ARTICLE IN PRESS

THE JOURNAL OF PEDIATRICS • www.jpeds.com ORIGINAL

ARTICLES
Early Antibiotic Exposure and Adverse Outcomes in Preterm, Very Low
Birth Weight Infants
Joseph B. Cantey, MD1,2,3, Alaina K. Pyle, MD1,2,4, Phillip S. Wozniak, BS1,5, Linda S. Hynan, PhD6, and
Pablo J. Sánchez, MD1,2,5

Objectives To determine whether antibiotic use in the first 14 postnatal days in preterm, very low birth weight
(birth weight of ≤1500 g) infants is associated with risk after 14 days of age for late-onset sepsis, necrotizing en-
terocolitis (NEC), or death after controlling for severity of illness using the Clinical Risk Index in Babies II score,
and determine whether duration of antibiotic exposure was associated with risk of adverse outcomes.
Study design This retrospective cohort study included very low birth weight infants born at ≤326/7 weeks of ges-
tation admitted to the neonatal intensive care unit from September 2010 to June 2014. Infants were excluded if
they had major congenital anomalies or culture-proven sepsis, NEC, or death during the first 14 days of life. An-
tibiotic exposure was recorded as days of therapy and length of therapy in days.
Results Of 374 infants, 70 (19%) had late-onset sepsis, NEC, or death after 14 days of age. The median number
of antibiotic days of therapy and length of therapy were 5.5 and 3.0, respectively. In multivariate analysis after con-
trolling for severity of illness, each antibiotic day of therapy was associated with a 1.24 times increased risk of sepsis,
NEC, or death (OR, 1.24; 95% CI, 1.17-1.31). Risk was similar when length of therapy was used (OR, 1.47; 95%
CI, 1.32-1.64).
Conclusions After controlling for severity of illness, each day of antibiotic therapy provided to preterm, very low
birth weight infants in the first 2 weeks of age is associated with an increased risk of late-onset sepsis, NEC, or
death. (J Pediatr 2018;■■:■■-■■).

ntibiotics are among the most prescribed medications in the neonatal intensive care unit (NICU).1,2 Preterm infants

A exposed to antibiotics develop a less diverse microbiome with a predominance of Enterobacteriaceae.3,4 The resultant
dysbiosis has been shown to precede episodes of late-onset sepsis and necrotizing enterocolitis (NEC),5,6 and clinical
studies have described an increased risk of these adverse outcomes in preterm infants exposed to prolonged antibiotic
therapy.7-11
Because adverse clinical outcomes are more likely to occur among critically ill preterm infants who at the same time receive
more broad-spectrum antibiotic therapy than healthier infants, there is concern for confounding by indication.12 Previous studies
have used a variety of approaches to control for severity of illness including proxy clinical variables (eg, gestational age, days of
mechanical ventilation) or a validated prediction model (Score for Neonatal Acute Physiology II).13 The Clinical Risk Index
for Babies II (CRIB II) score has not been used previously. The CRIB II is a validated risk-adjustment instrument for predic-
tion of mortality before NICU discharge among very low birth weight (VLBW; birth weight ≤1500 g) that incorporates gesta-
tional age, birth weight, sex, temperature on NICU admission, and initial base
deficit.14 Possible scores range from 0 to 27, with higher scores predictive of greater
mortality risk. For example, scores of 0, 5, 10, and 15 predict a mortality risk of
0.2%, 1.4%, 12.2%, and 56.8%, respectively. From the 1Department of Pediatrics, University of Texas
Southwestern Medical Center, Dallas; 2Division of
The aims of this study were to determine whether antibiotic use in the first 14 Neonatal-Perinatal Medicine and Pediatric Infectious
Diseases, University of Texas Southwestern Medical
postnatal days in preterm, VLBW infants is associated with risk of late-onset sepsis, Center, Dallas; 3University of Texas Health Science
NEC, or death after 14 days of age when controlling for severity of illness using Center San Antonio, San Antonio, TX; 4Division of
Neonatology, Yale School of Medicine, New Haven, CT;
the CRIB II score and to determine whether duration of antibiotic exposure was 5
Division of Neonatology and Pediatric Infectious
Diseases, Center for Perinatal Research, Nationwide
associated with risk of adverse outcomes. Children’s Hospital—The Ohio State University College of
Medicine, Columbus, OH; and 6Clinical Sciences and
Psychiatry, University of Texas Southwestern Medical
Center, Dallas, TX
Supported by a Gerber Novice Researcher Award
(#5200762201). The Gerber Foundation had no role or
input into the study design; the collection, analysis, and
interpretation of data; the writing of the report; or the
AUC Area under the curve decision to submit the paper for publication. The authors
CRIB-II Clinical Risk Index for Babies II declare no conflicts of interest.
DOT Days of therapy Portions of this study were presented at the Pediatric
LOT Length of therapy Academic Societies annual meeting, April 25-28, 2015,
San Diego, California.
NEC Necrotizing enterocolitis
NICU Neonatal intensive care unit 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights
VLBW Very low birth weight reserved.
https://doi.org10.1016/j.jpeds.2018.07.036

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■ • ■■ 2018

day) by 7 days of age. Umbilical venous catheters were removed

Methods by 7-10 days of age, and peripherally inserted central cath-
eters were used if central venous access was still required.
This prospective cohort study included all VLBW infants born
at ≤326/7 weeks of gestation who were admitted to the NICU
Statistical Analyses
at Parkland Memorial Hospital in Dallas, Texas, from Sep-
Group comparisons for categorical measures were con-
tember 1, 2010, to August 1, 2014. Infants with major con-
ducted using c2 or Fisher exact tests, as appropriate; the number
genital anomalies were excluded, as were those who were
and percentage are provided for each measure. Group com-
transferred or developed an outcome of interest (culture-
parisons for continuous measures were conducted using Mann-
proven sepsis, NEC, or death) within 14 days of life. Infants
Whitney U tests. Medians and IQR or means and SDs are
with spontaneous intestinal perforation but no signs of NEC
provided for each measure as appropriate. Stepwise logistic re-
in the first 14 days were included. Demographic, clinical, labo-
gression models were used to predict the composite score using
ratory, and other relevant data were collected from the elec-
either DOT or LOT and the following list of possible risk ad-
tronic health record by study personnel and a random sample
justed predictors: CRIB II score, receipt of antenatal cortico-
(10%) was reviewed by a second study member to ensure the
steroids, prolonged rupture of membranes (≥18 hours),
accuracy of abstracted data. The study was approved by the
intrapartum antibiotic prophylaxis, chorioamnionitis, mul-
Institutional Review Board at the University of Texas South-
tiple gestation, and 1- and 5-minute Apgar scores, days of me-
western Medical Center.
chanical ventilation, days of receipt of a vasopressor agent,
Outcomes central line days, and days to reach full enteral feeding
All antibiotic doses that each infant received in the first 14 days (≥120 mL/kg/day). Gestational age, birth weight, and sex were
of life were recorded. For infants who received an antibiotic not included in the models because they are components of
course that extended beyond age 14 days, only doses admin- and highly co-linear with the CRIB II score. Risk-adjusted pre-
istered up to and including the 14th day were counted. Anti- dictors were included in the stepwise models if P < .15. Models
biotic therapy was analyzed using 2 different metrics: (1) dose- used 3 different groups of data: all observations, gestational
days of antibiotic exposure, subsequently referred to as days age ≤296/7 weeks and gestational age >300/7 weeks. Model fit for
of therapy (DOT), which is an aggregate sum of the days of logistic regression was assessed using the Hosmer-Lemeshow
exposure for each antibiotic determined by multiplying the statistic with P > .20 providing adequate fit of the model to
number of antibiotic doses by the dosing interval, then divid- the data. In addition, receiver operating characteristic analy-
ing by 24 hours, and (2) length of therapy (LOT), the number sis was performed to determine the predictive ability of the
of calendar days that an infant received ≥1 dose of antibiotic, final DOT and LOT models for the composite outcome after
regardless of the number of agents administered during that controlling for severity of illness, and the area under the curve
period.15,16 Severity of illness was calculated for all infants using (AUC) was calculated. IBM SPSS V21.0 (IBM Corp, Armonk,
the CRIB II score. The primary outcome was a composite of New York) was used for all statistical analyses, and signifi-
late-onset sepsis, NEC, or death after 14 days of age. Late- cance was evaluated by P < .05.
onset sepsis was defined as recovery of a bacterial or fungal
pathogen from a normally sterile site (eg, blood, urine, cere- Results
brospinal fluid). Coagulase-negative staphylococci were con-
sidered pathogens if recovered from ≥2 blood cultures obtained Of 429 VLBW infants with a gestational age of ≤32 weeks who
within 2 days of each other.17 NEC was defined by the modi- were admitted to the NICU during the study period, 374 (87%)
fied Bell classification of stage IIa or greater.18 met our inclusion criteria. All were inborn and were in-
cluded in the analysis (Figure 1). The demographic and clini-
NICU Protocols cal characteristics of the 374 infants and their mothers are
The NICU used a standard protocol for antibiotic use. Am- shown in Table I. The mean ± SD antibiotic exposure was 8.8
picillin and gentamicin were used empirically for early onset ± 6 DOT and 4.6 ± 2.8 LOT, with 39% of infants (145 of 374)
sepsis. Oxacillin and gentamicin were used for late-onset sepsis receiving ≥5 calendar days of antibiotic therapy (LOT). Of the
unless the infant was known to be colonized with methicillin- 374 infants, 70 (19%) developed the composite outcome of late-
resistant Staphylococcus aureus, in which case vancomycin was onset sepsis (n = 52 [14%]), NEC (n = 24 [6%]), or death
used in lieu of oxacillin. Empiric therapy for NEC was oxa- (n = 11 [3%]; Table I). Eleven infants died from late-onset sepsis
cillin or ampicillin plus gentamicin; piperacillin/tazobactam (n = 6) or NEC (n = 5). Of the 59 surviving infants, 40 had
was given for more complicated gram-negative infections that late-onset sepsis alone, 13 had NEC alone, and 6 had NEC with
did not involve the central nervous system. Definitive therapy a positive blood culture. Among infants who developed the
for proven infection was modified based on culture results, but composite outcome, the median age at that time was 29 days
in general the use of third-generation cephalosporins, (range, 15-101 days); the median age for sepsis was 29 days
carbapenems, and vancomycin was avoided. Infectious dis- (range, 19-77 days), NEC was 35 days (range, 16-55 days), and
eases consultation is available.19 In addition, a standardized death was 31 days (range, 15-101 days). No outbreaks or
feeding protocol was in place throughout the study period that clonal spread of pathogens were detected during the study
aimed to have infants reach full enteral feeding (≥120 mL/kg/ period. The median CRIB II score was 8 (IQR, 5-10), which
2 Cantey et al

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018

■■ 2018 ORIGINAL ARTICLES

Figure 1. Study population and infant outcomes.

corresponded with a mortality risk of 5% (IQR, 1%-12%). The In bivariate analysis (Table II), gestational age, birth weight,
CRIB II score accounted for approximately 60% of the vari- male sex, Apgar scores at 1 and 5 minutes, CRIB II score, and
ability of the primary outcome (R2 = 0.593) but correlated antibiotic exposure (both DOT and LOT) were associated with
weakly with antibiotic exposure (r = 0.28). an increased risk for the composite outcome of late-onset sepsis,
NEC, or death. In multivariable analysis, after controlling for
CRIB II score, each DOT of antibiotic therapy was associ-
Table I. Characteristics of study infants and their ated with increased risk for sepsis, NEC, or death after 14 days
mothers of age (OR, 1.24; 95% CI, 1.17-1.31). The risk was similar when
infants were stratified by gestational age of ≤296/7 weeks of ges-
No. of mothers 335
Age, years 28 (22-33) tation (OR, 1.20; 95% CI, 1.12-1.29) or ≥300/7 weeks of ges-
Parity 2 (1-3) tation (OR, 1.30; 95% CI, 1.17-1.43). After CRIB-II score was
Hypertension 160 (48%) controlled for, the variation in antibiotic use accounted for the
Chorioamnionitis 36 (11%)
Antenatal betamethasone 168 (50%) majority of the remaining risk for sepsis, NEC, or death after
Prolonged rupture of membranes (≥18 hours) 91 (27%) 14 days of age (R2 = 0.29). When LOT was used instead of DOT
Intrapartum antibiotics 194 (58%) in the same model, the risk per day of exposure also in-
Vaginal delivery 98 (29%)
No. of infants 374 creased (OR, 1.47; 95% CI, 1.32-1.64). Receiver operating char-
Birth weight, grams 1130 (885-1310) acteristic analyses comparing DOT and LOT were similar, with
Multiple gestation 90 (24%) good AUC for both DOT (AUC = 0.82; 95% CI, 0.78-0.89) and
Male sex 199 (53%)
1-minute Apgar 4 (3-6) LOT (AUC 0.81; 95% CI, 0.75-0.87; Figure 2).
5-minute Apgar 7 (6-8)
CRIB II score 8 (5-10)
CRIB II score predicted mortality risk 5% (1-12%) Discussion
DOT, mean ± SD 8.8 ± 6
LOT, mean ± SD 4.6 ± 2.8
Length of stay, days 81 (64-110) In this analysis of all antibiotic use among VLBW infants born
Number with composite outcome >14 days of age* 70 (18%) at ≤326/7 weeks of gestation who were admitted to a Level 3
Sepsis >14 days of age 52 (14%)
NEC (modified Bell stage ≥IIa) 24 (6%) inborn NICU during a 4-year period, 18% developed the com-
Death 11 (3%) posite outcome of late-onset sepsis, NEC, or death after 14 days
Data shown as median (IQR) or percentage unless otherwise indicated.
of age. Each additional day of antibiotic therapy was associ-
*Infants could have more than 1 outcome. ated with a 24% increase in the risk for late-onset sepsis, NEC,
Early Antibiotic Exposure and Adverse Outcomes in Preterm, Very Low Birth Weight Infants 3

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Table II. Characteristics of infants with or without the

composite outcome of late-onset sepsis, NEC, and death
Composite outcome
(sepsis, NEC, or death
>14 days of age)
Yes No P value
n 70 304
Maternal age, years 28 (22-32) 28 (22-33) .66
Maternal parity 2 (1-3) 2 (1-3) .40
Maternal hypertension 33 (47%) 133 (44%) .69
Maternal chorioamnionitis 6 (9%) 32 (11%) .71
Maternal betamethasone 35 (50%) 139 (41%) .41
Prolonged rupture of 12 (17%) 58 (19%) .66
membranes (≥18 hours)
Intrapartum antibiotic 35 (50%) 161 (46%) .66
prophylaxis
Vaginal delivery 20 (29%) 79 (26%) .62
Birth weight, grams 895 (787-1225) 1160 (940-1330) <.0001
Gestational age, weeks 28 (25-30) 29 (27-30) .001
Multiple gestation 15 (21%) 75 (25%) .18
Male sex 50 (71%) 149 (49%) .002
1-minute Apgar 4 (2-5) 5 (3-6) .0013
5-minute Apgar 6 (4-8) 7 (6-8) <.0001
Spontaneous intestinal 2 (3%) 6 (2%) .65
perforation
Days of mechanical 9 (4-16) 6 (2-12) <.0001
Figure 2. Receiver operating characteristic curves compar-
ventilation
Days of vasopressor 0 (0-11) 0 (0-6) .22 ing DOT (black line) and LOT (gray line) as predictors for com-
medication (range) posite outcome of late-onset sepsis, NEC, or death after
Central line days 14 (9-23) 10 (7-18) .003 correction for CRIB II score. AUC for DOT: 0.82 (95% CI, 0.77-
Days to full enteral feeding 13 (8-19) 9 (7-15) <.0001 0.88) and for LOT: 0.81 (95% CI, 0.75-0.87), not significantly
CRIB II score 10 (7-12) 7 (5-10) .0001
CRIB II score predicted 12% (4-25%) 4% (1-12%) .0001 different (P = .24).
mortality risk
DOT ≤14 days of age, 14 (4-42) 5.5 (0-25) <.0001
median (range)
LOT ≤14 days of age, 7 (2-14) 3 (0-14) <.0001
median (range) morbidity, including periventricular leukomalacia, broncho-
Length of stay, days 91 (73-127) 70 (51-97) .0001 pulmonary dysplasia, and retinopathy of prematurity.11,20
Data shown as median (IQR) or percentage unless otherwise indicated. The association of each additional day of antibiotic use with
the composite adverse outcome in this study was not surpris-
ing, because antibiotics are known to reduce or eliminate com-
or death. This association was evident even after adjusting for mensal flora in the neonatal gut. Infants exposed to antibiotics
severity of illness using the CRIB II score, a validated predic- suffer a loss of diversity in their intestinal bacterial microbiome
tor of mortality in this population. with a surge in the concentration of proteobacteria3,21 (eg, Es-
The current study supports previous studies that found cherichia coli, Pseudomonas spp.)—changes that have been
similar associations using different markers of disease severity.7-11 shown to immediately precede episodes of NEC or late-
In a subgroup analysis, Cotten et al used mechanical ventila- onset sepsis.5,6,22,23 In the case of bacterial sepsis, the causative
tion begun within 24 hours of delivery and continued through agent often is the most prevalent organism in the intestinal
at least the first 7 days or until death (if before day 7) to support microbiome.24 Prebiotics and probiotics have been suggested
the association between prolonged (≥5 days) empiric antibi- as a way to mitigate the disruption of the microbiome and,
otic therapy and increased risk of NEC or death among ex- although early data are encouraging, neither the optimal dosing
tremely low birth weight (≤1000 g) infants.7 Characterizing the strategy nor safety of commercially available products has been
severity of illness by using respiratory indices through day of established in VLBW infants.25-27 Antibiotic stewardship,
life 7 and previously identified clinical predictors of mortal- however, can be implemented now to minimize the adverse
ity in the multivariate logistic regression models, Kuppala et al effects of prolonged or unnecessary antibiotic exposure.15 Ef-
showed that, among VLBW infants, those who received ≥5 days fective stewardship in preterm, VLBW infants is challenging
of early antibiotic therapy were more likely to have late- because these infants are at an increased risk for sepsis and can
onset sepsis and a composite of late-onset sepsis, NEC, or death manifest only nonspecific signs.28,29 Moreover, the early ini-
compared with infants treated for <5 days.8 Recently, after con- tiation of antibiotic therapy is a cornerstone of the effective
trolling for severity of illness using the Score for Neonatal Acute management of sepsis and therefore high-risk infants will con-
Physiology II score among VLBW infants, Ting et al found an tinue to be exposed to empiric antibiotic therapy. However,
association between total antibiotic exposure during the NICU this study demonstrates the potential harm associated with each
admission and an increased risk for mortality or major additional day of continued therapy and highlights the need
4 Cantey et al

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018

■■ 2018 ORIGINAL ARTICLES

for further research on the minimum effective duration of an- estimate of the dose-effect response was possible. However, the
tibiotic therapy for both empiric therapy and proven infection. dose effect may not be linear; risk per day of antibiotic expo-
The optimal metric for antibiotic consumption in the NICU sure may increase or decrease depending on the duration of
is not known. Antibiotic stewardship programs most com- exposure. Finally, we excluded infants with proven sepsis before
monly use DOT per 1000 patient-days as their metric of choice. 14 days of age (n = 14 [3.3%]) and were unable to assess their
However, clinicians generally speak in terms of LOT (eg, “10 subsequent risk for the outcomes of interest.
days of antibiotics for NEC”) and most studies investigating In conclusion, this study supports the potential risk of pro-
adverse outcomes after prolonged antibiotic exposure have used longed antibiotic therapy in preterm, VLBW infants. Impor-
LOT in their analysis.7-11 In this study, the adjusted risk for the tantly, a clear dose-effect relationship was seen with each day
composite outcome was approximately twice as high for LOT of antibiotic therapy in the first 2 weeks of life being associ-
compared with DOT (OR, 1.47 vs OR, 1.24), possibly because ated with a 24% increase in the risk of late-onset sepsis, NEC,
1 calendar day generally was equivalent to 2 DOT (eg, 1 DOT or death. Our findings suggest that the risk of prolonged an-
of ampicillin and 1 DOT of gentamicin). An analysis of the tibiotic exposure for infants without proven infection is sub-
receiver operating characteristic curve did not show a differ- stantial. Antibiotic stewardship efforts to limit unnecessary
ence between DOT and LOT (AUC of 0.82 vs AUC of 0.81; antibiotic exposure by minimizing the duration of therapy are
P = .24). It is not known whether these findings would be con- needed urgently. ■
sistent in NICUs with different prescribing patterns.
A limitation of this study that is consistent with other ob- We thank John Gard, PharmD, Melody Bush, RPh, Sara Mureeba,
servational studies investigating treatment effects is confound- PharmD, and Sheeba Tharayil, PharmD, of the Parkland NICU phar-
ing by indication.12,30 Conceptually, the infants at greatest risk macy for assistance with this study and dedication to antimicrobial stew-
ardship efforts.
for adverse outcomes are likely to be the same infants who
receive prolonged antibiotic therapy, leading to a false asso- Submitted for publication Apr 2, 2018; last revision received Jun 28, 2018;
ciation between treatment and adverse outcomes. This study accepted Jul 11, 2018
aimed to minimize the impact of confounding by indication Reprint requests: Joseph B. Cantey, MD, Department of Pediatrics, University
by using the CRIB II score, an accurate measure of severity of of Texas Health Science Center San Antonio, San Antonio, TX 78229. E-mail:
cantey@uthscsa.edu
illness that predicts survival to NICU discharge with an AUC
for the receiver operating characteristic curve of 0.91-0.95. This
AUC is above the 0.76-0.80 threshold needed to prevent con- References
founding by indication in modeling studies.30,31 Second, the du-
1. Clark RH, Bloom BT, Spitzer AR, Gerstmann DR. Reported medication
ration of empiric therapy only correlated weakly with CRIB use in the neonatal intensive care unit: data from a large national data
II score (r = 0.28). This lack of correlation between treat- set. Pediatrics 2006;117:1979-87.
ment and risk has been shown in previous studies and makes 2. Hsieh EM, Hornik CP, Clark RH, Laughon MM, Benjamin DK, Smith
confounding by indication less likely.12,32,33 Notably, CRIB II PB. Medication use in the neonatal intensive care unit. Am J Perinatol
has not been used previously to predict risk for sepsis or NEC, 2014;31:811-21.
3. Greenwood C, Morrow AL, Lagomarcino AJ, Altaye M, Taft DH, Yu Z,
the other 2 contributors to the composite outcome in this study. et al. Early empiric antibiotic use in preterm infants is associated with
However, multiple studies using different strategies to adjust lower bacterial diversity and higher relative abundance of Enterobacter.
for confounding have reported similar risk for adverse out- J Pediatr 2014;165:23-9.
comes after prolonged antibiotic exposure.7-11 Finally, the clini- 4. Arboleya S, Sanchez B, Milani C, Duranti S, Solis G, Fernandez N, et al.
cal findings are biologically plausible within our increasing Intestinal microbiota development in preterm neonates and effect of peri-
natal antibiotics. J Pediatr 2015;166:538-44.
understanding of antibiotic-driven disruption of the normal 5. Mai V, Young CM, Ukhanova M, Wang X, Sun Y, Casella G, et al. Fecal
intestinal bacterial microbiome. Although residual confound- microbiota in premature infants prior to necrotizing enterocolitis. PLoS
ing undoubtedly exists, the current study supports previous ONE 2011;6:e20647.
work suggesting that antibiotics may play an important role 6. Mai V, Torrazza RM, Ukhanova M, Wang X, Sun Y, Li N, et al. Distor-
tions in development of intestinal microbiota associated with late onset
in the increased risk for subsequent adverse outcomes among
sepsis in preterm infants. PLoS ONE 2013;8:e52876.
preterm infants. 7. Cotten CM, Taylor S, Stoll B, Goldberg RN, Hansen NI, Sanchez PJ, et al.
Other limitations include those inherent to a retrospective Prolonged duration of initial empirical antibiotic treatment is associ-
study, such as the inability to discern all postnatal practices, ated with increased rates of necrotizing enterocolitis and death for ex-
some of which may have influenced the risk for sepsis, NEC, tremely low birth weight infants. Pediatrics 2009;123:58-66.
8. Kuppala VS, Meinzen-Derr J, Morrow AL, Schibler KR. Prolonged initial
or death. The low rate (50%) of antenatal corticosteroid use
empirical antibiotic treatment is associated with adverse outcomes in pre-
also may have impacted the composite outcome. The study in- mature infants. J Pediatr 2011;159:720-5.
cluded patients only from a single center. The standardiza- 9. Shah P, Nathan E, Doherty D, Patole S. Prolonged exposure to antibiot-
tion of feeding practices and use of central venous catheter ics and its associations in extremely preterm neonates—the Western Aus-
bundles reduces some of these potential effects. In addition, tralian experience. J Matern Fetal Neonatal Med 2013;26:1710-4.
10. Alexander VN, Northrup V, Bizzarro MJ. Antibiotic exposure in the
the limited sample size did not provide sufficient power to
newborn intensive care unit and the risk of necrotizing enterocolitis. J
analyze late-onset sepsis, NEC, or death as individual out- Pediatr 2011;159:392-7.
comes. By analyzing antibiotic exposure as a continuous rather 11. Ting JY, Synnes A, Roberts A, Deshpandey A, Dow K, Yoon EW, et al.
than a dichotomous variable (ie, ≥5 days vs <5 days) an Associations between antibiotic use and neonatal mortality and

Early Antibiotic Exposure and Adverse Outcomes in Preterm, Very Low Birth Weight Infants 5

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

morbidities in very low-birth-weight infants without culture-proven sepsis 23. Zhou Y, Shan G, Sodergren E, Weinstock G, Walker WA, Gregory KE. Lon-
or necrotizing enterocolitis. JAMA Pediatr 2016;170:1181-7. gitudinal analysis of the premature infant intestinal microbiome prior to
12. Kyriacou DN, Lewis RJ. Confounding by indication in clinical research. necrotizing enterocolitis: a case-control study. PLoS ONE 2015;10:e0118632.
JAMA 2016;316:1818-9. 24. Madan JC, Salari RC, Saxena D, Davidson L, O’Toole GA, Moore JH, et al.
13. Richardson DK, Corcoran JD, Escobar GJ, Lee SK. SNAP-II and SNAPPE- Gut microbial colonisation in premature neonates predicts neonatal sepsis.
II: simplified newborn illness severity and mortality risk scores. J Pediatr Arch Dis Child Fetal Neonatal Ed 2012;97:F456-62.
2001;138:92-100. 25. Dilli D, Aydin B, Fettah ND, Ozyazici E, Zenciroglu A, Okumus N, et al.
14. Parry G, Tucker J, Tarnow-Mordi W. CRIB II: an update of the Clinical The propre-save study: effects of probiotics and prebiotics alone or com-
Risk Index for Babies score. Lancet 2003;361:1789-91. bined on necrotizing enterocolitis in very low birth weight infants. J Pediatr
15. Cantey JB, Wozniak PS, Pruszynski JE, Sanchez PJ. Reducing unneces- 2015;166:545-51, e541.
sary antibiotic use in the neonatal intensive care unit (SCOUT): a pro- 26. Manzoni P, Meyer M, Stolfi I, Rinaldi M, Cattani S, Pugni L, et al. Bovine
spective interrupted time-series study. Lancet Infect Dis 2016;16: lactoferrin supplementation for prevention of necrotizing enterocolitis
1178-84. in very-low-birth-weight neonates: a randomized clinical trial. Early Hum
16. Ibrahim OM, Polk RE. Antimicrobial use metrics and benchmarking to Dev 2014;90:S60-5.
improve stewardship outcomes: methodology, opportunities, and chal- 27. Vallabhaneni S, Walker TA, Lockhart SR, Ng D, Chiller T, Melchreit R,
lenges. Infect Dis Clin North Am 2014;28:195-214. et al. Notes from the field: fatal gastrointestinal mucormycosis in a pre-
17. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA, mature infant associated with a contaminated dietary supplement—
et al. Late-onset sepsis in very low birth weight neonates: the experience Connecticut, 2014. MMWR Morb Mortal Wkly Rep 2015;64:155-6.
of the NICHD Neonatal Research Network. Pediatrics 2002;110:285- 28. Bekhof J, Reitsma JB, Kok JH, Van Straaten IH. Clinical signs to identify
91. late-onset sepsis in preterm infants. Eur J Pediatr 2013;172:501-8.
18. Walsh MC, Kliegman RM. Necrotizing enterocolitis: treatment based on 29. Fischer JE. Physicians’ ability to diagnose sepsis in newborns and criti-
staging criteria. Pediatr Clin North Am 1986;33:179-201. cally ill children. Pediatr Crit Care Med 2005;6:S120-5.
19. Cantey JB, Wozniak PS, Sanchez PJ. Prospective surveillance of antibi- 30. Sjoding MW, Luo K, Miller MA, Iwashyna TJ. When do confounding by
otic use in the neonatal intensive care unit: results from the SCOUT study. indication and inadequate risk adjustment bias critical care studies? A simu-
Pediatr Infect Dis J 2015;34:267-72. lation study. Crit Care 2015;19:195.
20. Ting JY, Synnes A, Roberts A, Deshpandey AC, Dow K, Yang J, et al. As- 31. Ezz-Eldin ZM, Hamid TA, Youssef MR, Nabil Hel D. Clinical Risk Index
sociation of antibiotic utilization and neurodevelopmental outcomes among for Babies (CRIB II) scoring system in prediction of mortality in pre-
extremely low gestational age neonates without proven sepsis or necro- mature babies. J Clin Diagn Res 2015;9:SC8-11.
tizing enterocolitis. Am J Perinatol 2018;doi:10.1055/s-0038-1632390. Avail- 32. Cordero L, Ayers LW. Duration of empiric antibiotics for suspected early-
able online Feb 23, 2018. onset sepsis in extremely low birth weight infants. Infect Control Hosp
21. Torrazza RM, Neu J. The altered gut microbiome and necrotizing en- Epidemiol 2003;24:662-6.
terocolitis. Clin Perinatol 2013;40:93-108. 33. Spitzer AR, Kirkby S, Kornhauser M. Practice variation in suspected neo-
22. Patel RM, Denning PW. Intestinal microbiota and its relationship with natal sepsis: a costly problem in neonatal intensive care. J Perinatol
necrotizing enterocolitis. Pediatr Res 2015;78:232-8. 2005;25:265-9.

6 Cantey et al

FLA 5.5.0 DTD ■ YMPD10130_proof ■ August 29, 2018