*Manuscript (revised), with pgs numbered Click here to view linked References

Early antibiotic exposure and adverse outcomes in very preterm infants at low risk
of early-onset sepsis: the EPIPAGE-2 cohort study

M.Letouzey MD a,b, E.Lorthe RM,PhD a,c, L.Marchand-Martin MSc a, G.Kayem MD,PhD a,d,
C.Charlier MD,PhD e, M.Butin MD,PhD f, A.Mitha MD,PhD g, M.Kaminski MSc a,
V.Benhammou PhD a, PY.Ancel MD,PhD a,h, P.Boileau MD,PhD b,i, L.Foix-L’Hélias
MD,PhD a,j + EPIPAGE-2 Infectious diseases working group.

Affiliations:
a Université de Paris, CRESS, Obstetrical Perinatal and Pediatric Epidemiology Research

team, EPOPé, INSERM, INRA, F-75004 Paris, France

b
Department of Neonatal Pediatrics, Poissy Saint Germain Hospital, France
c
Unit of Population Epidemiology, Department of Primary Care Medicine, Geneva
University Hospitals, 1205 Geneva, Switzerland
d
Department of Gynecology and Obstetrics, Armand Trousseau Hospital, APHP, Paris
Sorbonne University, France
e
Université de Paris, Hôpital Universitaire Necker-Enfants Malades, Division of Infectious
Diseases and Tropical Medicine, APHP, Paris, Institut Pasteur, biology of Infection Unit,
French National Reference Center and WHO Collaborating center for listeria, Inserm U1117,
France
f
Department of Neonatal Pediatrics, Hôpital Femme Mère Enfant, Hospices Civils de Lyon,
France
g
Department of Neonatal Medicine, Jeanne de Flandre Hospital, CHRU Lille, F-59000 Lille,
France
h
URC - CIC P1419, Cochin Hotel-Dieu Hospital, APHP, Paris, France
i
Université Paris-Saclay, UVSQ, UFR Simone Veil Santé
j
Department of Neonatal Pediatrics, Armand Trousseau Hospital, APHP, Paris Sorbonne
University, France

Corresponding author:
Mathilde LETOUZEY UMR 1153, Obstetrical, Perinatal and Pediatric Epidemiology
Research Team, Hôpital Tenon Bâtiment Recherche, 4 rue de la Chine, 75020 PARIS,
FRANCE. Tel: +33156018372 Mail: [email protected]

No reprint

Key words: empirical antibiotics; prematurity; preterm birth; neonatal outcome; late-onset
sepsis; necrotizing enterocolitis; neonatal death; bronchopulmonary dysplasia; severe cerebral
lesions.
Short title: Early antibiotics and adverse outcomes in very preterm infants

Funding
The EPIPAGE-2 study has been funded with support from
1) The French Institute of Public Health Research/Institute of Public Health and its partners:
the French Health Ministry, the National Institute of Health and Medical Research
(INSERM), the National Institute of Cancer, and the National Solidarity Fund for Autonomy
(CNSA).

1
 2) The National Research Agency through the French EQUIPEX program of investments in
the future (reference ANR-11-EQPX-0038 and ANR-19-COHO-001).
3) The PREMUP Foundation
4) Fondation de France (reference 11779)
5) Fondation pour la Recherche Médicale (SPF20160936356)

The funding source had no role in the design and conduct of the study; collection,
management, analysis and interpretation of the data; preparation, review or approval the
manuscript; and the decision to submit the manuscript for publication.

We declare no potential, perceived or real conflict of interest relevant to this article to

disclose.
There are no prior publications or submissions with any overlapping information, including
studies and patients.

2
 ABSTRACT
Objective: To assess the association between early empirical antibiotics and neonatal

adverse outcomes in very preterm infants without risk factors of early-onset sepsis.

Study design: This is a secondary analysis of the EPIPAGE-2 study, a prospective national

population-based cohort that included all liveborn infants at 22-31 completed weeks of

gestation in France in 2011. Infants at high risk of early-onset sepsis (born after preterm labor

or preterm premature rupture of membranes, or from mother who had clinical chorioamnionitis

or received antibiotics during the last 72 hours) were excluded. Early antibiotic exposure was

defined as antibiotics started at Day 0 or Day 1 of life, regardless of the duration and type of

antibiotics. We compared treated and untreated patients by using inverse probability of

treatment weighting based on estimated propensity score.

Results: Among 648 very preterm infants at low risk of early-onset sepsis, 173 (26.2%) had

early antibiotics treatment. Early antibiotic exposure was not associated with death or late-onset

sepsis or necrotizing enterocolitis (OR=1.04 [95%CI=0.72-1.50]). However, it was associated

with higher odds of severe cerebral lesions (OR=2.71 [95%CI=1.25-5.86]) and moderate-to-

severe bronchopulmonary dysplasia (OR=2.30 [95%CI=1.21-4.38]).

Conclusion: Early empirical antibiotics administrated in very preterm infants at low risk of

early-onset sepsis were associated with a higher risk of severe cerebral lesions and moderate-

to-severe bronchopulmonary dysplasia.

3
 INTRODUCTION

Early-onset sepsis (EOS) remains frequent in very preterm infants with a prevalence of 18

per 1000 live births before 28 weeks, which is 10 to 30-fold higher than in term infants.(1)

Infection-associated mortality is close to 30% in these vulnerable infants.(1) When facing initial

clinical instability in preterm infants, neonatologists are trained to suspect EOS and widely

prescribe empirical antibiotic treatments upon admission to neonatal intensive care units

(NICU).(2) However, the risk of EOS differs according to the cause of preterm birth. At birth,

antibiotic administration is widely performed in the context of preterm labor or preterm

premature rupture of membranes due to the high risk of EOS. Nevertheless, these antibiotic

treatments are questionable in the context of maternal hypertensive disorders or fetal growth

restriction associated with a low risk of EOS.(2,3)

Promoting appropriate use of antibiotics in NICUs is a major concern in neonatology mainly

because of their potential adverse impact on antibiotic resistance.(4,5) In addition, antibiotic

exposure in early life disrupted the gut microbiota colonization (4,5) and can have long-term

effects on health such as obesity, allergic diseases, or inflammatory bowel disease.(6–8)

Studies on early antibiotics treatment in very preterm infants with or without risk factors for

EOS have shown that prolonged antibiotic exposure may be associated with neonatal adverse

outcomes such as necrotizing enterocolitis (NEC), late-onset sepsis (LOS), bronchopulmonary

dysplasia (BPD), and death.(9–12) However, these studies did not usually take into account the

cause of preterm birth which is associated with both early antibiotic exposure and some adverse

outcomes such as death or LOS.(11,13–16) Moreover, many of these adverse outcomes occur

among critically ill preterm infants who received antibiotics due to their severe clinical status

at birth. Therefore, controlling for the initial severity of illness would limit the indication

bias.(10,11)

4
 To better understand the impact of early antibiotics treatment and reduce the indication bias,

one way is to limit the study to very preterm infants at low risk of EOS, namely born after

maternal hypertensive disorders and/or fetal growth restriction, and controlling for severity of

illness.

The aim of our study was to assess the association between early antibiotic exposure and

adverse neonatal outcomes in very preterm infants at low risk for EOS.

METHODS

Study population

This is a secondary analysis of the EPIPAGE-2 cohort (Etude éPIdémiologique sur les Petits

Ages GEstationnels 2), a prospective national population-based cohort of preterm infants that

included liveborn and stillborn infants as well as all terminations of pregnancy at 22-31

completed weeks of gestation in all maternity units from 25 regions in France in 2011.(17)

Participants were recruited over different periods according to gestational age at birth: eight

months for births at 22-26 weeks and six months for births at 27-31 weeks. Maternal, obstetric,

and neonatal data were collected from medical records following a standardized protocol as

previously reported.(18) Recruitment took place at birth after families received information and

agreed to participate. During the recruitment period, 8400 births were eligible, among whom

7804 (93%) were enrolled in the study at baseline.(17) EPIPAGE-2 study was approved by the

National Data Protection Authority (Commission Nationale de l’Informatique et des Libertés,

CNIL n°911009) and by appropriate ethics committees (Consultative Committee on the

Treatment of Data on Personal Health for Research Purposes, reference 10.626) and the

Committee for the Protection of People participating in biomedical research (reference CPP

SC-2873).

5
 All infants enrolled in the EPIPAGE-2 cohort, born alive between 22 and 31 completed

weeks were included in this analysis (figure 1). Exclusion criteria were: death in the delivery

room or during the first day of life after antenatal decision of withholding care, severe

congenital malformation, and twin pregnancies with twin-twin transfusion syndrome. We also

excluded infants with a high risk of EOS having an undisputable indication for early antibiotics

administration (infants born after preterm labor, preterm premature rupture of membranes or

placental abruption, infants whose mother had clinical chorioamnionitis or received antibiotics

during the last 72 hours before birth or at birth). Preterm onset of labor and preterm rupture of

membranes are considered a clinical indicator of potential intrauterine infection which is

associated with a risk of EOS.(3) Very preterm infants born in these contexts were therefore

excluded. Clinical chorioamnionitis was defined as maternal fever (≥ 37.8°C) associated with

at least two of the following criteria: maternal leukocytosis (white blood cell count > 15,000

cells/mm3), maternal tachycardia (heart rate > 100 bpm), fetal tachycardia (heart rate > 160

bpm), uterine tenderness and foul-smelling vaginal discharge.(19) Therefore, very preterm

infants included in the present study were born after placental-mediated pregnancy

complications (hypertensive disorders during pregnancy or fetal growth restriction)(14), and

had a very low risk for EOS. Fetal growth restriction was defined by an estimated fetal weight

below the 10th percentile (according to the reference curves used by the hospital where the

antenatal ultrasound scans were performed), in conjunction with growth arrest and/or relevant

fetal Doppler abnormalities.

Finally, we excluded infants with missing data for early antibiotics treatment (n=151).

Maternal and neonatal characteristics of the study population and the population with missing

data for early antibiotics treatment were similar except for the rate of multiple pregnancy which

was significantly higher in case of missing data (Table 1 - Online).

6
 Main exposure and outcomes

Early antibiotic exposure was defined as antibiotics started at Day 0 or Day 1 of life,

regardless of the duration and type of antibiotics.

Neonatal outcomes included death, LOS, NEC, severe cerebral lesions, and moderate-to-

severe BPD during initial hospitalization in NICU. LOS was defined as a positive blood culture

after 72 hours of life, associated with antibiotic administration for five days or more, or death

within five days following positive blood culture.(20) NEC was defined by the Bell

classification of stage ≥ 2.(21) Severe cerebral lesions were defined as intraventricular

hemorrhage with ventricular dilatation or parenchymal hemorrhage,(22) or periventricular

leukomalacia (i.e., periventricular white matter echolucensies at ultrasonography). Moderate-

to-severe BPD was defined as oxygen requirement for at least 28 days and persistent need for

oxygen or ventilatory support at 36 weeks’ postmenstrual age (mechanical ventilation or

positive pressure).(23) Severe neonatal complication was a composite outcome including any

of the following: LOS; NEC; severe cerebral lesions; and/or moderate to severe BPD.

We defined the main outcome as a composite outcome accounting for the competing risk

between death and adverse outcomes, including at least one of the following: death, LOS and/or

NEC. We also considered the following composite outcomes: death or severe neonatal

complication, death or severe cerebral lesions, death or moderate-to-severe BPD.

Other characteristics considered

Gestational age at birth was determined as the best obstetrical estimate from early ultrasound

assessment and/or last menstrual period.(24) Initial severity of illness was assessed for all

infants using the Clinical Risk Index for Babies II (CRIB2) score (25). The CRIB2 score (range

0-27) is based on gestational age, birth weight, neonatal sex, temperature at NICU admission,

and initial base deficit. Higher scores of CRIB2 are predictive of higher mortality risk.(25)

Newborns were considered small-for-gestational-age if their birth weight was ≤ 10th percentile
7
according to French EPOPé intrauterine growth curves adjusted for fetal sex and gestational

age (26).

EOS was defined by positive blood or cerebrospinal fluid cultures before 72 hours of life.

Bacillus and coagulase-negative staphylococci grown on culture of blood <72h were considered

contaminants instead of EOS, they were found in 1 and 4 newborns, respectively.(2)

Maternity units were considered as type 3 when associated with a NICU. Inborn status was

defined as delivery in a type 3 maternity unit. The volume of the NICU was defined by the

number of infants included in the EPIPAGE 2 study in the unit, divided into terciles.

Statistical analysis

We first compared maternal, obstetrical, and neonatal characteristics by early antibiotic

treatment (yes vs no). Categorical variables were compared using χ2 or Fisher exact tests as

appropriate, and medians of quantitative variables by Wilcoxon’s test. All descriptive statistics

were weighted according to the duration of recruitment. Infants born at 22-26 weeks were

recruited over an eight-month period i.e. the equivalent of 35 weeks. Infants born at 27-31

weeks were recruited over a period of only six months i.e. the equivalent of 26 weeks. We

weighted the population of infants born at 27-31 weeks by a factor of 1.34 (35/26) in order to

have a population equivalent to what a 35-week recruitment period would have generated. This

weighting therefore allowed us to account for the sampling scheme of the cohort and to ensure

representativeness.(18) All tests were 2-sided; p<0.05 was considered statistically significant.

We used a propensity score analysis to minimize the indication bias in estimating the

antibiotic exposure effect.(27) The first step in the analysis consisted of estimating the

propensity score, defined as the newborn’s probability of receiving early antibiotic treatment

based on his/her individual observed characteristics at admission to NICU. The score was

estimated using a logistic regression model with the administration of early antibiotic treatment

as the dependent variable, regressed on baseline characteristics clinically associated with the

8
exposure and/or the outcome, excluding covariates that might be a consequence of the

treatment.(28) We included the following covariates: maternal age (<25, 25-34 or ≥35 years)

(29), maternal place of birth (30), overweight (body mass index (BMI) ≥ 25) (31), smoking

status during pregnancy (32), antenatal corticosteroids administration (at least one injection

administered before delivery), single or multiple pregnancy, inborn status, gestational age at

birth, birth weight (<1000, 1000-1499, or ≥1500 grams), neonatal sex, CRIB2 score, intubation

in the delivery room, surfactant administration in the delivery room and NICU characteristics

(level and volume of activity of the unit where the infant was hospitalized during the first 48

hours of life.).

The second step in the analysis involved inverse probability of treatment weighting (IPTW).

IPTW was based on estimated propensity scores, to obtain a synthetic sample (IPTW sample),

in which antibiotic treatment assignment was independent of measured baseline

covariates.(27,33,34) Balance in the observed baseline covariates between treated and untreated

patients in the initial and IPTW samples was assessed by examining standardized differences

(Figure 2 - Online). Finally, we estimated the association between early antibiotics exposure

and outcomes by a multilevel logistic regression within the weighted sample, with NICU as

level 2 to account for the wide variation in the proportion of newborns that received early

antibiotic treatment by center (Figure 3 - Online). We adjusted for gestational age because it

was considered a confounding factor (a common ancestor of both the exposure and outcomes),

and was not perfectly balanced in the propensity score (Figure 2 - Online). Results were

presented as ORs with a 95% confidence interval (95%CI).

We performed a sensitivity analysis after multiple imputations for missing data. The proportion

of missing data ranged from 0.0% to 8.9% for each covariate, and missing data were considered

missing at random. Multiple imputations were used including all baseline variables and

outcomes of the propensity-score model. Categorical variables were imputed using logistic or

9
multinomial regression as appropriate. We generated 50 independent imputed datasets. A

propensity score was estimated for each of the 50 imputed data sets generated, and results were

pooled for a final analysis according to Rubin’s rules.(35)

All statistical analyses were performed with STATA statistical software, version 16.1

(StataCorp).

RESULTS

We included 648 very preterm infants at low risk for EOS among which 173 (26.2%)

received early antibiotic treatment started at Day 0 or Day 1 of life (figure 1).

Maternal, obstetrical, and neonatal characteristics are presented in Table 2. In this sample of

pregnant women with hypertensive disorders and/or with fetal growth restriction, median

maternal age was 30 years (interquartile range [IQR] 26-35), 45.5% were overweight at the

beginning of pregnancy, 58.5% primipara and 20.3% smoked during pregnancy. Antenatal

corticosteroids were widely administered, almost all babies were delivered by cesarean section.

Infants exposed to early antibiotics were more frequently born from multiple pregnancies but

less frequently in a level 3 maternity unit. Other maternal and obstetrical characteristics did not

differ by early exposure to antibiotics.

The median gestational age at birth was 30.0 weeks (IQR 28.4-31.0) and the median birth

weight was 1040 grams (IQR=820-1260). Infants who received early antibiotic treatment had

a slightly lower median gestational age at birth (29.6 versus 30.3 weeks) and lower median birth

weight (1010 versus 1060 grams). Delivery room intubation and surfactant administration were

more frequent in the group of infants with early antibiotics. However, the severity of the initial

illness, measured by the CRIB2 score at admission in NICU did not differ significantly (Table

2).

10
 Among the very preterm infants who received early antibiotics, 88.1% had treatment

initiated at Day 0. Details about antibiotic treatment received were available for 165 patients.

All children received a combination of at least 2 molecules. Amoxicillin was used in 112/165

patients, third-generation cephalosporin in 129/165 patients, and aminoglycosides in 140/165

patients. Vancomycin was also used in combination in 11 patients, and metronidazole in 5. The

median duration of antibiotic administration was 3 days (IQR=2-6).

Four infants presented an episode of EOS diagnosed at the second or third day of life. None

had received antibiotic treatment within the first day of life. Identified bacteria were

Streptococcus agalactiae in one case and Staphylococcus aureus in the three others (including

one child who died on Day 3).

Neonatal outcomes and their association with early exposure to antibiotics are described in

table 3. After IPTW, standardized differences between treated and untreated patients were

below 0.10 for all covariates, except for a few gestational age groups (Figure 2 - Online).

Mortality was more frequent in the early exposure to antibiotics group (12.0% versus 6.3%).

The frequency of NEC was similar in both groups (5.9% in the early antibiotics group vs 4.9%).

In our population, 25.3% presented at least one episode of LOS at a median age of 12 days

(IQR=7-20), with no difference between the two groups. There were 165 episodes of LOS

including 41 in the group with early exposure to antibiotics and 124 in the group without

exposure to antibiotics. The most common pathogens found in blood cultures collected during

LOS were coagulase-negative staphylococci (75.5%), Staphylococcus aureus (15.7%),

Enterococcus sp. (3.1%), and Gram-negative organisms (2.6%), with no difference between the

two groups. We did not identify any episode of LOS due to a bacterium resistant to the initial

early antibiotics administered. Late-onset sepsis characteristics did not differ according to early

neonatal antibiotics (Table 4 – online). In the complete cases analyses, we did not find any

association between early exposure to antibiotics and death (OR=1.78 [95% CI=0.82-3.88]),

11
LOS (OR=0.93 [95%CI=0.64-1.34]) or NEC (OR=1.47 [95%CI=0.65-3.33]). Results were

consistent after multiple imputations. The composite outcome (death or LOS or NEC) did not

significantly differ between the two groups (35.8% vs 32.5%) and we did not find an association

between early exposure to antibiotics and this composite outcome (OR=1.04 [95%CI=0.72-

1.50]).

However, severe cerebral lesions, isolated or combined with death, were more frequent when

newborns received early antibiotics, (9.1% vs 4.5%, and 16.6% vs 9.9%, respectively). We

found an association between severe cerebral lesions and early exposure to antibiotics

(OR=2.71 [95%CI=1.25-5.86]). The trend was the same, but at the limit of significance

(OR=2.99 [95%CI=0.99-9.04]), after multiple imputations. The associations were less

pronounced for the composite outcome death or severe cerebral lesions (Table 3).

Moderate-to-severe BPD was also more frequent in the early exposure to antibiotics group,

whether isolated or associated with death (19.7% vs 10.2% and 28.6% vs 15.4%, respectively).

We found an association between moderate-to-severe BPD and early exposure to antibiotics

(OR=2.30 [95%CI=1.21-4.38]). The trend was the same, but at the limit of significance

(OR=1.87 [95%CI=0.96-3.65]), after multiple imputations. Results were similar for the

composite outcome death or moderate-to-severe BPD (Table 3).

Finally, we studied the composite outcome “severe neonatal complication” (which included

any of the following: LOS; NEC; severe cerebral lesions; and/or moderate to severe BPD), and

did not find an association with early exposure to antibiotics (OR=1.37 [95%CI=0.94-1.98]).

In a sensitivity analysis after multiple imputations for missing data, results were consistent

for the composite outcome (death or LOS or NEC) (Table 5). For severe cerebral lesions and

severe BPD, the trend was the same than complete cases analysis, but at the limit of significance

(OR=2.99 [95%CI=0.99-9.04] and OR=1.87 [95%CI=0.96-3.65], respectively).

12
 DISCUSSION

Among the 648 very preterm infants at low risk for EOS, 26% were exposed to antibiotics

during the first day of life. This early antibiotic exposure was not associated with death, LOS

and/or NEC, but was associated with a higher risk of severe cerebral lesions and moderate-to-

severe BPD.

The strengths of our study included the use of data from a national prospective population-

based cohort study. Data about antibiotic exposure and neonatal outcomes were prospectively

collected using a detailed and standardized questionnaire. We had a relatively small sample,

but our study provided a unique opportunity to evaluate the impact of antibiotic prescription on

neonatal outcomes, independently of indication biases: firstly, we chose to study a

homogeneous sample of very preterm infants at low risk of EOS. We therefore excluded those

born following: spontaneous preterm labor, rupture of membranes, maternal antibiotic

treatment during the last 72 hours before birth, placental abruption or clinical chorioamnionitis.

These infants had none of the EOS risk factors that would require empirical antibiotic treatment

at birth, and only four (in the group without initial exposition to antibiotics) were finally

diagnosed with EOS. Among them, three infants were infected with S. aureus. In one case, the

infection started on day 2, and in the two other cases on day 3. In this context of preterm birth

with cesarean section, vertical transmission is very unlikely and these 3 infections were

probably hospital-acquired infection.

Altogether, this empirical administration of antibiotics that involved 26% of our population was

probably started out of standard protocols, with wide variations between NICUs. This rate was

consistent with previous reports and should be addressed with specific antibiotic stewardship

interventions.(2,11) Then, we controlled for the initial severity of illness including the CRIB2

score in the propensity score, a validated predictor of mortality in this population.(25) We also

added intubation and surfactant administration in the delivery room to account for clinical

13
severity at birth. The initial clinical severity is a confounding factor, associated with both early

antibiotics treatment (empirical treatment of possible sepsis) and neonatal adverse outcomes.(2)

Finally, to limit the effect of the indication bias, we used a propensity score, based on variables

available at the time of decision to initiate or not an early antibiotic treatment. Using IPTW

enabled us to estimate the average treatment effect on the whole sample.(34) In contrast, using

the matching on propensity score, patients for whom no match existed are excluded, which

results in a significant loss of information that ultimately leads to a selection bias, in particular

in this population of infants with low risk of EOS at baseline. It provided a way to balance

measured covariates across exposed and non-exposed groups independently of the clinical

status at admission in NICU that may have influenced the decision to administrate early

antibiotics treatment.

Limitations should be kept in mind. We had a limited sample size due to our choice of a low

risk of EOS study population. As a part of this epidemiological study, we could not analyze the

gut microbiota nor the impact of early antibiotic exposure on subsequent colonization with

multi-drug resistant organisms.

In our study, early antibiotic exposure was not associated with a higher risk of death, LOS,

and/or NEC. Several studies found an association between early empirical exposition to

antibiotics and these outcomes, isolated or combined.(9–11) However, these studies have

included preterm infants alive at day 5 (9), at day 7 (11) or day 14 (10), to have sufficient time

to be exposed to antibiotics and focused on consequences of duration of antibiotics treatment.

We chose to study mortality and morbidities from the first day of life to obtain a complete

assessment of the putative consequences of early antibiotics exposure. In a subgroup of very

low birthweight infants with low risk of EOS, Ting et al found an increased risk of a composite

outcome (intraventricular hemorrhage, NEC, BPD, severe retinopathy of prematurity, hospital-

acquired infections, or death) in case of long antibiotic exposure (4-7 days) versus none

14
(aOR=1.51 [95%CI=1.22-1.87]) but no statistically significant difference in case of short

antibiotic exposure (1-3 days) (aOR=1.03 [95%CI=0.84-1.26]). (11) More recently, Li et al

found a lower incidence of NEC in case of early antibiotic administration in an international

cohort of 2831 preterm very low birthweight infants (aOR=0.25 [95%CI=0.12-0.47]).(16) In

this study, almost 90% of infants received antibiotics within the first three days after birth.

Moreover, in the small comparison group that did not receive early antibiotics treatment

(269/2831), baseline characteristics differed markedly such as higher gestational age, higher

rate of cesarean delivery, and higher proportion of small-for-gestational-age infants. The

relatively high NEC incidence in this control group could lead to a potential confounding bias

and requires caution in interpretation.

In our study, early antibiotic exposure was associated with a higher risk of moderate-to-severe

BPD and severe cerebral lesions, despite adjustment for the severity of initial illness. Our

findings were similar to those reported in some studies.(11,12) Puopolo et al find also similar

results: among 5334 low-EOS risk extremely preterm infants, those empirically treated with

early prolonged antibiotics in the absence of culture-confirmed infection had a higher adjusted

risk of severe cerebral lesions and higher adjusted risk of BPD compared to the infants who

were not given early prolonged antibiotics (adjusted for multiple predictors of illness

severity).(2)

We cannot totally rule out a lack of power to explain our observed lack of associations between

early antibiotics and death, LOS, and/or NEC. Nevertheless, concerning the association

between early antibiotics and severe cerebral lesions and bronchopulmonary dysplasia that we

have found, the pathophysiological mechanisms remain to be understood. Consequences of

early antibiotics should be different according to the type of antibiotics used or the number of

courses during the hospitalization for example. Studying the consequences of early antibiotic

treatment in very preterm infants is complex, and requires assessing not only neonatal mortality

15
and morbidity but also effects on microbiota and later health. Indeed, postnatal exposure to

antibiotics has been shown to impact the diversity of the gut microbiota.(4) Other concerns also

include a potential impact on host immunity and subsequent auto-inflammatory disorders,

according to the suggested consequence of microbiota on inflammation and in the pathogenesis

of a wide variety of inflammatory disease processes.(36) Tracheal microbiota seems also to be

different between infants who develop bronchopulmonary dysplasia and those who do not and

postnatal exposition to antibiotics could be involved in these differences.(36)

In order to reach a more robust conclusion on the consequences of early antibiotics in preterm

infants, a randomized controlled trial would be optimal. This study should include very preterm

infants, born between 26 and 31 weeks, without risk factor for early-onset sepsis and the

primary outcome will be the composite outcome of LOS, NEC, severe cerebral lesions and

BPD or death.

CONCLUSION

Early exposure to antibiotics was not associated with an increased risk of a composite

outcome including death, NEC and/or LOS in very preterm infants at low risk for EOS.

However, our results question whether exposure to early antibiotics is warranted since it was

associated with a higher risk of severe cerebral lesions and moderate-to-severe BPD.

16
 Abbreviations

BPD bronchopulmonary dysplasia, EOS Early-onset sepsis, IPTW inverse probability of

treatment weighting, LOS Late-onset sepsis, NEC necrotizing enterocolitis, NICU Neonatal

intensive care unit

CONTRIBUTORS

All authors approved the final manuscript as submitted and agree to be accountable for all

aspects of the work.

The corresponding author had full access to all the data in the study and had final

responsibility for the decision to submit for publication.

AKNOWLEDGMENT

We are grateful for the participation of all families of preterm infants in the EPIPAGE-2 cohort

study and for cooperation of all maternity and neonatal units in France.

17
 REFERENCES

1. Stoll BJ, Puopolo KM, Hansen NI, Sánchez PJ, Bell EF, Carlo WA, et al. Early-Onset

Neonatal Sepsis 2015 to 2017, the Rise of Escherichia coli, and the Need for Novel Prevention

Strategies. JAMA Pediatr. 2020;174:e200593.

2. Puopolo KM, Mukhopadhyay S, Hansen NI, Cotten CM, Stoll BJ, Sanchez PJ, et al.

Identification of Extremely Premature Infants at Low Risk for Early-Onset Sepsis. Pediatrics.

2017;140:e20170925.

3. Benitz WE, Gould JB, Druzin ML. Risk factors for early-onset group B streptococcal

sepsis: estimation of odds ratios by critical literature review. Pediatrics. 1999;103:e77.

4. Arboleya S, Sánchez B, Milani C, Duranti S, Solís G, Fernández N, et al. Intestinal

microbiota development in preterm neonates and effect of perinatal antibiotics. J Pediatr.

2015;166:538–44.

5. de Man P, Verhoeven BA, Verbrugh HA, Vos MC, van den Anker JN. An antibiotic

policy to prevent emergence of resistant bacilli. Lancet. 2000;355:973–8.

6. Bailey LC, Forrest CB, Zhang P, Richards TM, Livshits A, DeRusso PA. Association

of antibiotics in infancy with early childhood obesity. JAMA Pediatr. 2014;168:1063–9.

7. Hoskin-Parr L, Teyhan A, Blocker A, Henderson AJW. Antibiotic exposure in the first

two years of life and development of asthma and other allergic diseases by 7.5 yr: a dose-

dependent relationship. Pediatr Allergy Immunol. 2013;24:762–71.

8. Kronman MP, Zaoutis TE, Haynes K, Feng R, Coffin SE. Antibiotic exposure and IBD

development among children: a population-based cohort study. Pediatrics. 2012;130:e794-803.

9. Cotten CM, Taylor S, Stoll B, Goldberg RN, Hansen NI, Sánchez PJ, et al. Prolonged

duration of initial empirical antibiotic treatment is associated with increased rates of necrotizing

enterocolitis and death for extremely low birth weight infants. Pediatrics. 2009;123:58–66.

10. Cantey JB, Pyle AK, Wozniak PS, Hynan LS, Sánchez PJ. Early Antibiotic Exposure

18
and Adverse Outcomes in Preterm, Very Low Birth Weight Infants. J Pediatr. 2018;203:62–7.

11. Ting JY, Roberts A, Sherlock R, Ojah C, Cieslak Z, Dunn M, et al. Duration of Initial

Empirical Antibiotic Therapy and Outcomes in Very Low Birth Weight Infants. Pediatrics.

2019;143:e20182286.

12. Cantey JB, Huffman LW, Subramanian A, Marshall AS, Ballard AR, Lefevre C, et al.

Antibiotic Exposure and Risk for Death or Bronchopulmonary Dysplasia in Very Low Birth

Weight Infants. J Pediatr. 2017;181:289-293.e1.

13. Esaiassen E, Fjalstad JW, Juvet LK, van den Anker JN, Klingenberg C. Antibiotic

exposure in neonates and early adverse outcomes: a systematic review and meta-analysis. J

Antimicrob Chemother. 2017;72:1858–70.

14. Delorme P, Goffinet F, Ancel P-Y, Foix-LʼHélias L, Langer B, Lebeaux C, et al. Cause

of Preterm Birth as a Prognostic Factor for Mortality. Obstet Gynecol. 2016;127:40–8.

15. Letouzey M, Foix-L’Hélias L, Torchin H, Mitha A, Morgan AS, Zeitlin J, et al. Cause

of preterm birth and late-onset sepsis in very preterm infants: the EPIPAGE-2 cohort study.

Pediatr Res. 2021; 24:1–9.

16. Li Y, Shen RL, Ayede AI, Berrington J, Bloomfield FH, Busari OO, et al. Early Use of

Antibiotics Is Associated with a Lower Incidence of Necrotizing Enterocolitis in Preterm, Very

Low Birth Weight Infants: The NEOMUNE-NeoNutriNet Cohort Study. J Pediatr.

2020;227:128-134.e2.

17. Lorthe E, Benhammou V, Marchand-Martin L, Pierrat V, Lebeaux C, Durox M, et al.

Cohort Profile: The Etude Epidémiologique sur les Petits Ages Gestationnels-2 (EPIPAGE-2)

preterm birth cohort. Int J Epidemiol. 2021:1-15

18. Ancel P-Y, Goffinet F, EPIPAGE 2 Writing Group. EPIPAGE 2: a preterm birth cohort

in France in 2011. BMC Pediatrics. 2014;14:97.

19. Gibbs RS. Diagnosis of intra-amniotic infection. Semin Perinatol. 1977;1:71–7.

19
20. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA, et al. Late-

onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research

Network. Pediatrics. 2002;110:285–91.

21. Bell MJ, Ternberg JL, Feigin RD, Keating JP, Marshall R, Barton L, et al. Neonatal

necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg.

1978;187:1–7.

22. Volpe JJ. Brain injury in premature infants: a complex amalgam of destructive and

developmental disturbances. Lancet Neurol. 2009;8:110–24.

23. Jobe AH, Bancalari E. Bronchopulmonary dysplasia. Am J Respir Crit Care Med.

2001;163:1723–9.

24. Ancel P-Y, Goffinet F, EPIPAGE-2 Writing Group. Survival and morbidity of preterm

children born at 22 through 34 weeks’ gestation in France in 2011: results of the EPIPAGE-2

cohort study. JAMA Pediatr. 2015;169:230–8.

25. Parry G, Tucker J, Tarnow-Mordi W, UK Neonatal Staffing Study Collaborative Group.

CRIB II: an update of the clinical risk index for babies score. Lancet. 2003;361:1789–91.

26. Ego A, Prunet C, Lebreton E, Blondel B, Kaminski M, Goffinet F, et al. [Customized

and non-customized French intrauterine growth curves. I - Methodology]. J Gynecol Obstet

Biol Reprod. 2016;45:155–64.

27. Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of

Confounding in Observational Studies. Multivar Behav Res. 2011;46:399–424.

28. Garrido MM, Kelley AS, Paris J, Roza K, Meier DE, Morrison RS, et al. Methods for

Constructing and Assessing Propensity Scores. Health Serv Res. 2014;49:1701–20.

29. Klemetti R, Gissler M, Sainio S, Hemminki E. At what age does the risk for adverse

maternal and infant outcomes increase? Nationwide register-based study on first births in

Finland in 2005-2014. Acta Obstet Gynecol Scand. 2016;95:1368–75.

20
30. Howell EA, Janevic T, Hebert PL, Egorova NN, Balbierz A, Zeitlin J. Differences in

Morbidity and Mortality Rates in Black, White, and Hispanic Very Preterm Infants Among

New York City Hospitals. JAMA Pediatr. 2018;172:269–77.

31. Declercq E, MacDorman M, Cabral H, Stotland N. Prepregnancy Body Mass Index and

Infant Mortality in 38 U.S. States, 2012-2013. Obstet Gynecol. 2016;127:279–87.

32. Dietz PM, England LJ, Shapiro-Mendoza CK, Tong VT, Farr SL, Callaghan WM. Infant

morbidity and mortality attributable to prenatal smoking in the U.S. Am J Prev Med.

2010;39:45–52.

33. Thomas L, Li F, Pencina M. Using Propensity Score Methods to Create Target

Populations in Observational Clinical Research. JAMA. 2020;323:466–7.

34. Austin PC, Stuart EA. Moving towards best practice when using inverse probability of

treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in

observational studies. Stat Med. 2015;34:3661–79.

35. Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and

some applications. Stat Med. 1991;10:585–98.

36. Underwood MA, Sohn K. The Microbiota of the Extremely Preterm Infant. Clin

Perinatol. 2017;44:407–27.

21
 FIGURE LEGENDS

Figure 1: Flow chart

Figure 2 (online): Standardized differences in the initial and IPTW samples in complete-

cases (A) and after multiple imputation (B)

Each covariate included in the propensity score is on the y axis.

Figure 3 (online): Variations between NICUs in the proportion of very preterm infants at

low risk for early-onset sepsis treated with early antibiotics: number of infants admitted to each

NICU (blue bars, from 1 to 24) and proportion (%) of infants treated with early antibiotics

(orange solid line, from 0 to 100%).

22
Table 1, online only

Table 1 (online): Comparison of maternal and obstetrical characteristics between the

study population and the population with missing data for early neonatal antibiotics

Total Early neonatal antibiotics p

N=799
Complete cases Missing data
N=648 (80.9%) N=151 (19.1%)
Maternal characteristics
Mother's age (N=799) median (IQR) 30 (26-35) 30 (26-35) 30 (26-33) 0.95
Body mass index ≥ 25 (N=740) 343 (46.3) 274 (45.5) 69 (49.6) 0.39
Mother's place of birth (N=790) 0.07
Europe 595 (75.3) 474 (73.9) 121 (81.2)
Northern Africa 70 (8.9) 55 (8.6) 15 (10.2)
Subsaharian Africa 77 (9.9) 69 (10.9) 8 (5.4)
Other 48 (5.9) 43 (6.6) 5 (3.2)
Primiparous (N=791) 467 (59.0) 375 (58.5) 92 (61.0) 0.19
Smoking during pregnancy (N=767) 151 (19.8) 125 (20.3) 26 (17.8) 0.50
Multiple pregnancy (N=799) 131 (16.5) 83 (12.8) 48 (32.0) <0.001
Antenatal corticosteroids (N=799) 702 (87.7) 571 (88.3) 131 (85.6) 0.38
Cesarean section (N=799) 801 (99.4) 647 (99.4) 154 (99.4) 0.98
Neonatal characteristics
Gestational age (weeks) median (IQR)
30 (28-31) 30 (28-31) 30 (28-31) 0.93
(N=799)
Male sex (N=799) 375 (47.2) 304 (47.1) 71 (47.3) 0.96
Birth weight (grams) median (IQR)
1040 (830-1255) 1040 (820-1260) 1050 (880-1250) 0.79
(N=799)
CRIB2 score (N=678) 6 (4-9) 6 (4-9) 6 (5-8) 0.91
Intubation in delivery room (N=775) 396 (50.1) 329 (51.5) 67 (44.4) 0.12
Early-onset sepsis (N=798) 5 (0.6) 4 (0.6) 1 (0.7) 0.96
Death (N=799) 66 (7.6) 55 (7.8) 11 (6.8) 0.62
Late-onset sepsis (N=783) 199 (25.0) 165 (25.4) 34 (23.2) 0.59
Necrotizing enterocolitis (N=786) 38 (4.8) 33 (5.2) 5 (3.5) 0.40
Severe cerebral lesions (N=784) 44 (5.4) 38 (5.7) 6 (4.1) 0.43
Bronchopulmonary dysplasia (N=719) 99 (13.0) 78 (12.6) 21 (15.2) 0.41

IQR interquartile range

Table 2

Table 2: Maternal, obstetrical, and neonatal characteristics according to early neonatal

antibiotics administration in very preterm infants

Total Early neonatal antibiotics p

N=648 No Yes
N=475 (73.8%) N=173 (26.2%)
Maternal characteristics
Mother's age (N=648) 0.50
< 25 years 117 (18.1) 87 (18.4) 30 (17.2)
25-34 years 364 (56.0) 262 (55.0) 102 (58.9)
≥ 35 years 167 (25.9) 126 (26.5) 41 (24.0)
BMI ≥ 25 (N=601) 274 (45.5) 206 (46.6) 68 (42.5) 0.38
Mother's place of birth (N=641) 0.58
Europe 474 (73.9) 346 (73.8) 128 (74.3)
Northern Africa 55 (8.6) 41 (8.7) 14 (8.3)
Subsaharian Africa 69 (10.9) 54 (11.6) 15 (8.9)
Other 43 (6.6) 28 (5.9) 15 (8.5)
Primiparous (N=640) 375 (58.5) 278 (59.1) 97 (56.9) 0.74
Smoking during pregnancy (N=621) 125 (20.3) 91 (19.9) 34 (21.6) 0.65
Multiple pregnancy (N=648) 87 (12.9) 51 (10.9) 32 (17.5) 0.006
Antenatal corticosteroids* (N=639) 561 (88.1) 419 (89.3) 142 (84.4) 0.09
Cesarean section (N=641) 637 (99.4) 466 (99.6) 171 (98.8) 0.29
Inborn** (N=648) 600 (92.7) 449 (94.6) 151 (87.4) 0.002
Neonatal characteristics
Gestational age (weeks) (N=648) 0.09
25-27 127 (17.6) 84 (16.1) 43 (21.6)
28-29 196 (31.0) 140 (30.0) 56 (33.8)
30-31 325 (51.4) 251 (53.8) 74 (44.6)
Male sex (N=648) 304 (47.1) 216 (45.6) 88 (51.5) 0.18
Birth weight (grams) (N=648) 0.14
<1000 286 (42.7) 199 (40.8) 87 (48.1)
1000-1499 316 (50.0) 244 (52.3) 72 (43.4)
≥1500 46 (7.3) 32 (6.9) 14 (8.4)
SGA (<10th) (N=648) 521 (80.2) 386 (81.1) 135 (77.7) 0.34
Apgar score <5 at 5’ (N=625) 54 (8.5) 34 (7.2) 20 (12.4) 0.04
Intubation in delivery room (N=626) 329 (51.5) 213 (45.8) 116 (67.3) <0.001
Surfactant in delivery room (N=635) 195 (29.9) 132 (27.7) 63 (36.1) 0.04
Chest compressions in delivery room
28 (4.3) 18 (3.8) 10 (5.7) 0.29
(N=617)
CRIB2 (N=555) 6 (4-9) 6 (4-9) 7 (4-9) 0.18
Early-onset sepsis (N=648) 4 (0.6) 4 (0.9) 0 (0.0) 0.23
Neonatal unit characteristics ***
Level 3 (N=648) 629 (97.0) 463 (97.4) 166 (95.8) 0.29
Volume of activity (N=648) 0.13
Low 203 (31.6) 145 (30.7) 58 (34.1)
Medium 234 (35.9) 165 (34.5) 69 (39.8)
High 211 (32.6) 165 (34.8) 46 (26.2)
 Data are presented as number (percentage) of patients. Percentages are weighted according to the

differences between length of recruitment of gestational age groups.

For each variable, percentages might not sum up to 100%, owing to rounding.

IQR interquartile range; SGA small-for-gestational-age; CRIB Clinical risk index for babies score

* Antenatal corticosteroids were defined as at least one injection of corticosteroids before delivery

** Inborn was defined as birth in a hospital with a neonatal intensive care unit

*** Characteristics of the unit in which the newborn was hospitalized during the first 48 hours of

life.
Table 3

Table 3: Neonatal outcomes according to early neonatal antibiotics

Total Early neonatal antibiotics p Complete-cases analysis

(N=465)
aOR (95%CI)
N=648 No Yes
N=475 (73.8) N=173 (26.2)
Death (N=648) 55 (7.8) 32 (6.3) 23 (12.0) 0.01 1.78 (0.82-3.88)
LOS (N=641) 165 (25.4) 124 (26.0) 41 (23.3) 0.35 0.93 (0.64-1.34)
NEC (N=641) 33 (5.2) 23 (4.9) 10 (5.9) 0.62 1.47 (0.65-3.33)
Severe cerebral lesions (N=635) 38 (5.7) 21 (4.5) 17 (9.1) 0.02 2.71 (1.25-5.86)
Moderate to severe BPD (N=585) 78 (12.6) 47 (10.2) 31 (19.7) <0.01 2.30 (1.21-4.38)
Composite outcomes
Severe neonatal complication* (N=604) 252 (40.8) 174 (38.6) 78 (46.9) 0.07 1.37 (0.94-1.98)
Death or severe neonatal complication 1.33 (0.92-1.93)
274 (41.3) 188 (38.9) 86 (48.0) 0.04
(N=647)
Death or severe cerebral lesions (N=640) 79 (11.7) 48 (9.9) 31 (16.6) 0.02 2.16 (1.21-3.85)
Death or moderate to severe BPD (N=634) 128 (18.9) 76 (15.4) 52 (28.6) <0.001 2.25 (1.31-3.87)
Death or LOS or NEC (N=639) 218 (33.3) 154 (32.5) 64 (35.8) 0.43 1.04 (0.72-1.50)

Data are presented as number (percentage) of patients. Percentages are weighted according to the differences between length of recruitment of gestational

age groups.

For each variable, percentages might not sum up to 100%, owing to rounding.

LOS Late-onset sepsis; NEC necrotizing enterocolitis; BPD bronchopulmonary dysplasia; aOR adjusted odd ratio

Adjusted odds ratio (aOR) obtained from multilevel analysis with inverse probability of treatment weighting, adjusted for gestational age.
 *Severe neonatal complication was defined as any of the following outcomes: late-onset sepsis; severe cerebral lesions (intraventricular hemorrhage with

ventricular dilatation, parenchymal hemorrhage, and/or cystic periventricular leukomalacia); Bell’s stage ≥ 2 necrotizing enterocolitis; and/or moderate-to-severe

bronchopulmonary dysplasia (BPD), defined as administration of oxygen for at least 28 days plus need for 21% or more oxygen and/or mechanical ventilatory

support or continuous positive airway pressure at 36 weeks’ postmenstrual age.

Table 4 online only

Table 4 (online): Late-onset sepsis characteristics according to early neonatal antibiotics

Total Early neonatal antibiotics p

N=648 No Yes
N=475 (73.8) N=173 (26.2)
LOS (N=641) 0.35
0 476 (74.6) 347 (74.0) 129 (76.7)
1 episode 141 (21.7) 109 (22.9) 32 (18.6)
2 or 3 episodes 24 (3.6) 15 (3.2) 9 (4.7)
Age at 1st episode of LOS (days) median
12 (7-20) 11 (7-18) 12 (8-20) 0.86
(IQR) (N=164)
Pathogens implicated in LOS 0.75
coagulase-negative staphylococci 120 (75.5) 90 (75.6) 30 (75.0)
Staphylococcus aureus 25 (15.7) 20 (16.6) 5 (12.9)
Enterococcus sp. 5 (3.1) 3 (2.6) 2 (4.7)
Gram-negative organisms 4 (2.6) 2 (1.7) 2 (5.4)
Others 4 (2.5) 3 (2.6) 1 (2.0)

Data are presented as number (percentage) of patients unless otherwise indicated. Percentages are

weighted according to the differences between length of recruitment of gestational age groups.

For each variable, percentages might not sum up to 100%, owing to rounding.

LOS Late-onset sepsis; IQR interquartile range;

Table 5

Table 5: Sensitivity analysis: Association between neonatal outcomes and early neonatal
antibiotics after multiple imputation.

Complete-cases analysis Multiple imputation

(N=465) (N=648)
aOR (95%CI) aOR (95%CI)
Death 1.78 (0.82-3.88) 1.51 (0.57-3.99)
LOS 0.93 (0.64-1.34) 1.12 (0.62-2.03)
NEC 1.47 (0.65-3.33) *
Severe cerebral lesions 2.71 (1.25-5.86) 2.99 (0.99-9.04)
Moderate-to-severe BPD 2.30 (1.21-4.38) 1.87 (0.96-3.65)
Composite outcomes
Any severe neonatal 1.37 (0.94-1.98) 1.54 (0.87-2.72)
complication
Death or severe neonatal 1.33 (0.92-1.93) 1.53 (0.85-2.73)
complication
Death or severe cerebral 2.16 (1.21-3.85) 1.85 (0.82-4.18)
lesions
Death or moderate-to- 2.25 (1.31-3.87) 1.76 (0.91-3.39)
severe BPD
Death or LOS or NEC 1.04 (0.72-1.50) 1.28 (0.72-2.28)

Adjusted odds ratio (aOR) obtained from multilevel analysis with inverse probability of treatment

weighting, adjusted for gestational age.

OR Odds ratio; LOS late-onset sepsis, NEC necrotizing enterocolitis, BPD bronchopulmonary

dysplasia. Severe neonatal complication was defined as any of the following outcomes: late-onset sepsis;

severe cerebral lesions (intraventricular hemorrhage with ventricular dilatation, parenchymal

hemorrhage, and/or cystic periventricular leukomalacia); Bell’s stage ≥ 2 necrotizing enterocolitis;

and/or moderate-to-severe BPD

* Convergence not achieve due to the low number of cases of NEC.

Figure 1 Click here to access/download;Figure;Figure 1.jpeg
Figure 2 online only Click here to access/download;Figure;Figure 2 online only.docx

Figure 2 (online): Standardized differences in the initial and IPTW samples in complete-

cases (A) and after multiple imputation (B)

Each covariate included in the propensity score is on the y axis.

B
Figure 3 online only Click here to access/download;Figure;figure 3; online only.jpg
Online Appendix

Online appendix: Members of the EPIPAGE-2 Infectious diseases writing group and of the
EPIPAGE-2 Study Group

Members of the EPIPAGE-2 Infectious diseases writing group: Pascal Boileau, MD, PhD
(Department of Neonatal Pediatrics, Poissy Saint Germain Hospital, France, UVSQ, France),
Marine Butin (Department of neonatal pediatrics, Hospices civils de Lyon, Hôpital Femme
Mère Enfant, Lyon, France), Laurence Foix-L’Hélias, MD, PhD (Inserm UMR 1153,
Obstetrical, Perinatal and Pediatric Epidemiology Research Team (Epopé), Center for
Epidemiology and StatisticS ; Department of Neonatal Pediatrics, Trousseau Hospital, AP-
HP, Sorbonne University Paris, France), Christèle Gras-Le Guen, MD PhD (Department of
Pediatrics, Pediatric Emergency, Clinical investigation Center 1413 INSERM1043, CHU
Nantes, Nantes, France), Gilles Kayem, MD, PhD (Inserm UMR 1153, Obstetrical, Perinatal
and Pediatric Epidemiology Research Team (Epopé), Center for Epidemiology and StatisticS,
Paris France, Department of Obstetrics and Gynecology, Trousseau Hospital, AP-HP,
Sorbonne University, Paris, France), Pierre Kuhn, MD, PhD (Department of neonatal
medicine, CHU Strasbourg, Strasbourg, France), Mathilde Letouzey, MD, MSc (Inserm
UMR 1153, Obstetrical, Perinatal and Pediatric Epidemiology Research Team (Epopé),
Center for Epidemiology and StatisticS, Paris, France ; Department of Neonatal Pediatrics,
Poissy Saint Germain Hospital, France), Elsa Lorthe, RM, PhD (Inserm UMR 1153,
Obstetrical, Perinatal and Pediatric Epidemiology Research Team (Epopé), Center for
Epidemiology and StatisticS), Emeline Maisonneuve, MD, MSc (Department of Obstetrics
and Gynecology, Trousseau Hospital, APHP, Paris, France), Ayoub Mitha, MD PhD,
(Inserm UMR 1153, Obstetrical, Perinatal and Pediatric Epidemiology Research Team
(Epopé), Center for Epidemiology and StatisticS ; Department of neonatal medicine, Jeanne
de Flandres Hospital, CHRU Lille, Lille, France), Jeanne Sibiude (Department of
gynecology and obstetrics, Hôpital Louis Mourier, APHP, Colombes, France), Héloïse
Torchin, MD, PhD (Inserm UMR 1153, Obstetrical, Perinatal and Pediatric Epidemiology
Research Team (Epopé), Center for Epidemiology and StatisticS, Neonatal Medicine and
Resuscitation Service in Port-Royal, Cochin Hospital, APHP, Paris, France).
All the collaborators of the EPIPAGE-2 Infectious diseases writing group have no conflict
of interest or compensation in relation with this article to disclose. All of them consented to
such acknowledgment.

The authors acknowledge the collaborators of the EPIPAGE-2 Study Group: Alsace: D Astruc,
P Kuhn, B Langer, J Matis, C Ramousset; Aquitaine: X Hernandorena, P Chabanier, L Joly-
Pedespan, MJ Costedoat, A Leguen; Auvergne: B Lecomte, D Lemery, F Vendittelli; Basse-
Normandie: G Beucher, M Dreyfus, B Guillois, Y Toure; Bourgogne: A Burguet, S Couvreur,
JB Gouyon, P Sagot, N Colas; Bretagne: J Sizun, A Beuchée, P Pladys, F Rouget, RP Dupuy,
D Soupre, F Charlot, S Roudaut; Centre: A Favreau , E Saliba, L Reboul; Champagne-
Ardenne: N Bednarek, P Morville, V Verrière; Franche-Comté: G Thiriez, C Balamou;
Haute-Normandie: L Marpeau, S Marret, C Barbier; Ile-de-France: G Kayem, X Durrmeyer,
M Granier, M Ayoubi, O Baud, B Carbonne, L Foix L’Hélias, F Goffinet, PH Jarreau, D
Mitanchez, P Boileau, C Duffaut, E Lorthe, L Cornu, R Moras; Languedoc-Roussillon: P
Boulot, G Cambonie, H Daudé, A Badessi, N Tsaoussis; Limousin: A Bédu, F Mons, C
Bahans; Lorraine: MH Binet, J Fresson, JM Hascoët, A Milton, O Morel, R Vieux, L Hilpert;
Midi-Pyrénées: C Alberge, C Arnaud, C Vayssière, M Baron; Nord-Pas-de-Calais: ML
Charkaluk, V Pierrat, D Subtil, P Truffert, S Akowanou; PACA et Corse: C D’Ercole, C Gire,
U Simeoni, A Bongain, M Deschamps; Pays de Loire: B Branger, JC Rozé, N Winer, V
Rouger, C Dupont; Picardie: J Gondry, G Krim, B Baby; Rhône-Alpes: M Debeir, O Claris,
JC Picaud, S Rubio-Gurung, C Cans, A Ego, T Debillon, H Patural, A Rannaud; Guadeloupe:
E Janky, A Poulichet, JM Rosenthal, E Coliné; Guyane: A Favre, N Joly; Martinique: S
Châlons, J Pignol, PL Laurence; La Réunion : PY Robillard, S Samperiz, D Ramful.
Inserm UMR 1153: PY Ancel, V Benhammou, B Blondel, M Bonet, A Brinis, A Coquelin,
M Durox, M Kaminski, K Khemache, B Khoshnood, C Lebeaux, L Marchand-Martin, J
Rousseau, MJ Saurel-Cubizolles, D Tran, J Zeitlin.
Reporting guideline checklist

STROBE Statement—Checklist of items that should be included in reports of cohort studies

Item Page
No Recommendation No
Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the 3
abstract
(b) Provide in the abstract an informative and balanced summary of what was 3
done and what was found
Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being 4
reported
Objectives 3 State specific objectives, including any prespecified hypotheses 5

Methods
Study design 4 Present key elements of study design early in the paper 5
Setting 5 Describe the setting, locations, and relevant dates, including periods of 5
recruitment, exposure, follow-up, and data collection
Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of 5-6
participants. Describe methods of follow-up
(b) For matched studies, give matching criteria and number of exposed and
unexposed
Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and 6-7
effect modifiers. Give diagnostic criteria, if applicable
Data sources/ 8* For each variable of interest, give sources of data and details of methods of 6-7
measurement assessment (measurement). Describe comparability of assessment methods if
there is more than one group
Bias 9 Describe any efforts to address potential sources of bias 8-9
Study size 10 Explain how the study size was arrived at 8
Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, 8
describe which groupings were chosen and why
Statistical methods 12 (a) Describe all statistical methods, including those used to control for 8
confounding
(b) Describe any methods used to examine subgroups and interactions 8
(c) Explain how missing data were addressed 9
(d) If applicable, explain how loss to follow-up was addressed
(e) Describe any sensitivity analyses 9

Results
Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially Fig1
eligible, examined for eligibility, confirmed eligible, included in the study,
completing follow-up, and analysed
(b) Give reasons for non-participation at each stage Fig1
(c) Consider use of a flow diagram Fig1
Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) Tab2
and information on exposures and potential confounders
(b) Indicate number of participants with missing data for each variable of interest Tab2
(c) Summarise follow-up time (eg, average and total amount)
Outcome data 15* Report numbers of outcome events or summary measures over time Tab3

1
 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their Tab3
precision (eg, 95% confidence interval). Make clear which confounders were adjusted for
and why they were included
(b) Report category boundaries when continuous variables were categorized
(c) If relevant, consider translating estimates of relative risk into absolute risk for a
meaningful time period
Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity Tab5
analyses
Discussion
Key results 18 Summarise key results with reference to study objectives 13
Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. 13
Discuss both direction and magnitude of any potential bias
Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, 14-
15
multiplicity of analyses, results from similar studies, and other relevant evidence
Generalisability 21 Discuss the generalisability (external validity) of the study results 14

Other information
Funding 22 Give the source of funding and the role of the funders for the present study and, if 1
applicable, for the original study on which the present article is based

*Give information separately for exposed and unexposed groups.

Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and
published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely
available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at
http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is
available at http://www.strobe-statement.org.