ORIGINAL RESEARCH
Randomized Clinical Trial of Metronidazole Plus
Erythromycin to Prevent Spontaneous Preterm
Delivery in Fetal FibronectinPositive Women
William W. Andrews, PhD, MD, Baha M. Sibai, MD, Elizabeth A. Thom, PhD,
Donald Dudley, MD, J. M. Ernest, MD, Donald McNellis, MD, Kenneth J. Leveno, MD,
Ronald Wapner, MD, Atef Moawad, MD, Mary J. OSullivan, MD, Steve N. Caritis, MD,
Jay D. Iams, MD, Oded Langer, MD, Menachem Miodovnik, MD, and Mitchell Dombrowski, MD,
for the National Institute of Child Health & Human Development MaternalFetal Medicine
Units Network*
OBJECTIVE: To estimate whether antibiotic treatment of
asymptomatic women with a positive cervical or vaginal
fetal fibronectin test in the second trimester would reduce
the risk of spontaneous preterm delivery.
METHODS: Women were screened between 21 weeks 0 days
and 25 weeks 6 days of gestation with cervical or vaginal
swabs for fetal fibronectin. Women with a positive test (50
ng/mL or more) were randomized to receive metronida-
zole (250 mg orally three times per day) and erythromycin
(250 mg orally four times per day) or identical placebo pills
for 10 days. The primary outcome was spontaneous deliv-
From the Department of Obstetrics and Gynecology, Center for Research in
Womens Health, University of Alabama at Birmingham, Birmingham, Alabama;
Department of Obstetrics and Gynecology, University of Tennessee, Memphis,
Tennessee; The Biostatistics Center, George Washington University, Rockville,
Maryland; Department of Obstetrics and Gynecology, University of Utah, Salt
Lake City, Utah; Department of Obstetrics and Gynecology, Wake Forest Univer-
sity School of Medicine, Winston-Salem, North Carolina; The National Institute of
Child Health and Human Development, Bethesda, Maryland; Department of
Obstetrics and Gynecology, University of Texas Southwestern Medical Center,
Dallas, Texas; Department of Obstetrics and Gynecology, Thomas Jefferson
University, Philadelphia, Pennsylvania; Department of Obstetrics and Gynecology,
University of Chicago, Chicago, Illinois; Department of Obstetrics and Gynecology,
University of Miami, Miami, Florida; University of Pittsburgh/Magee-Womens
Hospital, Pittsburgh, Pennsylvania; Department of Obstetrics and Gynecology,
Ohio State University, Columbus, Ohio; Department of Obstetrics and Gynecology,
University of Texas at San Antonio, San Antonio, Texas; Department of Obstetrics
and Gynecology, University of Cincinnati, Cincinnati, Ohio; and Department of
Obstetrics and Gynecology, Wayne State University, Detroit, Michigan.
Supported by grants from the National Institute of Child Health and Human
Development (U10-HD27869, U10-HD21414, U10-HD36801, U10-
HD34208, U10-HD27860, U10-HD34116, U10-HD34136, U10-
HD27861, U10-HD34122, U10-HD21410, U10-HD27915, U10-
HD34210, U10-HD27905, and U10-HD27917). Fetal fibronectin assays
were performed at no cost to the project by Adeza Biomedical, Sunnyvale,
California.
*For other members of the Network, see the Appendix.
VOL. 101, NO. 5, PART 1, MAY 2003
2003 by The American College of Obstetricians and Gynecologists. Published by Elsevier.
ery before 37 weeks gestation after preterm labor or pre-
mature membrane rupture.
RESULTS: A total of 16,317 women were screened for fetal
fibronectin, and 6.6% had a positive test; 715 fetal fibronec-
tin testpositive women consented to randomization. Out-
come data were available for 703 women: 347 in the anti-
biotic group and 356 in the placebo group. The antibiotic
and placebo groups were not significantly different for
maternal age (P .051), ethnicity (P .849), marital status
(P .127), education (P .244), and bacterial vaginosis (P
.236). No difference was observed in spontaneous pre-
term birth before 37 weeks (odds ratio [OR] 1.17, 95%
confidence interval [CI] 0.80, 1.70), less than 35 weeks (OR
0.92, 95% CI 0.54, 1.56), or less than 32 weeks (OR 1.94,
95% CI 0.83, 4.52) gestation in antibiotic- compared with
placebo-treated women. Among women with a prior spon-
taneous preterm delivery, the rate of repeat spontaneous
preterm delivery at less than 37 weeks gestation was sig-
nificantly higher in the active drug compared with the
placebo group (46.7% versus 23.9%, P .039).
CONCLUSION: Treatment with metronidazole plus erythro-
mycin of asymptomatic women with a positive cervical or
vaginal fetal fibronectin test in the late second trimester
does not decrease the incidence of spontaneous preterm
delivery. (Obstet Gynecol 2003;101:84755. 2003 by
The American College of Obstetricians and Gynecolo-
gists.)
Preterm birth complicates 11% of all pregnancies and
remains the primary cause of perinatal mortality and
long-term neurologic morbidity.1,2 Published literature
has strongly linked clinically silent upper genital tract
bacterial infection and/or inflammation with preterm
birth and adverse pregnancy outcomes.37 Although
randomized trials of systemic antibiotics to reduce pre-
0029-7844/03/$30.00 847
doi:10.1016/S0029-7844(03)00172-8
term birth in women with altered vaginal flora have
yielded mixed results,6,8 12 genital tract infections repre-
sent a potentially treatable cause of spontaneous preterm
delivery.
Fetal fibronectin is a placental and membrane protein
that is a unique epitope of the fibronectins. When de-
tected in asymptomatic women (those without signs or
symptoms of vaginal infection or preterm labor) as early
as 2324 weeks gestation, a positive fetal fibronectin test
is associated with a 59-fold increased risk of spontaneous
preterm delivery at less than 28 weeks gestation.13 One
hypothesis holds that intrauterine infection causes dis-
ruption of the extracellular choriodecidual basement
membrane, causing leakage of fetal fibronectin into cer-
vical or vaginal secretions, in which it can then be
detected.14 16
Our objective in this study was to estimate whether
treatment with metronidazole plus erythromycin of
asymptomatic women with a positive cervical or vaginal
fetal fibronectin test before 26 weeks gestation would
reduce the risk of spontaneous preterm delivery.
MATERIALS AND METHODS
Women undergoing routine prenatal care at each of 13
participating centers were screened between 21 weeks 0
days and 25 weeks 6 days of gestation for cervical and
vaginal fetal fibronectin. Separate Dacron swab speci-
mens were collected from the exocervix and posterior
vaginal fornix with a vaginal speculum without a lubri-
cant and before collection of other specimens. These
specimens were immediately stored at 2 8C and
shipped overnight to a central laboratory (Adeza Bio-
medical, Sunnyvale, CA), where fetal fibronectin was
determined by an enzyme-linked immunosorbent assay.
Based on previously published data,17 a positive test was
defined as a concentration of 50 ng/mL or more in either
the cervical or the vaginal swab. Other details of the
specimen collection, handling, and assay have been pre-
viously published.13,14 Vaginal smears obtained at
screening were evaluated (with Nugent criteria18) in a
central laboratory for the presence of bacterial vaginosis.
Women were not selected based on the presence or
absence of any specific risk factor (other than a positive
fetal fibronectin test) for a spontaneous preterm birth.
Exclusion criteria included the following: fetal death,
major fetal anomalies, multiple gestation, allergy to
study medication, antibiotic use at randomization, use of
topical antifungals within 24 hours of screening, vaginal
symptoms (unelicited, spontaneously reported vaginal
symptoms, including itching, burning, or malodor indi-
cating need for specific treatment), vaginal bleeding (de-
fined as active bleeding at screening, not spotting and not
848 Andrews et al Fetal Fibronectin and Prematurity
bleeding due to a friable cervix), ethanol abuse, preterm
labor, rupture of membranes, tocolytic use, cerclage,
preeclampsia, hypertension requiring medication, pla-
centa previa, uterine anomaly, insulin-requiring diabetes
mellitus, renal insufficiency, collagen vascular disease,
isoimmunization, acquired immunodeficiency syn-
drome defining illness, and anticipated prenatal care or
delivery at a nonparticipating center.
The study was approved by the institutional review
boards at each clinical site, and all women gave written,
informed consent before enrollment in the study. Eligi-
ble women with a positive fetal fibronectin test were
randomized (double-masked) between 21 weeks 0 days
and 25 weeks 6 days of gestation to receive either met-
ronidazole 250 mg three times daily plus erythromycin
250 mg four times daily or identical placebo capsules
orally for 10 days. The simple urn method of random-
ization19 with stratification according to clinical center
was used to create a computer-generated random se-
quence, and study drugs were packaged accordingly in
vials for distribution to the participants. Subjects were
instructed to take one metronidazole capsule (or identi-
cal placebo) three times daily with meals and one eryth-
romycin capsule (or placebo) four times daily (at each
meal and at bedtime) for 10 days. Medication was begun
on the day of randomization.
One follow-up visit was scheduled 14 days after ran-
domization, at which fetal fibronectin and bacterial vagi-
nosis status were reassessed. Medication compliance was
estimated at this visit by examining the pill containers for
remaining capsules. Subjects who failed to return the pill
container were asked whether any capsules remained. If
unconsumed capsules remained, the cervical and vaginal
specimens were still obtained, and the women were
encouraged to take the remainder of the study medica-
tion.
Establishment of the gestational age at randomization
was determined by the date of the last menstrual period
and by ultrasound results with a previously published
scheme.11,12 Preterm birth was defined as delivery be-
fore 37 completed weeks gestation (259 days), and
spontaneous preterm birth was defined as preterm deliv-
ery due to the spontaneous onset of preterm labor or
spontaneous preterm premature rupture of membranes.
The primary outcome was spontaneous delivery before
37 weeks gestation after preterm labor or premature
membrane rupture. Usual prenatal care was provided at
the respective institutions, but enrollment continued un-
til pregnancy outcomes were assessed.
The target sample size of 700 women was calculated
assuming a 30% rate of spontaneous preterm delivery
before 37 weeks gestation in the placebo group and an
estimated effect size of a one third reduction of this
OBSTETRICS & GYNECOLOGY
Figure 1. Summary of screening for fetal fibronectin and randomization to active drugs or placebo.
Andrews. Fetal Fibronectin and Prematurity. Obstet Gynecol 2003.
outcome in the active drug group. A type I error (two-
sided) of 5% and a power of 80% was chosen, and an
adjustment for a dropout rate of 5% was made.
Continuous variables were compared using the Wil-
coxon rank-sum test, and categoric variables were com-
pared using 2 and Fisher exact tests. Before the study
was started, the group sequential method of Lan and
DeMets with the modified OBrian-Fleming spending
function was chosen for adjustment of the significance
VOL. 101, NO. 5, PART 1, MAY 2003
level in interim analyses.20 An independent data and
safety-monitoring committee reviewed the interim re-
sults.
RESULTS
A total of 26,203 women were evaluated for a screening
examination between July 1996 and May 2000 (Figure
1). Of these, 9886 were either ineligible or refused con-
Andrews et al Fetal Fibronectin and Prematurity 849
Table 1. Baseline Characteristics of the Study Subjects
Active drug group Placebo group
Characteristic (n 353) (n 362) P
Marital status
Married 49.6 42.5
Divorced 3.1 4.7
Never married 47.3 52.8 .127
Ethnicity
Black 49.6 52.8
White 16.1 16.3
Hispanic 31.7 29.3
Asian 0.8 0.6
Other/unknown 1.7 1.1 .849
Medical funding source
Government 73.1 75.7
Private 10.2 11.3
None 16.7 13.0 .359
Smoked during this pregnancy 14.7 16.6 .498
Alcohol during this pregnancy 6.8 6.9 .955
Illicit drugs this pregnancy 4.2 4.1 .944
Nulliparous 39.7 44.2 .219
Prior spontaneous preterm delivery 8.5 13.3 .041
Infection during this pregnancy 28.6 31.2 .447
Syphilis 0 0.8 .249
Gonorrhea 4.2 2.2 .141
Chlamydia 3.4 4.4 .565
Trichomonas 5.4 3.6 .280
Bacterial vaginosis at screening 27.4 31.7 .236
Human immunodeficiency virus 0.6 0.3 .620
Maternal age (y) 24.2 5.4 23.4 5.1 .051
Gestational age at randomization (wk) 23.7 1.1 23.7 1.1 .689
Educational level (y) 11.2 2.6 11.1 2.4 .244
Prepregnancy weight (kg) 65 15.9 64.5 16.4 .781
Data are presented as % or mean standard deviation.

sent. Thus, 16,317 women agreed to a screening pelvic
examination; 1079 of them were determined to be fetal
fibronectin positive. A total of 715 women were random-
ized to either the active drug (353) or placebo (362)
groups (Figure 1). Only 15 women (2%) were lost to
follow-up. The treatment groups were similar with re-
spect to demographics and socioeconomic indicators
(Table 1). A history of a previous spontaneous preterm
delivery was significantly more common among place-
bo-treated compared with antibiotic-treated women (Ta-
ble 1).
Outcome data were available for 703 of the 715 ran-
domized women. Among the 15 women lost to follow-
up, the gestational age at which the subject was last
known to be pregnant was greater than 37 weeks gesta-
tion for three. Therefore, these three women were in-
cluded as term births in the analysis. No significant
differences were observed between the treatment groups
for spontaneous preterm delivery or preterm delivery
due to any reason before 37, 35, or 32 weeks gestation
(Table 2). Likewise, the mean gestational age at delivery
(38.1 3.0 versus 38.1 2.8 weeks, P .954) and the
frequency of infant birth weights less than 2500 g and
less than 1500 g were no different between the two
treatment groups (Table 2). Because a history of a pre-
vious spontaneous preterm delivery was significantly
more common among placebo-treated compared with
antibiotic-treated women, a regression analysis including
prior preterm birth history was performed. This regres-
sion model revealed no significant association between
treatment group and spontaneous preterm birth at less
than 37 weeks gestation (odds ratio 1.3, 95% confidence
interval 0.82, 2.05).
No statistically significant difference in spontaneous
preterm delivery before 37 weeks gestation was ob-
served between the two groups among women in terms
of race, a low body mass index, or the presence of
bacterial vaginosis at screening (Table 3). Among
women with a history of a spontaneous preterm delivery
in a previous pregnancy, the frequency of spontaneous
preterm delivery before 37 weeks gestational age was
nearly twice as high in the antibiotic- compared with the
placebo-treated group (46.7% versus 23.9%, P .039;
Table 3). Although not statistically significant among

850 Andrews et al Fetal Fibronectin and Prematurity OBSTETRICS & GYNECOLOGY

Table 2. Pregnancy Outcomes According to Treatment Group
Active drug group (%) Placebo group (%)
Outcome (n 347) (n 356) Relative risk (95% CI)
Spontaneous* preterm delivery
Before 37 wk 14.4 12.4 1.17 (0.80, 1.70)
Before 35 wk 6.9 7.5 0.92 (0.54, 1.56)
Before 32 wk 4.3 2.2 1.94 (0.83, 4.52)
Preterm delivery
Before 37 wk 16.4 16.6 0.99 (0.71, 1.38)
Before 35 wk 8.1 9.5 0.85 (0.53, 1.37)
Before 32 wk 4.9 3.1 1.60 (0.76, 3.36)
Birth weight
Less than 2500 g 12.7 14.3 0.88 (0.60, 1.29)
Less than 1500 g 3.5 3.2 1.12 (0.50, 2.50)
CI confidence interval
* Due to spontaneous onset of preterm labor or spontaneous preterm premature rupture of membranes.

women with a prior spontaneous preterm delivery, a
higher risk of repeat spontaneous preterm birth in the
antibiotic- compared with the placebo-treated women
was observed for delivery before 35 weeks gestation
(23.3% versus 14.6%, P .327) and before 32 weeks
gestation (16.7% versus 8.3%, P .294).
Because black women are known to be at increased
risk for spontaneous preterm delivery and also have a
higher prevalence of bacterial vaginosis, a consistent risk
factor for preterm birth,7 a subanalysis within this ethnic
group was performed. Among black women, no differ-
ence in the frequency of spontaneous preterm delivery

Table 3. Spontaneous Preterm Delivery* Stratified by Selected Risk Clinical Factors

Active drug group Placebo group
Outcome (n, %) (n, %) Relative risk (95% CI)
Previous spontaneous preterm delivery
Yes 14/30 (46.7) 11/46 (23.9) 1.95 (1.03, 3.71)
No 36/317 (11.4) 33/310 (10.7) 1.07 (0.68, 1.67)
Ethnicity
Black 30/172 (17.4) 25/187 (13.4) 1.30 (0.80, 2.13)
Hispanic 7/109 (6.4) 6/104 (5.8) 1.11 (0.39, 3.20)
White/other 13/66 (19.7) 13/65 (20.0) 0.98 (0.49, 1.96)
Body mass index (kg/m2)
19.8 9/60 (15.0) 13/71 (18.3) 0.82 (0.38, 1.78)
19.8 40/268 (14.9) 30/269 (11.2) 1.34 (0.86, 2.08)
Bacterial vaginosis at screening examination
Negative 30/229 (13.1) 23/217 (10.6) 1.24 (0.74, 2.06)
Positive 11/86 (12.8) 15/99 (15.2) 0.84 (0.41, 1.74)
Timing of FFN test (gestational wk)
2022 28/205 (13.7) 23/212 (10.9) 1.26 (0.75, 2.11)
2324 17/127 (13.4) 19/124 (15.3) 0.87 (0.48, 1.60)
25 5/15 (33.3) 2/20 (10.0) 3.33 (0.75, 14.9)
Recent intercourse
No 21/94 (22.3) 18/115 (15.7) 1.43 (0.81, 2.52)
Yes 3/61 (4.9) 7/63 (11.1) 0.44 (0.12, 1.63)
Screening to randomization interval (d)
07 32/176 (18.2) 22/176 (12.5) 1.45 (0.88, 2.40)
814 11/108 (10.2) 16/121 (13.2) 0.77 (0.37, 1.59)
1521 5/43 (11.6) 5/42 (11.9) 0.98 (0.30, 3.13)
22 2/20 (10.0) 1/17 (5.9) 1.70 (0.17, 17.2)
FFN Fetal Fibronectin; other abbreviation as in Table 2.
* Less than 37 weeks gestational age.

Less than the tenth percentile.

Within 24 hours.

VOL. 101, NO. 5, PART 1, MAY 2003 Andrews et al Fetal Fibronectin and Prematurity 851
Table 4. Selected Neonatal Outcomes According to Treatment Group
Active drug group Placebo group
Outcome (n 347) (n 356) P
Congenital malformations 1.5 1.4 .999
1-min Apgar score 5 5.9 2.9 .056
5-min Apgar score 7 2.4 1.2 .234
Umbilical artery pH 7.3 0.1 7.3 0.1 .601
Transient tachypnea 3.0 3.8 .558
Persistent pulmonary hypertension 0.3 0.0 .495
Culture-proven sepsis 1.5 1.4 .999
Pneumonia 1.2 1.2 .999
Necrotizing enterocolitis 0.9 0.0 .121
Grade III or IV intraventricular hemorrhage 0.6 0.3 .621
Respiratory distress syndrome 3.8 3.8 .958
Days of ventilator support 0.3 2.7 0.2 1.6 .797
Days of supplemental oxygen 0.8 5.0 0.5 3.1 .407
Bronchopulmonary dysplasia 0.9 0.3 .369
Patent ductus arteriosus 0.6 0.6 .999
Seizures 0.3 0.3 .999
Retinopathy of prematurity 0.9 0.0 .122
Fetal death 0.8 0.3 .369
Infant death 0.6 0.6 .999
Data are presented as % or mean standard deviation.

before 37 weeks gestation was observed between the
antibiotic- and placebo-treated groups, regardless of
whether bacterial vaginosis was present (13.2% versus
19.3%, P .388) or absent (17.5% versus 10.4%, P
.138) at randomization.
Stratification by other clinical factors, including the
gestational timing of the fetal fibronectin test, intercourse
within 24 hours of the fetal fibronectin test, and the
interval between obtaining the fetal fibronectin screening
test and subsequent randomization, revealed no statisti-
cally significant difference in spontaneous preterm deliv-
ery before 37 weeks gestation between the antibiotic-
treated and placebo-treated groups (Table 3). No
significant differences in spontaneous preterm delivery
between the two groups were observed when the out-
comes were stratified by clinical center (data not shown).
The antibiotic- and placebo-treated groups did not
differ significantly with respect to the number of admis-
sions to the hospital for preterm labor or premature
rupture of membranes (16.3% versus 13.4%, P .291),
use of tocolytic drugs (6.4% versus 6.0%, P .821),
frequency of antepartum bleeding (0.9% versus 1.1%, P
1.00), occurrence of preeclampsia or gestational hy-
pertension (7.9% versus 8.3%, P .842), oligohydram-
nios (5.0% versus 5.2%, P .911), need for cesarean
delivery (15.4% versus 15.9%, P .868), presence of
meconium in labor (20.9% versus 16.4%, P .134),
placental abnormalities (2.3% versus 1.7%, P .567), or
gestational diabetes (3.8% versus 2.6%, P .360).
The number of women who developed vaginal infec-
tions requiring treatment during the pregnancy (14.7%
versus 21.3%, P .025), postpartum endometritis (2.1%
versus 5.2%, P .028), and clinical chorioamnionitis
(3.8% versus 6.6%, P .099) was lower in the antibiotic-
than in the placebo-treated group.
Selected neonatal outcomes were also similar in the
group treated with active drugs compared with the group
treated with placebo (Table 4). Of particular importance
in this antibiotic-intervention study, neonatal infection
related morbidities were similar in the antibiotic- and
placebo-treated groups, including culture-proven sepsis,
pneumonia, and necrotizing enterocolitis (Table 4).
A full course of treatment consisted of 70 capsules,
including three metronidazole plus four erythromycin
(or respective placebo) capsules daily for 10 days. Com-
pliance data were available for 666 women. Overall,
68.4% (n 489) of the women completed more than 1
week of treatment (56 70 capsules), and 53.1% con-
sumed all 70 capsules. There were no statistically signif-
icant differences between the active drug and placebo
groups regarding the percentage of women who com-
pleted 8 or more days of treatment (67.4% versus 69.3%,
P 0.582) or all 10 days of treatment (50.7% versus
55.5%, P 0.197). The primary reason reported for not
completing the treatment course was the occurrence of
one or more adverse effects; 44.3% and 36.7% of those
who did not complete treatment were in the active drug
and placebo groups, respectively (P .157). Commonly
reported adverse effects in the active drug and placebo
groups, respectively, included any adverse effect (40.0%
versus 29.7%, P .005), nausea (28.6% versus 19.8%, P
.008), vomiting (15.1% versus 7.9%, P .003), diar-

852 Andrews et al Fetal Fibronectin and Prematurity OBSTETRICS & GYNECOLOGY

rhea (6.8% versus 3.8%, P .084), abdominal pain
(5.5% versus 2.6%, P .056), and loss of appetite (4.0%
versus 4.1%, P .957). Also, women in both treatment
groups who did not complete their treatment reported
that they often simply forgot to take their fourth (bed-
time) dose of erythromycin or placebo (active drug ver-
sus placebo: 28.7% versus 30.6%, P .695). Overall,
29.5% of women were determined to have bacterial
vaginosis at the randomization visit, and no difference in
the frequency of this condition was observed between
the active drug and placebo groups (27.4% versus 31.7%,
P .236). Bacterial vaginosis was significantly lower in
the active drug group than in the placebo group at the
follow-up visit (9.9% versus 27.3%, P .001).
By design, all women were positive for fetal fibronec-
tin at the randomization visit. The frequency of a posi-
tive result for fetal fibronectin overall was substantially
lower at the follow-up visit (19.5%). The frequency of a
positive fetal fibronectin test at the follow-up visit was
significantly higher in the active drug group than in the
placebo group (22.9% versus 16.2%, P .040).
DISCUSSION
In this randomized, clinical trial, treatment of asymptom-
atic women with a positive cervical or vaginal fetal
fibronectin test between 21 and 25 weeks gestational age
with metronidazole plus erythromycin did not decrease
the risk of spontaneous delivery and did not result in
improved neonatal outcomes. The ineffectiveness of the
intervention was observed despite the use of a prolonged
course of broad-spectrum combination antibiotics. Ra-
tionale for this treatment regimen was derived from a
prior study, in which treatment of more than a week with
this same combination of antibiotics was associated with
a lower rate of spontaneous preterm birth among high-
risk women with asymptomatic bacterial vaginosis.9 Ad-
ditionally, the combination of both metronidazole and
erythromycin seemed best suited to cover the major
organisms typically isolated from the upper genital tract
among women with a spontaneous preterm delivery.5
Compliance with the study medication was good ac-
cording to verbal reports by participants and capsule
counts conducted by study personnel. One objective
measure of compliance was the frequency of bacterial
vaginosis detected at the follow-up visit, which was ob-
served to be significantly lower in the women who took
active drugs (including metronidazole) compared with
those given placebo capsules. Some adverse effects were
more commonly reported by women taking active drug
compared with placebo-treated women. Also, the pri-
mary reason given by subjects for not completing the
treatment course was the occurrence of one or more
VOL. 101, NO. 5, PART 1, MAY 2003
adverse effects. However, the percentage of women who
completed 7 or more days of treatment was almost
identical in both groups. Therefore, noncompliance in
the groups assigned to metronidazole plus erythromycin
compared with the group assigned to placebo did not
explain the failure of the active intervention to reduce
spontaneous preterm delivery.
The assumptions used to calculate the sample size for
this investigation were based on extensive data collected
during the Preterm Prediction Study, an observational
study of 2929 pregnant women previously conducted by
the National Institute of Child Health and Human De-
velopment MaternalFetal Medicine Units Network.13
Based on this study, we assumed that an approximate
30% rate of spontaneous preterm delivery before 37
weeks gestation would occur in the placebo group.13
Instead, we observed a much lower incidence of sponta-
neous preterm delivery, 13.7% overall. Therefore, it
could be argued that this study lacked sufficient statistical
power and that a significantly lower incidence of sponta-
neous preterm birth might have been observed if the
sample size had been larger. However, given the almost
identical percentage of spontaneous preterm delivery
observed in the active drug and placebo groups (14.4%
and 12.4%, respectively) providing no suggestion of
even a trend toward a lower incidence of the primary
outcome in either group, this seems highly unlikely.
Nevertheless, explanation for the lower incidence of
spontaneous delivery before 37 weeks gestation ob-
served in this investigation compared with that observed
in the Preterm Prediction Study remains speculative.
Although the frequency of a positive fetal fibronectin
test at the follow-up visit was significantly higher in the
active drug group than in the placebo group, the magni-
tude of this difference does not likely explain the failure
to observe a benefit to antibiotic treatment. It is difficult
to speculate as to whether or not the antibiotic treatment
actually increased choriodecidual matrix disruption,
leading to higher rate of fetal fibronectin leakage from
the upper to the lower genital tract. Notably, the fetal
fibronectin test was negative in 80% of the women at the
follow-up visit.
An intriguing finding was the nearly two-fold higher
risk of spontaneous preterm delivery observed in the
active drug group compared with the placebo group
among women with a history of a spontaneous preterm
delivery in a previous pregnancy. Interpretation of this
observation must be guided by caution because this is a
secondary analysis of data, and the explanation remains
a subject of speculation. It is unclear whether antibiotic
treatment actually increased the risk of preterm delivery
in this subset of women by altering the vaginal flora or by
some other mechanism. However, this observation is
Andrews et al Fetal Fibronectin and Prematurity 853
consistent with an increased risk of preterm delivery in
certain subgroups of women treated with antibiotics
during pregnancy (Hauth JC, Cliver SP, Hodgkins P,
Andrews WW, Schwebke JR, Hooke EW, et al. Mid-
trimester metronidazole and azithromycin did not pre-
vent preterm birth in women at increased risk: A double-
blind trial [abstract]. Am J Obstet Gynecol 2001;185:
S86).6,9,12 Such data underscore the need for clinical
caution and argue against the indiscriminate use of anti-
biotics to prevent preterm birth.
Although vaginal infections and postpartum endome-
tritis were less common in the active drug group com-
pared with the placebo group, these observations are
secondary outcomes of this study and do not, in our
view, warrant a recommendation for second trimester
fetal fibronectin screening and antibiotic treatment solely
to reduce the risk of these outcomes. Despite the previ-
ously reported performance of the fetal fibronectin test to
identify women at increased risk for subsequent sponta-
neous preterm delivery,13,14 based on the results of this
trial, we discourage the practice of cervical or vaginal
fetal fibronectin screening to direct an antibiotic inter-
vention for preterm birth prevention.
REFERENCES
1. Hack M, Fanaroff AA. Outcomes of children of extremely
low birthweight and gestational age. Semin Neonatol
2000;5:89106.
2. Ventura SJ, Martin JA, Curtin SC, Mathews TJ. Report of
final natality statistics, 1995. Monthly vital statistics report,
vol 45, no 11 supplement. Hyattsville, Maryland: National
Center for Health Statistics, 1997:181.
3. Andrews WW, Goldenberg RL, Hauth JC. Preterm labor:
Emerging role of genital tract infections. Infect Agents Dis
1995;4:196211.
4. Gibbs RS, Romero R, Hillier SL, Eschenbach DA, Sweet
RL. A review of premature birth and subclinical infection.
Am J Obstet Gynecol 1992;166:151528.
5. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine
infection and preterm delivery. N Engl J Med 2000;342:
15007.
6. Andrews WW, Hauth JC, Goldenberg RL. Infection and
preterm birth. Am J Perinatol 2000;17:35765.
7. Kimberlin DF, Andrews WW. Bacterial vaginosis: Asso-
ciation with adverse pregnancy outcome. Semin Perinatol
1998;22:24250.
8. Morales WJ, Schorr S, Albritton J. Effect of metronidazole
in patients with preterm birth in preceding pregnancy and
bacterial vaginosis: A placebo-controlled, double-blind
study. Am J Obstet Gynecol 1994;171:3457.
9. Hauth JC, Goldenberg RL, Andrews WW, DuBard MB,
Copper RL. Reduced incidence of preterm delivery with
854 Andrews et al Fetal Fibronectin and Prematurity
metronidazole and erythromycin in women with bacterial
vaginosis. N Engl J Med 1995;333:17326.
10. McDonald HM, OLoughlin JA, Vigneswaran R, Jolley
PT, Harvey JA, Bof A, et al. Impact of metronidazole
therapy on preterm birth in women with bacterial vagi-
nosis flora (Gardnerella vaginalis): A randomized, pla-
cebo controlled trial. Br J Obstet Gynaecol 1997;104:
13917.
11. Carey JC, Klebanoff MA, Hauth JC, Hillier SL, Thom EA,
Ernest JM, et al. Metronidazole to prevent preterm deliv-
ery in pregnant women with asymptomatic bacterial vagi-
nosis. N Engl J Med 2000;342:53440.
12. Klebanoff MA, Carey JC, Hauth JC, Hillier SL, Nugent
RO, Thom EA, et al. Failure of metronidazole to prevent
preterm delivery among pregnant women with asymptom-
atic Trichomonas vaginalis infection. N Engl J Med 2001;345:
48793.
13. Goldenberg RL, Mercer BM, Meis PJ, Copper RL, Das A,
McNellis D. The Preterm Prediction Study: Fetal fibronec-
tin testing and spontaneous preterm birth. Obstet Gynecol
1996;87:6438.
14. Lockwood CJ, Senyei AE, Dische MR, Casal D, Shah KD,
Thung SN, et al. Fetal fibronectin in cervical and vaginal
secretions as a predictor of preterm delivery. N Engl J Med
1991;325:66974.
15. Feinberg RF, Kliman HJ, Lockwood CJ. Is oncofetal
fibronectin a trophoblast glue for human implantation?
Am J Pathol 1991;138:53743.
16. Goldenberg RL, Thom E, Moawad AH, Johnson F, Rob-
erts J, Caritus SN, et al. The Preterm Prediction Study:
Fetal fibronectin, bacterial vaginosis and peripartum infec-
tion. Obstet Gynecol 1996;87:65660.
17. Goepfert AR, Goldenberg RL, Andrews WW, Hauth JC,
Mercer B, Iams J, et al. The Preterm Prediction Study:
Quantitative fetal fibronectin values and the prediction of
spontaneous preterm birth. Am J Obstet Gynecol 2000;
183:14803.
18. Nugent RP, Krohn MA, Hillier SL. Reliability of diag-
nosing bacterial vaginosis by a standardized method of
Gram stain interpretation. J Clin Microbiol 1991;29:
297301.
19. Wei LJ, Lachin JM. Properties of the urn randomization in
clinical trials. Control Clin Trials 1988;9:34564.
20. Lan KKG, DeMets DL. Discrete sequential boundaries for
clinical trials. Biometrika 1983;70:65963.
Address reprint requests to: William W. Andrews, PhD, MD,
University of Alabama at Birmingham, Department of Obstet-
rics and Gynecology, 619 19th Street SouthOHB 458,Bir-
mingham, AL 35249-7333; E-mail: wandrews@uab.edu.
Received January 21, 2003. Received in revised form February 7, 2003.
Accepted February 13, 2003.
OBSTETRICS & GYNECOLOGY
APPENDIX
Other members of the National Institute of Child Health
and Human Development MaternalFetal Medicine
Units Network are as follows: University of Alabama at
Birmingham, R. Goldenberg, A. Northen, R. Cooper, C.
Morgan; University of Tennessee at Memphis, R. Ram-
sey; George Washington University Biostatistics Cen-
ter, L. Leuchtenburg, A. Das, M. Fisher; University of
Utah, E. Taggart; Wake Forest University, M. Swain, A.
Luper; National Institute of Child Health and Human
Development, M. Klebanoff, C. Spong; University of
VOL. 101, NO. 5, PART 1, MAY 2003
Texas Southwestern Medical Center, J. McCampbell,
M. L. Sherman; Thomas Jefferson University, M. Di-
Vito, A. Sciscione; University of Chicago, P. Jones, G.
Mallet, D. Scott, M. R. Brinson, M. Brown; University
of Miami, S. Beydoun, C. Alfonso, F. Doyle, R. Wash-
ington, G. Diaz; University of Pittsburgh/Magee-Wom-
ens Hospital, M. Cotroneo, T. Kamon, E. Daugherty,
P. Heine; Ohio State University, M. Landon, C.
Latimer, S. Meadows, F. Johnson; University of Texas
San Antonio, S. Nicholson, S. Barker, D. Skiver; Univer-
sity of Cincinnati, H. How; Wayne State University, Y.
Sorokin, G. Norman.
Andrews et al Fetal Fibronectin and Prematurity 855