BACTERIAL VAGINOSIS (BV) CONTINUES TO be one of the most common vaginal disorders in reproductive-aged women. It represents 40% to 50% of all cases of vaginitis and is thought to be more common than either vulvovaginal candidiasis or trichomoniasis.1 In addition to the symptoms that it causes, BV has been associated with many disorders of the female reproductive tract, including urinary tract infections, postpartum endometritis, increased risk of infection after gynecologic surgery, pelvic inflammatory disease, and adverse pregnancy outcomes.2,3 BV is characterized microbiologically by a shift from the normal Lactobacillus-dominant vaginal flora to an excess of other bacteria that are primarily anaerobic.4 The cause of this shift is not fully understood. Recommended antimicrobial treatments for nonpregnant women often include intravaginal therapy with the antimicrobial agents clindamycin or metronidazole. Several intravaginal antimicrobial therapies containing these antimicrobials have been approved for use in various patient populations for the treatment of BV, including 2% clindamycin single-dose vaginal cream and 0.75% metronidazole vaginal gel.
The presence of curved rods in vaginal fluid is one of the indicators for BV on Gram stain, and these curved rods have been shown to be one of the components of the pathogenic flora in BV.5,6 These organisms are thought to represent members of the genus Mobiluncus, a group of anaerobic, motile, Gram-negative curved rods that are strongly associated with BV; they have been reported in up to 50% of women with BV but in only 0% to 5% of women without BV.5–9 Such reports have generated a great deal of interest in Mobiluncus species, yet only a limited number of efforts have been made to evaluate the effect of antimicrobial agents on these species. In vitro studies have demonstrated that clindamycin is more effective against Mobiluncus than is metronidazole and that many Mobiluncus strains may be resistant to metronidazole.5,7,10–13 However, the in vivo effects of clindamycin and metronidazole on Mobiluncus in women with BV are not well characterized in the scientific literature.
A retrospective analysis was performed on data from 3 similarly conducted clinical studies to compare the effects of clindamycin and metronidazole on vaginal Mobiluncus in patients with BV. The purpose of these retrospective analyses was to compare the effects of intravaginal clindamycin and metronidazole on vaginal Mobiluncus morphotypes in vivo in patients with clinical diagnoses of BV and without other infections.
Materials and Methods
A retrospective analysis of data from 3 similarly conducted studies was performed. Over the course of these studies, 2% clindamycin single-dose vaginal cream (CVSDC; Clindesse [clindamycin phosphate] vaginal cream 2%; Ther-Rx, St. Louis, MO) and 0.75% metronidazole vaginal gel (MVG; MetroGel-Vaginal Gel; 3 mol/l Pharmaceuticals, St. Paul, MN) were evaluated in a population of women with BV. In this analysis, Gram stain scores for Mobiluncus morphotypes (Gram-negative curved rods) were compared between the treatments.
All 3 studies were designed according to U.S. Food and Drug Administration (FDA) guidance on developing drugs for the treatment of BV14 and conducted in accordance with International Conference on Harmonization Good Clinical Practice Guidelines and the principles outlined in the Declaration of Helsinki. The studies were multicenter, masked, controlled studies conducted in the United States at a pool of 75 clinical sites, each with its respective Institutional Review Board/ethics board approval for each study. Some, but not all, of the sites participated in more than one of the studies. Patients were screened for study eligibility from the overall patient population at each clinical site. All patients were voluntary participants and provided signed informed consent before any study-related procedure was performed.
In all 3 studies, eligible patients were nonpregnant women at least 18 years of age with a clinical diagnosis of BV, which was defined as meeting all of the following requirements, based on a modification of Amsel's criteria15: ≥20% clue cells; off-white (milky or gray), thin, homogenous vaginal discharge; vaginal pH >4.5; and a fishy amine odor on the addition of 10% KOH to vaginal fluid (a positive “whiff” test). The ≥20% clue cell requirement was used in light of evidence that this is more predictive of BV than the simple presence or absence of clue cells put forth in Amsel's criteria.16 A diagnosis of BV also required patients to have a Gram stain Nugent score17 of ≥4 at baseline. Although a minimum Nugent score of 7 is usually required to support a diagnosis of BV,17 this study also enrolled patients considered to be “intermediate” (scores of 4–6) to maintain compliance with FDA guidance.14
Patients were excluded if they were pregnant or nursing, had sexually transmitted infections, had vulvovaginal infections other than BV (including a culture for the presence of Candida), had vulvovaginal or cervical abnormalities or disorders, were actively menstruating, had received antifungal or antimicrobial treatment within 14 days of the study, were using intrauterine devices, were taking anticoagulants, lithium, or disulfiram, or were hypersensitive to any component of the drug formulations. It should be noted that patients were allowed to be enrolled in the study before the Nugent score and yeast culture results became available. If they were found to have a nonqualifying Nugent score or a positive yeast culture, they were not included in the analyses although they had received study medication.
In each study, eligible patients were randomly assigned to a treatment group in a masked fashion according to a computer-generated randomization schedule. CVSDC was self-administered by patients in a single dose, which applied 100 mg clindamycin. MVG was self-administered by patients twice daily for 5 days with 37.5 mg metronidazole applied in each dose. In all 3 studies, BV was evaluated at baseline and at a test-of-cure (TOC) visit 21 to 30 days after the start of treatment.
In addition to the study entry criteria, all patients included in the analyses met the following requirements: 1) administered a single dose of CVSDC or administered at least one dose of MVG on each of the first 3 days of treatment starting no less than 48 hours after the baseline visit; 2) had Gram stain Nugent score data both at baseline and at a TOC visit not less than 21 days and not more than 40 days after the start of treatment; 3) had no antimicrobial therapy for conditions other than BV during the study; 4) did not use intravaginal products such as diaphragms, intrauterine devices, and douches during the study; 5) did not have vaginal intercourse during or within 7 days after the completion of study therapy; and 6) had no significant violations of the study protocol.
In all 3 clinical studies, the vaginal flora was assessed by direct microscopic evaluation with a Gram stain for Nugent score.17 The Nugent score is determined based on the weighted sum of scores indicating the presence or absence of each of 3 bacterial morphotypes in Gram-stained vaginal smears (large Gram-positive rods, small Gram-negative or variable rods, and curved Gram-variable rods). The Nugent score was proposed in 1991 as a means to expand and improve the reliability of Gram stain interpretation from the previous Spiegel criteria.18 It is endorsed by many as the gold standard for quantifying the most critical bacteria to determine whether an infection exists (i.e., Lactobacillus, Gardnerella/Bacteroides/ other anaerobes, and Mobiluncus morphotypes).17,19 In addition to the composite score, the 3 bacterial morphotypes can each be scored on an individual basis, allowing the presence or absence of any of the 3 morphotypes in a vaginal smear to be determined. Mobiluncus morphotypes (Gram-negative curved rods) have been shown to be identified with Gram stain with excellent reliability.17 The scoring system used to determine the presence of Mobiluncus morphotypes in each of the studies is as follows: a Mobiluncus score of 2 reflects an average of 5 or more Mobiluncus morphotypes observed per oil immersion field; a score of 1 reflects an average of <1 to <5 Mobiluncus morphotypes observed per oil immersion field; and a score of 0 reflects no Mobiluncus morphotypes observed in any examined field. All microbiologic evaluations in all 3 studies were performed at the Mayo Clinic (Rochester, MN). The evaluations were performed using a standard procedure by evaluators who were masked about treatment. A minimum of 10 oil immersion fields were examined in each evaluation.
In the 3 clinical studies, patients were evaluated for microbiologic, clinical, and therapeutic cure of BV. Microbiologic cure was defined as a Nugent score of 3 or less on the 10-point scale. Clinical cure was defined as resolution of the modified Amsel's criteria (≤20% clue cells, normal vaginal discharge, vaginal pH ≤4.5, and a negative “whiff” test). Therapeutic cure was defined as being both clinically and microbiologically cured.
The statistical analyses compared the changes from baseline to the TOC visit in mean Mobiluncus scores within each treatment group and mean Mobiluncus scores at baseline and at the TOC visit across groups. Within-group changes from baseline were evaluated using a t-test. Between-group differences in the mean Mobiluncus scores at baseline and the TOC visit were compared using a one-way analysis of variance (ANOVA) F test. Percentages were compared using Fisher exact test. Cure rates were compared between groups using a χ2 test.
A total of 580 patients were enrolled in the 3 studies (357 in the CVSDC group and 223 in the MVG group). Of these, 284 (49.0%) met all of the entry criteria and analysis requirements outlined in “Materials and Methods” (176 in the CVSDC group and 108 in the MVG group). The reasons that patients were not included in the analyses were as follows: 99 patients (17.1% of the total patients enrolled) were found to have a baseline Nugent score of <4, 62 patients (10.7%) did not have data from a TOC visit more than 21 days after the start of treatment without being considered treatment failures (including patients who were found to have a positive baseline Candida culture at baseline), 49 patients (8.4%) were noncompliant with study drug administration (did not administer CVSDC, did not administer MVG on each of the first 3 days of treatment, or did not administer study medication within 48 hours of baseline), 27 patients (4.7%) were found after study medication dosing to have not met study entry requirements (other than the Gram stain Nugent score and Candida culture result requirements), 25 patients (4.3%) took an antimicrobial therapy for a condition other than BV, 20 patients (3.4%) used disallowed intravaginal products or had vaginal intercourse within 7 days of the completion of treatment, 8 patients (1.4%) had no TOC visit Gram stain results, and 6 patients (1.0%) did not have a TOC visit within 40 days of the start of study treatment. Among those patients who met the outlined requirements, a total of 55 had Mobiluncus morphotypes present in Gram-stained vaginal fluid samples at baseline (Mobiluncus scores of 1 or 2; 40 in the CVSDC group and 15 in the MVG group).
The demographic characteristics of the analyzed patients were similar across the treatment groups among patients with Mobiluncus morphotypes at baseline. Demographic characteristics were also similar between patients with Mobiluncus morphotypes at baseline and those without with the exception of race and baseline composite Nugent score. A significantly higher percentage of patients with baseline Mobiluncus morphotypes were black than patients without baseline Mobiluncus morphotypes (P <0.001), and the mean baseline Nugent score was higher among patients with Mobiluncus morphotypes at baseline than among those without (P <0.001). Baseline patient demographic characteristics for each treatment group are shown in Table 1.
In the CVSDC group, only 2.5% of patients did not achieve a reduction in Mobiluncus score from 1 or 2 at baseline to 0 at the TOC visit, whereas 20.0% in the MVG group remained with a score of 1 or 2 at the TOC visit (P = 0.0574).
The distributions of Mobiluncus scores at baseline and at the TOC visit are shown in Table 2. The baseline distributions of Mobiluncus scores were similar across the treatment groups. In both groups, all patients with a Mobiluncus score of 1 at baseline had a score of 0 at the TOC visit. Twenty-one of 22 patients in the CVSDC group with a baseline Mobiluncus score of 2 had a score of 0 at the TOC visit (95.5%), whereas 6 of the 9 patients in the MVG group with a Mobiluncus score of 2 at baseline had a score of 0 at the TOC visit (66.7%).
Like with the distributions of the Mobiluncus scores, mean Mobiluncus scores were similar between the 2 treatment groups at baseline. There were significant reductions in mean Mobiluncus scores from baseline to the TOC visit in both treatment groups (P <0.0001 for both groups). This represents significantly reduced presence of Mobiluncus morphotypes at TOC in both groups. The baseline mean Mobiluncus scores were not significantly different between the treatment groups (1.55 ± 0.08 [mean ± standard error of the mean] and 1.60 ± 0.13 in the CVSDC and MVG groups, respectively; P = 0.7448). However, the mean Mobiluncus score in the CVSDC group was less than in the MVG group at the TOC visit (0.05 ± 0.05 and 0.33 ± 0.19 in the CVSDC and MVG groups, respectively; P = 0.0471), indicating a significantly greater decrease in the presence of Mobiluncus in the CVSDC group compared with the MVG group. The mean scores at baseline and at the TOC visit for both groups are shown in Figure 1.
Because this analysis was performed on a subset of a larger database, we also sought to determine whether there were differences in acquisition of Mobiluncus among women treated for BV who did not have Mobiluncus on Gram stain at baseline. In the 3 studies, there were 229 patients with BV (136 treated with CVSDC and 93 with MVG) who had a Mobiluncus score of 0 at baseline but who met all other criteria for inclusion, including all diagnostic criteria for BV, in the analyses. The demographic characteristics of these patients were similar between the CVSDC and MVG groups. Among patients with a baseline Mobiluncus score of 0 treated with CVSDC, the mean score at the TOC visit was 0.02 ± 0.02, indicating that a small number of patients with a baseline score of 0 had increases to scores of 1 or 2. However, the mean score among these patients did not increase significantly from the baseline score of 0 (P = 0.1807). Among patients with baseline Mobiluncus scores of 0 treated with MVG, the mean score at the TOC visit was 0.09 ± 0.04. The mean score among these patients was significantly higher than the baseline score of 0 at the TOC visit (P = 0.0449). Although the difference between groups was not statistically significant, a trend was observed with a larger percentage of patients having a baseline Mobiluncus score of 0 increased to scores of 1 or 2 in the MVG group compared with those in the CVSDC group (4 of 94 [4.3%] and 2 of 135 [1.5%], respectively; P = 0.1172).
The increased clearance of Mobiluncus morphotypes with CVSDC treatment appeared to correlate with an improvement in clinical efficacy. Among the 55 patients with Mobiluncus morphotypes present at baseline, significantly more patients in the CVSDC group achieved microbiologic and clinical cure of BV than in the MVG group (P = 0.04 for both comparisons). Although there was also a notable difference in patients achieving therapeutic cure, the difference only trended toward statistical significance (P = 0.09). There were no significant between-treatment differences in the frequency of any type of cure among patients with no Mobiluncus morphotypes at baseline (P >0.20 for all comparisons). Microbiologic, clinical, and therapeutic cure rates among patients in the CVSDC group were comparable between patients with baseline Mobiluncus and those without. Microbiologic, clinical, and therapeutic cure rates for each group are shown in Table 3.
Among patients with Mobiluncus morphotypes present at baseline (scores of 1 or 2), there were no significant differences between the treatment groups in small Gram-negative or variable rod scores (another component of the Nugent score) at baseline (mean score 3.5 ± 0.10 in the CVSDC group and 3.5 ± 0.22 in the MVG group, P = 0.969). However, the mean Gram-negative or variable rod score was significantly higher in the MVG group (indicating a larger presence of these organisms) at the TOC visit (mean scores on a 0–4 scale: 1.1 ± 0.27 in the CVSDC group and 2.8 ± 0.46 in the MVG group, P <0.002). In patients without Mobiluncus morphotypes at baseline, there was no significant between-treatment difference in mean Gram-negative or variable rod scores (mean scores at the TOC visit: 1.2 ± 0.15 in the CVSDC group and 1.4 ± 0.19 in the MVG group, P = 0.481).
Although assessment of clinical manifestations is the most common method for diagnosing and evaluating BV,17 the direct microscopic examination of Gram-stained slides of vaginal secretions is endorsed by many as the gold standard for such evaluations.17–20 Clindamycin and metronidazole have been demonstrated to be similar in efficacy in the treatment for BV,21–26 but there are differences in their antimicrobial spectra on microbiologic examination, which may have implications for the clinical management of BV.9 One such potential difference is in the effects of clindamycin and metronidazole on Mobiluncus species. In vitro studies demonstrate that clindamycin is a more potent inhibitor of Mobiluncus than is metronidazole (MIC50 of 0.03–0.125 μg/mL for clindamycin compared with an MIC50 of 32–128 μg/mL for metronidazole)9,12 and that Mobiluncus species may be resistant to metronidazole.5,7,10–12 It has been postulated that, despite its lower activity against Mobiluncus, Gardnerella, and Mycoplasma compared with clindamycin, metronidazole may effectively eradicate other susceptible organisms, leading ultimately to a decrease in nonsusceptible organisms.27 Although this would account for the equivalent activity found with clindamycin and metronidazole,21–25 it remains unknown what, if any, impact this phenomenon might have in patients with BV. To our knowledge, our study represents one of the first attempts to evaluate whether these in vitro results found by other investigators can be reproduced in vivo in women with BV.
Although a focus on the individual Mobiluncus morphotypes may appear superfluous, it should be noted that microbiologic variations that can be characterized on Gram stain do exist among women with BV and that these differences may be of clinical importance. In a study of over 6,800 pregnant women, Hauth and colleagues found that those with a Nugent score of 9 or 10 were at greater risk of preterm birth than women with BV who have a lower Nugent score and women who did not have BV by Nugent criteria.28 Such a score can only be achieved if Mobiluncus morphotypes are present on Gram stain. Pereira and colleagues have reported significant variations in microflora of pregnant women with BV enrolled in a large, clinical prevalence study assessing the association among maternal stress, BV, and risk of preterm birth.29 They found that pregnant women with BV who had Mobiluncus morphotypes on Gram stain tended to be older, non-Hispanic black, and to have more lifetime sexual partners compared with women with BV who did not have Mobiluncus on Gram stain. Furthermore, women with Mobiluncus were also more likely to be given a clinical diagnosis of BV by their healthcare provider. It has been suggested that the presence of Mobiluncus morphotypes with Gram stain has been reported to be associated with higher numbers of clue cells and positive “whiff” tests.29 Similar to Pereira's findings, in the population studied in our present analyses, there was a significantly higher percentage of black women among those whom had Mobiluncus morphotypes at baseline than among those who did not. Thus, with clinical information to suggest that not all BV infections are the same and that Mobiluncus morphotypes are associated with more adverse outcomes in pregnancy and with more obvious disease, determining the differences in antimicrobial activity in vivo against Mobiluncus morphotypes becomes potentially important to clinical care. Mobiluncus has been characterized as being part of the end-stage process with BV.18 Furthermore, although unstudied, it is also possible that Mobiluncus resistance to antibiotics, particularly metronidazole, may play a role in recurrent BV, which occurs in as many as 15% to 30% of women with BV.30 Despite the suggestive in vitro data, there are few data on the effects of clindamycin and metronidazole on Mobiluncus in women with BV.
The purpose of these retrospective analyses was to compare the effects of intravaginal clindamycin and metronidazole on vaginal Mobiluncus morphotypes in patients with clinical diagnoses of BV. These retrospective analyses examined similar patient populations that were given the recommended treatment regimens of CVSDC or MVG. In these analyses, patients with BV who had Mobiluncus demonstrated a decrease in Mobiluncus scores with either clindamycin or metronidazole vaginal preparations. However, those who received CVSDC exhibited a better clearance of Mobiluncus morphotypes than those who were given MVG. Furthermore, among those women who did not have Mobiluncus present at baseline, those who received MVG had a slight but statistically significant increase in Mobiluncus after treatment. The results of the present analyses also showed that patients with Mobiluncus morphotypes at baseline who were treated with metronidazole had a significantly higher mean Gardnerella score (indicating a greater presence of Gardnerella and other BV-associated flora) at the TOC visit than those treated with clindamycin. There was no significant between-treatment difference in Gardnerella score among patients without Mobiluncus. This suggests that the presence of Mobiluncus in patients with BV may have an effect on the ability of metronidazole to eradicate Gardnerella and other BV-associated flora, which may lead to reduced initial antimicrobial activity and increased recurrence of infection in these patients.
The results of these analyses of data from patients with BV indicate that clindamycin reduces the presence of vaginal Mobiluncus to a greater extent than metronidazole, confirming the antimicrobial spectra for clindamycin and metronidazole observed in vitro. This reduced presence of Mobiluncus at TOC in CVSDC-treated patients appears to correlate with increases in microbiologic, clinical, and therapeutic cure of BV in patients that have Mobiluncus. Other reports have examined the effects of clindamycin on Lactobacillus as a single component of the Nugent score and found that although clindamycin does have in vitro activity against Lactobacillus (contrary to metronidazole), restoration of Lactobacillus at 21 to 30 days posttreatment is virtually identical.31,32 This analysis, which to our knowledge is one of the first to look specifically at the issue of the effect of antibiotic treatment on Mobiluncus morphotypes in women with BV, is potentially limited by the fact that only 55 patients with Mobiluncus morphotypes present at baseline were evaluated and the fact that the data were obtained from 3 separate clinical studies. However, this latter limitation is mitigated by the similarities in the study populations, study design, patient evaluation, and masked performance of the microscopic evaluations. Although current data suggest that there are important differences between women diagnosed with BV with or without the presence of Mobiluncus morphotypes, it is currently unknown whether these variations can be directly attributed to Mobiluncus species or to some other unknown organisms for which these Gram stain findings are a marker. Further studies will be necessary to determine whether addressing Mobiluncus morphotypes specifically when treating women with BV will result in better long-term cure rates and a decrease in morbidities associated with BV. Of particular interest is the trend noted in this study for metronidazole treatment to actually result in increased Mobiluncus colonization after treatment among women who did not have detectable Mobiluncus morphotypes at baseline. Our data suggest that there exist important differences in how these morphotypes respond to various antimicrobial therapies; such differences should be carefully considered when designing future studies of therapy to reduce BV-associated morbidities.
1. Sobel JD. Vaginitis. N Engl J Med 1997; 337:1896–1903.
2. Beigi RH, Austin MN, Meyn LA, Krohn MA, Hillier SL. Antimicrobial resistance associated with the treatment of bacterial vaginosis. Am J Obstet Gynecol 2004; 191:1124–1129.
3. Koumans EH, Kendrick JS. CDC Bacterial Vaginosis Working Group. Preventing adverse sequelae of bacterial vaginosis: A public health program and research agenda. Sex Transm Dis 2001; 28:292–297.
4. Ness RB, Hillier SL, Richter HE, et al. Douching in relation to bacterial vaginosis, lactobacilli, and facultative bacteria in the vagina. Obstet Gynecol 2002; 100:765–772.
5. Spiegel CA, Eschenbach DA, Amsel R, Holmes KK. Curved anaerobic bacteria in bacterial (nonspecific) vaginosis and their response to antimicrobial therapy. J Infect Dis 1983; 148:817–822.
6. Phillips I, Taylor E. Anaerobic curved rods in vaginitis. Lancet 1982; 1:221.
7. Sprott MS, Ingham HR, Pattman RS, et al. Characteristics of motile curved rods in vaginal secretions. J Med Microbiol 1983; 16:175–182.
8. Thomason JL, Schreckenberger PC, Spellacy WN, Riff LJ, LeBeau LJ. Clinical and microbiological characterization of patients with nonspecific vaginosis associated with motile, curved anaerobic rods. J Infect Dis 1984; 149:801–808.
9. Tohill BC, Heilig CM, Klein RS, et al. Vaginal flora morphotypic profiles and assessment of bacterial vaginosis in women at risk for HIV infection. Infect Dis Obstet Gynecol 2004; 12:121–126.
10. Spiegel CA. Susceptibility of Mobiluncu
s species to 23 antimicrobial agents and 15 other compounds. Antimicrob Agents Chemother 1987; 31:249–252.
11. Carlone GM, Thomas ML, Arko RJ, et al. Cell wall characteristics of Mobiluncus
species. Int J Syst Bacteriol 1986; 36:635–637.
12. Skarin A, Holst E, Mårdh PA. Antimicrobial susceptibility of comma-shaped bacteria isolated from the vagina. Scand J Infect Dis Suppl 1983; 40:81–84.
13. Paupermpoonsiri S, Watanabe K, Kato N, Ueno K. In vitro activities of 10 antimicrobial agents against bacterial vaginosis-associated anaerobic isolates from pregnant Japanese and Thai women. Antimicrob Agents Chemother 1997; 41:2297–2299.
14. Guidance for Industry: Bacterial Vaginosis—Developing Antimicrobial Drugs for Treatment. Draft Guidance. US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), July 1998.
15. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis: Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
16. Thomason JL, Gelbart SM, Anderson RJ, Walt AK, Osypowski PJ, Broekhuizen FF. Statistical evaluation of diagnostic criteria for bacterial vaginosis. Am J Obstet Gynecol 1990; 162:155–160.
17. 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; 29:297–301.
18. Spiegel CA, Amsel R, Holmes KK. Diagnosis of bacterial vaginosis by direct Gram stain of vaginal fluid. J Clin Microbiol 1983; 18:170–177.
19. Mastrobattista JM, Bishop KD, Newton ER. Wet smear compared with Gram stain diagnosis of bacterial vaginosis in asymptomatic pregnant women. Obstet Gynecol 2000; 96:504–506.
20. McGregor JA, Hager WD, Gibbs RS, Schmidt L, Schulkin J. Assessment of office-based care of sexually transmitted diseases and vaginitis and antibiotic decision-making by obstetrician–gynecologists. Infect Dis Obstet Gynecol 1998; 6:247–251.
21. Andres FJ, Parker R, Hosein I, Benrubi GI. Clindamycin vaginal cream versus oral metronidazole in the treatment of bacterial vaginosis: A prospective double-blind clinical trial. South Med J 1992; 85:1077–1080.
22. Schmitt C. Bacterial vaginosis: Treatment with clindamycin cream versus oral metronidazole. Obstet Gynecol 1992; 79:1020–1023.
23. Ferris DG, Litaker MS, Woodward L, Mathis D, Hendrich J. Treatment of bacterial vaginosis: A comparison of oral metronidazole, metronidazole vaginal gel, and clindamycin vaginal cream. J Fam Pract 1995; 4:443–449.
24. Paavonen J, Mangioni C, Martin MA, Wajszczuk CP. Vaginal clindamycin and oral metronidazole for bacterial vaginosis: A randomized trial. Obstet Gynecol 2000; 96:256–260.
25. Fischbach F, Petersen EE, Weissenbacher ER, et al. Effect of clindamycin cream versus oral metronidazole in the treatment of bacterial vaginosis. Obstet Gynecol 1993; 82:405–410.
26. Faro S, Skokos CK, et al. The efficacy and safety of a single dose of Clindesse™ vaginal cream versus a seven-dose regimen of Cleocin® vaginal cream in patients with bacterial vaginosis. Infect Dis Obstet Gynecol 2005; 13:155–160.
27. Austin MN, Beigi RH, Meyn LA, et al. Microbiologic response to treatment of bacterial vaginosis with topical clindamycin or metronidazole. J Clin Microbiol 2005; 43:4492–4497.
28. Hauth JC, MacPherson C, Carey C, et al. Early pregnancy threshold vaginal pH and Gram stain scores predictive of subsequent preterm birth in asymptomatic women. Am J Obstet Gynecol 2003; 188:831–835.
29. Pereira L, Culhane J, McCollum K, Agnew K, Nyirjesy P. Variation in microbiologic profiles among pregnant women with bacterial vaginosis. Am J Obstet Gynecol 2005; 193:746–751.
30. Wilson J. Managing recurrent bacterial vaginosis. Sex Transm Infect 2004; 80:8–11.
31. Agnew KJ, Hillier SL. The effect of treatment regimens for vaginitis and cervicitis on vaginal colonization by lactobacilli. Sex Transm Dis 1995; 22:269–273.
32. Nyirjesy P, McIntosh MJ, Gattermeir DJ, Schumacher RJ, Steinmetz JI, Joffrion JL. The effects of intravaginal clindamycin and metronidazole therapy on vaginal lactobacilli in patients with bacterial vaginosis. Am J Obstet Gynecol 2006; 194:1277–1282.