Perinatal morbidity associated with late preterm deliveries compared with deliveries between 37 and 40 weeks of gestation

Authors

Errata

This article is corrected by:

  1. Errata: Errata Volume 118, Issue 13, 1687, Article first published online: 11 November 2011

Dr YW Cheng, Division of Maternal–Fetal Medicine, Department of Obstetrics, Gynecology and Reproductive Sciences, University of California, San Francisco, 505 Parnassus Avenue Box 0132, San Francisco, CA 94143-0132, USA. Email yvecheng@hotmail.com

Abstract

Please cite this paper as: Cheng Y, Kaimal A, Bruckner T, Hallaron D, Caughey A. Perinatal morbidity associated with late preterm deliveries compared with deliveries between 37 and 40 weeks of gestation. BJOG 2011;118:1446–1454.

Objective  To estimate the risk of short-term complications in neonates born between 34 and 36 weeks of gestation.

Design  This is a retrospective cohort study.

Setting  Deliveries in 2005 in the USA.

Population  Singleton live births between 34 and 40 weeks of gestation.

Methods  Gestational age was subgrouped into 34, 35, 36 and 37–40 completed weeks of gestation. Statistical comparisons were performed using chi-square test and multivariable logistic regression models, with 37–40 weeks of gestation designated as referent.

Main outcome measures  Perinatal morbidities, including 5-minute Apgar scores, hyaline membrane disease, neonatal sepsis/antibiotics use, and admission to the intensive care unit.

Results  In all, 175 112 neonates were born between 34 and 36 weeks in 2005. Compared with neonates born between 37 and 40 weeks, neonates born at 34 weeks had higher odds of 5-minute Apgar <7 (adjusted odds ratio [aOR] 5.51, 95% CI 5.16–5.88), hyaline membrane disease (aOR 10.2, 95% CI 9.44–10.9), mechanical ventilation use >6 hours (aOR 9.78, 95% CI 8.99–10.6) and antibiotic use (aOR 9.00, 95% CI 8.43–9.60). Neonates born at 35 weeks were similarly at risk of morbidity, with higher odds of 5-minute Apgar <7 (aOR 3.42, 95% CI 3.23–3.63), surfactant use (aOR 3.74, 95% CI 3.21–4.22), ventilation use >6 hours (aOR 5.53, 95% CI 5.11–5.99) and neonatal intensive-care unit admission (aOR 11.3, 95% CI 11.0–11.7). Neonates born at 36 weeks remain at higher risk of morbidity compared with deliveries at 37–40 weeks of gestation.

Conclusions  Although the risk of undesirable neonatal outcomes decreases with increasing gestational age, the risk of neonatal complications in late preterm births remains higher compared with infants delivered at 37–40 weeks of gestation.

Introduction

Late preterm birth, defined as delivery between 34 (0/7) and 36 (6/7) weeks of gestation,1 accounted for more than 70% of all preterm births in the USA in 2006.2 The risk of infant death among late preterm births is three-fold higher than the risk among term births. Moreover, late preterm births comprise almost 10% of all infant deaths in the USA.3 Although there appears to be a small (3%) decrease between 2006 and 2008,4 late preterm delivery has increased by 25% since 1990 in the USA, and it continues to comprise a significant portion of the overall rise in the preterm birth rate.5 Whereas the majority of late preterm births result from spontaneous preterm labour, premature rupture of the membranes, as well as preterm deliveries caused by maternal or fetal indications,6 it has been estimated that in Latin America up to 18% of births at this gestational age are iatrogenic.7

The American College of Obstetricians and Gynecologists does not define an upper limit of gestational age for the use of tocolysis,8 but most authorities do not recommend the use of tocolysis or antenatal corticosteroids9,10 beyond 34 weeks of gestation because of the expectation of favourable outcomes for infants delivered at this gestational age.1,7,11 With improved neonatal care, infants born beyond 34 weeks have been considered ‘near term’ or ‘functionally full term’, and less emphasis has been placed on potential neonatal morbidity when making decisions regarding delivery. Although perinatal outcomes in neonates born after 34 weeks are certainly improved when compared with infants born before this gestational age, recent studies suggest that neonates born in the late preterm period are less mature both physiologically and metabolically when compared with neonates delivered at term. Late-preterm neonates are therefore at higher risk of morbidity and mortality than term neonates.1,12–14

To further explore the association between late preterm births and perinatal outcome by gestational age, we conducted a population-based study of all low-risk, singleton, live-born deliveries that occurred in the USA in 2005 using birth certificate data. We hypothesised that the risk of perinatal morbidity associated with preterm delivery is a continuum, and that perinatal complications decrease with increasing gestational age in a continuous, rather than a threshold manner. Hence, compared with neonates born at 37–40 weeks of gestation, neonates born in the late preterm period remain at risk of complications.

Methods

This is a retrospective cohort study of low-risk women with singleton live births that were delivered in 2005 in the USA using the Vital Statistics Natality birth certificate registry provided by the Centers for Disease Control and Prevention (CDC). The 2005 natality data include births to US and non-US residents that occurred in the 50 United States, the District of Columbia, the Virgin Islands and US territories. Details regarding compilation of these data have been published elsewhere.15 We excluded multiple gestations and deliveries before 340/7 weeks of gestation or after 406/7 weeks of gestation. Pregnancies complicated by the following medical or obstetric conditions were also excluded: cardiac, pulmonary or renal diseases, chronic hypertension, pregnancy-associated hypertension (includes gestational hypertension and pre-eclampsia), eclampsia, pre-gestational and gestational diabetes mellitus, premature rupture of the membranes, cord prolapse, placental abruption and placenta praevia. Institutional Review Board approval was obtained from the Committee on Human Research at the University of California, San Francisco.

In the 2005 Natality data, there were two entries for gestational age, one based on menstrual dates, the other based on obstetric/clinical dates. For this study, the gestational age was based on the obstetric/clinical dating because studies have shown that obstetric/clinical estimates provide a good approximation to the menstrual dating, and when ultrasound dating is designated as the ‘gold standard’, menstrual dating tends to overestimate gestational age.16,17 The gestational age at delivery was subgrouped into 34, 35 and 36 completed weeks of gestation; infants delivered between 37 and 40 weeks of gestation were designated as the referent group.

We examined maternal outcomes and neonatal outcomes, including 5-minute Apgar scores. In the 2005 Natality data, California did not collect information on Apgar scores and as a result 525 904 births (16.0%) were excluded from the analysis on 5-minute Apgar scores.14 The definition and diagnostic criteria of outcomes in the birth data were based on definitions compiled by a committee from the Federal and State Health Statistics.18,19 For example, ‘Fetal intolerance of labor’ is one of the fields that the National Center for Health Statistics collects as a check field under ‘Characteristics of Labor’. It is characterised as ‘Fetal intolerance of labor such that one or more of the following actions was taken: in-utero resuscitative measures, further fetal assessment, or operative delivery’.18

Incidence proportions of these outcomes were examined and compared by gestational age using the Cochran–Armitage test for trend of linearity, or dose–response fashion, with < 0.05 as the threshold for statistical significance. Multivariable logistic regression models were used to control for potential confounding bias. Model building and selection were based on a stepwise backward elimination process starting with a full model, which includes all potential confounding variables to derive a restricted model, with P < 0.10 as the threshold. Second-order interaction terms were generated but not included in the final model because these did not reach statistical significance for model selection. Births with missing outcomes of interest were excluded from such analyses. Deliveries that occurred between 37 and 40 weeks of gestation were designated as the referent group because we aimed to compare outcomes associated with late preterm deliveries with those of infants delivered at term. We did not designate 40 weeks as referent and compare 34 weeks with 40 weeks, 35 weeks with 40 weeks, 36 weeks with 40 weeks, and extending this to term gestations (37 weeks to 40 weeks, and so forth) because such comparisons have previously been reported in term pregnancies.20 Statistical analysis was performed using stata v9.0 (StataCorp, College Station, TX, USA). Statistical significance was indicated using P < 0.05 and 95% CI that did not contain the null value.

Results

There were 3 167 615 live, singleton births in the USA in 2005 that delivered between the gestational ages of 34 and 40 weeks and that met study criteria; these served as the study population. Of these, 175 112 neonates were born between 340/7 and 366/7 weeks of gestation: 23 574 (13.6%) delivered at 34 weeks, 44 705 (25.4%) delivered at 35 weeks and 106 833 (61.0%) delivered at 36 weeks. The maternal characteristics are shown in Table 1.

Table 1.   Maternal characteristics associated with gestational age at delivery*
 34 weeks35 weeks36 weeks37–40 weeks
(n = 23 574)(n = 44 705)(n = 106 833)(n = 2 992 503)
%%%%
  1. Source: National Center for Health Statistics (2005).

  2. *P < 0.0001 for all comparisons using chi-square test.

Age
≤19 years (n = 327 026)1.001.823.9293.26
20–34 years (n = 2 410 494)0.711.373.3194.61
35–40 years (n = 382 813)0.721.333.2594.70
≥41 years (n = 47 282)0.971.513.6693.87
Parity
Nulliparas (n = 1 200 976)0.821.483.3494.46
Multipara (n = 1 949 268)0.681.353.3794.60
Race/Ethnicity
Non-Hispanic White (n = 1 654 044)0.691.383.4394.50
African American (n = 491 124)1.222.094.5992.10
Latino/Hispanic (n = 717 254)0.611.152.6695.57
Asian (n = 149 945)0.490.932.3896.20
Other (n = 66 644)0.751.413.5194.33
Education
0–8 years (n = 679 484)0.891.623.5793.92
9–11 years (n = 902 617)0.791.513.6194.10
12 years (n = 647 963)0.721.403.4394.45
13–16 + years (n = 896 308)0.571.132.9095.39
Not stated/unknown (n = 36 449)0.821.423.1694.59
Prenatal care visits
≤6 visits (n = 283 863)1.863.005.7089.44
7–14 visits (n = 2 310 368)0.621.253.1494.99
≥15 visits (n = 472 469)0.481.032.7795.72

A small fraction of women received tocolysis beyond 34 weeks of gestation, with decreased frequency of tocolysis with advancing gestational age, whereas induction of labour became more frequent with increasing gestational age (Table 2). The frequency of primary caesarean delivery and caesarean delivery for fetal intolerance of labour as an indication was highest at 34 weeks, then decreased with increasing gestation at 35, 36 and 37 weeks (Table 2). A similar pattern was seen when primary caesarean delivery was stratified by parity. In contrast, the frequency of operative vaginal deliveries increased with advancing gestational age (Table 2).

Table 2.   Maternal outcomes by gestational age at delivery*
 34 weeks35 weeks36 weeks37–40 weeks
(n = 23 574)(n = 44 705)(n = 100 833)(n = 2992 503)
%%%%
  1. CD, caesarean delivery; VD, vaginal delivery.

  2. Source: National Center for Health Statistics (2005).

  3. *P < 0.0001 for all outcomes by Cochran-Armitage test for trend.

Primary CD (n = 459 688)24.619.917.014.3
Nulliparas (n = 249 649)32.828.126.124.8
Multiparas (n = 207 177)20.315.912.89.43
CD for fetal intolerance of labour (n = 46 709)11.911.08.677.64
Operative VD (n = 144 806)2.163.013.554.67
Nulliparas (n = 82 795)3.575.296.528.42
Multiparas (n = 61 439)1.421.912.232.93
Tocolysis (n = 10 438)5.203.762.600.92
Antenatal corticosteroids6.422.811.220.1
(n = 2 732)
Labour induction (n = 658 576)12.313.115.021.2
Nulliparas (n = 230 398)14.816.318.823.3
Multiparas (n = 425 481)11.011.513.420.3
Febrile morbidity (n = 27 541)1.010.871.001.28

When neonatal outcomes were examined by gestational age at delivery, we observed that neonates delivered at 34 weeks had a higher frequency of low 5-minute Apgar scores (3.4% for Apgar <7 and 1.5% for Apgar <4) than those delivered at 35, 36 or 37 weeks of gestation (Table 3). The frequency of hyaline membrane disease (3.9%) and need for mechanical ventilation for >6 hours (5.8%) was highest at 34 weeks (Table 3), as was the frequency of antibiotic use (11%) and admissions to the neonatal intensive care unit compared with greater gestational age subgroups (Table 3).

Table 3.   Neonatal outcomes by gestational age at delivery*
 34 weeks35 weeks36 weeks37 weeks
(n = 23 574)(n = 44 705)(n = 106 833)(n = 2 992 503)
%%%%
  1. NICU, neonatal intensive care unit.

  2. Source: National Center for Health Statistics (2005).

  3. *P < 0.0001 for all outcomes by Cochran–Armitage test for trend.

  4. **Available in 3 237 630 (86.0%) birth records (excluding California).

5-minute Apgar <7** (n = 19 408)3.422.201.540.65
5-minute Apgar <4** (n = 1712)1.470.920.510.18
Hyaline membrane disease (n = 5708)3.932.531.260.17
Mechanical ventilation
>30 minutes (n = 6966)3.932.421.170.24
>6 hours (n = 4324)5.763.141.490.31
Surfactant administration (n = 848)1.980.960.340.04
Antibiotics administration (n = 11 980)10.86.363.220.97
Neonatal seizures (n = 1107)0.090.080.060.03
NICU admission (n = 34 604)47.024.311.02.49

The association between gestational age and perinatal outcomes in late preterm births was further examined using multivariable logistic regression; in these analyses, the adjusted odds ratio approximates the relative risk as neonatal complications are often rare outcomes. Compared with deliveries that occurred between 37 and 40 weeks of gestation, women who delivered at 34 weeks had a nearly two-fold increase in the risk of primary caesarean delivery (adjusted odds ratio [aOR] 1.86; 95% CI 1.80–1.93). A subgroup analysis by parity indicated that this association was present for both nulliparous and multiparous women, and the risk of primary caesarean delivery in multiparous women was particularly higher at 34 weeks and 35 weeks of gestation (Table 4). Compared with deliveries between 37 and 40 weeks, neonates born at 34, 35 and 36 weeks were more likely to have fetal intolerance of labour as an indication for caesarean delivery (Table 4). In contrast, preterm deliveries at 34–36 weeks were protective against operative vaginal delivery for both nulliparous and multiparous women (Table 4). Women were also more likely to receive antenatal corticosteroids at 34 weeks (aOR 25.5, 95% CI 23.5–27.7) than at 37 weeks; this risk decreased with increasing gestational age (Table 4).

Table 4.   Adjusted odds ratios of perinatal outcomes using multivariable logistic regression analyses*
 34 weeks35 weeks36 weeks
aOR (95% CI)aOR (95% CI)aOR (95% CI)
  1. aOR, adjusted odds ratio; CD, caesarean delivery; NICU, neonatal intensive care unit; VD, vaginal delivery.

  2. Source: National Center for Health Statistics (2005).

  3. *Adjusting for maternal age, parity, race/ethnicity, maternal education, gestational weight gain, prenatal visits and cigarette smoking during pregnancy.

  4. Reference comparison group: women who delivered between 37 and 40 weeks of gestation.

  5. **Available in 3 237 630 (86.0%) birth records (excluding California).

Maternal outcomes
1° caesarean delivery1.86 (1.80–1.93)1.43 (1.39–1.46)1.19 (1.17–1.21)
 1° CD nulliparas1.47 (1.42–1.53)1.17 (1.13–1.21)1.04 (1.02–1.07)
 1° CD multiparas2.60 (2.48–2.73)1.90 (1.83–1.98)1.44 (1.40–1.48)
 CD-fetal intolerance of labour1.75 (1.65–1.86)1.55 (1.48–1.62)1.25 (1.20–1.29)
Operative vaginal delivery0.51 (0.47–0.55)0.68 (0.64–0.72)0.79 (0.76–0.82)
 OpVD nulliparas0.48 (0.43–0.53)0.66 (0.62–0.71)0.79 (0.76–0.82)
 OpVD multiparas0.58 (0.49–0.68)0.72 (0.65–0.78)0.82 (0.76–0.85)
Antenatal corticosteroids25.5 (23.5–27.7)10.7 (9.80–11.6)4.63 (4.26–5.04)
Labour induction0.56 (0.54–0.58)0.61 (0.59–0.62)0.70 (0.69–0.71)
Febrile morbidity0.88 (0.77–1.01)0.76 (0.69–0.85)0.89 (0.84–0.95)
Neonatal outcomes
5-minute Apgar <7**5.51 (5.16–5.88)3.42 (3.23–3.63)2.34 (2.24–2.46)
5-minute Apgar <4**6.97 (6.11–7.95)4.08 (3.61–4.63)2.41 (2.16–2.68)
Hyaline membrane disease10.2 (9.44–10.9)6.49 (6.08–6.93)3.61 (3.41–3.82)
Mechanical ventilation9.09 (8.46–9.77)5.46 (5.11–5.83)3.03 (2.87–3.21)
>30 minutes
Mechanical ventilation9.78 (8.99–10.6)5.53 (5.11–5.99)2.94 (2.74–3.15)
>6 hours
Surfactant6.31 (5.55–7.18)3.74 (3.21–4.22)1.88 (1.68–2.10)
Antibiotics9.00 (8.43–9.60)5.24 (4.93–5.55)2.84 (2.70–2.99)
Seizure2.08 (1.82–2.39)1.59 (1.42–1.79)1.47 (1.36–1.59)
NICU admission32.3 (30.9–33.6)11.3 (11.0–11.7)4.43 (4.29–4.57)

Neonates delivered at 34 weeks of gestation had a more than five-fold increased risk of having a low 5-minute Apgar score (aOR 5.51. 95% CI 5.16–5.88 for 5-minute Apgar <7 and aOR 6.97, 95% CI 6.11–7.95 for 5-minute Apgar <4) compared with those delivered between 37 and 40 weeks. The risk of a low 5-minute Apgar score was also higher for neonates delivered at 35 and 36 weeks compared with 37–40 weeks of gestation (Table 4). Neonates born at 34 weeks of gestation had a much higher risk of having hyaline membrane disease (aOR 10.2, 95% CI 9.44–10.9), requiring assisted mechanical ventilation for >6 hours (aOR 9.78, 95% CI 8.99–10.6) and having neonatal seizures (aOR 2.08, 95% CI 1.82–2.39) than those delivered between 37 and 40 weeks. This increased risk of respiratory complications was also seen in neonates born at 35 and 36 weeks compared with those delivered at 37–40 weeks of gestation (Table 4).

Neonatal outcomes were further examined with stratification by mode of delivery (vaginal deliveries and caesarean deliveries). In neonates who were delivered vaginally, the risks of low 5-minte Apgar scores (<7 and <4), hyaline membrane disease, and need for mechanical ventilation use were all higher in neonates delivered late preterm compared with deliveries between 37 and 40 weeks (Table 5). Similar associations were seen for neonates who were delivered by caesarean section: neonates delivered late preterm had higher risks of morbidity compared with those delivered between 37 and 40 weeks of gestation (Table 5). We further performed multivariable logistic regression analysis with stratification by gender to estimate the effect of gestational age on perinatal outcomes and observed similar directions and magnitudes of the associations as with the entire cohort (results not shown).

Table 5.   Stratification by mode of delivery: adjusted odds ratios of neonatal outcomes using multivariable logistic regression analyses*
Vaginal delivery34 weeks35 weeks36 weeks
(n = 16 019)(n = 32 777)(n = 81 101)
aOR (95% CI)aOR (95% CI)aOR (95% CI)
  1. aOR, adjusted odds ratio; NICU, neonatal intensive care unit.

  2. Source: National Center for Health Statistics (2005).

  3. *Adjusting for maternal age, parity, race/ethnicity, maternal education, gestational weight gain, prenatal visits and cigarette smoking during pregnancy.

  4. Reference comparison group: women who delivered between 37 and 40 weeks of gestation.

  5. **Available in 3 237 630 (86.0%) birth records (excluding California).

5-minute Apgar <7**4.27 (3.88–4.70)2.91 (2.67–3.16)2.14 (2.01–2.27)
5-minute Apgar <4**5.32 (4.36–6.50)3.21 (2.68–3.85)2.11 (1.82–2.44)
Hyaline membrane disease9.87 (8.94–10.9)6.13 (5.62–6.69)3.32 (3.08–3.58)
Mechanical ventilation >30 minutes9.20 (8.35–10.1)4.88 (4.46–5.35)2.67 (2.48–2.89)
Mechanical ventilation >6 hours8.85 (7.80–10.0)4.21 (3.73–4.76)2.53 (2.29–2.80)
Surfactant4.95 (4.07–6.03)2.56 (2.13–3.09)1.47 (1.26–1.72)
Antibiotics9.40 (8.60–10.3)4.71 (4.34–5.11)2.55 (2.37–2.73)
Neonatal seizure2.28 (1.91–2.72)1.41 (1.20–1.64)1.37 (1.24–1.51)
NICU admission33.7 (31.9–35.6)10.7 (10.2–11.3)4.14 (3.97–4.33)
Caesarean delivery34 weeks35 weeks36 weeks
(n = 10 268)(n = 17 126)(n = 37 074)
aOR (95% CI)aOR (95% CI)aOR (95% CI)
5-minute Apgar <7**6.74 (6.16–7.37)4.09 (3.74–4.46)2.66 (2.47–2.87)
5-minute Apgar <4**8.04 (6.74–9.59)5.06 (4.25–6.01)2.78 (2.36–3.26)
Hyaline membrane disease10.8 (9.68–12.1)7.46 (6.74–8.26)4.34 (3.98–4.75)
Mechanical ventilation >30 minutes9.05 (8.10–10.1)6.58 (5.95–7.27)3.79 (3.47–4.13)
Mechanical ventilation >6 hours9.91 (8.82–11.1)6.78 (6.09–7.54)3.39 (3.06–3.75)
Surfactant8.32 (6.98–9.91)5.77 (4.90–6.78)2.63 (2.23–3.09)
Antibiotics7.99 (7.25–8.79)5.69 (5.22–6.20)3.20 (2.96–3.46)
Neonatal seizure1.86 (1.46–2.37)2.10 (1.76–2.50)1.72 (1.50–1.97)
NICU admission27.7 (26.0–29.5)11.3 (10.7–11.9)4.56 (4.46–4.76)

Discussion

In spite of being ‘near-term’, neonates delivered between 340/7 and 366/7 weeks remained at increased risk of perinatal complications compared with those women who delivered at 37–40 weeks. In particular, neonates delivered at 34 weeks had the highest risk of respiratory complications, low Apgar scores, neonatal seizures and required admission to the intensive care nursery. Although these perinatal morbidities decreased with increasing gestational age, neonates born throughout the late preterm period had a higher risk of complications when compared with infants born at 37–40 weeks of gestation.

Studies have demonstrated that caesarean delivery can be associated with increased risk of respiratory complications.21,22 To examine the association of late preterm delivery and perinatal morbidity independent of mode of delivery, the study cohort was stratified by mode of delivery. Compared with deliveries at 37–40 weeks of gestation, neonates born late preterm consistently had higher risks of lower 5-minute Apgar scores, respiratory complications, and need for ventilatory support regardless of whether they were delivered vaginally or by caesarean section, suggesting that in this cohort, this effect arises primarily from the gestational age at delivery rather than from the mode of delivery.

Several studies find long-term cognitive and developmental sequelae among infants born late preterm. Key neurological, behavioural, and cognitive conditions in childhood that appear more prevalent among late preterm births include cerebral palsy, antisocial behaviour, attention problems and sub-optimal academic performance.23–26 Talge and colleagues,27 for example, find that children age 6 years who were born late preterm exhibit an over two-fold increased risk of borderline intellectual functioning (IQ < 85). In addition, the risk of developing cerebral palsy is three times more likely in children born late preterm than in children born at term.28 These circumstances underscore the high social cost of late preterm birth well into childhood and adulthood.

Women who deliver between 34 and 36 weeks of gestation are at increased risk of primary caesarean delivery compared with women delivered at 37–40 weeks of gestation. This increased risk applies to both nulliparous and multiparaous women and is even more pronounced in the multiparous, with the indication for caesarean section probably less frequently the result of labour dystocia and more likely to be attributable to fetal intolerance of labour, because women undergoing caesarean for malpresentation and praevia were excluded from the cohort in this study. Women undergoing caesarean delivery incur higher risks of maternal and neonatal complications both in the index pregnancy and in future pregnancies.29,30 In an era of declining trial of labour after previous caesarean delivery, once the first caesarean has been performed, women are likely to face the increased morbidity from repeat caesareans in the future.31

In addition to the increase in perinatal morbidity during the late preterm period observed in this study, previous studies have shown that infants born between 34 and 36 weeks of gestation are at increased risk of grade 1–2 intraventricular haemorrhage, temperature instability, hypoglycaemia, hyperbilirubinaemia, feeding difficulty, and longer hospital stay.11–13,32,33 Recent studies also report that beyond the immediate neonatal period, infant mortality (death from 28 days of life to 1 year of life) of infants born in the late preterm period is twice that of full-term births.32

The healthcare costs required to care for neonates born at 33 weeks of gestation are estimated as ten times that for term neonates: the average spending per case was $7200 for 34 weeks of gestation, $2600 for 36 weeks and $1100 for term births in California in 1996.34 In a recent study of the financial costs associated with clinical problems and extended hospital stays among late preterm births, Wang et al.13 estimated a mean excess of $2630 ($3081 in 2010 dollars) in total healthcare costs for each late preterm, relative to term, birth. If we apply this value to the total number of late preterm births in the USA in 2005 (i.e. 173 819), the excess hospital costs for late preterm births add up to approximately $535 million per year. Given that Wang et al. followed late preterm births only to age 1 year, this estimate represents a lower bound of the total healthcare costs incurred over the life course. As healthcare expenditures continue to skyrocket and financial and social resources are finite, strategies to prevent preterm birth should not only focus on curtailing preterm delivery between 24 and 34 weeks but should also include strategies for the prevention of late preterm births.

The emerging evidence of significant perinatal, neonatal and postnatal morbidity and mortality associated with late preterm births is particularly alarming given the 25% increase in the incidence rate of late preterm births between 1990 and 2006 in the USA. Most obstetric interventions to reduce perinatal morbidity and mortality associated with preterm birth are tertiary prevention (intervention initiated after the parturition process has begun), because it is difficult to identify women at risk (secondary prevention) or institute population-level interventions for women before or during pregnancy (primary prevention).35 As the majority of late preterm births result from spontaneous preterm labour,5,6 obstetricians/clinicians are in a key position to provide preventive care, including: preconception counselling, public education, nutritional supplementation, smoking cessation, prenatal care, screening for at-risk women, treatment for high-risk women, and early diagnosis and treatment of preterm labour.28 For women who have an obstetric or fetal indication for delivery the benefits and risks of intervention and the resulting iatrogenic late preterm delivery with its associated immediate neonatal effects as well as long-term ramifications should be carefully considered. For example, in the setting of preterm premature rupture of the membranes (PROM), the American College of Obstetricians and Gynecologists Practice Bulletin states that ‘delivery is recommended when PROM occurs at or beyond 34 weeks of gestation’.36 Perhaps, in the absence of clinical or subclinical infection, the benefit of expectant management may be cautiously weighed against potential morbidity associated with late preterm delivery to optimise perinatal outcome. Hence, this topic deserves further investigation.

Although this population-based study reflects obstetric and neonatal outcomes of low-risk births between 34 and 36 weeks of gestation in the USA in 2005, it has limitations. As this study examines perinatal outcomes associated with late preterm deliveries, the accuracy of gestational age dating is essential. The issue of gestational age dating in the Natality data by menstrual or obstetric/clinical estimates has been examined in depth.16,17 Obstetric/clinical estimates reportedly provide a close approximation to menstrual dating. Whereas we chose to use obstetric/clinical dating for this analysis to minimise such errors in estimation, some women may have been misclassified such that they were assigned a higher gestational age than they should have been. Although misclassification bias is usually unidirectional, the bias affects all groups so the comparisons of each week’s sub-strata with the others are still valuable. The advantage of using the Vital Statistics Natality data is the representation of all live births in the USA, which truly reflects the obstetric care in this country. Although maternal and neonatal outcomes were reported with detailed definitions and routinely verified by the Federal and State maternal and child health personnel for quality control checks to ensure accuracy and completeness, missing data and reporting error may still exist; however, this typically represents a very small proportion (<1%) of the population and they were excluded from the analysis. We assumed that censorship occurred in a random fashion and probably would not have biased the study findings given the small number. One exception was the examination of 5-minute Apgar scores, which the state of California did not report, representing 15% of the births.

Conclusion

We report on one of the largest cohorts in the literature, which examines the perinatal outcomes associated with late preterm births. Consistent with previous studies, we found that deliveries at 34 weeks of gestation remain at risk of perinatal morbidity compared with deliveries at 37–40 weeks of gestation. This risk decreases with increasing gestational age but is still significant at 34, 35 and 36 weeks of gestation. This evidence suggests that as neonates born late preterm are not physiologically as mature as term infants, they should not be considered ‘functionally term’. As late preterm births continue to rise, there is an urgent need for clinical and research efforts to focus on the prevention of preterm delivery even at 34 weeks and beyond. In the meantime, the findings of our large population-based study may aid clinicians in the management of late preterm labour and in counselling women at risk for late preterm births.

Disclosure of interests

There are no conflicts of interest to disclose for any authors.

Contribution to authorship

YWC performed the statistical analysis and composed the majority of this manuscript. AJK helped with the statistical analysis and composition of the manuscript. TAB and DRH helped with the composition of this manuscript and provided specialty/subject-specific advice as well as knowledge regarding statistical analysis. ABC helped with developing the design of this research question and with editing/composition of the manuscript as well as knowledge regarding statistical analysis.

Details of ethics approval

This study was approved by the Committee on Human Research at the University of California, San Francisco on 4 November 2009.

Funding

YWC is supported by the UCSF Women’s Reproductive Health Research Career Development Award, NIH, the Eunice Kennedy Shriver National Institute of Child Health and Human Development (K12 HD001262) and DRH is supported by the Washington University CTSA (KL2RR024994-01).

Acknowledgements

The authors would like to acknowledge Sanae Nakagawa for statistical programming assistance.

Ancillary