Best Practice & Research Clinical Obstetrics and Gynaecology

Vol. 21, No. 3, pp. 375e390, 2007

doi:10.1016/j.bpobgyn.2006.12.005
available online at http://www.sciencedirect.com

3
Asymptomatic bacterial vaginosis and
intermediate flora as risk factors for
adverse pregnancy outcome
Harald Leitich *

MD, PhD

Associate Professor of Obstetrics and Gynecology

Herbert Kiss

MD

Associate Professor of Obstetrics and Gynecology

Department of Obstetrics and Gynecology, Medical University of Vienna, Wahringer Gurtel 18-20,
1090 Vienna, Austria

We updated a previously published meta-analysis to evaluate bacterial vaginosis (BV) and intermediate vaginal flora as risk factors for adverse pregnancy outcome. Selection criteria were original, published, English-language reports of cohort studies or control groups of clinical trials
including women <37 weeks gestation with intact amniotic membranes. All women had to
be screened for BV, diagnosed either by clinical criteria or by criteria based on Gram-stain findings. Outcomes were preterm delivery, late miscarriages, maternal or neonatal infections, and
perinatal mortality. Fourteen new studies with results for 10,286 patients were included, so
that results for 30,518 patients in 32 studies were available for this meta-analysis. BV more
than doubled the risk of preterm delivery in asymptomatic patients (OR: 2.16, 95% CI:
1.56e3.00) and in patients with symptoms of preterm labor (OR: 2.38, 95% CI: 1.02e5.58).
BV also significantly increased the risk of late miscarriages (OR: 6.32, 95% CI: 3.65e10.94)
and maternal infection (OR: 2.53, 95% CI 1.26e5.08) in asymptomatic patients. No significant
results were calculated for the outcomes of neonatal infection or perinatal mortality. Also, intermediate vaginal flora was not significantly associated with any outcome included. The results
of this meta-analysis confirm that BV is a risk factor for preterm delivery and maternal infectious
morbidity and a strong risk factor for late miscarriage.
Key words: bacterial vaginosis; intermediate vaginal flora; preterm delivery; late miscarriage;
risk factor; meta-analysis.

* Corresponding author. Tel.: 43 1 40400 2822; Fax: 43 1 40400 2861.

E-mail address: harald.leitich@meduniwien.ac.at (H. Leitich).
1521-6934/$ - see front matter 2006 Elsevier Ltd. All rights reserved.

376 H. Leitich and H. Kiss

Although it is now clear that intrauterine infection plays a major role in the pathogenesis of preterm birth, and even though the association between bacterial vaginosis (BV)
and preterm birth has been confirmed in two recent independent meta-analyses1,2, the
role of BV itself in the pathogenesis of preterm labor and preterm delivery remains
unexplained.3 More recently, it has been hypothesized that the risk of prematurity
may not depend on the type of vaginal flora alone, but also on the type of immune
response that is mounted to control an infectious process. A good overview of this
hypothesis can be found in an article by Romero et al.4 This hypothesis also includes
evidence that the differences in the types of immune responses may have a genetic
explanation. This topic will be specifically addressed in Chapter 12 of this issue.
We hope to support the discussion and research concerning BV with epidemiological data about the association between BV and preterm delivery by an update of our
previous meta-analysis2 covering this topic.
A specific purpose of this article will be to summarize evidence about the association between intermediate vaginal flora and adverse pregnancy outcome. Although intermediate vaginal flora appears to be a heterogeneous condition, it has been identified
as a strong risk factor for preterm delivery.5 Intermediate vaginal flora is an area of active research and is specifically addressed by Gilbert Donders in Chapter 2 of this issue.
Again, we hope that epidemiological data will support this discussion and research.
MATERIAL AND METHODS
In May 2005, we updated the literature search of our previous meta-analysis2 using
Medline and Science Citation Index Expanded to identify all literature included under
vaginitis or vaginosis or vagina* infection or vagina* inflammation and pregnant* or
gravid* or birth? or labour?
The following criteria were used to select studies for inclusion: article e original
published English-language report; study design e cohort study or control group of
randomized clinical trial; population e women entered before 37 weeks of gestation
with intact amniotic membranes; intervention e screening for bacterial vaginosis (BV),
diagnosed either by clinical criteria or by criteria based on Gram-stain findings (positive cultures for BV-associated microflora alone were not sufficient); observational
character e the diagnosis of BV was not used for the clinical management of patients;
and preterm delivery by any available definition as outcome.
We scanned all abstracts from the computer printouts (excluding those that we
have already screened for our previous meta-analysis)2, the retrieved full-text reports,
review articles, and the references from each retrieved report to determine whether
studies met our inclusion criteria. We then excluded all studies in which the relationship between the presence of BV and the outcome of preterm delivery was not
reported in numerical form.
To determine the internal validity of the trials included, we extracted information
about the blinding of the attending clinicians to the results of BV assessment and of
the microbiologists (or other persons evaluating the criteria for a diagnosis of BV)
to clinical information, reasons for excluding patients from the final analyses, and
the rate of follow-up losses. To determine the clinical homogeneity of the studies included, we extracted patient inclusion and exclusion criteria, mean gestational ages at
BV screening, criteria for a diagnosis of bacterial vaginosis and/or intermediate vaginal
flora, information on whether samples for BV assessment were repeatedly collected,
management of patients with concomitant genitourinary infections, and definitions of

Asymptomatic bacterial vaginosis 377

outcomes. If mean gestational ages at BV screening were not reported in individual

studies, we estimated them from patient inclusion criteria.
The primary outcome of this meta-analysis was preterm delivery by any available
definition. Secondary outcomes included late miscarriage, perinatal mortality, and maternal or neonatal infectious morbidity.
If patients were classified in a BV group, a group with intermediate vaginal flora and
a group with normal vaginal flora in an individual trial, we pooled patients with intermediate and normal vaginal flora as non-BV controls. Thus, (1) patients with BV were
compared to patients with no BV (including intermediate and normal vaginal flora), and
(2) patients with intermediate vaginal flora were compared to patients with normal
vaginal flora.
To compare patients with BV and patients without BV, data were extracted only
from studies that were not already included in our previous meta-analysis.2 To compare patients with intermediate versus normal vaginal flora, all studies e including
those that we have already scanned for our previous meta-analysis e were screened
to determine whether numbers of patients with intermediate and normal vaginal flora
were reported.
For all outcomes in each individual study we extracted total numbers of patients
with results, numbers of true-positive, true-negative, false-positive, and false-negative
results. When numbers of true-positive, true-negative, false-positive, and falsenegative results were not available, we tried to recalculate them from the reported
results.
For all outcomes in each individual study, we calculated odds ratios with 95% confidence intervals.6 We also calculated7 odds ratios and 95% confidence intervals for all
studies combined and for subgroups of studies according to the following characteristics: patient inclusion criteria (patients admitted with symptoms of preterm labor,
symptom-free patients with or without a previous preterm delivery or any other
risk factor for preterm delivery, and inclusion of patients with multifetal gestation), criteria for a diagnosis of BV and/or intermediate vaginal flora, and gestational age at BV
screening.
A statistical test of homogeneity7 was carried out for each analysis to address the
statistical validity of aggregating the trials. This test assumes that differences in the results of the individual trials are due to chance alone, i.e., that all trial results are
homogenous.
To calculate7 overall results for each outcome and all subgroups, we generally used
a fixed-effects model. In such models the overall effect is obtained as a weighted average of the individual trial effects. However, if significant heterogeneity was observed
(which we defined as a P-value of the homogeneity test of 0.10 because of the
low power of the test), we used a random-effects model instead. Trials with zero cells
in both the study and control groups were not included in the calculation of combined
results.
RESULTS
Fourteen new studies with results for 10,286 patients were included.8e21 Combined
with the studies22e41 that we included in our previous meta-analysis, results for
30,518 patients in 32 studies8e41 were available to calculate overall results.
Eight reports42e49 about patients already included in one of the 14 studies were not
included in this review.

378 H. Leitich and H. Kiss

Two of the studies included8,18 were placebo-controlled trials of antibiotic treatment of pregnant women including subgroups of patients with BV. From these, only
the patients receiving placebos were included. From one study14 in which two cohorts
of patients with preterm or term labor were included, only the cohort of patients with
preterm labor was included in this review.
A total of 16 studies were excluded.50e65 In one of them62 patients were classified
in BV, non-BV, and intermediate groups, but results were not shown for the intermediate group. In two other studies50,60 patients were classified in BV, non-BV, and intermediate groups, but only results for BV pooled with intermediate vaginal flora versus
normal vaginal flora were presented. In two studies of antibiotic treatment of pregnant
women52 or patients positive for cervicovaginal fetal fibronectin59, subgroups of control patients with BV were identified but results were not reported for these. In
11 studies51,53e58,61,63e65 the relationship between the presence of BV and the outcome of preterm delivery was not reported in numerical form.
In seven of the 14 studies included, blinding of the attending clinicians to the results
of BV assessment9,13,15,18,20,21 or blinding of the microbiologists (or other persons
evaluating the criteria for a diagnosis of BV) to clinical data19 were described. In
one study19 the patients general practitioners were informed of the results of infection screening at 16 weeks gestation. They were advised to refer women with chlamydial infections to the local genitourinary clinic, but no recommendation was
made on treatment of asymptomatic bacterial vaginosis. Apart from this intervention,
we did not find any explicit information that the results of BV assessment were used
for the clinical management of study patients in any of the 14 studies included.
Follow-up losses in individual trials were reported in 12.6%19, 6.8%20, 6.6%11,
6.3%21, 5.9%9, 5%17, 3.2%18, 2%8, 0.4%12, or 0%10,13,14e16, of patients.
Patient inclusion criteria, criteria for a diagnosis of BV, mean gestational ages at BV
screening, outcome definitions, and results of the individual studies included are
shown in Table 1. In two studies only high-risk patients with a history of at least
one preterm delivery or a positive fetal fibronectin test were included. In one study
only low-risk patients without a prior preterm delivery were included, and in eight
studies asymptomatic patients at average risk of preterm delivery were included.
Patients with symptoms of preterm labor were included in three studies.
Results to compare the pregnancy outcome of patients with intermediate and normal vaginal flora were available in seven studies including 2358 patients. Patient inclusion criteria, criteria for a diagnosis of intermediate flora, mean gestational ages at BV
screening, outcome definitions, and the results of those studies are shown in Table 2.
Patients in individual studies were either primarily excluded or excluded from the
final analyses for a variety of reasons listed in Table 3.
In one study8, treatment of study patients with clinical chorioamnionitis or symptomatic vaginal infections was described and patients were not excluded from the final
analysis.
Criteria for a diagnosis of BV in individual trials were clinical in one study, Grambased in 12 studies, and both clinical and Gram-based in one study. In one study23, intermediate vaginal flora was defined as abnormal vaginal bacterial flora, but two or
fewer of the diagnostic criteria of Amsel et al.66 Criteria for the diagnosis of intermediate vaginal flora were Gram-based in the remaining six studies. In two of the latter,
the definitions of intermediate vaginal flora were based on the definition of Spiegel
et al67, with the addition of an intermediate category which was defined as the
presence of reduced lactobacilli mixed with other morphotypes29 or as the presence
of intermediate microbial flora, predominantly Mycoplasma hominis, Ureaplasma

Asymptomatic bacterial vaginosis 379

urealyticum, Gardnerella vaginalis, gram-negative anaerobic rods, Chlamydia trachomatis

and few Lactobacillus spp.16
Mean gestational ages at BV screening ranged from 7 to 30 weeks. The collection of
samples for BV assessment was done once in ten studies9,11e14,16,17,19e21, repeated
once within 14 days after the first sampling8, and repeated once at 24e26 weeks15
or at 28e32 weeks10 gestation in two studies. In one study18, BV assessment was repeated only in the intervention, but not in the control group, from which patient data
were extracted for this review.
Outcomes included preterm delivery by any available definition in all studies, late
miscarriage in two studies, and maternal or neonatal infectious morbidity in one study.
Outcomes to compare patients with intermediate and normal vaginal flora were preterm delivery in all studies, late pregnancy losses in four and maternal or neonatal infectious morbidity in one study.
A significant association between BV and preterm delivery was reported in three
studies, with odds ratios ranging from 2.31 to 7.55. BV was significantly associated
with the occurrence of midtrimester pregnancy losses in two studies, with odds ratios
in the range 3.50e4.90. A significant association between intermediate vaginal flora
and preterm delivery was reported in two studies, with odds ratios in the range
6.01e8.00.
In Tables 4 and 5, the combined results of all studies included are shown. The main
analyses and the subgroup analyses were done as planned and described in the
Methods section. Subgroup analyses according to different criteria for a diagnosis of
BV or intermediate vaginal flora, however, were not done because of the large variability of criteria used in the different studies included.
Four studies using the outcomes of delivery <35 weeks13,25,36e38 or 34 weeks24
gestation were pooled with other studies as the combined outcome of delivery
<34 weeks in patients with or without BV, and one study30 using the outcome of
delivery <33 weeks was pooled with other studies as the combined outcome of
delivery <32 weeks in asymptomatic patients with or without BV.
To increase consistency within studies included in the calculations of combined results, the results based on Gram-stain evaluation were used instead of the results
based on clinical criteria in one study.22 In five other studies, results for BV screening
at 12.5 instead of 16.5 weeks29, 18 instead of 30 weeks39, 23 instead of 27.5
weeks36e38, and 8.5 instead of 25.0 weeks15 were preferred, with the exception of
sub-analyses according to gestational age at BV screening, for which the appropriate
results were used.
When all trials were pooled, a significant association between BV in asymptomatic
patients and preterm delivery <37 weeks gestation was calculated (OR: 2.16, 95% CI:
1.56e3.00). Considerable and highly significant heterogeneity, however, was seen in
this main analysis, and a random-effects model was used.
Although the sub-analysis of patients with singleton gestations was practically
identical to the main analysis, subgroup analyses of studies including multifetal gestations showed no significant heterogeneity and no significant association between
BV and preterm delivery by any definition. Heterogeneity was also largely diminished when other subgroups of studies according to patient inclusion criteria
were pooled. In patients with or without a previous preterm delivery, a significant
association between BV and preterm delivery was calculated, but the strength of
association was comparable to that of the main analysis. No significant association
between BV and preterm delivery was seen in asymptomatic patients with a positive
fetal fibronectin test.

Study

Patient inclusion
criteria

Diagnosis
of BV

Mean
gestational
age at BV
screening
(weeks)

Outcome

Patients
with BV

Patients
without BV

OR (95% CI)

Andrews et al, 2003

(RCT)8

Singleton pregnancy
at 21e25 weeks and
positive fetal
fibronectin test

Gram stain (Nugent

score 7)68

23.7

Delivery <37 weeks

15/99a

23/217a

1.51 (0.75e3.03)

Daskalakis et al, 20069

Singleton pregnancy
at 22e25 weeks and
no previous preterm
delivery

Gram stain (Nugent

score 7)68 and
vaginal pH > 4.5

23.5

Delivery <37 weeks

16/95

88/1102

2.33 (1.31e4.17)

De Seta et al, 200510

Singleton pregnancy
at 13e18 weeks

Gram stain (Nugent

score 7)68

15.5

Delivery <37 weeks

14/95

35/503

2.31 (1.19e4.49)

Edwards et al, 200611

Singleton pregnancy
at 23e32 weeks and
preterm labor

Gram stain (Nugent

score 7)68

28.3

Delivery <37 weeks

9/23

25/105

2.06 (0.80e5.32)

Genc et al, 200412

Singleton pregnancy
at 18e22 weeks
and no previous
preterm delivery
<24 weeks

Gram stain (Nugent

score 7)68

20.0

Delivery <37 weeks

4/30

19/177

1.28 (0.40e4.06)

Goffinet et al, 200313

Pregnancy at 24e34
weeks and preterm labor

Gram stain (Nugent

score 7)68

29.0

Delivery <35 weeks

Delivery <33 weeks
Chorioamnionitis
Neonatal infection

6/24
6/24
1/24
2/24

71/330
41/330
7/330
19/330

1.22
2.35
2.01
1.49

(0.47e3.18)
(0.88e6.26)
(0.24e17.01)
(0.33e6.80)

380 H. Leitich and H. Kiss

Table 1. Bacterial vaginosis (BV) and adverse pregnancy outcome: patients, methods and results of individual studies. (Studies that were already included in our
previous meta-analysis2 are not shown.)

Pregnancy at 23e36
weeks and preterm
labor

Gram stain (Spiegel

definition)67

30.2

Delivery <37 weeks

18/19

30/41

6.60 (0.79e55.48)

Guerra et al, 200615

Singleton pregnancy
at <10 weeks
and 1 previous
preterm delivery

Gram stain (Nugent

score 7)68

8.5

33/72
23/95

45/138
9/147

1.75 (0.97e3.14)
4.90 (2.15e11.14)

25.0

Delivery <37 weeks

Late miscarriage
25 weeks
Delivery <37 weeks

12/36

60/168

0.90 (0.42e1.93)

Kalinka et al, 200316

Singleton pregnancy
at 8e16 weeks

Gram stain (Spiegel

definition)67

12.3

Delivery <37 weeks

9/55

14/141

1.77 (0.72e4.38)

Kalinka et al, 200517

Singleton pregnancy
at 22e34 weeks

Gram stain (Spiegel

definition)67

29.0

Delivery <37 weeks

5/31

10/83

1.40 (0.44e4.49)

Kiss et al, 2004 (RCT)18

Singleton pregnancy
at 15e19 weeks

Gram stain (Nugent

score 7)68

17.0

Delivery <37 weeks

10/179a

116/1918a

0.92 (0.47e1.79)

Oakeshott et al, 200419

Singleton pregnancy
at <10 weeks

Gram stain (Nugent

score 7)68

7.0

Delivery <37 weeks

Late miscarriage
at 13e23 weeks

7/112
5/117

47/785
10/795

1.05 (0.46e2.38)
3.50 (1.18e10.44)

Purwar et al, 200120

Singleton pregnancy
at 12e28 weeks

Gram stain (Nugent

score 7)68

21.5

Delivery <37 weeks

32/115

40/823

7.55 (4.50e12.66)

Thorsen et al, 200621

Pregnancy at
<24 weeks

Clinical BV
(3 Amsel et al
criteria)66

17.0

Delivery <37 weeks

Delivery <34 weeks
Delivery <32 weeks

13/401
7/401
2/401

99/2526
39/2526
22/2526

0.82 (0.46e1.48)
1.13 (0.50e2.55)
0.57 (0.13e2.44)

RCT, randomized controlled trial; OR, odds ratio; CI, confidence interval.
a
Control patients of RCT only.

Asymptomatic bacterial vaginosis 381

Goyal et al, 200414

Study

Patient inclusion
criteria

Diagnosis of
intermediate
vaginal flora

Mean gestational
age at BV
screening (weeks)

Outcome

Patients with
intermediate
vaginal flora

Patients with
normal
vaginal flora

OR (95% CI)

Donders
et al, 200023

Singleton pregnancy
at <14 weeks

Clinical criteria
(2 Amsel criteria)66
and abnormal
vaginal bacterial flora

9.1

Delivery <37 weeks

Late miscarriage
<20 weeks

5/10
1/11

20/180
12/192

8.00 (2.13e30.06)
1.50 (0.18e12.71)

Edwards
et al, 200611

Singleton pregnancy
at 23e32 weeks
and preterm labor

Gram stain (Nugent

score 4e6)68

28.3

Delivery <37 weeks

5/19

20/86

1.18 (0.38e3.67)

Goffinet
et al, 200313

Pregnancy at 24e34
weeks and preterm
labor

Gram stain (Nugent

score 4e6)68

29.0

Delivery <35 weeks

Delivery <33 weeks
Chorioamnionitis
Neonatal infection

21/76
14/76
3/76
6/76

50/254
27/254
4/254
13/254

1.56
1.90
2.57
1.59

Guerra
et al, 200615

Singleton pregnancy
at <10 weeks

Gram stain (Nugent

score 4e6)68

8.5

Delivery <37 weeks

Late miscarriage
25 weeks
Delivery <37 weeks

28/48
4/52

17/90
5/95

6.01 (2.76e13.11)
1.50 (0.38e5.85)

17/23

Hay
et al, 199429

Singleton pregnancy
at 9e24 weeks

Gram stain (Spiegel

definition with addition
of an intermediate
category, see text)67

Kalinka
et al, 200316

Singleton pregnancy
at 8e16 weeks

Gram stain (Spiegel

definition with addition
of an intermediate
category, see text)67

Oakeshott
et al, 200419

Singleton pregnancy
at <10 weeks

Gram stain (Nugent

score 4e6)68

25.0

(0.86e2.81)
(0.94e3.84)
(0.56e11.74)
(0.58e4.33)

43/145

6.72 (2.48e18.21)

0/19
1/20

9/380
4/384

0.00 (n.c.)
5.00 (0.53e46.93)

16.5

Delivery <37 weeks

Late miscarriage
at 16e24 weeks
Delivery <37 weeks

1/32

17/584

1.08 (0.14e8.35)

12.3

Delivery <37 weeks

8/71

6/70

1.35 (0.44e4.13)

7.0

Delivery <37 weeks

Late miscarriage
at 13e23 weeks

1/39
2/41

46/746
8/754

0.40 (0.05e2.98)
4.78 (0.98e23.28)

12.5

BV, bacterial vaginosis; OR, odds ratio; CI, confidence interval; n.c., CI not calculable.

382 H. Leitich and H. Kiss

Table 2. Intermediate vaginal flora and adverse pregnancy outcome: patients, methods, and results of individual studies.

Asymptomatic bacterial vaginosis 383

Table 3. Patient exclusion criteria in individual studies. (Studies that were already included in our
previous meta-analysis2 are not shown.)
Exclusion criteria
Multiple gestation
Major fetal anomalies
Maternal or fetal conditions leading to indicated preterm delivery
Diabetes mellitus
Hypertension or preeclampsia
Placenta previa
Abruptio placentae
Intrauterine growth retardation
Maternal renal or cardiac disease
Rh-isoimmunization
Oligo- or polyhydramnios
Autoimmune disease
Threatened spontaneous preterm delivery
Advanced preterm labor
Cervical cerclage
Cervical incompetence, uterine or cervical anomalies
Previous preterm delivery
Previous late miscarriage
Fetal death
Miscarriage at <13 weeks gestation
Pregnancy termination at 16e23 weeks gestation
Induced or late miscarriage
Use of antibiotics before the study
Use of antibiotics during the study
Vaginal symptoms
Genitourinary infection
Clinical chorioamnionitis
HIV positive
Drug abuse

References of studies
8e12,15e20
8,9,13,14
10,12,13,16,17
8,10,12,14
8,10,12,14
8,9,11,12,14,20
11,14
14
8,10,12
10,12,14
9
8,12
8,10,17,18,20
11,13,14
8e10,12,14
8,9,15,20
9,14
12
8,12,21
19
19
21
8,10e12,14,17,20
9,19
8,18,20
10
11,13
8,11
15

When subgroups of studies were pooled according to the gestational age at BV

screening, heterogeneity was reduced only among studies screening for BV at
16 weeks gestation or less, and the calculated risk for preterm delivery was slightly
larger than in the main analysis. Among studies screening for BV at <20 weeks gestation or at 20 weeks gestation, considerable and highly significant heterogeneity remained, and the strength of association between BV and preterm delivery was slightly
weaker within studies in which BV was screened later in pregnancy.
The association between BV and adverse pregnancy outcome was much stronger
when the occurrence of late miscarriage was used as outcome, and no heterogeneity
was seen in this analysis. BV was also significantly associated with maternal infectious
morbidity, with absent heterogeneity. BV, on the other hand, was not significantly associated with perinatal mortality or neonatal infectious morbidity or with preterm delivery
when it was more restrictively defined as delivery <34, <33, or <32 weeks gestation.
No significant association between intermediate vaginal flora and adverse pregnancy outcome for any outcome was calculated, with the exception of a strong association between intermediate flora and preterm delivery that was seen in one
study including high-risk patients with a previous preterm delivery.

384 H. Leitich and H. Kiss

Table 4. Bacterial vaginosis (BV) and adverse pregnancy outcome: results of all studies combined in
patients with or without preterm labor.
Patients

Patients without preterm labor

Delivery <37 weeks
All patients
Singletons
Twins
Patients with no previous PTD
Patients with 1 previous PTD
Patients with positive FFN test
Screening at <16 weeks
Screening at <20 weeks
Screening at 20 weeks
Delivery <34 weeks
All patients
Singletons
Twins
Delivery <32 weeks
All patients
Singletons
Twins
Late miscarriages
All patients
Maternal infection
All patients
Perinatal mortality
All patients
Neonatal infection
All patients
Patients with preterm labor
Delivery <37 weeks
All patients
Delivery <34 weeks
All patients
Delivery <33 weeks
All patients
Maternal infection
All patients
Neonatal infection
All patients

Number Number
Test of
of
of
heterogeneity
studies patients
included included

Model
used

OR (95% CI)

P < 0.000001
P < 0.000001
P 0.37
P 0.58
P 0.27
e
P 0.001
P 0.00006
P < 0.000001

Random
Random
Fixed
Fixed
Fixed
e
Random
Random
Random

2.16
2.40
1.16
2.63
2.22
1.51
2.97
2.15
1.89

6088
3014
147

P 0.83
P 0.61
e

Fixed
Fixed
e

1.29 (0.92e1.82)
1.42 (0.94e2.15)
1.00 (0.43e2.32)

4
2
1

11042
7968
147

P 0.03
P 0.02
e

Random 1.34 (0.59e3.06)

Random 1.89 (0.57e6.23)
e
1.09 (0.28e4.21)

2010

P 0.14

Fixed

6.32 (3.65e10.94)

3067

P 0.98

Fixed

2.53 (1.26e5.08)

1089

P 0.33

Fixed

1.03 (0.40e2.63)

3775

P 0.09

Random 1.96 (0.54e7.14)

350

P 0.08

Random 2.38 (1.02e5.58)

466

P 0.47

Fixed

1.59 (0.74e3.42)

354

2.35 (0.88e6.26)

354

2.01 (0.24e17.01)

354

1.49 (0.33e6.80)

24
21
2
2
2
1
7
11
15

24190
21015
248
1365
400
316
3292
9688
15415

4
2
1

(1.56e3.00)
(1.69e3.41)
(0.59e2.29)
(1.55e4.44)
(1.46e3.37)
(0.75e3.03)
(1.48e5.98)
(1.34e3.43)
(1.27e2.83)

Random, random effects model; Fixed, fixed effects model; FFN, fetal fibronectin; PTD, preterm delivery.

DISCUSSION
The results of this meta-analysis confirm that BV is significantly associated with adverse pregnancy outcome and that the risk of preterm delivery <37 weeks gestation,

Asymptomatic bacterial vaginosis 385

Table 5. Intermediate vaginal flora and adverse pregnancy outcome: results of all studies combined in
patients with or without preterm labor.
Patients

Patients without preterm labor

Delivery <37 weeks
All patients
Patients with 1 previous PTD
Late miscarriages
All patients
Patients with preterm labor
Delivery <37 weeks
All patients
Delivery <35 weeks
All patients
Delivery <33 weeks
All patients
Maternal infection
All patients
Neonatal infection
All patients

Number Number
Test of
of
of
heterogeneity
studies patients
included included

Model
used

OR (95% CI)

5
1

1653
138

P 0.0007
e

Random 2.41 (0.63e9.20)

e
6.01 (2.76e13.11)

1549

P 0.26

Fixed

2.77 (0.94e8.16)

105

1.18 (0.38e3.67)

330

1.56 (0.86e2.81)

330

1.90 (0.94e3.84)

330

2.57 (0.56e11.74)

330

1.59 (0.58-4.33)

Random, Random effects model; Fixed, fixed effects model; PTD, preterm delivery.

as calculated in the main analysis, is more than doubled in asymptomatic women with
BV. As in our previous meta-analysis2, considerable and highly significant statistical heterogeneity was seen in the main analysis. By pooling subgroups of studies sharing specific clinical characteristics, we hoped to reduce heterogeneity and to calculate risks
that were specific to patients with those characteristics.
In contrast to our previous meta-analysis, screening early in pregnancy was not associated with a risk for preterm delivery that was distinctively different from the result
of the main analysis. Also, substantial heterogeneity was seen in all subgroup analyses
according to gestational age at BV screening. Thus, the theory that BV in early pregnancy carries a risk that is much higher than BV diagnosed later in pregnancy is no longer supported by the available data. This lack of a strong association between BV in
early pregnancy and preterm delivery was also seen in a re-analysis of patients that
were screened for a large trial of antibiotic treatment of BV in pregnancy.69 This article
was not included in our meta-analysis because patients with BV were entered into the
trial and received study medications.
The present meta-analysis confirmed, however, that BV is a strong risk factor for
a midtrimester miscarriage. Also, heterogeneity was not detected in this sub-analysis,
and this supports the validity of this association. It appears that BV in early pregnancy
has a strong short-term adverse effect leading to late miscarriage, but does not
strongly increase the risk of preterm delivery in continuing pregnancies.
There was no significant association between BV and preterm delivery <34 or <32
weeks gestation. However, only a few studies tested these outcomes. There is a lack
of evidence assessing BV as a risk factor for preterm delivery in specific subgroups of

386 H. Leitich and H. Kiss

pregnant women. Only two studies specifically included patients with or without a previous preterm delivery, respectively, and although the results were significant in both
sub-analyses, they did not differ much from the result of the main analysis.
Only small numbers of patients with multiple gestations or patients with a positive
fetal fibronectin test were available for this analysis, and the results were not significant
for any outcome included.
In contrast to our previous meta-analysis, in this analysis BV was significantly associated with preterm delivery in symptomatic patients admitted with a diagnosis of preterm labor, and the test of heterogeneity was not significant. Thus, BV as a marker of
intrauterine infection may help to identify patients with true preterm labor, who subsequently deliver preterm. BV in patients with preterm labor may also be a target for
intervention studies.
Other outcomes studied in this meta-analysis included maternal infection and fetal
outcome. The risk of maternal infection was significant and more than two-fold in
women with BV, with no heterogeneity. Results for neonatal outcome, including perinatal mortality and neonatal infection, were non-significant. Trends, however, were
positive in these analyses, and the rarity of the outcomes leading to insufficient power
might be responsible for the lack of statistical significance in these and other analyses.
There was no significant association between intermediate vaginal flora and either
preterm delivery, late miscarriage or maternal or neonatal infectious morbidity. One
explanation for this lack of a significant association may be that the numbers of patients
included in the analyses were too small to reach statistical significance. This explanation is supported by the fact that positive trends were seen in all analyses and that the
association of intermediate flora with late miscarriage was almost significant, with no
heterogeneity. Another explanation might be that patients with intermediate vaginal
flora represent a very heterogeneous group of patients with different conditions,
and that we need to know more about these conditions before we can pool those patients in a meta-analysis.
Even though the association between abnormal vaginal flora and adverse pregnancy
outcome can be consistently confirmed, abnormal vaginal flora is a common condition
and, as a single factor, does not sufficiently explain all spontaneous preterm births. As
mentioned before, current research tries to identify cofactors that determine whether
or not abnormal vaginal flora will lead to adverse pregnancy outcome.
Seen from a larger perspective, the focus on specific organisms may be insufficient
to determine whether interventions can help to avoid adverse pregnancy outcome. In
a recent community-based randomized trial, screening of pregnant women for BV, Candida, or T. vaginalis reduced the rate of prematurity, even though BV was not associated
with preterm delivery in this population (as shown in Table 2); Candida is not known as
a risk factor for prematurity, and the incidence of T. vaginalis was low in this population.18 Maybe intervention studies should not focus on specific organisms or conditions, but rather evaluate complete screening programmes. A detailed discussion
about interventions in patients with abnormal vaginal flora can be found in Chapter 4
of this issue.
SUMMARY
This meta-analysis included original English-language reports of cohort studies or control groups of randomized controlled trials including women <37 weeks gestation
with intact amniotic membranes. All women had to be screened for BV, diagnosed

Asymptomatic bacterial vaginosis 387

either by clinical criteria or by criteria based on Gram-stain findings. Outcomes were

preterm delivery, late miscarriages, maternal or neonatal infections, and perinatal
mortality.
Thirty-two studies with results for 30,518 patients were included in this metaanalysis. BV increased the risk of preterm delivery more than two-fold in asymptomatic patients (OR: 2.16, 95% CI: 1.56e3.00) and in patients with symptoms of preterm
labor (OR: 2.38, 95% CI: 1.02e5.58). BV also significantly increased the risk of miscarriage (OR: 6.32, 95% CI: 3.65e10.94) and maternal infection (OR: 2.53, 95% CI
1.26e5.08) in asymptomatic patients. No significant results were calculated for the
outcomes of neonatal infection or perinatal mortality. Also, intermediate vaginal flora
was not significantly associated with any outcome included.
The results of this meta-analysis confirm that BV is a risk factor for preterm delivery and maternal infectious morbidity and a strong risk factor for miscarriage. In contrast to our previous meta-analysis, screening early in pregnancy was not associated
with a risk for preterm delivery that was distinctively different from the result of
the main analysis. Also, substantial heterogeneity was seen in all subgroup analyses according to gestational age at BV screening. Thus, the theory that BV in early pregnancy
carries a risk that is much higher than BV diagnosed later in pregnancy is no longer
supported by the available data. Although intermediate flora was not significantly associated with any outcome included, trends were positive in all of these analyses.
Thus, the role of intermediate vaginal flora in the pathogenesis of preterm labour
and preterm delivery needs to be explored.

Practice points
 bacterial vaginosis (BV) is a risk factor for preterm delivery in asymptomatic
patients and in patients with symptoms of preterm labour
 BV is a risk factor for maternal infections
 BV is a strong risk factor for late miscarriage

Research agenda
 the role of abnormal vaginal flora in the pathogenesis of preterm labour and
delivery
 the identification of cofactors that determine whether abnormal vaginal flora
leads to adverse pregnancy outcome

ACKNOWLEDGEMENT
We would like to thank Alexandra Kaider, MSc, for her statistical assistance.
REFERENCES
1. Honest H, Bachmann LM, Knox EM et al. The accuracy of various tests for bacterial vaginosis in predicting preterm birth: a systematic review. BJOG 2004; 111: 409e422.

388 H. Leitich and H. Kiss

2. Leitich H, Bodner-Adler B, Brunbauer M et al. Bacterial vaginosis as a risk factor for preterm delivery:
a meta-analysis. Am J Obstet Gynecol 2003; 189: 139e147.
3. Goldenberg RL, Hauth JC & Andrews WW. Intrauterine infection and preterm delivery. N Engl J Med
2000; 342: 1500e1507.
*4. Romero R, Chaiworapongsa T, Kuivaniemi H & Tromp G. Bacterial vaginosis, the inflammatory response and the risk of preterm birth: a role for genetic epidemiology in the prevention of preterm
birth. Am J Obstet Gynecol 2004; 190: 1509e1519.
5. Donders GGG, Odds A, Vereecken A et al. Abnormal vaginal flora in the first trimester, but not
full-blown bacterial vaginosis, is associated with preterm birth. Prenat Neonatal Med 1998; 3:
588e593.
6. Demets DL. Methods for combining randomized clinical trials: strengths and limitations. Stat Med 1987;
6: 341e348.
7. Whitehead A & Whitehead J. A general parametric approach to the meta-analysis of randomized clinical trials. Stat Med 1991; 10: 1665e1677.
8. Andrews WW, Sibai BM, Thom EA et al. Randomized clinical trial of metronidazole plus erythromycin
to prevent spontaneous preterm delivery in fetal fibronectin-positive women. Obstet Gynecol 2003; 101:
847e855.
9. Daskalakis G, Papapanagiotou A, Mesogitis S et al. Bacterial vaginosis and group B streptococcal colonization and preterm delivery in a low-risk population. Fetal Diagn Ther 2006; 21: 172e176.
10. De Seta F, Sartore A, Piccoli M et al. Bacterial vaginosis and preterm delivery: an open question. J Reprod
Med 2005; 50: 313e318.
11. Edwards RK, Ferguson RJ & Duff P. The interleukin-1 beta 3953 single nucleotide polymorphism: Cervical protein concentration and preterm delivery risk. Am J Reprod Immunol 2006; 55: 259e264.
12. Genc MR, Witkin SS, Delaney ML et al. A disproportionate increase in IL-1beta over IL-1ra in the cervicovaginal secretions of pregnant women with altered vaginal microflora correlates with preterm
birth. Am J Obstet Gynecol 2004; 190: 1191e1197.
13. Goffinet F, Maillard F, Mihoubi N et al. Bacterial vaginosis: prevalence and predictive value for premature
delivery and neonatal infection in women with preterm labour and intact membranes. Eur J Obstet
Gynecol Reprod Biol 2003; 108: 146e151.
14. Goyal R, Sharma P, Kaur I et al. Bacterial vaginosis and vaginal anaerobes in preterm labour. J Indian Med
Assoc 2004; 102: 548e553.
15. Guerra B, Ghi T, Quarta S et al. Pregnancy outcome after early detection of bacterial vaginosis. Eur J
Obstet Gynecol Reprod Biol 2006; 128: 40e45.
16. Kalinka J, Laudanski T, Hanke W & Wasiela M. Do microbiological factors account for poor pregnancy
outcome among unmarried pregnant women in Poland? Fetal Diagn Ther 2003; 18: 345e352.
17. Kalinka J, Sobala W, Wasiela M & Brzezinska-Blaszczyk E. Decreased proinflammatory cytokines in cervicovaginal fluid, as measured in midgestation, are associated with preterm delivery. Am J Reprod Immunol 2005; 54: 70e76.
18. Kiss H, Petricevic L & Husslein P. Prospective randomised controlled trial of an infection screening programme to reduce the rate of preterm delivery. BMJ 2004; 329: 371.
19. Oakeshott P, Kerry S, Hay S & Hay P. Bacterial vaginosis and preterm birth: a prospective communitybased cohort study. Br J Gen Pract 2004; 54: 119e122.
20. Purwar M, Ughade S, Bhagat B et al. Bacterial vaginosis in early pregnancy and adverse pregnancy outcome. J Obstet Gynaecol Res 2001; 27: 175e181.
21. Thorsen P, Vogel I, Olsen J et al. Bacterial vaginosis in early pregnancy is associated with low birth
weight and small for gestational age, but not with spontaneous preterm birth: a population-based study
on Danish women. J Matern Fetal Neonatal Med 2006; 19: 1e7.
22. Crane JMG, Armson BA, Dodds L et al. Risk scoring, fetal fibronectin, and bacterial vaginosis to predict
preterm delivery. Obstet Gynecol 1999; 93: 517e522.
23. Donders GGG, Van Bulck B, Caudron J et al. Relationship of bacterial vaginosis and mycoplasmas to the
risk of spontaneous abortion. Am J Obstet Gynecol 2000; 183: 431e437.
24. Goffeng AR, Holst E, Milsom I et al. Fetal fibronectin and microorganisms in vaginal fluid of women with
complicated pregnancies. Acta Obstet Gynecol Scand 1997; 76: 521e527.
25. Goldenberg RL, Iams JD, Miodovnik M et al. The preterm prediction study: Risk factors in twin gestations. Am J Obstet Gynecol 1996; 175: 1047e1053.

Asymptomatic bacterial vaginosis 389

26. Govender L, Hoosen AA, Moodley J et al. Bacterial vaginosis and associated infections in pregnancy. Int J
Gynaecol Obstet 1996; 55: 23e28.
27. Gratacos E, Figureas F, Barranco M et al. Spontaneous recovery of bacterial vaginosis during pregnancy
is not associated with an improved perinatal outcome. Acta Obstet Gynecol Scand 1998; 77: 37e40.
28. Hauth JC, Goldenberg RL, Andrews WW et al. Reduced incidence of preterm delivery with metronidazole and erythromycin in women with bacterial vaginosis. N Engl J Med 1995; 333: 1732e1736.
29. Hay PE, Lamont RF, Taylor-Robinson D et al. Abnormal bacterial colonisation of the genital tract and
subsequent preterm delivery and late miscarriage. BMJ 1994; 308: 295e298.
30. Heath VCF, Daskalakis G, Zagaliki A et al. Cervicovaginal fibronectin and cervical length at 23 weeks of
gestation: relative risk of early preterm delivery. Br J Obstet Gynaecol 2000; 107: 1276e1281.
31. Hillier SL, Nugent RP, Eschenbach DA et al. Association between bacterial vaginosis and preterm delivery of a low-birth-weight infant. N Engl J Med 1995; 333: 1737e1742.
32. Holst E, Goffeng AR & Andersch B. Bacterial vaginosis and vaginal microorganisms in idiopathic preterm labor and association with pregnancy outcome. J Clin Microbiol 1994; 32: 176e186.
33. Kurki T, Sivonen A, Renkonen O-V et al. Bacterial vaginosis in early pregnancy and pregnancy outcome.
Obstet Gynecol 1992; 80: 173e177.
34. McGregor JA, French JI, Richter R et al. Antenatal micobiologic and maternal risk factors associated
with prematurity. Am J Obstet Gynecol 1990; 163: 1465e1473.
35. McGregor JA, French JI & Seo K. Adjunctive clindamycin therapy for preterm labor: Results of a doubleblind, placebo-controlled trial. Am J Obstet Gynecol 1991; 165: 867e875.
36. Goldenberg RL, Thom E, Moawad AH, et al, for the NICHD Maternal Fetal Medicine Units Network.
The preterm prediction study: Fetal fibronectin, bacterial vaginosis, and peripartum infection. Obstet
Gynecol 1996; 87: 656e660.
37. Goldenberg RL, Iams JD, Mercer BM et al. The preterm prediction study: The value of new versus standard risk factors in predicting early and all spontaneous preterm births. Am J Public Health 1998; 88:
233e238.
38. Meis PJ, Goldenberg RL, Mercer B et al. The preterm prediction study: Significance of vaginal infections.
Am J Obstet Gynecol 1995; 173: 1231e1235.
39. Riduan JM, Hiller SL, Utomo B et al. Bacterial vaginosis and prematurity in Indonesia: Association in
early and late pregnancy. Am J Obstet Gynecol 1993; 169: 175e178.
40. Wennerholm U-B, Holm B, Mattsby-Baltzer I et al. Interleukin-1a, interleukin-6 and interleukin-8 in
cervico/vaginal secretion for screening of preterm birth in twin gestation. Acta Obstet Gynecol Scand
1998; 77: 508e514.
41. Vermeulen GM & Bruinse HW. Prophylactic administration of clindamycin 2% vaginal cream to reduce
the incidence of spontaneous preterm birth in women with an increased recurrence risk: a randomised
placebo-controlled double-blind trial. Br J Obstet Gynaecol 1999; 106: 652e657.
42. Edwards RK, Ferguson RJ, Shuster JJ et al. Assessment of cervical antibody concentrations fails to enhance
the value of cervical length as a predictor of preterm delivery. Am J Obstet Gynecol 2005; 192: 887e891.
43. Jamie WE, Edwards RK, Ferguson RJ & Duff P. The interleukin-6-174 single nucleotide polymorphism:
Cervical protein production and the risk of preterm delivery. Am J Obstet Gynecol 2005; 192:
1023e1027.
44. Genc MR, Onderdonk AB, Vardhana S et al. Polymorphism in intron 2 of the interleukin-1 receptor
antagonist gene, local midtrimester cytokine response to vaginal flora, and subsequent preterm birth.
Am J Obstet Gynecol 2004; 191: 1324e1330.
45. Genc MR, Vardhana S, Delaney ML et al. Relationship between a toll-like receptor-4 gene polymorphism, bacterial vaginosis-related flora and vaginal cytokine responses in pregnant women. Eur J Obstet
Gynecol Reprod Biol 2004; 116: 152e156.
46. Goffinet F, Kayem G, Maillard F et al. Detection of interleukin 6 mRNA by RT-PCR in vaginal secretions:
association with preterm delivery and neonatal infection in women with preterm labour and intact
membranes. Eur J Obstet Gynecol Reprod Biol 2005; 123: 167e173.
47. Kalinka J, Hanke W, Wasiela M & Laudanski T. Socioeconomic and environmental risk factors of bacterial vaginosis in early pregnancy. J Perinat Med 2002; 30: 467e475.
48. Oakeshott P, Hay P, Hay S et al. Association between bacterial vaginosis or chlamydial infection
and miscarriage before 16 weeks gestation: prospective community based cohort study. BMJ
2002; 325: 1334.

390 H. Leitich and H. Kiss

49. Oakeshott P, Hay P, Taylor-Robinson D et al. Prevalence of Mycoplasma genitalium in early pregnancy
and relationship between its presence and pregnancy outcome. BJOG 2004; 111: 1464e1467.
50. Cauci S, Hitti J, Noonan C et al. Vaginal hydrolytic enzymes, immunoglobulin A against Gardnerella vaginalis toxin, and risk of early preterm birth among women in preterm labor with bacterial vaginosis or
intermediate flora. Am J Obstet Gynecol 2002; 187: 877e881.
51. Dowd J, Permezel M, Garland S & de Crespigny L. Is there an interaction between cervical length and
cervical microbiology in the pathogenesis of preterm labour? Aust N Z J Obstet Gynaecol 2001; 41:
177e181.
52. Goldenberg RL, Mwatha A, Read JS et al. The HPTN 024 Study: the efficacy of antibiotics to prevent
chorioamnionitis and preterm birth. Am J Obstet Gynecol 2006; 194: 650e661.
53. Jeffcoat MK, Hauth JC, Geurs NC et al. Periodontal disease and preterm birth: results of a pilot intervention study. J Periodontol 2003; 74: 1214e1218.
54. Kim YH, Kim CH, Cho MK et al. Hydrogen peroxide-producing Lactobacilli in the vaginal flora of pregnant women with preterm labor with intact membranes. Int J Gynaecol Obstet 2006; 93: 22e27.
55. McMahon KS, Neerhof MG, Haney EI et al. Prematurity in multiple gestations: identification of patients
who are at low risk. Am J Obstet Gynecol 2002; 186: 1137e1141.
56. Onderdonk AB, Lee ML, Lieberman E et al. Quantitative microbiologic models for preterm delivery.
J Clin Microbiol 2003; 41: 1073e1079.
57. Ruiz RJ, Fullerton J & Brown CE. The utility of fFN for the prediction of preterm birth in twin gestations. J Obstet Gynecol Neonatal Nurs 2004; 33: 446e454.
58. Ruiz RJ, Fullerton J, Brown CE & Schoolfield J. Relationships of cortisol, perceived stress, genitourinary
infections, and fetal fibronectin to gestational age at birth. Biol Res Nurs 2001; 3: 39e48.
59. Shennan A, Crawshaw S, Briley A et al. A randomised controlled trial of metronidazole for the prevention of preterm birth in women positive for cervicovaginal fetal fibronectin: the PREMET Study. BJOG
2006; 113: 65e74.
60. Steyn PS, Odendaal HJ, Schoeman J et al. A randomised, double-blind placebo-controlled trial of ascorbic acid supplementation for the prevention of preterm labour. J Obstet Gynaecol 2003; 23: 150e155.
61. Schoeman J, Steyn PS, Odendaal HJ & Grove D. Bacterial vaginosis diagnosed at the first antenatal visit
better predicts preterm labour than diagnosis later in pregnancy. J Obstet Gynaecol 2005; 25: 751e753.
62. Tripathi R, Dimri S, Bhalla P & Ramji S. Bacterial vaginosis and pregnancy outcome. Int J Gynaecol Obstet
2003; 83: 193e195.
63. Vogel I, Gronbaek H, Thorsen P & Flyvbjerg A. Insulin-like growth factor binding protein 1 (IGFBP-1) in
vaginal fluid in pregnancy. In Vivo 2004; 18: 37e41.
64. Wenman WM, Joffres MR & Tataryn IV. A prospective cohort study of pregnancy risk factors and birth
outcomes in Aboriginal women. CMAJ 2004; 171: 585e589.
65. Wilks M, Wiggins R, Whiley A et al. Identification and H2O2 production of vaginal lactobacilli from
pregnant women at high risk of preterm birth and relation with outcome. J Clin Microbiol 2004; 42:
713e717.
66. Amsel R, Totten PA, Spiegel CA et al. Nonspecific vaginitis. Diagnostic criteria and microbiologic and
epidemiologic associations. Am J Med 1983; 74: 14e22.
67. Spiegel CA, Amsel R & Holmes KK. Diagnosis of bacterial vaginosis by direct Gram stain of vaginal fluid.
J Clin Microbiol 1983; 18: 170e177.
68. Nugent RP, Krohn MA & Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of Gram stain interpretation. J Clin Microbiol 1991; 29: 297e301.
69. Klebanoff MA, Hillier SL, Nugent RP et al. Is bacterial vaginosis a stronger risk factor for preterm birth
when it is diagnosed earlier in gestation? Am J Obstet Gynecol 2005; 192: 470e477.