796-807.qxd

CLINICAL OBSTETRICS AND GYNECOLOGYVolume 47, Number 4, 796–807 Ó 2004, Lippincott Williams & Wilkins Treatment of VaginalInfections to PreventPreterm Birth:A Meta-Analysis MARGARET A. RIGGS, PhD andMARK A. KLEBANOFF, MD Division of Epidemiology, Statistics and Prevention Research,National Institute of Child Health and Human Development,National Institutes of Health, Department of Health andHuman Services, Bethesda, Maryland Approximately 10% of all pregnancies in the United States end in preterm birth, but this 10% accounts for approximately 70% of the perinatal deaths and half of the long-term delivery with and without labor at various neurologic morbidity.1,2 A majority of mor- gestational ages, there was an inverse asso- bidity and mortality is concentrated in those ciation between positive cultures and gesta- infants delivered at less than 32 weeks ges- tional age, suggesting that microbial coloni- tation and birth weight ,1500 g.2,3 Intra- zation is more frequent among spontaneous uterine infection has been found to play a preterm births at lower gestational ages.8–11 major role in preterm birth and is thought It is likely that maternal and fetal signals in to be responsible for preterm birth in up to response to infection are what initiate pre- term labor after silent chorioamnionitis.7 of intrauterine infection and resultant amni- An active inflammatory response occurs in otic infection is thought to be the ascending brane cultures as shown by high concentra- Intrauterine infection may occur early in interleukin-6 in amniotic fluid.12,13 In vivo studies in women have linked increasedexpression of proinflammatory cytokinesin vaginal secretions, amniotic fluid, and Correspondence: Mark A. Klebanoff, MD, DESPR, maternal and fetal blood with adverse preg- NICHD, NIH, 6100 Bldg, Room 7B05, Bethesda, MD20892. E-mail: mk90h@nih.gov In the normal vaginal environment, lacto- bacilli protect against overgrowth of endog- (TNF), interleukin (IL)-1a and IL-1b, IL-6, Gardnerella vaginalis by producing hydro- activating a local inflammatory reaction. In gen peroxide, lactic acid, and bacteriocins.16 response, prostaglandin synthesis and re- lease is stimulated. Prostaglandins are re- characterized by reduction or elimination in the normal lactobacilli and overgrowth and are known to stimulate uterine contrac- sponsible for the release of metalloproteases though a substantial fraction of women with that are responsible for deterioration of cho- this condition are asymptomatic. In clinical rioamniotic membranes, leading to rupture.4 settings, it is usually defined as the presence of 3 of the following: elevated vaginal pH the association between other organisms and and/or amniotic fluid of pregnancies with fluid on a glass slide; and the presence of spontaneous preterm labor with intact mem- ‘‘clue cells’’ (vaginal epithelial cells with in- distinct borders due to attached bacteria) on microscopic examination of vaginal fluid.17 In research settings, a gram stain of vaginal micro-organisms most often associated with clinical chorioamnionitis and fetal infection et al18 is the preferred diagnostic method.
cocci (GBS) and Escherichia coli, are only occasionally found in the uterus.7 However, asymptomatic college students to 60% among the association between preterm birth and women attending an STD clinic.19 Bacterial lower genital tract colonization with these vaginosis has been consistently associated organisms is less clear. Up to 30% of preg- with an increased risk of preterm birth, but it is not known whether BV actually causes Group B streptococci is a leading cause of preterm birth, or is only a marker for intra- neonatal sepsis, and a substantial fraction preterm delivery have been examined in ani- mal, in vitro, and human studies.7 Bacterial during pregnancy has been associated with vaginosis is associated with increased con- preterm delivery.30 In the Vaginal Infections centrations of elastase, mucinase, and siali- dase in the vagina and cervix,21,22 and, most colonized with T. vaginalis had a 30% higher importantly, may be a marker for microbial risk of delivering an infant with low birth colonization of the upper genital tract.7 The weight or delivering before term and a 40% higher risk of giving birth to an infant who was both preterm and of low birth weight.
sulting in phospholipase and prostaglandin Eleven percent of preterm deliveries among ered potentially attributable to trichomonia- the decidua and fetal membranes to produce sis.30 Trichomoniasis during pregnancy is or yeast) and (antibiotic or treatment) and (preterm or premature) and (randomized or randomised) and (clinical trial). Bibliogra- phies of all relevant articles were reviewed for further potential references. Studies with fected.32 It has been associated with adverse titles or abstracts discussing antibiotic treat- pregnancy outcomes in several reports.32,33 ment of vaginal infections and prematurity were retrieved. Any study with primary data investigating a relevant exposure and out- genitourinary tract of women and are con- come was submitted for further review. The sidered a potential cause of intrauterine in- following criteria were used to select studies fection.34 Patients with amniotic fluid cul- for inclusion: article (original published), tures positive for only U. urealyticum had a study design (randomized placebo-controlled significantly higher rate of adverse perinatal clinical trial), population (women at less than 37 weeks gestation, not in labor with intact amniotic membranes), intervention (antibiotic treatment) and relevant outcome, gestational weight, neonatal morbidity, and perinatal age at birth, as well as any of the following death. However, although isolation of M.
hominis from the lower genital tract is asso- births when the mother had a prior preterm.
birth,36 isolation of U. urealyticum from the lower tract does not appear to increase the studies in languages other than English were In response to the association between in- stract or upon translation that indicated they fectious micro-organisms and adverse preg- were not clinical trials. When one report dif- nancy outcome, many investigators have tried to show that treatment of these infections can patients into the same study, we included prevent preterm birth. Several different or- the final report. We evaluated total preterm ganisms have been targeted, using a variety births as the outcome, unless only data on of regimens with different antibiotics and spontaneous preterm birth were presented, routes of administration. Some studies have in which case we evaluated spontaneous pre- term birth. Some studies treated women re- others enrolled women who have had a pre- gardless of the presence of infection (eg, mass community treatment trials; treatment of women who had a prior preterm birth re- to evaluate the role of antibiotic treatment of gardless of the current presence of specific a variety of vaginal infections in pregnancy infections). If those studies presented results specifically for a particular infection, we in-cluded those results; otherwise the studywas excluded. Two studies enrolled women with either BV or intermediate flora. One(McDonald et al40) presented most results those specific results in this review. The other (Ugwumadu et al41) presented nearly using keywords (vaginitis or vaginosis or va- all results only for the entire group, and there- gina* infection) and (group B strep* or GBS fore, we used the results for all randomized Quality of the trials was assessed on a 6- point scale using a modified Jadad Scale.42 evaluated ORs and confidence intervals for Studies were given 1 point each for descrip- subgroups defined by antibiotic used, route tion of randomization, description of double of administration, and obstetrical history.
dropouts, analyzed as intent to treat, and an was assessed to determine the appropriate- additional point each was given for both ap- ness of combining trial results.43 To calcu- late overall results for each outcome and all blinding or a point was deducted for each if subgroups, we generally used the fixed ef- they were inappropriately done. Studies with fects model. If significant heterogeneity was a score less than 3 were considered of poor used a random effects model instead. Trialswith zero cells in both the study and control groups were not included in the calculation The studies identified by the criteria de- scribed above were reviewed independentlyby both authors. Disagreements were settledby discussion and consensus. The following inclusion criteria were applied: 1) appropri- A total of 155 articles were retrieved.
ate exposure and outcome measures, as de- fined above; 2) randomized clinical trial; and 3) relative risk (RR) or odds ratio (OR) with phy review. Sixty-seven publications were 95% confidence interval (CI) provided or submitted for review. The remaining articles able to be calculated from the data presented were discarded because of redundancy with in the article. The following data were ab- other articles, data reporting, lack of original stracted onto standardized forms: publication data, Jadad score of 0, or lack of infection- year, gestational age when treatment began, specific results. The remaining 16 studies antibiotic therapy, micro-organisms identi- are summarized in Table 1.44–54 Only one fied, birth weight, gestational age at birth, study was included in the analysis with a Jadad score of ,3 (Odendaal et al51). All births when the mother had a prior preterm, studies that were included followed a ran- whether it was analyzed as intent to treat, ex- tent of blinding, extent of controlling for po- studies were stratified first by type of bacte- tential confounding, and relevant risk ratio rial infection and then by type of antibiotic therapy used for treatment of BV and by his-tory of a previous preterm birth.
STATISTICAL ANALYSISFor the outcomes in each individual study, we determined treatment effectiveness by cal- There has only been 1 randomized, placebo- controlled double-blind study of treatment of 95% CIs. This produced a weighted average GBS colonization to prevent preterm birth (Klebanoff et al28). In that study, pregnant version 4.2.5. There were few clinical trials placebo continuously for up to 10 weeks in the third trimester. No statistically significant ing subanalysis. However, there were many benefit was seen for birth weight ,2500 g studies of treatment of bacterial vaginosis, ,37 weeks (OR 0.91, 95% CI 0.61–1.36), or others. Therefore, in addition to an overall preterm PROM (OR 1.0, 95% CI 0.46–2.11).
TABLE 1. Summary of Study Characteristics birth, or low birth weight infant in the prior Only 1 clinical trial of treatment of Chla- pregnancy, no reduction in low birth weight birth was noted by treatment in the current done (Martin et al32). The design and treat- pregnancy (1.19, 95% CI 0.50–2.85).
ment was identical to the GBS trial notedabove. There was no statistically significantreduction in low birthweight (OR 0.74, 95% TRICHOMONAS VAGINALISThere have been 2 randomized controlled CI 0.38–1.45), preterm birth (OR 0.89, 95% trials of treatment of trichomoniasis to pre- CI 0.49–1.62), or preterm PROM (OR 0.70, vent adverse pregnancy outcome. One trial (Klebanoff et al54) screened asymptomaticwomen for trichomoniasis and treated them with 2 doses of 2 g each of metronidazole or placebo at 16 to 23 weeks and again at 24 to Ureaplasma urealyticum using the same de- 29 weeks. The other (Kigozi et al53) was a sign as the studies above.52 No statistically treatment trial in which villages in Uganda birth weight (1.36, 95% CI 0.85–2.17) or in preterm birth (1.02, 95% CI 0.67–1.54).
treatment with metronidazole, azithromycin, Among women who had a miscarriage, still- Both studies found an elevated risk of low Eleven studies evaluated the association be- birth weight among the treated women; the tween treatment of BV and gestational age elevation was statistically significant in one.
to 2.26 (Fig. 1). Eight of these 11 found the 2 trials was 1.68 (95% CI 1.11–2.53), no significant association, and again the and there was no significant heterogeneity between the studies (P ¼ 0.20). Results were was not statistically significant. However, there was significant heterogeneity between found an elevated risk in the treated women, studies (P ¼ 0.007). Five studies evaluated which was significant in one. The combined OR for preterm birth in the 2 trials was 1.71 (95% CI 1.19–2.46); there was no signifi- (95% CI 0.33–1.81), but there was signifi- was not reduced in the one study reporting this outcome (OR 1.1, 95% CI 0.5–2.3).
Six studies evaluated the association be- Most studies of antibiotic treatment of lower tween treatment of BV with any antibiotic genital tract infection to prevent preterm prior preterm birth, with ORs ranging from bacterial vaginosis; there have been 12 such 0.11 to 2.44 (Fig. 2). Three of these studies studies, one of which was a subset of a sec- found a significant benefit of treatment, 1 ond,55 for a total of 11 included in this re- found significant harm, and 2 found no sig- view. Six studies evaluated the association nificant difference with treatment. The com- but the results were extremely variable (het- 3.45. Five of these 6 studies found no signif- icant association, and the combined OR of for the 4 of these studies that used metroni- 0.92 (95% CI 0.60–1.40) was not statistically dazole alone was 0.75 (95% CI 0.25–2.24), significant. However, there was a suggestion but these studies were heterogeneous in their of heterogeneity between studies (P ¼ 0.07).
results (P ¼ 0.002). The fifth study (Hauth FIGURE 1. Treatment of bacterial vaginosis to prevent preterm birth.
FIGURE 2. Treatment of bacterial vaginosis to prevent preterm birth inwomen with a previous preterm birth.
et al47) used metronidazole and erythromy- birth, but several studies have evaluated oral between treatment of BV with vaginal clin- and there was no significant heterogeneity 0.89, which was not statistically significant between studies (P ¼ 0.11). Four studies evaluated the association between treatment some possible evidence of statistical hetero- of BV with vaginal clindamycin and gesta- geneity between these studies (P ¼ 0.06).
Four studies evaluated the association be- from 0.75 to 2.26. None of the studies found tween metronidazole treatment and preterm a significant association. The combined OR birth; one study found a significant improve- was 1.18 (95% CI 0.83–1.69), and there was no significant heterogeneity between studies (P ¼ 0.59). Results for preterm PROM were not statistically significant (OR 0.92, 95% reported by 2 studies, with ORs of 1.14 and CI 0.52–1.62). Nevertheless, there was sig- nificant heterogeneity between studies (P ¼ 0.66–5.38), and there was no significant het- 0.009). Three studies evaluated the associa- tion between treatment with metronidazole FIGURE 3. Metronidazole treatment of bacterial vaginosis to prevent pre-term birth.
those who have a history of a previous pre- 0.10 to 1.13. One study found a significant term delivery). Current evidence does not support routine testing for BV.’’56 Similarly, significant, although the confidence limits the U.S. Preventive Services Task Force57 notes that current evidence does not support and there was significant heterogeneity be- screening for and treating BV in a general obstetrical population, and the heterogeneity between studies precludes making a defini- tive recommendation on screening and treat- ing women with a prior preterm birth.
significant improvement in low birth weight with treatment of BV (OR 0.81; 95% CI 0.43– most study results, not just in the studies that 1.51); there was an improvement in sponta- treated women with a prior preterm birth.
neous preterm birth (OR 0.26; 95% CI 0.12– Although several studies reported a benefit 0.58), although total preterm birth was not of treating BV, other studies reported an overall adverse effect, and several more re- 0.31–1.04). Two studies used a combination ported significant increases in preterm birth in various subgroups of women. Therefore, et al47 and Andrews et al44) administered at it is possible that differences in the popula- an average gestation of 23 weeks. There was tions enrolled in the various studies and/or an overall significant reduction in preterm count for the highly variable results. Until no heterogeneity between studies (P ¼ 0.29).
there is more understanding of the sourceof these different results, caution is advisedin treating BV to prevent preterm birth.
In spite of a consistent association between duced interesting results, with one study a variety of lower genital tract infections and (Ugwumadu et al41) reporting a benefit of adverse pregnancy outcome, results of this treatment of prevention of preterm delivery.
meta-analysis show that screening for and However, this is the only trial to date exam- treating vaginal infections in a general pop- ining treatment with this antibiotic. Data on the occurrence of PROM were not reported, does not appear to reduce the rate of preterm and randomization did not balance the base- delivery or low birth weight. This is in con- line history of previous miscarriage between treatment groups. In addition, the lack of an no reduction in preterm birth by screening effect of treatment on birth weight, which is for bacterial vaginosis. The only exception gestational age, seems inconsistent with the may be for those with a prior history of pre- dramatic reduction in preterm birth that was term birth, but the heterogeneity between observed. These inconsistencies need to be re- studies precludes a definitive conclusion.
solved before recommendations can be made.
vention has recognized this in their 2002 surprising finding that preterm birth and by treatment. There is no obvious explana- ‘‘Evaluation for bacterial vaginosis (BV) tion for this surprising finding, although it first prenatal visit for asymptomatic patients T. vaginalis are infected with M. hominis, who are at high risk for preterm labor (eg, which might be released when the parasite is killed.58 Trichomonads are often noted in though no specific infections were reported, Pap smears obtained during pregnancy from neither study found a statistically significant asymptomatic women, and these trial results benefit of treatment on preterm birth. (OR raise the difficult question of whether to treat the infection in this circumstance. Perhaps 57% preterm in treated and placebo groups, a reasonable course would be for the physi- P ¼ 0.515.61 The Andrews et al study found cian to explain the evidence to the woman a substantial reduction in mean gestational and offer the option of treating, or of delay- age and birth weight among treated women; ing treatment until after pregnancy unless the birth weight difference was statistically symptoms develop before then. If the latter significant. The latter study in particular course is taken, it seems reasonable to ad- should give pause to a general recommenda- tion in favor of early pregnancy treatment not prevent preterm birth, there are other reasons to screen and treat this organism.
have shown an association between a variety First, there is the public health benefit of de- of vaginal infections and preterm birth, there tection and treatment of this STD. Second, is little evidence that treatment of these infections lowers the risk. The statistical het- prevents maternal postpartum morbidity, as erogeneity of the bacterial vaginosis studies, well as transmission to the infant at delivery, with the attendant morbidity. The Centers groups of women who benefit, and possibly for Disease Control (CDC) currently recom- mend screening at the first prenatal visit, during pregnancy. Therefore, future research during the third trimester.56 The recommen- also have upper tract infection and might screening and treatment allows for reinfec- therefore benefit from treatment; on inflam- tion later and also acknowledges that evi- matory response modifiers such as cytokine dence is lacking that treatment early in preg- nancy can prevent preterm birth. Similarly, anti-inflammatory therapy in addition to the CDC’s recommendations for screening antibiotics in the presence of infection.63 and treatment of group B streptococci areintended to prevent neonatal sepsis, not toprevent preterm birth.59 birth weight to infant mortality and child- but were not included in this review because they did not report results for women with any specific infection. One study treated women who had either U. urealyticum or M. hominis tremely immature infants—a perinatal di- with either erythromycin or clindamycin.60 Unfortunately, the results of erythromycin treatment were not presented separately for 3. Ventura SJ, Martin JA, Curtin SC, et al.
Births: final data for 1997. Natl Vital Stat each infection, and the clindamycin results were not presented at all. The other, pre- 4. Romero R, Mazor M. Infection and preterm labor. Clin Obstet Gynecol. 1988;31:553– who recently had a very preterm birth with metronidazole and azithromycin and waited 5. Lettieri L, Vintzileos AM, Rodis JF, et al.
Does ‘‘idiopathic’’ preterm labour resulting in preterm birth exist? Am J Obstet Gynecol.
microbial and epidemiologic associations.
6. Goncalves LF, Chaiworapongsa T, Romero 18. Nugent RP, Krohn MA, Hillier SL. Reliabil- R. Intrauterine infection and prematurity.
ity of diagnosing bacterial vaginosis is im- stain interpretation. J Clin Microbiol. 1991; 7. Goldenberg RL, Hauth JC, Andrews WC. In- trauterine infection and preterm delivery.
19. Mead PB. Epidemiology of bacterial vagino- sis. Am J Obstet Gynecol. 1993;169:446– 8. Cassell G, Hauth J, Andrews W, et al. Cho- 20. Krohn MA, Hillier SL, Nugent RP, et al. The gestational age in women delivered follow- genital flora of women with intraamniotic ing spontaneous labor versus indicated deliv- infection. J Infect Dis. 1995;171:1475– ery [abstract]. Am J Obstet Gynecol. 1993; 21. McGregor JA, French JI, Jones W, et al.
Bacterial vaginosis is associated with pre- et al, eds. Sexually Transmitted Diseases sialidase: results of a controlled trial of top- ical clindamycin cream. Am J Obstet Gyne- 22. Andrews WW, Tsao J, Goldenberg RL, et al.
10. Hauth JC, Andrews WW, Goldenberg RL.
Infection-related risk factors predictive of midtrimester cervical sialidase level elevea- spontaneous labor and birth. Prenat Neonat tion to predict subsequent spontaneous pre- term birth. Am J Obstet Gynecol. 1999;180: 11. Andrews WW, Hauth JC, Goldenberg RL, et al. Amniotic fluid interleukin-6: correlation 23. Schwebke JR. Bacterial vaginosis. Curr In- with upper genital tract microbial coloniza- tion and gestational age in women delivered 24. Lamont RF, Rose M, Elder MG. Effect of after spontaneous labor versus indicated de- bacterial products on prostaglandin E pro- 12. Andrews WW, Goldenberg RL, Hauth JC.
25. Mazor M, Wiznitzer A, Maymon E, et al.
Preterm labor: emerging role of genital tract Changes in amniotic fluid concentrations of prostaglandins E2 and F2 alpha in women with preterm labor. Isr J Med Sci. 1990;26: 13. Romero R, Mazor M, Sepulveda W, et al.
Tumor necrosis factor in preterm and term la- 26. Leitich H, Brunbauer M, Bodner-Adler B, bor. Am J Obstet Gynecol. 1992;166:1576– et al. Antibiotic treatment of bacterial vagi- nosis in pregnancy: a meta-analysis. Am J ductive failure I: immunological factors.
27. Regan JA, Chao S, James LS. Premature rup- Hum Reprod Update. 2000;7:113–134.
15. Saji F, Samejima Y, Kamiura S, et al. Cyto- group B streptococcal colonization of moth- kine production in chorioamnionitis. J Re- ers. Am J Obstet Gynecol. 1981;141:184– 16. Hillier SL, Witkin SS, Krohn MA, et al. The 28. Klebanoff MA, Regan JA, Rao AV, et al.
relationship of amniotic fluid cytokines and preterm delivery, amniotic fluid infection, prematurity study: results of a clinical trial histologic chorioamnionitis, and chorion in- fection. Obstet Gynecol. 1993;81:941–948.
colonized with group B streptococci.
17. Amsel R, Totten PA, Spiegel CA, et al. Non- specific vaginitis: diagnostic criteria and 29. Weisman LE, Stoll BJ, Creuss DF, et al. Early- terial vaginosis: a randomised controlled onset group B streptococcal sepsis: a current assessment. J Pediatr. 1992;121:428–433.
42. Jadad A. Randomised Controlled Trials.
30. Cotch MF, Pastorek JG 2nd, Nugent RP, et al.
Trichomonas vaginalis associated with low 43. Whitehead A, Whitehead J. A general para- metric approach to the meta-analysis of ran- domized clinical trials. Stat Med. 1991;10: 31. Gulmezoglu AM. Interventions for tricho- moniasis in pregnancy [Cochrane Review].
44. Andrews WW, Sibai BM, Thom EA, et al.
In: The Cochrane Library, Issue 1. Chiches- Randomized clinical trial of metronidazole ter, UK: John Wiley & Sons; 2004.
32. Martin DH, Eschenbach DA, Cotch FA, et al.
preterm delivery in fetal fibronectin-positive Double-blind placebo-controlled treatment women. Obstet Gynecol. 2003;101:847–855.
trial of Chlamydia trachomatis endocervical 45. Carey JC, Klebanoff MA, Hauth JC, et al.
33. Andrews WW, Hauth JC, Goldenberg RL.
terial vaginosis. N Engl J Med. 2000;342: Infection and preterm birth. Am J Perinatol.
46. Gauschino S, Ricci E, Franchi M, et al.
34. Taylor-Robinson D. Infections due to species Treatment of asymptomatic bacterial vagi- of mycoplasma and ureaplasma: an update.
nosis to prevent pre-term delivery: a ran- domised trial. Eur J Obstet Gynecol Repro 35. Yoon BH, Romero R, Park JS, et al. Micro- bial invasion of the amniotic cavity with Ureaplasma urealyticum is associated with et al. Reduced incidence of preterm deliv- a robust host response in fetal, amniotic, in women with bacterial vaginosis. N Engl 36. Hillier SL, Nugent RP, Eschenbach DA, et al.
48. Joesoef MR, Hillier SL, Wiknjosastro G, Association between bacterial vaginosis and et al. Intravaginal clindamycin treatment preterm delivery of a low-birth-weight in- for bacterial vaginosis: effects on preterm fant. N Engl J Med. 1995;333:1737–1742.
delivery and low birth weight. Am J Obstet et al. Antepartum cultures for Ureaplasma 49. Kekki M, Kurki T, Pelkonen J, et al. Vaginal clindamycin in preventing preterm birth and pregnancy outcome. Am J Obstet Gynecol.
women with bacterial vaginosis: a random- 38. Mertz HL, Ernest JM. Antibiotics and pre- ized, controlled trial. Obstet Gynecol. 2001; 50. Morales WJ, Schorr S, Albritton J. Effect of 39. Lamont RF. The role of infection in preterm metronidazole in patients with preterm birth in preceding pregnancy and bacterial vagi- nosis: a placebo controlled, double blind 40. McDonald HM, O’Loughlin JA, Jolley PT, study. Am J Obstet Gynecol. 1994;171:345–349.
et al. Impact of metronidazole therapy on 51. Odendaal HJ, Popov I, Schoeman J, et al.
preterm birth in women with bacterial vagi- Preterm labour—is bacterial vaginosis in- nosis flora (Gardnerella vaginalis): a ran- volved? S Afr Med J. 2002;92:231–234.
domised, placebo controlled trial. Brit J 52. Eschenbach DA, Nugent RP, Rao AV, et al.
Obstet Gynecol. 1997;104:1391–1397.
41. Ugwumadu A, Manyonda I, Reid F, et al. Ef- of erythromycin for the treatment of Urea- fect of early oral clindamycin on late miscar- riage and preterm delivery in asymptomatic women with abnormal vaginal flora and bac- cells. Arch Microbio. 2001;175:70–74.
59. Schrag S, Gorwitz R, Fultz-Butts K, et al.
Prevention of perinatal group B streptococ- 54. Klebanoff MA, Carey JC, Hauth JC, et al.
Failure of metronidazole to prevent preterm 60. McCormack WM, Rosner B, Lee Y-H, et al.
Effect on birth weight of erythromycin treat- asymptomatic Trichomonas vaginalis infec- ment of pregnant women. Obstet Gynecol.
tion. N Engl J Med. 2001;345:487–493.
55. Kurkinen-Raty M, Vuopala S, Koskela M, 61. Andrews W, Goldenberg R, Hauth J, et al. In- et al. A randomised controlled trial of vagi- terconceptional antibiotics to prevent spon- nal clindamycin for early pregnancy bacterial taneous preterm birth: a randomizedtrial.
vaginosis. Br J Obstet Gynecol. 2000;107: 62. Macones G, Parry S, Marder S, et al. Evi- 56. Workowski KA, Levine WC. Sexually trans- dence of a gene-environment interaction in mitted disease treatment guidelines—2002.
the etiology of spontaneous preterm birth.
57. Guise J-M, Mahon SM, Aickin M, et al.
63. Gravett M, Sadowsky D, Witkin S, et al.
Screening for bacterial vaginosis in preg- Immunomodulation plus antibiotics to pre- nancy. Am J Prev Med. 2001;20:62–72.
vent preterm delivery in experimental intra- 58. Rappelli P, Carta F, Delogu G, et al. Myco- amniotic infection. Am J Obstet Gynecol.

Source: http://www.qtnpr.ca/files/treatment_of_vaginal_infection_to_prevent_preterm_birth.pdf