If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
We sought to determine the proportion of evidence-based (EB), vs non-EB (NEB) iatrogenic late preterm birth, and to compare corresponding rates of neonatal intensive care unit (NICU) admission.
We performed a retrospective cohort study. Cases were categorized as EB or NEB. NICU admission was compared between groups in both univariate and multivariate analysis.
Of 2693 late preterm deliveries, 32.3% (872/2693) were iatrogenic; 56.7% were delivered for NEB indications. Women with NEB deliveries were older (30.0 vs 28.6 years, P = .001), and more likely to be pregnant with twins (18.8% vs 7.9%, P < .001), have private insurance (80.3% vs 59.0%, P < .001), or have a second complicating factor (27.5% vs 10.1%, P < .001). A total of 56% of EB deliveries resulted in NICU admissions. After controlling for confounders, early gestational age (34 vs 36 weeks: odds ratio, 19.34; 95% confidence interval, 4.28–87.5) and mode of delivery (cesarean: odds ratio, 1.88; 95% confidence interval, 1.15–3.05) were most strongly associated with NICU admission.
Over half of nonspontaneous late preterm births were NEB. EB guidelines are needed.
We now understand that late preterm birth is associated with significantly higher rates of respiratory morbidity, but also results in other morbidities such as intraventricular hemorrhage, necrotizing enterocolitis, neonatal intensive care unit (NICU) admission, and sepsis, when compared to infants born at term. McIntire and Leveno
compared the different types of morbidities associated with late preterm birth to birth at 39 weeks, the gestational age with the lowest morbidity in their cohort of 21,771 deliveries over an 18-year period. They found that ventilator use, transient tachypnea of the newborn, sepsis, phototherapy for hyperbilirubinemia, and intraventricular hemorrhage were all significantly higher in late preterm infants compared to term. Similarly, Yoder et al
reviewed the epidemiology of respiratory disease in late preterm infants. They found that respiratory morbidity from all causes was higher at 34 weeks (22%), 35 weeks (8.5%), and 36 weeks (3.9%) when compared to 39 and 40 weeks (0.7% and 0.8%, respectively, P < .001). These findings were corroborated recently by the Safe Labor Consortium.
found that neonatal mortality and infant mortality were 9.5- and 5.4-fold higher following birth at 34 weeks compared to 39 weeks. The increased morbidity and mortality associated with late preterm birth when compared to term is concerning and raises the question as to whether the indications for these births are justified.
Additionally, long-term outcomes have been found to be poorer in late preterm infants compared to term. Tagle et al
found that IQ scores were lower at 6 years of age in children of women with a late preterm birth compared with similar term children. Another study of preschool and kindergarten children born late preterm compared with term children showed that the late preterm children were more likely to have developmental delay and suspension and retention in kindergarten.
found that adults born late preterm in Norway were 2.7 times more likely to have cerebral palsy and 1.6 times more likely to have mental retardation. These data further question the need to deliver these infants in the late preterm period.
In an effort to understand the etiology of late preterm births at our respective institutions, we sought to define a cohort of these deliveries to review indications and neonatal outcomes. Our specific objectives were 3-fold: first, to evaluate the proportion of late preterm deliveries that were nonspontaneous (iatrogenic); second, to evaluate the proportion of nonspontaneous deliveries that were based on evidence as opposed to common local practice; and third, to identify the indications for late preterm birth that require further research.
Materials and Methods
Using 2 contemporaneously maintained obstetrical databases, all women from Columbia University Medical Center and Christiana Care Health System who delivered between 34 0/7-36 6/7 weeks from January 2003 through July 2007 were identified. Both datasets are entered by individuals who solely perform data abstraction from medical charts. Prior to data collection the institutional review boards from both institutions were consulted and this study was approved by both via expedited review. Data regarding indication for delivery were abstracted from direct chart review.
The primary outcome of this study was the rate of nonspontaneous, late preterm delivery with an indication that was supported by current evidence as endorsed by the American College of Obstetricians and Gynecologists (ACOG) or published expert opinion (level III evidence). Spontaneous causes of preterm delivery were considered to have occurred for women with preterm birth preceded by either preterm premature rupture of membranes (PPROM) or preterm labor with intact membranes. Women with PPROM were considered to have spontaneous preterm labor whether or not labor induction was performed. Women with triplet or higher-order multiple gestations were excluded from all analysis as the optimal period of delivery for these women remains controversial. Also excluded were women for whom the indication for delivery was indeterminate. Cases were designated as indeterminate under the following circumstances: (1) the indication was not identified in the medical record; (2) the indication was an appropriate indication but the clinical documentation failed to support the given reason (eg, severe preeclampsia with no record of blood pressure or other support of the diagnosis); or (3) the 2 reviewers were unable to reach consensus as to the appropriateness of the indication. We did not exclude fetuses with documented fetal lung maturity because this has not been shown to improve neonatal outcomes.
Indications for nonspontaneous deliveries were assessed through review of the medical records, and 2 authors (M.K.H. and C.G-B.) independently reviewed each indication for consensus. Outpatient records were not reviewed. Those indications supported by ACOG guidelines and/or expert opinion (level C evidence and consensus from obstetric texts) were defined as evidence based (EB), while indications not supported by either of these were labeled non-EB (NEB). Where multiple indications were listed, if at least one of the indications was EB, the patient would be classified as such. For example, patients with severe preeclampsia and a prior myomectomy would be classified as EB for severe preeclampsia diagnosis even though a history of myomectomy was considered NEB.
A list of diagnoses and their classifications are found in Table 1. Diagnoses considered EB for the late preterm period included severe preeclampsia/eclampsia, growth restriction with abnormal fetal testing (abnormal testing included a biophysical profile of 6/10 or worse, abnormal umbilical artery or ductus venosus Doppler, or coexisting oligohydramnios), frank abruption, and a nonreassuring fetal heart tracing (a category II or III fetal heart tracing requiring immediate delivery). Deliveries defined as NEB included stable patients with the following diagnoses: chronic hypertension, mild preeclampsia, gestational hypertension, oligohydramnios (amniotic fluid index <5), intrauterine growth restriction with normal testing, and prior myomectomy/classic cesarean, not in labor.
We compared the rate of NICU admission for the 2 groups. At both institutions, infants are admitted to the NICU based primarily on morbidity. However, infants weighing <1800 g (Columbia) and <2200 g (Christiana) are automatically admitted to the NICU. There are no other protocol-based admission criteria. To detect the variable most likely associated with nonweight-based NICU admission, we excluded infants who weighed <1800 and <2200 g, respectively, at either institution. We also controlled for site (Columbia vs Christiana), mode of delivery, and gestational age in the multivariable logistic regression model that was created.
Statistical analysis was performed using Stata, Version 10.0 (Stata Corp, College Station, TX). Univariate analysis was performed using Student t test with unequal variance, Wilcoxon rank sum, χ2, and Fisher exact test when appropriate. A P value of < .05 was considered statistically significant.
We identified 2756 late preterm pregnancies over the study period. We excluded 9 women with triplets and 54 women who had indeterminate causes of delivery (Figure). Of the remaining 2693 women, 1429 delivered due to spontaneous labor with intact membranes and 392 delivered for PPROM. Thus, the rate of nonspontaneous, iatrogenic deliveries in this late preterm cohort was 32.3% (n = 872/2693). We then categorized these deliveries by indication and found that the majority of iatrogenic deliveries, 56.7% (n = 494/872), were deemed NEB (Table 2) for a total of 18.3% (494/2693) NEB deliveries in the entire late preterm cohort.
TABLE 2Indication for delivery and proportion of nonevidence based
Women in the NEB group were older (30.0 vs 28.6 years, P = .001), more likely to have twin gestations (18.8% vs 7.9%, P < .001), and more likely to have private insurance (80.3% vs 59.0%, P < .001) (Table 3). The most common NEB indication for delivery was for mild preeclampsia (31.4%), followed by oligohydramnios (16.7%), and gestational hypertension (15.0%) (Table 2).
TABLE 3Characteristics of evidence-based vs nonevidence-based deliveries
Maternal age, y
Prior preterm delivery
Maternal length of stay, d
Gestational age, wk
Apgar 1 (% ≥8)
Apgar 5 (% ≥9)
Amniocentesis prior to delivery
Infant to NICU
NICU, neonatal intensive care unit.
Gyamfi-Bannerman. Iatrogenic late preterm birth. Am J Obstet Gynecol 2011.
The mean gestational age at delivery was higher in the NEB group (35.7 weeks) compared to the EB group (35.1 weeks, P < .001). Also, neonates in the NEB group were more likely to be delivered in the late preterm period ≥36 weeks (70.6% vs 29.4%, P < .001). These infants were also more likely to be heavier at birth (2665 vs 2306 g, P < .001), and to have an amniocentesis for fetal lung maturity (8.7% vs 0.8%). In the univariate analysis, neonates in the EB group were more likely to be admitted to the NICU (56.0% vs 31.0%, P < .001). After excluding infants admitted to the NICU for weight criteria and controlling for other confounders, early gestational age (34 vs 36 weeks: odds ratio [OR], 19.34; 95% confidence interval [CI], 4.28–87.5) and mode of delivery (cesarean delivery: OR, 1.88; 95% CI, 1.15–3.05) were most strongly associated with NICU admission (Table 4). Babies delivered at Columbia were less likely to be admitted to the NICU (OR, 0.38; 95% CI, 0.19–0.77). Importantly, an EB indication was no longer an independent risk factor for NICU admission.
TABLE 4Adjusted odds for neonatal intensive care unit admission
CI, confidence interval.
Gyamfi-Bannerman. Iatrogenic late preterm birth. Am J Obstet Gynecol 2011.
There were 54 cases classified as indeterminate. The majority of these cases (36/54, 66.7%) were because data to support the indication listed could not be found (eg, severe preeclampsia with normal bloods pressures). The other indications were quite varied, included from 1-4 patients, and included diagnoses such as vasa previa, neonatal alloimmune thrombocytopenia, maternal cancers, cerebral infarction, and alcoholic seizures.
We found that the majority of nonspontaneous late preterm deliveries in our cohort were NEB. Although women with EB deliveries were more likely to have infants admitted to the NICU, this was not an independent risk factor for NICU admission in the multivariable analysis, and the primary factor resulting in NICU admission was early gestational age. Overall, 18.3% of our late preterm cohort was delivered for NEB, and potentially avoidable, indications. Our findings demonstrate the need to evaluate the indications warranting late preterm delivery, particularly due to the morbidity associated with birth in this gestational age window.
As previously noted, the rate of late prematurity is increasing steadily in the United States, and there is concern that some of these deliveries may not be indicated.
The indications for delivery vary widely in the late preterm literature. On one end of the spectrum, Lubow et al found that only 8% of 149 late preterm births over 2 years in their institution were delivered due to nonspontaneous indications.
; however, there is an inherent flaw in this designation. First, the terminology implies that an indicated delivery is necessary, and second, it does not take into account different pathophysiologic mechanisms that lead to indicated preterm birth. Meis and colleagues described risk factors for indicated preterm birth after defining such deliveries as nonspontaneous.
This resulted in 19 varied etiologies of indicated preterm birth including risk factors such as mullerian abnormalities, race, and chronic hypertension. While these data are helpful from an epidemiologic perspective, they do not address pathophysiology or appropriateness of the indication.
Whether a nonspontaneous preterm delivery is described as “indicated” or “elective” is subjective. Because there is an obvious stigma associated with elective preterm delivery, it would be hard to find “elective” listed as an indication for late preterm birth. However, when Reddy et al,
In their review they excluded births for obstetric complications, maternal medical conditions, major congenital abnormalities, and spontaneous labor without induction or coexisting obstetrical/medical indications. They speculated that many of these births without listed indications were elective due to perceptions of similar morbidity and mortality risks by either the patient or the provider. In contrast to these findings, as previously noted, Holland et al
found that <10% of those deliveries were elective. The authors acknowledged that their classification of the appropriateness of delivery was based on limited scientific evidence and commented that within their own group, they debated the appropriateness of each indication.
It is particularly because of the lack of consensus regarding accepted indications for preterm delivery and the stigma associated with elective preterm delivery that we chose an alternative approach in designating our deliveries. Although it could be implied that a NEB delivery is not indicated, this is far from true. We simply are saying that there is not enough evidence to support one practice over another. Take, for example, timing of delivery for mild preeclampsia, the most common NEB diagnosis in our cohort. There are institutional and regional differences with regard to advocacy for both late preterm and term delivery of these pregnancies, yet there are no data to show improved neonatal outcomes with either practice. Nor is there evidence to show that maternal outcomes are improved. A recent clinical trial evaluating labor induction vs expectant management for mild preeclampsia and gestational hypertension at ≥36 weeks found that maternal outcomes were improved with labor induction >37 weeks.
The authors did not find this trend with delivery between 36-37 weeks, but they cited low numbers in that subgroup as a reason for further study at that gestational age. Barton and colleagues specifically reviewed outcomes with gestational hypertension and late preterm delivery using a retrospective database.
They found that elective delivery from 34-36 weeks resulted in increased neonatal morbidity without maternal benefit. No other clinical trials have addressed optimal timing of delivery in this setting. The same argument can be made for our other NEB indications. Oligohydramnios is another diagnosis that commonly leads to iatrogenic late prematurity, however, the available literature does not support delivery for this indication. Zhang et al
found a population of women with isolated oligohydramnios and evaluated the correlation with this finding and perinatal morbidity. Using data from the RADIUS trial in which women underwent routine ultrasound screening, they identified a population of 113 women with isolated oligohydramnios. There was no association with adverse perinatal outcomes in this group of women. Because the largest percentage of NEB deliveries was related to hypertensive disorders of pregnancy and oligohydramnios, this is where research as to the validity of these indications should be focused.
There are several limitations to this study. Clearly there are indications for delivery that are not amenable to an EB review. For instance, we included uterine rupture in our list of EB deliveries. We are not anticipating that a prospective randomized trial on expectant management of uterine rupture will ever, or should ever, be attempted. Therefore, as in the article by Holland et al,
our designations of NEB vs EB deliveries can be criticized. Next, we classified twin deliveries as NEB. We did not have information on chorionicity, and there is literature that both supports and refutes delivery at <37 weeks in the setting of monochorionic twins.
Finally, we could not comment on neonatal outcomes beyond NICU admission rates because the neonatal data could not be linked at 1 of the 2 participating institutions. The strengths of this study are the large numbers of late preterm deliveries we were able to evaluate. To our knowledge, this is the largest study looking at indications for delivery where medical records were reviewed and diagnoses were verified. Also, because of institutional variations in practice, we were able to evaluate a variety of listed indications rather those only specific to one center.
In conclusion, we found that the majority of nonspontaneous late preterm deliveries in our cohort were NEB. Although babies from EB deliveries were more likely to be admitted to the NICU, this finding did not hold true after controlling for gestational age at delivery and mode of delivery. Although the individual indication for delivery likely plays a role, due to the number of indications, we could not reliably control for these in a logistic regression model. Regardless, because the majority of deliveries in this period were NEB, there is a pressing need to systematically reevaluate indications for delivery in this period since data for poorer neonatal outcomes associated with late preterm birth cannot be refuted.
Neonatal mortality and morbidity rates in late preterm births compared with births at term.