OBSTETRICS

Use of antibiotics for the treatment of preterm parturition

and prevention of neonatal morbidity: a metaanalysis
Carolyn E. Hutzal, MD; Elaine M. Boyle, MD; Sara L. Kenyon, MA; Jennifer V. Nash, BHSc;
Stephanie Winsor, MD; David J. Taylor, MD; Haresh Kirpalani, BM, MSc
BACKGROUND AND OBJECTIVE
Despite major advances inperinatal care,
preterm delivery remains the predomi-
nant cause of perinatal mortality and a
major cause of neurologic morbidity in
surviving infants. Evidence suggests that
intrauterine infection is a contributing
factor to preterm delivery.
Potential benets of delaying delivery
may be greatest for infants born before
34 weeks. Even prolongation of preg-
nancy for 48 hours may facilitate admin-
istration of antenatal maternal cortico-
steroids, reducing neonatal respiratory
disease and improving outcomes.
The aim of our review was to deter-
mine whether antibiotic treatment of 24
hours or longer prolongs pregnancy and
reduces neonatal morbidity in women
presenting in preterm labor with intact-
membranes (PTL) or preterm prema-
ture rupture of membranes (PPROM) at
22-34 weeks gestation, representing the
subgroup of infants at highest risk.
MATERIALS AND METHODS
We conducted a literature search limited
to randomized placebo-controlled trials
published between 1988 and 2006, with
no language restriction. Hand searches
of conference proceedings (1995 to
2005) were performed. Inclusion criteria
were: (1) women presenting at 22-34
weeks of gestation with PPROM or in
PTL; (2) randomized comparison of any
antibiotic therapy regimenvs placebofor
24 hours or longer; and (3) data available
for latency period.
Studies using antenatal steroids or to-
colytics were not excluded. Outcomes
were dened a priori. Separate subgroup
analyses of PPROM and PTL were per-
formed. The primary outcome was la-
tency period.
RESULTS
Of 456 citations retrieved, 416 failed to
meet inclusion criteria. Nineteen articles
included infants born at longer than 34
weeks gestation, and authors were un-
able to provide separate data for the rel-
evant infants. The remaining 21 studies
were included: 9 studies of PPROM, 10
of PTL, and 2 of both.
For PPROM, 11 studies reported av-
erage latency period as a primary out-
come. Because studies used different
measures of latency, we conducted
separate subanalyses for average la-
tency period, delivery within 48 hours,
and delivery within 7 days. There was a
signicant prolongation in time to de-
livery with antibiotic use in PPROM
across all 3 measures of latency (Fig-
ure). Seven studies reported clinically
diagnosed chorioamnionitis (CA),
which was signicantly reduced in
women receiving antibiotics. There
was a trend toward decreased histolog-
ically diagnosed CA in the treatment
group.
Maternal antibiotics did not reduce
neonatal mortality, as reportedin9 stud-
ies of PPROM. Ten studies reported the
occurrence of respiratory distress syn-
drome (RDS) or hyaline membrane dis-
ease (HMD), with no difference between
groups. In 3 studies, no statistically sig-
nicant differences were seen in rates of
bronchopulmonary dysplasia (BPD).
Nine PPROM studies reporting on all
grades of intraventricular hemorrhage
(IVH) found a statistically signicant re-
duction in any grade of IVHin infants of
mothers given antibiotics.
One study reported periventricular
leukomalacia (PVL) in infants of moth-
ers who had PPROM. No differences
were seen between treatment groups.
Neonatal sepsis conrmed by culture
was reported in 5 studies of PPROM. A
trend toward reduced infection was seen
in infants whose mothers had received
antibiotics. Clinically diagnosed neona-
tal infection was reported in 9 studies of
PPROM. Maternal antibiotic treatment
was associated with reduced neonatal
infection.
Fromthe Departments of Pediatrics (Drs
Hutzal, Boyle, and Kirpalani and Ms Nash),
Obstetrics and Gynecology (Dr Winsor),
and Clinical Epidemiology and Biostatistics
(Dr Kirpalani), Faculty of Health Sciences,
McMaster University, Hamilton, ON,
Canada; the Reproductive Sciences Section,
Department of Cancer Studies and
Molecular Medicine, University of Leicester,
Leicester, United Kingdom(Ms Kenyon and
Dr Taylor); and the Department of
Neonatology, Childrens Hospital,
University of Pennsylvania, Philadelphia, PA
(Dr Kirpalani).
Presented at the 45th Annual Meeting of the
European Society for Paediatric Research,
Stockholm, Sweden, Sept. 18-21, 2004.
Cite this article as: Hutzal CE, Boyle EM, Kenyon SL, et al.
Use of antibiotics for the treatment of preterm parturition
and prevention of neonatal morbidity: a metaanalysis.
Am J Obstet Gynecol 2008;199:620.e1-620.e8.
0002-9378/free
2008 Mosby, Inc. All rights reserved.
doi: 10.1016/j.ajog.2008.07.008
For Editors Commentary,
see Table of Contents
See related editorial, page 583
OVERVIEW
Antibiotic therapy in preterm prema-
ture rupture of membranes at a ges-
tation of 34 weeks or less reduces
rates of neonatal infection and intra-
ventricular hemorrhage, but in preterm
labor with intact membranes the ben-
ets remain unclear.
Research www.AJOG.org
620 American Journal of Obstetrics &Gynecology DECEMBER 2008
In 8 PPROMstudies reporting on nec-
rotizing enterocolitis (NEC), no differ-
ences were observed between groups.
Five studies reported length of hospital-
ization. Treatment of women with anti-
biotics was not associated with reduced
hospitalization for neonates whose
mothers had PPROM.
Women with PTL showed no signi-
cant improvement in any measure of
prolongation of pregnancy for women
given antibiotics (Figure). Regarding
secondary outcomes, in 5 trials examin-
ing clinical CA, a trend toward decreased
CA was noted with antibiotic treatment
in studies using clinical denitions. In 2
studies that used histology, a signicant
difference paralleled these ndings.
Mortality was reported in 9 studies of
PTL, with no signicant difference be-
tween groups.
Eight studies of PTL reported on RDS
or HMD. There were no signicant dif-
ferences between groups. One study of
PTL reported BPD as an outcome. Both
accepted denitions were statistically in-
signicant, whether oxygen dependency
at 36 weeks corrected gestation or at 28
postnatal days.
Six studies of PTL included any grade
of IVH. No differences were seen be-
tween randomized groups. Two studies
examined the subgroup with severe IVH
(grades III and IV). There were no
differences.
One study of PTL included PVL as an
outcome. No differences were seen be-
tween treatment groups. Regarding neo-
natal infection, culture-positive infec-
tion was reported in 2 studies of PTL;
there was no decrease in neonatal infec-
tion with antibiotic therapy. Clinical
neonatal sepsis was reported in 9 PTL
studies. Analysis showed signicantly re-
duced infection in infants of women re-
ceiving antibiotics, although signicant
heterogeneity existedbetweenstudies. In
5 studies of PTL that included necrotiz-
ing enterocolitis as an outcome, no sig-
nicant differences were seen in rates of
NEC.
Four studies reported on length of
neonatal hospitalization. Maternal anti-
biotics were associated with signicantly
reduced hospital stay in the PTL group;
however, signicant heterogeneity was
seen between studies.
A potential advantage of delaying de-
livery is the ability to enhance fetal lung
maturation by administering maternal
corticosteroids. Steroids were given to
women in 1 study (9%) of PPROM and
5 studies (42%) of PTL. Two studies did
not report steroid use.
Publication bias was evaluated with
funnel plots of the 3 outcomes of latency.
All plots appeared symmetrical for both
PPROM and PTL.
COMMENT
Our results suggest that for women with
PPROMat a gestationof 34 weeks or less,
antibiotics were benecial in delaying
delivery and in reducing CA, neonatal
infection, and IVH. However, for
women in PTL at 34 weeks or less with
intact membranes, other than a reduc-
tion in clinically diagnosed infection, ev-
idence is insufcient to recommend ad-
ministration of antibiotics. More rened
testing for PTL, such as fetal bronectin,
might improve outcomes of antibiotic
use by more appropriate patient
selection.
Our ndings broadly agree with those
of prior metaanalyses. Although our re-
view contains fewer total numbers of
studies and mother-infant dyads ran-
domized, the included sample pool re-
mains large.
The presumption of the current re-
view was that even a 24-hour window
would allow improved respiratory sta-
tus, because steroids might be coadmin-
istered. Yet we do not showbenecial ef-
fects of antibiotics on respiratory
morbidity in the PPROM group, al-
though this was seen in the Cochrane re-
FIGURE
Average Latency Period: PPROM and PTL
Study Treatment Control WMD (fixed) Weight WMD (fixed)
or sub-category N Mean (SD) N Mean (SD) 95% CI % 95% CI
01 Preterm premature rupture of membranes
Kenyon PPROM 3081 16.62(23.65) 1055 16.40(24.58) 0.94 0.22 [-1.48 to 1.92]
Matsuda 34 7.91(8.55) 41 5.01(9.54) 0.16 2.90 [-1.20 to 7.00]
McCaul: PPROM group 41 8.10(11.20) 43 17.80(27.00) 0.04 -9.70 [-18.47 to -0.93]
Ovalle 50 0.60(0.52) 50 0.27(0.30) 98.84 0.33 [0.16-0.50]
Winkler 20 34.00(26.00) 20 25.00(23.00) 0.01 9.00 [-6.21 to 24.21]
Subtotal (95% CI) 3226 1209 100.00 0.33 [0.17-0.50]
Test for heterogeneity: Chi = 7.80, df = 4 (P = .10), I = 48.7%
Test for overall effect: Z = 3.92 (P < .0001)
02 Preterm labour
Kenyon PTL 4314 44.28(28.05) 1446 44.56(27.85) 89.27 -0.28 [-1.94 to 1.38]
Keuchkerian 47 43.99(25.61) 49 45.52(30.16) 1.97 -1.53 [-12.71 to 9.65]
McCaul: PTL group 21 35.20(29.60) 19 34.70(21.10) 0.98 0.50 [-15.32 to 16.32]
McGregor PTL 53 35.30(24.10) 50 25.40(20.00) 3.38 9.90 [1.37-18.43]
Nadisauskiene 59 51.10(34.30) 51 40.60(36.40) 1.40 10.50 [-2.78 to 23.78]
Newton 1989 48 34.20(21.00) 47 34.10(24.00) 2.99 0.10 [-8.98 to 9.18]
Subtotal (95% CI) 4542 1662 100.00 0.21 [-1.36 to 1.78]
Test for heterogeneity: Chi = 7.69, df = 5 (P = .17), I = 35.0%
Test for overall effect: Z = 0.26 (P = .79)
-10 -5 0 5 10
Favors control Favors treatment
Latency period was signicantly prolonged in studies of PPROM but was not prolonged in studies of PTL.
Hutzal. Antibiotics for preterm parturition and prevention of neonatal morbidity. Am J Obstet Gynecol 2008.
www.AJOG.org Obstetrics Research
DECEMBER 2008 American Journal of Obstetrics &Gynecology 621
view by Kenyon et al that included more
mature infants.
Our study period covered eras ank-
ing the full recognition of the benets of
antenatal steroids for preterm infants,
thus accounting for the variation in their
use. Pooled effects probably underesti-
mate the potential benets of combined
use of antenatal steroids and antibiotics
for PTL. Steroid administration is now
standard care. Possibly antenatal antibi-
otics confer a greater effect on lung ion
and water transport in this group rather
than on lung surfactant system matura-
tion in the preterm.
Results of our secondary outcomes are
also similar to the Cochrane metaanaly-
ses. We found that for both PPROMand
PTL, infection-related outcomes were
probably improved. However, plausible
trends toward reduced histological CA
are seen in PTL and PPROM, supported
by a signicant reduction in clinical CA
in both groups.
Overt neonatal infection was exam-
ined with the criterion standard of cul-
ture-positive infection and the less ro-
bust outcome of clinically diagnosed
infection. For PPROM, both were re-
duced, with clinically diagnosed infec-
tion achieving statistical signicance.
For PTL, however, we found no signif-
icant reduction in culture-proven in-
fections; the nding of reduced clini-
cally diagnosed infection is vitiated by
heterogeneity.
The signicant reduction in all grades
of IVHininfants of womeninPPROMis
not surprising, because cranial ultra-
sound abnormalities are more likely at
lower gestational ages. However, there is
no signicant reduction in clinically im-
portant types of IVH or PVL.
The relative merits of different antibi-
otic regimens in PPROM and PTL and
their impact on colonization and neona-
tal outcomes require further evaluation.
The reduced incidence in several as-
pects of neonatal morbidity in the
PPROM group did not translate into
shortened neonatal hospitalization or
reduced mortality. Conversely, treat-
ment during PTL demonstrated a signif-
icant reduction in infant hospital stay
despite the lack of benecial effect on
neonatal morbidity, apart from a reduc-
tion in clinically diagnosed sepsis. It is
difcult to be conclusive about a possible
link between the 2 outcomes. However,
our data appear to suggest that antenatal
therapy may reduce the rates of postnatal
sepsis in PPROM, but results were not
conclusive in PTL.
CLINICAL IMPLICATIONS
Labor is prolonged with antibiotic
treatment in preterm premature rup-
ture of membranes (PPROM) but not
in preterm labor (PTL).
Rates of infection and intraventricu-
lar hemorrhage in the neonate are re-
duced with the use of antibiotics dur-
ing PPROM.
Antibiotic use in PTL fails to demon-
strate improvement in neonatal out-
comes.
Prolongation of preterm labor allows
time to give antenatal steroids, al-
though the effects of concurrent ste-
roids and antibiotics in PPROM and
PTL remain unclear. f
Research Obstetrics www.AJOG.org
622 American Journal of Obstetrics &Gynecology DECEMBER 2008