Professional Documents
Culture Documents
METHODS
From the *Epidemiology and Health Systems Division, Menzies School of
Health Research, Darwin, Northern Territory, Australia; The Kirby
Institute, University of New South Wales (UNSW) Australia, Sydney,
New South Wales, Australia; and Robinson Institute, The University of
Adelaide, Adelaide, South Australia, Australia.
&
This review used the Preferred Reporting Items for Systematic Reviews and Meta Analyses guidelines,11 the Metaanalysis of Observational Studies in Epidemiology criteria,12
and the National Institute for Health and Clinical Excellence
guidelines.13
Search Strategy
We searched Medline, EMBASE, and BioMedCentral
to 16th May 2013 using the following terms: Trichomonas or
Trichomonas vaginalis AND adverse pregnancy outcome
or premature birth or premature delivery or premature labour or premature rupture of membranes or low birth weight
or intrauterine growth retardation or small for gestational age
or gestational age.
Study Selection
Studies were included if they were published before 16th
May 2013 and assessed 1 or more of the specified outcomes
in women with T. vaginalis and a control group of women
without the infection. The following perinatal outcomes were
369
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
Silver et al.
RESULTS
Study Characteristics
Table 1 describes the characteristics of included studies.
Briefly, the size of studies varied from 115 to 60,296 (median,
1038), with data collection spanning from 1973 to 2003. Only
one study focused on adolescents (defined as 13Y17 years),
the remaining studies recruited women of any reproductive
age (e.g., 13Y49 years).
Most (n = 7) studies were conducted in general antenatal clinics. Four studies classified their population as being
at high risk for adverse pregnancy outcomes, based on the presence of one or more of the following risk factors: young age
(G18 years), African American or Hispanic ethnicity, unmarried,
of low socioeconomic status, uninsured, or had a low income.
For the remaining studies, the study population included women
who were unselected in regard to risk of adverse pregnancy
outcome. Most (n = 8) studies excluded women with multiple pregnancies, and 4 studies included statistical adjustments
for a previous preterm birth. Of the 9 studies that reported on
preterm birth, 8 excluded multiple pregnancies, 2 excluded
women with an existing obstetric or medical condition known
to predict preterm birth, and 5 made adjustments to account
for differences in smoking and/or other recognized risk factors
for preterm birth.
The most frequently reported outcome was preterm
birth. Eight studies defined preterm birth as gestation less than
37 weeks and one study as gestation less than 36 weeks. Only
one study reported on preterm and very preterm birth. Four
studies reported on PROM and LBW. The remaining outcomes
were reported by 1 or 2 studies each. The methods used to estimate gestational age varied across the 11 studies (see Table S1,
http://links.lww.com/OLQ/A85), with the most frequently used
method being based on the date of the last menstrual period and
ultrasound (where available). Five studies did not provide any
information on method used.
The risk of methodological bias was rated as unclear for
most (n = 7) studies (Table S1, http://links.lww.com/OLQ/A85),
mostly due to a lack of information regarding performance and
detection bias. Three studies17Y19 were rated as being at low risk
for bias and one rated at high risk.20 A key issue was potential
confounding by other STI; coinfection was considered in the
analyses of 6 studies, no statistical adjustment was made in 4
studies, and it was unclear in 1 study. Three studies reported that
treatment was given to women with trichomoniasis (although
none reported the specific regimen), 4 studies stated that no
treatment was given, and a further 4 studies stated that it was
unclear whether treatment was given.
The timing of screening for trichomoniasis varied, with
2 studies collecting specimens during the first trimester only, 3
in the second only, 1 in the third trimester, 3 allowing for screening in any trimester, and 2 not stating when screening was done. In
2 studies, repeat specimens were collected on all women closer
to birth regardless of infection status, and in 1 study, positive
women only were repeat tested. The method of diagnosis was
microscopy in 4 studies, culture in 1, a combination of both in 5,
and unspecified in 1.
Perinatal Outcomes
Selection of Studies
The literature search identified 178 articles, of which 127
were excluded on the basis of information conveyed by the title
370
Nine studies (n = 81,001), including 4 cohort studies, reported on the RR of preterm birth. Because heterogeneity was
moderate (I2 = 62.7%, P = 0.006), a random-effects analysis
&
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
&
371
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
372
19
&
India (hospital)
NR
Iran (hospital)
2001Y2003 Tanzania/Zambia/
Malawi (NR)
1996Y02 USA (STI clinic)
Sweden (Pap
smear registry)
2001Y2002 Austria
(community)
2002Y2003 Iran (hospital)
1985Y1993 USA (community
hospital/health
center)
1973Y00
NR
Secondary analysis
of RCT
Retrospective case
control
Prospective cohort
Population-based
data linkage
Subgroup analysis
of RCT
Prospective cohort
Secondary analysis
of RCT and cohort
250
115
PROM, PTB
PTB, LBW,
PTB, LBW
PPROM, PTB
PTB
PTB
PROM
450
1531
2661
1223
1038
877
60,296
300
PROM, PTB
LBW, PTB, SB
Outcomes
NR
24.2/21.7 (13Y49)
24/25* (21Y28)
24.5/25* (NR)
24/23.8 (13Y47)
28.9 (NR)
NR (915)
23.3/23.8* (NR)
NR
27 T 5.7 (17Y39)
16.6/16.5* (13Y17)
Mean Age
(Target population)
NR
Underserved, uninsured
NR
56/76% unmarried
NR
SES
93% black
Race and/or
Ethnicity
NR
NR
NR
NR
70% Non-Hispanic
white or Asian
30% African American
NR
NR
NR
38% Black
30% Hispanic
32% White
Predominately black
*Infected/Uninfected.
Author provided missing information.
Non-Hispanic white women/African American.
With outcome/without outcome.
NR indicates not reported; RCT = randomized controlled trial; PTB, preterm birth; SB, stillbirth; Ed, education; PE, postpartum endometritis; NND, neonatal death.
Rasti et al.21
Johnson et al.17
Stringer et al.43
Azargoon et al.
French et al.42
Kiss et al.22
Buchmayer et al.
18
Prospective cohort
Minkoff et al.
Mathai et al.20
Prospective cohort
Prospective cohort
USA (hospital)
Study
Design
1982
1982
Country
(Setting)
Prospective cohort
Study
Year
USA (hospital)
Hardy et al.
Author(s)
TABLE 1.
G1
14.9
21.4
5.5
6.7
G1
G1
6.4
12.6
14.6
34.0
TV, %
Silver et al.
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
Publication Bias
A funnel plot to assess publication bias is presented in
Figure S1, http://links.lww.com/OLQ/A85. The graph plots the
log risk ratios for preterm birth against trial size as measured by
standard error of the log risk ratio. One study21 was removed
before the analysis due to small exposure numbers (n = 2). In
the bottom right quadrant, there is one small study with a large
standard error suggesting that the point estimate in this study
may be inflated, possibly because of poor methodological quality.
Of note, this study was removed in the sensitivity analyses
presented earlier. The plot also shows some asymmetry around
the log of the pooled estimate, with fewer studies with point
estimates below the pooled estimate and no studies in the bottom
left quadrant. This indicates that small studies demonstrating
Sexually Transmitted Diseases
&
DISCUSSION
To our knowledge, this is the first comprehensive review
and meta-analysis of the relationship between the presence of
infection with T. vaginalis and perinatal morbidity. The most
commonly reported end point, preterm birth, was 42% more
likely (95% CI, 15%Y75%) to occur in infected women. Based
on much fewer studies, there were also substantial increases in
the risk of PPROM (increase in risk, 41%; 95% CI, 10%Y82%)
and having a SGA baby (increase of 51%; 95% CI, 32%Y73%).
Because of insufficient data and/or concerns about heterogeneity, we were unable to draw reliable conclusions about the
impact of T. vaginalis on other outcomes including PROM,
LBW, still birth, neonatal death, or postpartum endometritis.
We followed a systematic process to identify studies and
extract data. It is nevertheless possible that we missed 1 or more
studies. Data were abstracted by 2 independent reviewers, to
minimize any subjectivity in coding. We also undertook a detailed assessment of the quality of each study.
There are several limitations. Moderate heterogeneity was
detected for the outcome preterm birth, and significant heterogeneity prevented a summary measure being calculated for several
important outcomes. The statistical heterogeneity found may
reflect differences in the design of eligible studies as well as
the characteristics of participants. The proportion of preterm
births in the control groups ranged from 5% to 33%, suggesting a
range in baseline maternal risk in eligible studies. There was
373
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
Silver et al.
374
&
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
&
pronounced with higher levels of T. vaginalis prevalence. To determine the ideal policy for screening in pregnancy, we must
consider prevalence, outcome, and treatment safety.
A single dose of 2 mg metronidazole provides parasitological cure34 and has not been associated with any teratogenic
effects in pregnancy,35 even when given in the first trimester.
However, concerns about the safety of treatment in pregnancy
arose after the publication of 2 studies a decade ago that reported
an increased risk of preterm birth and/or LBW associated with
metronidazole treatment. Both studies had potential methodological limitations. Klebanoff and colleagues9 administered a dose
4 times higher than the recommended regimen and the trial was
stopped early after reaching a third of the target sample when
an interim analysis revealed an increased risk of preterm birth in
the treatment arm. Kigozi and colleagues36 reported on a small
subgroup of women with trichomoniasis participating in a trial
of a combination of antimicrobials to prevent perinatal HIV transmission. Conversely, a recent chart review37 of metronidazole use
in pregnancy among 2829 women found no association with
preterm birth or LBW. There has been very little subsequent
research investigating appropriate strategies for managing trichomoniasis in pregnancy with most treatment studies focusing on
bacterial vaginosis,38 which is likely to elicit a different immunological response to T. vaginalis. Therefore, whether there are
indeed risks or benefits associated with treatment in pregnancy
remains unclear,39 and further studies are needed to answer this
important question40 to ensure clinical practice and guidelines are
supported by a solid evidence base.
Future research should be conducted in a high-prevalence
setting, where the impact of policy change will be greatest, and
there will be adequate numbers of infected women to provide
statistical accuracy. It is recommended that a trial of routine screening and treatment of T. vaginalis is conducted that includes either
a strict exclusion criteria for known confounders or, where unfeasible, statistical adjustment. Using a highly sensitive diagnostic technology such as nucleic acid amplification assays
would ensure all infections were detected. Swab collection at
multiple time points throughout pregnancy and measurement
of organism load would also need to be considered. Data should
be collected on a comprehensive range of perinatal outcomes,
and statistical approaches should be used to take account of
treatment of T. vaginalis and STI coinfection, in particular bacterial vaginosis.
CONCLUSIONS
Although our review provides strong evidence that T.
vaginalis in pregnancy is associated with preterm birth, there
remains a gap in evidence on treatment effects in pregnancy. If
treatment is found to be effective, this represents a significant
opportunity to make inroads into the prevention of perinatal
morbidity in populations where trichomoniasis is endemic. Preventing the acquisition of T. vaginalis, as well as early detection
and treatment, should remain a priority.
REFERENCES
1. World Health Organization. Prevalence and Incidence of Selected
Sexually Transmitted Infections, Chlamydia trachomatis, Neisseria
gonorrhoeae, Syphilis and Trichomonas vaginalis: Methods and
Results Used by WHO to Generate 2005 Estimates. Geneva, Switzerland:
World Health Organization, 2011.
2. Hobbs M, Sena A, Swygard H, et al. Trichomonas vaginalis and
trichomoniasis. In: Holmes K, Sparling P, Stamm W, eds. Sexually
Transmitted Diseases. New York: McGraw Hill, 2008:771Y794.
3. Poole DN, Scott McClelland R. Global epidemiology of Trichomonas
vaginalis. Sex Transm Infect 2013; 89:418Y422.
375
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.
Silver et al.
376
&
Copyright 2014 by the American Sexually Transmitted Diseases Association. Unauthorized reproduction of this article is prohibited.