The relationships between antenatal management, the cause of delivery and neonatal outcome in a large cohort of very preterm singleton infants

Authors

  • O. Baud,

    Registrar
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • V. Zupan,

    Senior Lecturer
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • T. Lacaze-Masmonteil,

    Professor, Corresponding author
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • F. Audibert,

    Senior Lecturer
    1. Department of Obstetrics and Gynaecology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • T. Shojaei,

    Registrar
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • B. Thebaud,

    Registrar
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • Y. Ville,

    Professor
    1. Department of Obstetrics and Gynaecology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • R. Frydman,

    Professor
    1. Department of Obstetrics and Gynaecology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author
  • M. Dehan

    Professor
    1. Department of Neonatology, Antoine-Béclére University Hospital, Assistance Publique/Hôpitaux de Paris, Clamart, France
    Search for more papers by this author

Correspondence: Dr T. Lacaze-Masmonteil, Réanimation et Pédiatrie Néonatales, Hôpital Antoine Béclére, 157, rue de Porte-de-Trivaux, 92141 Clamart, France.

Abstract

Objective To determine whether the cause of very preterm delivery influences neonatal outcome.

Design A cohort study of 685 consecutive singletons born before 33 weeks of gestation.

Methods Causes of birth and perinatal outcome variables were correlated for statistical significance by uni- and multi-variate analyses.

Results Intrauterine growth retardation or pre-eclampsia were associated with a higher rate of respiratory distress syndrome compared with prolonged rupture of membranes, after controlling for gestational age, antenatal corticosteroid therapy, antenatal antibiotic administration, mode of delivery and origin (inborn or outborn) (adjusted OR 3.12; 95% CI 1.55–6.28). The prevalence of grade 3–4 intraventricular haemorrhage or cystic periventricular leukomalacia was 25% in newborn babies born after intrauterine infection or prolonged rupture of membranes. Among infants born after intrauterine growth retardation/pre-eclampsia, the rate of severe intraventricular haemorrhage was 3.2% and the rate of periventricular leukomalacia was 0.9%. Compared with intrauterine infection and after controlling for potential confounding covariates, intrauterine growth retardation/pre-eclampsia was associated with a lower rate of periventricular leukomalacia (adjusted OR 0.08; 95% CI 0.02–0.41). In the same multiple logistic regression model, antenatal corticosteroid administration was associated with a lower incidence of periventricular leukomalacia (adjusted OR 0.36; 95% CI 0.16–0.79).

Conclusions The cause of very preterm delivery has an important influence on neonatal outcome.

INTRODUCTION

During the last decade prematurity has become an important issue in public health policies of developed countries. Recent improvements in neonatal intensive care of very preterm babies have led to substantial progress with an increased survival rate of the most immature infants1. In contrast, the prevalence of cerebral palsy following prematurity has not decreased, and it has been hypothesised that, irrespective of the level of perinatal care, morbidity of very premature infants is strongly influenced by the obstetric circumstances accounting for delivery2,3. Few published data are focused on neonatal outcome adjusted for perinatal events preceding very preterm birth4–6. Iannucci et al.4 found no difference in neonatal outcome in infants of extremely low birthweight classified according to the cause of premature delivery. In contrast, several recent observational studies have pointed out a relationship between intrauterine infection, the development of cystic periventricular leukomalacia and the subsequent occurrence of cerebral palsy2,7,8.

Although several antenatal therapies (e.g. corticosteroids, antibiotics, tocolytics) have an important effect on the immediate outcome, their impact on neonatal mortality and morbidity has not been hitherto assessed according to each main cause of delivery. This study was therefore undertaken to determine to what extent the cause of premature birth may influence neonatal outcome before discharge, and to assess the impact of several antenatal treatments on this influence. Multiple gestations were excluded from the analysis to avoid a major bias in both maternal care and immediate neonatal outcome.

METHODS

Antoine Béclère University Hospital is a regional level 3 perinatal centre, part of the Assistance Publique/Hôpitaux de Paris system. In 1997 the rate of severe prematurity (gestational age (33 weeks) and overall prematurity ges-tational age (37 weeks) was 7.2% and 17.6%, respectively. During the period 1993–1997, about 2400 infants with a postnatal age less than 24 hours, either inborn (65%) or outborn (35%), were admitted to the neonatal intensive care unit of this institution. Outborn patients were referred from surrounding maternity centres and cared for during transfer by a specialised neonatal transport team. The population in this study (n= 702) comprised consecutive singletons born before 33 weeks of gestation admitted to the neonatal intensive care unit during this five-year period. Seventeen babies were excluded from the analysis: eight were prematurely born to mothers suffering from severe nonobstetrical diseases; five had multiple malformations; and maternal data were not available for four. A computerised database providing data on the pregnancies along with details of neonatal progress in hospital was used to review the studied infants (n= 685).

Maternal variables

Gestational age was established according to the last menstrual period together with an ultrasound examination performed before 14 weeks of gestation. Babies and their mothers were classified according to the main cause of preterm delivery into one of the five groups. Group 1 (n= 170) included those with intrauterine infection associated or not with premature and prolonged (> 24 hours) rupture of membranes. Intrauterine infection was defined by the association of preterm labour and at least two pre-and/ or intrapartum criteria of maternal fever (temperature > 38°C on at least two occasions); blood inflammatory response (C-reactive protein plasma concentration > 40 mg/L or white blood cell count > 18,000/mm3); or bacteriological evidence of infection in amniotic fluid obtained by amniocentesis. Group 2 (n= 82) comprised those with prolonged rupture of the membranes without clinical and biological evidence of intrauterine infection. Group 3 (n= 68) comprised those with vaginal bleeding (placenta praevia and placenta abruptio); Group 4 (n= 219) included women with intrauterine growth retardation, detected by at least two consecutive ultrasonographic evaluations one week apart and confirmed after delivery (birthweight < 10th centile), or pre-eclampsia. Group 5 (n= 146) included those with isolated preterm labour. Antenatal corticosteroid administration, tocolysis and antibiotherapy were only included when birth occurred more than 24 hours after the mother received the first dose of treatment.

Neonatal outcome analysis

All outcomes, including death, were reported until discharge home. Respiratory morbidity included respiratory distress syndrome, bronchopulmonary dysplasia and chronic lung disease. The latter two were defined by the need for either assisted ventilation or oxygen supplementation at 28 days of life and 36 weeks of post-conceptional age, respectively. Respiratory distress syndrome was diagnosed if any two criteria were present in the first 24 hours of life: clinical symptoms (respiratory failure requiring assisted ventilation and administration of exogenous surfactant), typical radiological feature, and biological evidence of lung immaturity (fetal lung maturity test on tracheal aspirates).

Ultrasonographic scans and electroencephalography were performed at least twice during the first week of life and then at least once weekly until week 34 of gestational age. Neurological lesions noted in the study were severe intraventricular haemorrhage (Grades 3 to 4 of Papile classification9) and cystic periventricular leukomalacia diagnosed on cerebral ultrasound scan either before death or before the 6th week of life while late onset periventricular leukomalacias were excluded. Cystic leukomalacia was defined by hyperechoic lesions observed in periventricular areas, usually associated with positive rolandic sharp waves on electroencephalography10 and subsequently evolving into localised or extensive parenchymal cystic lesions.

Haemodynamic failure was defined by the need for both inotrope drugs and volaemic expansion during the first 24 hours of life. Neonatal sepsis was defined as bacteriological evidence of intrapartum infection associated with blood inflammatory response in neonates before 48 hours of life.

Statistical analysis

The mortality rate and the incidence of respiratory distress syndrome, bronchopulmonary dysplasia, chronic lung disease, haemodynamic failure, sepsis, intraventricular haemorrhage Grades 3–4 and periventricular leukomalacia were calculated for each circumstance of delivery and compared with their equivalent in the overall population (Pearson χ2 test, unadjusted Fisher's exact test and adjusted OR for gestational age, as appropriate). Continuous variables were compared by Student's t test. The relation between each circumstance of delivery and several outcome variables was estimated by multiple logistic regression models. Statistical significance was fixed at P < 0.05. Univariate and multivariate analyses were performed with Epi-Info Software (version 6) and STATA Software (version 6), respectively.

RESULTS

Premature birth

Gestational age was related to the cause of preterm delivery. Extremely immature infants were more frequently delivered after intrauterine infection or prolonged rapture of membranes. In contrast, intrauterine growth retardation/ pre-eclampsia was the main circumstance of delivery in babies born after 28 weeks of gestation. For each group, the odds of neonatal death were not statistically different from the odds in the total population. Severe brain damage (intraventricular haemorrhage Grades 3–4 or cystic periventricular leukomalacia) was the main cause of death (64.6%) among infants born after intrauterine infection/ prolonged rupture of membranes, whereas refractory respiratory failure accounted for 57.7% of death in infants delivered after intrauterine growth retardation/ pre-eclampsia (Table 1).

Table 1.  Univariate analysis of prenatal management and neonatal outcome according to the cause of birth. Values are given as % or median [range], unless otherwise stated.IUI = intrauterine infection;PROM = prolonged rupture of the membranes;IUGR = intrauterine growth retardation;PET = pre-eclampsia;RDS = respiratory distress syndrome;BPD = bronchopulmonary dysplasia;CLD = chronic lung disease;Grade3–4IVH = intraventricular haemorrhage (Papile grade3–4);PVL = periventricular leukomalacia.
 Total population (n= 685)IUI with or without PROM (n= 170)PROM without IUI (n= 82)Vaginal bleeding (n= 68)IUGR/PET (n= 219)Isolated preterm labour (n= 146)
  1. Comparison with the whole population using the χ2 test.

  2. *P ≤0.05; ≤0.01; *≤0.001.

Gestational age (wks)30[23–32.9]29[24–32.3]30[25–32.9]31[25–32.9]30[24–32.9]30[23–32.7]
≤28 weeks23.536.524.413.216*24
Birthweight (g)1210[420–2500]1210[530–2280]1420[610–2500]1500[630–2180]1030[420–1920]1350[600–2210]
Outborn3526.5*32.952.9*22.863
Antenatal steroids41.835.27223.5*53.423.3
Tocolytics23.838.82826.51.436.3
Caesarean section58.844.441.582.497.716.4
RDS50.940*26.876.554.360.3
RDS w/o antenatal steroids60.245.530.4*80.8*6170.5
BPD21.228.2*19.513.220.518.5
CLD7.210.69.51.56.45.5
Haemodynamic failure16.227.6*13.416.27.817.1
Neonatal sepsis23.451.8*20.70*0.537
Grade3–4IVH5.86.54.97.43.28.9
Cystic PVL8.914.7*13.411.70.910.3
Total death15.518.22214.711.415.1

Compared with the incidence of respiratory distress syndrome in the whole population (50.9%), the incidence of respiratory distress syndrome among infants born after either intrauterine infection with or without prolonged rupture of membranes, or prolonged rupture of membranes without intrauterine infection was significantly lower (P < 0.05 and P < 0.001, respectively). This difference was not abolished when infants were antenatally exposed to corticosteroids. The incidence of respiratory distress syndrome was particularly high (76.5%) in babies delivered after vaginal bleeding (P < 0.001). Among infants with a gestational age < 28 weeks, intrauterine growth retardation/ pre-eclampsia was associated with a very high incidence of respiratory distress syndrome and antenatal corticosteroid administration appeared to have no beneficial effect on respiratory distress syndrome incidence in this circumstance (Table 2). The incidence of chronic lung disease, strongly correlated with gestational age, was not statistically related to the cause of delivery.

Table 2.  Univariate analysis of prenatal management and neonatal outcome according to the cause of birth in infants with gestational age ≤28 weeks. Values are given as %. Definitions are as for Table 1.
 Total population n= 161IUI with or w/oPROM n= 62PROM w/oIUI n= 20Vaginal bleeding n= 9IUGR/PET n= 35Isolated preterm labour n= 35
  1. Comparison to the whole population with the χ2 test.

  2. *P ≤0.05; P≤0.001.

Antenatal steroids6.33.901.15.77.1*
RDS6.98.156.75.7*4.3
Antenatal steroids      
With5.22.9001.36.7
Without3.55.9051.75.9
CLD9.91401.17.111.4
Neonatal sepsis33.567.720005.7
Grade3–4IVH2.49.702.28.60
Cystic PVL4.36.103.30*7.1
Total death1.14.2533.34.337.1

Severe brain damage was a common complication among infants born after intrauterine infection: 25% of them suffered from severe intraventricular haemorrhage or cystic periventricular leukomalacia. Those two complications accounted for more than 50% of deaths in this group. In contrast, intrauterine growth retardation and pre-eclampsia were two perinatal events associated with a very low incidence of both severe intraventricular haemorrhage (3.2%) and periventricular leukomalacia (0.9%).

Compared with the overall population, neonatal sepsis was very frequent among infants born after intrauterine infection (51, 8%, P < 0.001) or isolated preterm labour (37%, P < 0.01). Haemodynamic failure was observed with a higher incidence among infants born after intrauterine infection with or without prolonged rupture of membranes (27.6%, P < 0.001). On the other hand, this complication was seldom encountered after intrauterine growth retardation/ pre-eclampsia (7.8%, P < 0.01).

The relation between the cause of delivery and the occurrence of respiratory distress syndrome, chronic lung disease and haemodynamic failure was estimated in a multiple logistic regression model, controlling for antenatal corticosteroid therapy, antenatal antibiotic administration, mode of delivery, gestational age and origin (inborn or outborn) (Table 3). Prolonged rupture of membranes, the circumstance of delivery associated with the lowest risk of respiratory distress syndrome in the univariate analysis, was considered as the subgroup of reference. Vaginal bleeding and intrauterine growth retardation/pre-eclampsia were associated with a higher rate of respiratory distress syndrome (adjusted OR 8.85, 95% CI 3.79–20.70 and adjusted OR 3.12; 95% CI 1.55–6.28, respectively).

Table 3.  Multiple logistic regression analysis of neonatal outcome (respiratory and haemodynamic complications; number of infants = 685). Values are given as adjusted OR (95%CI). Definitions are as for Table 1.
  RDS(n= 349)CLD(n= 49)Haemodynamic failure (n= 111)
 nAdjusted OR (95%CI)PAdjusted OR (95%CI)PAdjusted OR (95%CI)P
Causes of delivery       
  PROM w/o IUI821 1 1 
  IUI with or w/o PROM1701.23 (0.65–2.35)0.510.96 (0.36–2.59)0.932.05 (0.93–4.54)0.07
  Vaginal bleeding688.85 (3.79–20.70)≤0.0010.18 (0.02–1.64)0.131.48 (0.53–4.14)0.45
  IUGR/PET2193.12 (1.55–6.28)0.0010.50 (0.16–1.60)0.240.42 (0.16–1.09)0.07
  Isolated preterm labour1463.34 (1.68–6.68)0.0010.47 (0.14–1.55)0.221.40 (0.58–3.39)0.45
Antenatal corticosteroids       
  No3991 1 1 
  Yes2860.52 (0.35–0.76)0.0011.43 (0.70–2.94)0.3300.94 (0.55–1.59)0.55
Antenatal antibiotics       
  No5121 1 1 
  Yes1730.81 (0.50–1.34)0.420.60 (0.26–1.38)0.230.80 (0.44–1.45)0.44
Mode of delivery       
  Vaginal delivery2821 1 1 
  Caesarean section4031.35 (0.84–2.16)0.211.67 (0.71–3.90)0.241.69 (0.95–3.01)0.95
Gestational age (weeks)       
  30–32481 1 1 
  27–29463.10 (2.13–4.50)≤0.0018.69 (2.94–25.63)≤0.0012.66 (1.56–4.54)≤0.001
  ≤27918.52 (4.69–15.47)≤0.0010.86 (9.82–96.97)≤0.00111.31 (5.99–21.36)≤0.001
Outborn       
  No4451 1 1 
  Yes401.43 (0.97–2.12)0.071.62 (0.78–3.35)0.201.01 (0.60–1.71)0.97

The relation between the cause of delivery and the occurrence of severe neurological complications or death was evaluated in a separate multiple logistic regression model, controlling for the same set of covariates in addition to respiratory distress syndrome and haemodynamic failure (Table 4). Intrauterine infection, previously reported as the highest risk factor for periventricular leukomalacia, was defined as the subgroup of reference in this model. The occurrence of severe intraventricular haemorrhage was not related to the circumstance of delivery. In contrast, intrauterine growth retardation/ pre-eclampsia was associated with a lower rate of periventricular leukomalacia (adjusted OR 0.08; 95% CI 0.02–0.41).

Table 4.  Multiple logistic regression analysis of neonatal outcome (severe brain damage and neonatal death; number of infants = 685). Values are given as adjusted odds ratios (95%CI). Definitions are as in Table 1.
  Severe IVH (n= 40)Cystic PVL (n= 61)Death (n= 106)
 nAdjusted OR (95%CI)PAdjusted OR (95%CI)PAdjusted OR (95%CI)P
Causes of delivery       
  IUI with or w/oPROM1701 1 1 
  PROM w/o IUI22.77 (0.69–11.12)0.151.45 (0.61–3.44)0.402.91 (1.30–6.51)0.009
  Vaginal bleeding81.82 (0.44–7.52)0.410.88 (0.32–2.46)0.8146 (0.54–3.94)0.45
  IUGR/PET190.61 (0.17–2.25)0.460.08 (0.02–0.41)0.0021.53 (0.64–3.64)0.34
  Isolated preterm labour462.54 (0.84–7.67)0.100.49 (0.21–1.13)0.090.94 (0.43–2.06)0.88
Antenatal corticosteroids       
No3991 1 1 
  Yes860.22 (0.08–0.64)0.0060.36 (0.16–0.79)0.010.68 (0.38–1.22)0.19
Antenatal antibiotics       
No5121 1 1 
Yes730.67 (0.23–1.95)0.471.07 (0.53–2.15)0.841.33 (0.70–2.54)0.38
Mode of delivery       
  Vaginal delivery2821 1 1 
  Caesarean section031.66 (0.64–4.26)0.290.62 (0.31–1.22)0.160.78 (0.41–1.49)0.45
Gestational age (weeks)       
  30–323481 1 1 
  27–292466.36 (2.23–18.10)0.0012.51 (1.33–4.75)0.0053.59 (2.01–6.39)≤0.001
  ≤27913.60 (0.96–13.48)0.061.30 (0.51–3.31)0.583.88 (1.86–8.08)≤0.001
Outborn       
  No4451 1 1 
  Yes400.51 (0.22–1.20)0.121.20 (0.64–2.26)0.571.63 (0.95–2.80)0.07
RDS       
  No3361 1 1 
  Yes492.01 (0.72–5.64)0.181.32 (0.68–2.56)0.411.62 (0.91–2.91)0.10
Haemodynamic failure       
  No5741 1 1 
  Yes115.48 (2.40–12.50)≤0.0011.10 (0.52–2.35)0.805.57 (3.16–9.80)≤0.001

Antenatal corticosteroid therapy

Antenatal corticosteroid therapy was associated with a decrease in the incidence of respiratory distress syndrome (adjusted OR 0.41; 95% CI 0.29–0.56) (Table 5). The strongest association was observed in the group of infants born after isolated preterm labour (adjusted OR 0.10; 95% CI 0.04–0.30). The odds ratios for respiratory distress syndrome prevention in infants delivered after intrauterine infection with or without prolonged rupture of membranes (adjusted OR 0.53; 95% CI 0.26–1.10), prolonged rupture of membranes without intrauterine infection (adjusted OR 0.63; 95% CI 0.19–2.06) or intrauterine growth retardation/ pre-eclampsia (adjusted OR 0.63; 95% CI 0.33–1.17) suggest a trend towards a treatment effect which does not reach statistical significance for a risk of 5% (Table 5).

Table 5.  Analysis of the respiratory and neurological effects of antenatal steroid administration according to the cause of delivery. Odds ratios are adjusted to gestational age. Definitions are as for Table 1.
 Severe IVHCystic PVLRDS
Antenatal steroidsIncidenceAdjusted OR (95%CI)PIncidenceAdjusted OR (95%CI)PIncidenceAdjusted OR (95%CI)
  1. *Fisher exact test (No IVH or PVL in one group).

IUI with or w/o PROM        
  No (110)10%  20%  45.5% 
  Yes (60)0% 0.03*5%0.21 (0.05–0.71) 30%0.53 (0.26–1.10)
PROM w/o IUI        
  No (23)4.3%  21.7%  30.4% 
  Yes (59)5.1%1.04 (0.12–9.21) 10.2%0.26 (0.06–1.18) 25.4%0.63 (0.19–2.06)
Vaginal bleeding        
  No (52)9.6%  13.5% 80.8%  
  Yes (16)0% 0.44*6.3%0.46 (0.04–4.33) 62.5%0.37 (0.10–1.32)
IUGR/PET        
  No (102)6.9%  0%  61.4% 
  Yes (117)0% 0.013*1.7% 0.48*48.7%0.63 (0.33–1.17)
Isolated preterm labour        
  No (112)8.9%  13.4%  70.5% 
  Yes (34)8.8%0.88 (0.17–4.56) 0% 0.07*26.5%0.10 (0.04–0.30)
Total        
  No (399)8.5% 12.2%  60.2%  
  Yes (286)2.1%0.24 (0.10–0.61) 3.6%0.33 (0.17–0.63) 38.1%0.41 (0.29–0.56)

Antenatal corticosteroid exposure reduced the odds of both severe intraventricular haemorrhage (adjusted OR 0.24; 95% CI 0.1–0.61) and cystic periventricular leukomalacia (adjusted OR 0.33; 95% CI 0.17–0.63). A significant reduction in the incidence of severe intraventricular haemorrhage was observed not only in infants born after intrauterine infection with or without prolonged rupture of membranes (P= 0.03, Fisher's exact test) but also among infants delivered after intrauterine growth retardation or pre-eclampsia (P= 0 013, Fisher's exact test). The effect of antenatal steroid administration on the incidence of periventricular leukomalacia was independent of gestational age. The treatment was associated with a lower incidence of periventricular leukomalacia in infants born after intrauterine infection (adjusted OR 0.21; 95% CI 0.05–0.71).

The relation between antenatal corticosteroid therapy and the occurrence of respiratory distress syndrome, severe intraventricular haemorrhage and periventricular leukomalacia was estimated in the two previously described multiple logistic regression models (Tables 3 and 4). Compared with no treatment, antenatal corticosteroid therapy was associated with a lower rate of respiratory distress syndrome (adjusted OR 0.52; 95% CI 0.35–0.76), severe intraventricular haemorrhage (adjusted OR 0.22; 95% CI 0.08–0.64) and periventricular leukomalacia (adjusted OR 0.36; 95% CI 0.16–0.79).

Antenatal antibiotherapy

Antibiotics were administered in a large proportion of pregnancies complicated by prolonged rupture of membranes or intrauterine infection (60%). After adjustment for gestational age, antibiotics given for intrauterine infection at least 24 hours before birth significantly reduced the rates of severe intraventricular haemorrhage and cystic periventricular leukomalacia (adjusted OR 0.39; 95% CI 0.16–0.95 and adjusted OR 0.21; 95% CI 0.05–0.86, respectively) (Table 6).

Table 6.  Impact of antenatal antibiotics (≤24 h) on the prevalence of severe brain damage according to the cause of delivery. Odds ratios are adjusted to gestational age. Values are given as n (%) or adjusted odds ratios [95%CI]. Definitions are as for Table 1.
 Severe IVHCystic PVL
 With antibiotics (n= 173)W/o antibiotics (n= 512)Adjusted OR [95%CI]PWith antibiotics (n= 173)W/o antibiotics (n= 512)Adjusted OR [95%CI]P
  1. *Fisher exact test (No IVH or PVL in one group).

IUI with or w/o        
  PROM (n= 170)3/100 (3%)8/70 (11%)0.39 [0.16–0.95] 10/100 (10%)15/70 (21%)0.21 [0.05–0.86] 
PROM w/o IUI (n= 82)3/47 (6%)1/35 (3%)1.28 [0.34–4.76] 7/47 (15%)4/35 (11%)2.13 [0.20–22.20] 
Vaginal bleeding (n= 68)1/8 (12%)4/60 (7%)2.19 [0.20–24.09] 2/8 (25%)6/60 (10%)3.64 [0.51–25.81] 
IUGR/PET (n= 219)0/5 (0%)7/214 (3%) 0.47*0/6 (0%)2/214 (1%) 0.90*
Isolated preterm labour (n= 146)0/13 (0%)13/133 (10%) 0.56*1/13 (8%)14/133 (11%)0.83 [0.09–7.42] 

Mode of delivery

Prelabour caesarean section was associated with a decrease in the incidence of periventricular leukomalacia (adjusted OR 0.32; 95% CI 0.18–0.56) (Table 7). Caesarean section was often performed in the periventricular leukomalacia low risk factor group (intrauterine growth retardation/ pre-eclampsia). After exclusion of the intrauterine growth retardation/ pre-eclampsia subgroup, the difference was no longer significant (adjusted OR 0.63; 95% CI 0.34–1.16). In a separate analysis performed for infants born after intrauterine infection exclusively, caesarean section was associated with a lower rate of periventricular leukomalacia (adjusted OR 0.37; 95% CI 0.13–0.97).

Table 7.  Perinatal factors associated with the occurrence of PVL. Odds ratios are adjusted for gestational age. Values are given as n (%). Definitions are as for Table 1.
 With PVL (n= 61)W/o PVL (n= 624)Adjusted OR [95% CI]
  1. *IUGR/PET excluded.

  2. IUI with or w/o PROM only.

IUI24 (41)142 (23.2)1.96 [1.13–3.40]
PROM11 (18)71 (114)1.63 [0.81–3.29]
1 UGR/pre-eclampsia2 (3.3)217 (34.8)0.07 [0.02–0.28]
Vaginal bleeding8 (13.1)60 (9.6)1.76 [0.77–4.02]
Isolated preterm labour15 (24.6)131 (21)1.25 [0.66–2.33]
Antenatal steroids12 (19.7)274 (43.9)0.31 [0.17–0.63]
Caesarean section20 (32.8)383 (61.4)0.32 [0.18–0.56]
   0.63 [0.34–1.16]*
   0.37 [0.13–0.97]*
RDS36 (59)313 (50.2)1.17 [0.67–2.03]
Haemodynamic failure15 (24.6)96 (15.4)1.53 [0.79–2.97]

DISCUSSION

There are very few data on the relation between obstetrical events leading to premature birth and neonatal complications. The first main finding of this observational study is confirmation that intrauterine infection is an antenatal risk factor for periventricular leukomalacia. This association, previously pointed out by other investigators and our group7,11,12, is consistent with observations suggesting that an inflammatory process initiated before birth is associated with several short term complications of prematurity13. This hypothesis has been strengthened recently by the association found between high levels of interleukin-6 in neonatal blood and periventricular leukomalacia occurrence14. The low rate of respiratory distress syndrome among infants born after intrauterine infection or prolonged rupture of membranes in our study may be another consequence of the in utero exposure to mediators of inflammation. Lung exposure to interleukin-1β, a valuable marker of intrauterine infection in amniotic fluid15, has been shown to accelerate maturation of the surfactant system16.

In contrast, intrauterine growth retardation and preeclampsia were associated with a higher and gestational age-dependent risk of respiratory distress syndrome. These results are consistent with those reported by Ley et al.17. In this study, the incidence of respiratory distress syndrome below 29 weeks of gestation was higher in small for gestational age infants, compared with appropriate for gestational age infants. After adjustment for gestational age, odds ratios for respiratory distress syndrome prevention with corticosteroids also suggest that the hormonal treatment had a less beneficial effect in the intrauterine growth retardation/pre-eclampsia population. This observation is in agreement with the recent results of a retrospective study to assess the effectiveness of antenatal steroids according to obstetric subgroups18. In both studies growth retarded infants nevertheless benefited from antenatal corticosteroids by a substantial reduction in the incidence of severe postnatal intraventricular haemorrhage.

Antibiotic administration initiated at least 24 hours before birth is associated with a decreased risk of both severe intraventricular haemorrhage and periventricular leukomalacia among infants born after intrauterine infection. This supports the hypothesis of an inflammatory link between brain damage and preterm delivery after intrauterine infection or prolonged rupture of membranes19. Preterm labour followed by ‘unexpected’ delivery was a circumstance associated with both a high mortality rate and a high prevalence of brain damage. Though none of the women had clinical chorioamnionitis, the possibility cannot be excluded that infection escaped to biochemical and bacteriological detection in many of those pregnancies. A noninfection-related inflammatory process is another explanation which could be considered. In both cases, the fetal inflammatory response might have triggered the development of white matter injury by still unknown mechanisms.

The second main finding of this study is the demonstration that, independent of gestational age and obstetric management, antenatal exposure to corticosteroids is associated with significant effects on neonatal morbidity, including cystic periventricular leukomalacia occurrence. In her meta-analysis, Crowley20 suggested that antenatal corticosteroid administration may protect against neurologic impairment such as diplegia or quadriplegia. Our results are consistent with those of another study suggesting that antenatal exposure to betamethasone is associated with a lower risk of periventricular leukomalacia21. In the present study the benefit on immediate neurologic outcome observed in association with antenatal corticosteroids is more marked in the high risk group represented by intrauterine infection and prolonged rupture of membranes. Among infants born after intrauterine infection with or without prolonged rupture of membranes, antenatal exposure to corticosteroids reduced the incidence of both periventricular leukomalacia (5%vs 20%) and severe intraventricular haemorrhage (0%vs 10%). Of course, patients were not randomly assigned to hormonal treatment and it cannot be excluded that, despite adjustment for several covariates in multiple logistic regression models, other unknown factors may be responsible for this association.

An association between caesarean section and a reduction in cystic periventricular leukomalacia incidence has been previously reported in an observational study focused on preterm infants born after chorioamnionitis22. In the present study, which included more infants and only singletons, the beneficial effect of caesarean section remains significant in infants born from pregnancies complicated by intrauterine infection. Once again, conclusions should be drawn with caution as unknown confounding factors may account for this observation. Gomez et al.23 have suggested recently that the onset of preterm labour, initiated when the intrauterine environment becomes hostile, may represent an adaptative response developed to prevent maternal and fetal complications. It may be hypothesised that performing caesarean section after prolonged rupture of membranes or in case of intrauterine infection has anticipated this survival response.

In conclusion, neonatal outcome is dependent on prenatal causes of premature birth. This study strongly suggests that antenatal corticosteroid therapy protects against periventricular leukomalacia, the main cause of cerebral palsy. It also confirms the benefit of antibiotic administration to mothers with intrauterine infection or prolonged rupture of membranes. Whether contraindication of sustained tocolysis or more active intervention (caesarean section) should be considered for improving neurological outcome in these high risk situations remains to be assessed.

Ancillary