INTRODUCTION — Bacterial vaginosis (BV) is the most common cause of vaginal discharge in women of childbearing age, accounting for 40 to 50 percent of cases [1,2]. The worldwide prevalence ranges from 11 to 48 percent of women of reproductive age, with variations according to the population studied [3]. In the United States, the National Health and Nutrition Examination Survey, which included results from self-collected vaginal swabs from over 3700 women, estimated the prevalence of BV was 29 percent in the general population of women aged 14 to 49 years [4]. The prevalence in African-American women was 50 percent.

The absence of inflammation is the basis of the term "vaginosis" rather than "vaginitis."

PATHOGENESIS AND MICROBIOLOGY — BV is not due to a single organism. Instead it represents a complex change in the vaginal flora characterized by a reduction in concentration of the normally dominant hydrogen-peroxide producing lactobacilli and an increase in concentration of other organisms, especially anaerobes. These include Gardnerella vaginalis, Mycoplasma hominis, Prevotella species, Porphyromonas species, Bacteroides species, anaerobic Peptostreptococcus species, Fusobacterium species, and Atopobium vaginae [5,6]. The mechanism by which the floral imbalance occurs and the role of sexual activity in the pathogenesis of BV are not clear.

Hydrogen-peroxide producing lactobacilli appear to be important in preventing overgrowth of the anaerobes normally present in the vaginal flora. With the loss of lactobacilli, pH rises and massive overgrowth of vaginal anaerobes occurs. These anaerobes produce large amounts of proteolytic carboxylase enzymes, which break down vaginal peptides into a variety of amines that are volatile, malodorous, and associated with increased vaginal transudation and squamous epithelial cell exfoliation, resulting in the typical clinical features observed in patients with BV. (See 'Clinical features' below.)

The rise in pH also facilitates adherence of G. vaginalis to the exfoliating epithelial cells, thereby creating the "clue cells" that are diagnostic of the disorder. (See 'Diagnosis' below.)

The difference in vaginal flora between women with and without BV was illustrated in a study that used broad range DNA probes to determine the vaginal flora of 27 women with BV and 46 controls [7]. Overall, 35 bacterial phylotypes were identified in women with BV, including 16 which were newly recognized. Women with BV had a mean of 12.6 phylotypes (range 9 to 17) per sample compared to 3.3 phylotypes (range 1 to 6) per sample in women without BV.

RISK FACTORS — Risk factors for acquisition of BV include multiple or new sexual partners, douching, and cigarette smoking [1,2,8-12]. Although sexual activity is a risk factor for the condition, BV can occur in women who have never had vaginal intercourse [13]. There is a high occurrence of BV and concordance of flora in women who have sex with women, suggesting sexual transmission is important in this setting [14,15]. It is not clear, however, whether one type of sexual activity may be more important in the pathogenesis of infection than another. As an example, oral-genital sex may be a more important risk factor than penile intromission into the vagina. (See "Gynecologic and obstetric care for lesbians".)

CLINICAL FEATURES — Approximately 50 to 75 percent of women with BV are asymptomatic [13,16,17]. Those with symptoms present with an unpleasant, "fishy smelling" discharge that is more noticeable after coitus. The discharge is off-white, thin, and homogeneous.

Dysuria and dyspareunia are rare, while pruritus, erythema, and inflammation are typically absent. BV can be associated with cervicitis (endocervical mucopurulent discharge or easily induced bleeding), with or without concomitant chlamydial or gonococcal infection [18].

DIAGNOSIS — The general diagnostic approach to women with vaginal discharge is reviewed separately. (See "Diagnostic approach to women with vaginal discharge or vulvovaginal symptoms".)

The diagnosis of BV is based on clinical findings and laboratory testing; criteria have been established that are simple and useful in office practice [16]. Three of the four criteria (ie, Amsel criteria) listed below are necessary for diagnosis, although the first three findings are sometimes present in patients with trichomoniasis (table 1).

• Homogeneous, thin, grayish-white discharge that smoothly coats the vaginal walls
• Vaginal pH greater than 4.5
• Positive whiff-amine test, defined as the presence of a fishy odor when 10 percent potassium hydroxide (KOH) is added to a sample of vaginal discharge
• Clue cells on saline wet mount

The presence of clue cells diagnosed by experienced microscopists is the single most reliable predictor of BV [19]. Clue cells are vaginal epithelial cells studded with adherent coccobacilli that are best appreciated at the edge of the cell (picture 1). In women with BV, at least 20 percent of the epithelial cells on wet mount should be clue cells. Using gram stain as the gold standard, the sensitivity of Amsel criteria for diagnosis of BV is over 90 percent [20]. Specificity is 77 percent.

Gram staining of vaginal secretions [21] is a reliable method for the diagnosis of BV (picture 2A-B and table 2), but is mostly performed in research studies because it is more cumbersome to use in clinical practice than Amsel criteria [22-24]. If clinical criteria are used to define infection, then the sensitivity reported for Gram stain ranges from 62 to 100 percent [25].

Vaginal culture has no role in diagnosis because there are no bacteria that are specific for BV. Although cultures for G. vaginalis are positive in almost all women with symptomatic infection, the organism is detected in up to 50 to 60 percent of healthy asymptomatic women; thus, its presence alone is not diagnostic of BV.

DNA probes for G. vaginalis are expensive and suffer from the same limitation as culture, but provide fairly quick results and may be useful to practitioners unable to perform microscopy. In one study, as an example, the combination of a positive DNA probe (concentration of G. vaginalis ≥10(7) CFU/mL) and vaginal pH greater than 4.5 had a sensitivity and specificity of 95 and 99 percent, respectively, for diagnosis of BV when clinical criteria were used as the diagnostic standard [26].

Diagnostic cards (eg, Femcard, QuickVue, Pip Activity TestCard) are another rapid test for confirming the clinical suspicion of BV. These cards are particularly useful for practitioners not able to perform microscopy. One group reported these tests detected the presence of elevated vaginal pH and increased amines with sensitivity and specificity of 87 and 92 percent, respectively [27], although others have reported lower values [28].

The Papanicolaou smear is not reliable for diagnosis of BV (sensitivity 49 percent, specificity 93 percent) [29]. There is no information on the sensitivity and specificity of liquid based methods. If a cervical cytology smear suggests BV, women should undergo standard diagnostic testing (vaginal examination, vaginal pH, whiff-amine test, wet mount) and treatment, if appropriate.

Differential diagnosis — BV is usually suspected because of high vaginal pH (>4.5). Other causes of increased pH include trichomoniasis, atrophic vaginitis and desquamative inflammatory vaginitis. These four entities are easily distinguishable by clinical and microscopy features. Women with BV do not have dyspareunia or signs of vaginal inflammation; in contrast, women with atrophic vaginitis, desquamative inflammatory vaginitis, and trichomoniasis usually have these signs and symptoms. Both atrophic vaginitis and desquamative inflammatory vaginitis are associated with an increased number of parabasal cells on microscopy, which is not observed in women with BV. A large number of polymorphonuclear leukocytes on microscopy are characteristic of desquamative inflammatory vaginitis, trichomoniasis, and atrophic vaginitis with infection, but not BV. Visualization of trichomonads readily makes the diagnosis of trichomoniasis in the setting of an elevated pH.

COMPLICATIONS

• Pregnant women with BV are at higher risk of preterm delivery (see 'Treatment in pregnancy' below [30-33].
• There is a causal relationship between BV and endometrial bacterial colonization, plasma-cell endometritis, postpartum fever, post-hysterectomy vaginal cuff cellulitis, and postabortal infection [34-36].
• BV is a risk factor for HIV acquisition and transmission [37,38]. (See "HIV and women".)

• BV is a risk factor for acquisition of HSV-2, gonorrhea, and chlamydial infection [39,40].
• BV is more common among women with pelvic inflammatory disease (PID), but it is not clear if it is an independent risk factor for this disease [41-43]. In the largest series, diagnosis of BV based on gram stain did not show a strong, independent correlation with development of PID [42], but microbiology on the same patients indicated heavy growth of BV-associated microorganisms increased PID risk (OR 2.03, 95% CI 1.16-3.53) [43].

It has been hypothesized that the increased risk of acquisition of sexually transmitted infections may be due to lack of hydrogen peroxide–producing lactobacilli in the vaginal flora of women with BV; other factors associated with BV infection, such as local cytokine production, may also play a role.

• Some studies have found that the presence of G.vaginalis on the cervix, as detected on Papanicolaou smear, is associated with high grade squamous intraepithelial neoplasia (HSIL); however, a causal relationship has not been proven and others have reported no association [44-46].

TREATMENT OF NONPREGNANT WOMEN — BV resolves spontaneously in up to one-third of nonpregnant and one-half of pregnant women [47-50]. Treatment is indicated for relief of symptoms in women with symptomatic infection and to prevent postoperative infection in those with asymptomatic infection prior to abortion or hysterectomy (see 'Gynecologic procedures' below.

Treatment of BV may also reduce the risk of acquiring other STDs, including HIV [38,51,52]. For this reason, some experts now support the concept of treating all women with BV regardless of presence or absence of symptoms.

Asymptomatic pregnant women with previous preterm births may also benefit, but screening and treatment of these women is controversial (see 'Treatment in pregnancy' below.

Options — Metronidazole or clindamycin administered either orally or intravaginally will result in a high rate of clinical cure (70 to 80 percent at four weeks of follow-up) (table 3) [53-56]. Oral medication is more convenient, but associated with a higher rate of systemic side effects, than vaginal administration. Follow-up is unnecessary if symptoms resolve [51].

Metronidazole — Metronidazole is the most successful therapy. Most comparative studies using multiple divided-dose oral regimens for one week achieved early rates of clinical cure in excess of 90 percent, and cure rates (by Amsel criteria) of approximately 80 percent at four weeks [51,54,57-61]. A randomized trial showed that short-term cure rates were significantly higher when the initial course of metronidazole therapy was 14 days rather than 7 days [11]. However, long-term cure rates (21 days after completion of therapy) were similar for both treatment regimens.

The oral regimen we recommend is 500 mg twice daily for seven days [51]. Topical vaginal therapy with 0.75 percent metronidazole gel (5 g once daily for five days) is as effective as oral metronidazole [51,58,59,62]. The choice of oral versus topical therapy depends upon patient preference. The 2 g single dose therapy has lower efficacy [63] and is no longer recommended for treatment of BV [51].

Side effects of metronidazole include a metallic taste, nausea (in 10 percent of patients), transient neutropenia (7.5 percent), a disulfiram-like effect with alcohol, interaction with warfarin, and peripheral neuropathy. Allergy to metronidazole is uncommon; it manifests as rash, urticaria, pruritus, and rarely, anaphylaxis, which can be successfully treated by oral desensitization [64].

Clindamycin — The efficacy of clindamycin has been demonstrated in multiple randomized trials, both comparative and placebo controlled [54,65-68]. Topical vaginal therapy with 2 percent clindamycin cream (5 g of cream containing 100 mg of clindamycin phosphate) is recommended as a seven-day regimen, but may be less effective than the metronidazole regimens [51]; nevertheless, topical clindamycin cream is a reasonable therapeutic choice.

Clindamycin cream should not be used concurrently with latex condoms, which may be weakened.

Alternative regimens include oral clindamycin (300 mg twice daily for seven days) or clindamycin ovules (100 mg intravaginally once daily for three days) [51,69]. A one-day or single application of clindamycin as a bioadhesive is another option (Clindesse). These regimens have not been studied extensively and may have lower efficacy for eradicating BV.

Intravaginal clindamycin therapy has been associated with an increased prevalence of clindamycin resistant anaerobic bacteria in the vagina posttreatment (17 percent of bacterial isolates before versus 53 percent of isolates after therapy) [70]. This effect persisted in most women for at least 90 days after clindamycin treatment. In contrast, increased resistance to metronidazole was not observed in women treated with that drug. Further investigation is required to determine the clinical implications of these findings.

Pseudomembranous colitis has been reported with both oral and topical clindamycin.

Tinidazole — Tinidazole is a second generation nitroimidazole. It has a longer half-life than metronidazole (12 to 14 hours versus 6 to 7 hours) and fewer side effects [71]. Randomized trials in the United States and elsewhere have shown that it is at least as effective as metronidazole [72-74]. We suggest 1 g orally once daily for five days, as efficacy is slightly higher and side effects are slightly less frequent than with shorter course therapy (tinidazole 2 g orally daily for two days) [72].

Less effective therapies — Triple-sulfa creams, erythromycin, tetracycline, ampicillin, amoxicillin, acetic acid gel, azithromycin, and povidone-iodine vaginal douches are significantly less effective than metronidazole and clindamycin and should not be used [11,75-79]. Cure rates with ampicillin and amoxicillin are mediocre.

Efficacy of exogenous lactobacillus recolonization with probiotic lactobacilli vaginal gelatin capsules was reported in a single small trial [80]; these results should be reproduced in other trials before use of this therapy is considered.

Sexual partners — It is not necessary to treat male sexual partners of affected women, as there is no evidence that the woman's response to therapy and risk of relapse are influenced by treatment of her male sex partner(s) [51,81]. However, these data are limited and larger trials should be performed. Some studies have reported reduced rates of recurrence when male sexual partners used condoms routinely with coitus or the patient remained abstinent [11,82-85].

Women with BV who have sex with women should be encouraged to inform their sexual partner(s) regarding the need for screening, diagnosis, and treatment. (See "Gynecologic and obstetric care for lesbians".)

Relapse and recurrent infection — Approximately 30 percent of patients with initial responses to therapy have a recurrence of symptoms within three months [57] and more than 50 percent experience a recurrence within 12 months [82]. The explanation for this high rate of recurrence is unclear. Reinfection is possible, but recurrence more likely reflects a failure to eradicate the offending organisms or to reestablish the normal protective vaginal flora dominated by lactobacillus.

We suggest symptomatic relapse be first treated with prolonged induction therapy with oral or vaginal metronidazole or clindamycin for 10 to 14 days; the CDC suggests a treatment regimen different from the initial or previous treatment regimen [51].

Most women with a history of recurrent infection benefit from suppressive therapy to maintain an asymptomatic state. We feel any patient with more than three documented episodes of BV in the previous 12 months should be offered a long-term maintenance regimen consisting of maintenance metronidazole gel. Long-term clindamycin regimens, oral or topical, are not advised because of toxicity and lack of documented efficacy. Accordingly, if any of the aforementioned antimicrobials fail, we prescribe metronidazole gel 0.75 percent for 10 days followed by twice weekly applications for three to six months. In one multicenter prospective trial of this regimen, recurrent BV occurred in 25.5 percent of patients on suppressive therapy versus 59.1 percent of those receiving placebo [86]. However, secondary vaginal candidiasis was common.

As discussed above, no study has demonstrated reduced rates of recurrence in women whose partners were treated with metronidazole. However, some studies have reported reduced rates of recurrence when sexual partners used condoms routinely with coitus or the patient remained abstinent [11,82-85]. For this reason, some experts suggest these behavioral interventions for women with recurrent infection.

Exogenous lactobacillus recolonization with 30 days of oral probiotic Lactobacillus rhamnosus GR-1 and Lactobacillus reuteri RC-14 in addition to seven days of metronidazole therapy has been suggested, but there is minimal evidence of the efficacy of this approach to date [87]. These results should be reproduced in other trials before use of this therapy is considered.

A single trial has reported that use of condoms and combination metronidazole/nystatin therapy may reduce the frequency of recurrent infection [83].

Vaginal acidifying agents, although popular and widely used, have no role in the treatment of acute or chronic BV, as they have never been shown to enhance cure rates. Douching should be avoided.

Asymptomatic infection — Treatment of asymptomatic BV is typically avoided since patients often spontaneously improve over a period of several months and all antibacterial therapy is often followed by symptomatic vaginal yeast infection. This practice is supported by a double-blind, placebo controlled trial of 54 women with asymptomatic BV who were randomly assigned to receive intravaginal metronidazole or placebo gel [49]. There was no difference in the proportion of women in either group who noticed improvement in vaginal odor or discharge. In addition, 6 of 28 women receiving metronidazole developed symptomatic vaginal candidiasis compared with no women taking placebo.

However, some experts now recommend treatment of asymptomatic BV because affected women are more susceptible to acquiring other STDs (including HIV and increased expression of herpes simplex virus).

Treatment is indicated for asymptomatic women who are scheduled for certain gynecologic procedures (see below). Treatment of asymptomatic infection in pregnant women is more complex and is discussed in detail below (see 'Treatment in pregnancy' below.

Gynecologic procedures — Evidence linking asymptomatic BV with gynecologic complications has caused this issue to be reassessed, especially with the availability of topical therapy. It is reasonable to treat asymptomatic BV prior to hysterectomy and before pregnancy termination to prevent postprocedure infection (cuff infection after hysterectomy, endometritis after abortion). Reported reductions in postoperative infectious complications range from 10 to 75 percent [51,88-94].

TREATMENT IN PREGNANCY

Symptomatic BV infection — All women with symptomatic BV should be treated to relieve bothersome symptoms. Oral treatment is effective and has not been associated with adverse fetal or obstetrical effects [95-100]. The therapeutic options include [51]:

• Metronidazole 500 mg orally twice daily for seven days
• Metronidazole 250 mg orally three times daily for seven days
• Clindamycin 300 mg orally twice daily for seven days

Some clinicians avoid use of metronidazole in the first trimester because it crosses the placenta, with a potential for teratogenicity. In addition, the drug is mutagenic in bacteria and carcinogenic in mice. However, meta-analysis has not found any relationship between metronidazole exposure during the first trimester of pregnancy and birth defects [101]. The Centers for Disease Control and Prevention no longer discourage the use of metronidazole in the first trimester.

As mentioned above, topical therapy is as effective as oral therapy in the treatment of nonpregnant women with BV. In contrast, some experts avoid topical therapy in pregnant women because they believe oral treatment is more effective against potential subclinical upper genital tract infection [102-104].

Asymptomatic BV infection — As many as one-third of pregnant women in the United States have BV [30]. Of concern, a meta-analysis has reported a statistically significant increased risk of preterm birth in these women; the pooled odds ratio for prematurity was 2.19 (95% CI, 1.54-3.12) [105]. The increased risk of preterm birth attributable to BV appears to be linked to chorioamnionitis [31,106]. Other complications of BV are endometritis after cesarean delivery (55 versus 17 percent in uninfected controls) [107] and after vaginal birth (8 versus 1.5 percent in uninfected controls), and possibly an increased risk of miscarriage [105,108].

Despite the association between BV and preterm birth, screening and treatment of asymptomatic BV during pregnancy is controversial. Most studies in general obstetric populations have not found that treatment of asymptomatic infection reduces the incidence of preterm labor or delivery, but this may not apply to women at high risk for preterm birth.

General obstetrical population — Two meta-analyses have evaluated the efficacy of antibiotic therapy for asymptomatic BV infection in the general obstetrical population.

• A Cochrane meta-analysis included 15 trials involving 5888 pregnant women asymptomatic for BV and at average risk of preterm birth (a mixed population of women at low and high risk for preterm delivery) [95]. Antibiotic therapy was highly effective in eradicating infection, but did not significantly reduce the odds of preterm birth at less than 37 weeks (OR 0.91, 95% CI 0.78-1.06) or the risk of preterm premature rupture of membranes (OR 0.88, 95% CI 0.61-1.28). However, treatment initiated before 20 weeks of gestation appeared to reduce the risk of preterm birth before 37 weeks (OR 0.63, 95% CI 0.48 to 0.84; five trials, 2387 women).
• A United States Preventive Services Task Force (USPSTF) meta-analysis included three trials involving 526 pregnant women asymptomatic for BV and at low risk for preterm delivery [109]. Antibiotic treatment had no significant effect on delivery before 37 weeks.

When the USPSTF meta-analysis was performed on eight trials of 4972 average-risk women with asymptomatic BV, treatment still did not lead to a significant reduction in preterm delivery or preterm premature rupture of membranes .

Based upon these data, screening and treating all pregnant women with asymptomatic BV to prevent preterm birth and its consequences is NOT recommended.

Women at high risk for preterm birth — When the Cochrane reviewers separately analyzed the subgroup of women with a history of one or more prior preterm births (ie, women at high risk for preterm birth), the detection and treatment of asymptomatic BV appeared to markedly reduce the rate of preterm prelabor rupture of membranes (OR 0.14, 95% CI 0.05-0.38) and low birthweight (OR 0.31, 95% CI 0.13-0.75; two trials, 114 women), but the reduction in risk of subsequent preterm birth was not statistically significant (OR 0.83, 95% CI 0.59-1.17; five trials, 622 women) [95]. The lack of significant reduction in preterm birth was largely due to data from the largest (n = 1953) trial of screening and treatment of BV in pregnant women, which reported an absence of benefit in both average risk women and in the subgroup of women at high risk of preterm birth (n = 210) [110].

The USPSTF also evaluated the effect of treatment of asymptomatic BV in women at high risk for preterm birth [109]. Due to heterogeneity, results could not be pooled. Three trials showed fewer deliveries before 37 weeks of gestation with treatment, one showed no benefit, and one trial [111] reported an increase in preterm deliveries after treatment with metronidazole. For delivery before 34 weeks of gestation, meta-analysis was possible and showed no significant treatment benefit.

It is not possible to define specific features characterizing a subgroup of high risk women who might respond favorably to a screening and treatment protocol. Defining these features is an active area of investigation. As an example, women with polymorphisms of genes regulating cytokine production (eg, tumor necrosis factor variants) have a greater proinflammatory immune response to infectious stimuli, such as BV [112]. Enhanced induction of cytokines in these women could then lead to preterm labor or rupture of membranes. Other aspects of host response (eg, low levels of IgA to Gardnerella vaginalis) or the specific types of BV associated bacteria involved (eg, bacteria that produce high levels of sialidase or protease) may also play a role in placing some women with BV at high risk of preterm birth. (See "Risk factors for preterm labor and delivery", section on 'Infection' and "Risk factors for preterm labor and delivery", section on 'Genetic factors'.)

In summary, there may be benefits to screening and treating asymptomatic pregnant women who have a history of a previous preterm delivery, but there are insufficient data to recommend this as a routine practice [97,113]. In particular, the balance between the benefits and harms of this approach is unclear, given that in some settings [111,114], metronidazole treatment appeared to increase the risk of preterm birth.

The optimal choice of antibiotic, time for initiation of therapy, or duration of use remain unclear. Early diagnosis and treatment appear to be more effective in preventing preterm birth than late second trimester treatment [97,105,115,116]. For this reason, the Centers for Disease Control suggest that if screening of women with a previous preterm delivery is performed, it should be done at the first prenatal visit and treatment initiated at that time, with follow-up evaluation one month after completion of therapy [51].

The first trials that demonstrated a reduction in preterm birth in high risk patients treated women with positive test results with oral metronidazole or oral metronidazole and erythromycin (table 4) [102-104]. Subsequent studies suggested that seven days of treatment was more effective than two-day regimens [96] and that oral clindamycin or topical clindamycin cream (intravaginally each day for three to seven days) given early in pregnancy were effective alternatives [115-118].

INFORMATION FOR PATIENTS — Educational materials on this topic are available for patients. (See "Patient information: Bacterial vaginosis".) We encourage you to print or e-mail this topic, or to refer patients to our public web site www.uptodate.com/patients, which includes this and other topics.

SUMMARY AND RECOMMENDATIONS

• Bacterial vaginosis (BV) is the most common cause of vaginitis in women of childbearing age. (See 'Introduction' above.)

• BV represents a complex change in the vaginal flora characterized by a reduction in concentration of the normally dominant hydrogen-peroxide producing lactobacilli and an increase in concentration of other organisms, particularly anaerobes. (See 'Pathogenesis and microbiology' above.)

• Approximately 50 to 75 percent of women with BV are asymptomatic. Those with symptoms present with an unpleasant, "fishy smelling" discharge that is more noticeable after coitus. The discharge is off-white, thin, and homogeneous. (See 'Clinical features' above.)

• Sequelae of BV include an increased risk of preterm birth, plasma-cell endometritis, postpartum fever, post-hysterectomy vaginal cuff cellulitis, postabortal infection, pelvic inflammatory disease, and acquisition of other sexually transmitted infections. (See 'Complications' above.)

Diagnosis

• The diagnosis of BV is based on clinical findings and laboratory testing. Three of the following four criteria (Amsel criteria) are necessary for diagnosis:

  (1) Homogeneous, thin, grayish-white discharge that smoothly coats the vaginal walls,

  (2) Vaginal pH greater than 4.5,

  (3) Positive whiff-amine test, defined as the presence of a fishy odor when 10 percent potassium hydroxide (KOH) is added to a sample of vaginal discharge,

  (4) Clue cells on saline wet mount.

The presence of clue cells is the single most reliable predictor of BV. Vaginal culture is not useful. (See 'Diagnosis' above.)

Treatment of nonpregnant women

• Treatment of symptomatic women with bacterial vaginosis is indicated to reduce vaginal discharge and odor. We recommend metronidazole or clindamycin (Grade 1A). (See 'Treatment of nonpregnant women' above.) Options include:

• - Metronidazole 500 mg (twice daily orally for seven days)
• - Metronidazole gel 0.75 percent (5 g once daily vaginally for five days)
• - Clindamycin bioadhesive cream (Clindesse) 2 percent (a single vaginal dose of 5 g of cream containing 100 mg of clindamycin phosphate).

• We suggest treatment of asymptomatic women who are to undergo pregnancy termination or hysterectomy (Grade 2B). Preoperative treatment decreases the frequency of postoperative infectious complications. (See 'Gynecologic procedures' above.)

• We recommend not treating sexual partners of women with BV (Grade 1B). (See 'Sexual partners' above.)

• Approximately 30 percent of patients with an initial response to therapy have a recurrence of symptoms within three months and more than 50 percent experience a recurrence within 12 months. We treat symptomatic relapse with a longer course of therapy, using a different antibiotic than that used for the initial episode. For women who prefer preventive therapy instead of treatment of frequent episodes of BV, we recommend suppressive therapy with metronidazole gel 0.75 percent for 10 days followed by twice weekly applications for three to six months (Grade 1A). (See 'Relapse and recurrent infection' above.)

Treatment of pregnant women

• Pregnant women with BV are at increased risk of preterm birth. We recommend not screening all pregnant women for BV, given there is no evidence that screening and treatment of asymptomatic infection reduces the risk of preterm birth (Grade 1A). There may be benefits to screening and treating asymptomatic pregnant women who have a history of a previous preterm delivery. (See 'Asymptomatic BV infection' above.)

• We treat pregnant women with symptomatic BV infection to relieve symptoms. We suggest seven days of metronidazole 500 mg orally twice daily or clindamycin 300 mg orally twice daily (Grade 2C). (See 'Symptomatic BV infection' above.)