After taking these risk factors into account, cesarean delivery was associated with a 47% decrease in VLBW neonatal mortality rate (odds ratio [OR] 0.53, 95% CI 0.49–0.57). For non-VLBW neonates, cesarean delivery was associated with a 62% increase in neonatal mortality rate (OR 1.62, 95% CI 1.57–1.68).
Because 31.2% of VLBW neonates delivered by cesarean did not have a reported medical condition, we performed a multivariate analysis on this subcohort that controlled for maternal age, education, and prenatal care. This analysis also demonstrated a significant survival associated with cesarean delivery (OR 0.49, 95% CI 0.44–0.54).
In some neonates born at the limits of viability, care may be withheld due to the impression that the infant may not be salvageable.26 To partially take into account neonates for whom intensive care may have not been provided, we excluded VLBW neonates who died within the first day of life. These neonates accounted for 6.6% of the population and 53.9% of 28-day mortalities. Even with their exclusion, although the advantage was diminished, there remained a significant risk-adjusted advantage to cesarean delivery (OR 0.78, 95% CI 0.72–0.86) in the multivariate regression model. One-day mortalities accounted for 68.8% of the 28-day mortalities in the 500–599-g weight group and between 35.8% and 53.3% in 100-g weight groups from 600–1,499 g. One-day mortalities accounted for 37.9% of 28-day mortalities in the 1,400–1,499-g weight group.
In our analysis of a national cohort born 1999 to 2000, we found that a higher proportion of VLBW vertex neonates were being delivered by cesarean and for different reasons from those seen in non-VLBW vertex neonates. Cesarean delivery was associated with improved survival in VLBW neonates, even after adjustment for medical and sociodemographic factors.
Cesarean rates in preterm deliveries have increased dramatically in the past several decades.18 The reason for this large increase has not been fully investigated, but is likely to be multifactorial in origin and partly due to the increased medical intervention and increased survival for this group. Previous analyses have suggested that cesarean delivery may benefit outcomes in both breech and vertex preterm neonates. However, other studies have not shown a consistent advantage. Also, most of the previous studies have analyzed cohorts in a time preceding the use of routine antenatal steroids and surfactants. In order for cesarean delivery to be useful, there must be effective antenatal and postnatal therapies for premature neonates. Prior studies may not have found improved outcomes due to overall lower survival.
Although a prospective randomized trial may be able to answer this question in a more satisfying fashion, ethical and practical considerations make this an unlikely possibility.27–29 Legal ramifications, the biases of delivering obstetricians, as well as the large number needed to power the study, are barriers that would be difficult to overcome. Thus, a large retrospective cohort study may be a reasonable alternative to help clarify this issue.
We examined this issue by analyzing the U.S. national database of births for the years 1999 to 2000. In our analysis by birth weight of neonates in vertex presentation, the survival advantage associated with cesarean delivery was most prominent in the lowest birth weight categories and decreased for the higher birth weight groups. This advantage was not statistically significant in neonates above 1,300 g. A recent analysis of a similar national cohort in Israel found no survival advantage associated with cesarean delivery in premature vertex neonates.12 However, their analysis compared neonates by gestational age rather than by birth weight.
The top 3 factors associated with cesarean delivery in non-VLBW neonates could be considered subjective factors in that they are dependent on the opinion of the obstetrician: nonreassuring fetal status, cephalopelvic disproportion, and dysfunctional labor. These factors were less prominent in the VLBW group, with the top factors being maternal hypertension, nonreassuring fetal status, premature rupture, and placental abruption.
The survival advantage from cesarean delivery observed in this cohort is most prominent in the smallest of neonates. This advantage in the lowest birth weight groups is likely due to a variety of factors, including the motivations of the parents, obstetrician, and neonatologist as they make decisions regarding the pursuit of aggressive care for a neonate that may be at the limit of viability. An obstetrician taking care of a mother who has expressed interest in pursuing aggressive care may be more likely to perform a cesarean delivery for a fetus showing signs of distress than in a case where the mother has not desired aggressive care. Furthermore, a mother who has undergone cesarean delivery may have more of a vested interest in seeing her baby survive after delivery.
Due to the nature of medical care when outcomes are uncertain, there is the possibility of confounding by selection bias. Although decision making, in regard to physician and parent interest in pursuing intensive care, may play a role in these tiniest of neonates, these factors are not able to be determined from this database. For example, when a fetus is thought to be nonviable, or an obstetrician considers the fetus to have a poor prognosis, the vaginal delivery mode may be preferred to minimize maternal risk. We attempted to account for the possibility of this bias by eliminating neonates that did not survive past the first day from the analysis. This would be the most conservative estimate in considering the possibility of this bias. However, there was still a significant effect after eliminating the first-day deaths. Furthermore, it is possible that the advantage of cesarean compared with vaginal delivery may be very prominent in improving first-day survival. In the clinical experience of the authors, the large number of first-day deaths is unlikely to be attributable solely to withholding of intensive care, considering that up to 40% of deaths occur in the first day even for infants close to 1,500 g. The retrospective nature of this study can not fully address these issues.
The survival advantage is seen at birth weights that would be considered well above the limit of viability in this day and age. Although this limit is still controversial and not well defined, most obstetricians and neonatologists would consider a birth weight of greater than 750 g or gestational age greater than 24 weeks as one in which care should not be withheld.
Another factor that may be important in the advantage of cesarean delivery could be the presence of fetal or intrauterine growth restriction. Considering that many of the VLBW neonates were born to mothers with hypertensive disorders, neonates who are growth restricted secondary to compromised placental function could derive important benefit from cesarean delivery. We are currently beginning a study to examine risks, outcomes, and mode of delivery in VLBW neonates with various degrees of intrauterine growth restriction.
Our study being based upon national birth certificates has limitations with respect to both the depth and accuracy of the clinical information available for analysis. For example, the large number of VLBW neonates delivered by cesarean to mothers with no obstetric indications may be due to underreporting of conditions in this data set. Furthermore, due to the design of this study, a causal relationship of cesarean delivery with decreased mortality cannot be proven. Despite these limitations, it is unlikely that a sufficiently large randomized controlled study will be developed to address this issue.
Another limitation of this data set is the absence of information on morbidity and long-term outcome. Cesarean delivery improving survival, but increasing serious morbidities or long-term disabilities in survivors would be an undesirable outcome. It would be important to compare morbidities and long-term outcomes in a similar fashion with a more detailed database.
It is possible that further insight could be developed from an analysis of hospital level outcomes. Cesarean delivery rates by hospital have been demonstrated as a potential factor in accounting for differences in mortality across hospitals.30 The extent to which the VLBW cesarean rate could potentially account for survival differences deserves further investigation. We believe that the extent of the survival advantage seen in VLBW born by cesarean warrants that cesarean delivery be given clinical consideration in the very low birth weight neonate.
1. Lee KS, Khoshnood B, Sriram S, Hsieh HL, Singh J, Mittendorf R. Relationship of cesarean delivery to lower birth weight–specific neonatal mortality in singleton breech infants in the United States. Obstet Gynecol 1998;92(5):769–74.
2. Main DM, Main EK, Maurer MM. Cesarean section versus vaginal delivery for the breech fetus weighing less than 1,500 grams. Am J Obstet Gynecol 1983;146:580–4.
3. Weissman A, Blazer S, Zimmer EZ, Jakobi P, Paldi E. Low birthweight breech infant: short-term and long-term outcome by method of delivery. Am J Perinatol 1988;5:289–92.
4. Granati B, Rondinelli M, Capoti C, Carnielli V, Bottos M, Rubaltelli FF. The premature breech presentation: outcome of newborn infants born by vaginal or abdominal delivery. Am J Perinatol 1984;1:145–7.
5. Gorbe E, Chasen S, Harmath A, Patkos P, Papp Z. Very-low-birthweight breech infants: short-term outcome by method of delivery. J Matern Fetal Med 1997;6:155–8.
6. Suidan JS, Sayegh RA. Delivery of the low birthweight and the very low birthweight breech: cesarean section or vaginal delivery? J Perinat Med 1989;17:145–9.
7. Ingemarsson I, Westgren M, Svenningsen NW. Long-term follow-up of preterm infants in breech presentation delivered by caesarean section: a prospective study. Lancet 1978;2:172–5.
8. Kigawa J, Goto S, Narita K, Aoki S. Studies on the mode of delivery and the outcome of small preterm infants with less than 32 weeks of gestation. Nippon Sanka Fujinka Gakkai Zasshi 1984;36:2629–33.
9. Ment LR, Oh W, Ehrenkranz RA, Philip AG, Duncan CC, Makuch RW. Antenatal steroids, delivery mode, and intraventricular hemorrhage in preterm infants. Am J Obstet Gynecol 1995;172:795–800.
10. Harper RG, Rehman KU, Sia C, Buckwald S, Spinazzola R, Schlessel J, et al. Neonatal outcome of infants born at 500 to 800 grams from 1990 through 1998 in a tertiary care center.J Perinatol 2002;22:555–62.
11. Philip AG, Allan WC. Does cesarean section protect against intraventricular hemorrhage in preterm infants? J Perinatol 1991;11(1):3–9.
12. Riskin A, Riskin-Mashiah S, Lusky A, Reichman B; Israel Neonatal Network. The relationship between delivery mode and mortality in very low birthweight singleton vertex-presenting infants. BJOG 2004;111:1365–71.
13. Melchor JC, Aranguren G, Lopez JA, Avila M, Fernandez-Llebrez L, Linares A. Perinatal outcome of very low birthweight infants by mode of delivery. Int J Gynaecol Obstet 1992;38:181–7.
14. Jonas HA, Lumley JM. The effect of mode of delivery on neonatal mortality in very low birthweight infants born in Victoria, Australia: Caesarean section is associated with increased survival in breech-presenting, but not vertex-presenting, infants. Paediatr Perinat Epidemiol 1997;11:181–99.
15. Paul DA, Sciscione A, Leef KH, Stefano JL. Caesarean delivery and outcome in very low birthweight infants. Aust N Z J Obstet Gynaecol 2002;42:41–5.
16. Yang YC, Jou TJ, Wu CH, Wang KG, Lan CC, Shen EY. The obstetric management in very-low-birth weight infants. Asia Oceania J Obstet Gynaecol 1990;16:329–35.
17. Jonas HA, Khalid N, Schwartz SM. The relationship between Caesarean section and neonatal mortality in very-low-birthweight infants born in Washington State, USA. Paediatr Perinat Epidemiol 1999;13:170–89.
18. Fanaroff AA, Hack M, Walsh MC. The NICHD neonatal research network: changes in practice and outcomes during the first 15 years. Semin Perinatol 2003;27:281–7.
19. National Center for Health Statistics. 1999, 2000 birth cohort linked birth and infant death data set. CD-ROM Series 20, No. 17a/18a. Hyattsville (MD): National Center for Health Statistics; 2003, 2004.
20. Mathews TJ, Menacker F, MacDorman MF. Infant mortality statistics from the 2000 period linked birth/infant death data set. Natl Vital Stat Rep 2002;50:1–28.
21. Mathews TJ, MacDorman MF, Menacker F. Infant mortality statistics from the 1999 period linked birth/infant death data Set. Natl Vital Stat Rep 2002;50:1–27.
22. National Center for Health Statistics. Technical appendix from vital statistics of the United States 1999 natality. Hyattsville (MD): Centers for Disease Control and Prevention; 2001.
23. National Center for Health Statistics. Technical appendix from vital statistics of the United States 2000 natality. Hyattsville (MD): Centers for Disease Control and Prevention; 2002.
24. Kessner DM, Kalk CW. Infant death: an analysis by maternal risk and health care. contrasts in health status. Washington (DC): Institute of Medicine, National Academy of Sciences; 1973.
25. Friedman LM, Furberg CD, DeMetes DL. Fundamentals of Clinical Trials. 3rd ed. New York (NY): Springer; 1998. p. 99–105.
26. Tyson JE, Stoll BJ. Evidence-based ethics and the care and outcome of extremely premature infants. Clin Perinatol 2003;30:363–87.
27. Eller DP, VanDorsten JP. Route of delivery for the breech presentation: a conundrum. Am J Obstet Gynecol 1995;173:393–6.
28. Penn ZJ, Steer PJ, Grant A. A multicentre randomised controlled trial comparing elective and selective caesarean section for the delivery of the preterm breech infant. Br J Obstet Gynaecol 1996;103:684–9.
29. Grant A, Glazener CM. Elective caesarean section versus expectant management for delivery of the small baby. Cochrane Database Syst Rev 2001;(2):CD000078.
© 2006 The American College of Obstetricians and Gynecologists
30. Gould JB, Danielsen B, Korst LM, Phibbs R, Chance K, Main E, et al. Cesarean delivery rates and neonatal morbidity in a low-risk population. Obstet Gynecol 2004;104(1):11–9.