Bacterial vaginosis (BV) and Trichomonas vaginalis (TV) infection have been estimated to affect as many as one-quarter to one-third of sexually active females worldwide,1,2 and are often found concurrently.3 Both BV and TV have been associated with adverse birth outcomes such as prematurity and low birth weight,4,5 pelvic inflammatory disease,6,7 infertility,8,9 and acquisition of HIV and herpes simplex virus type 2 (HSV-2) infections.1,10–12 Surprisingly, given the well-described association between BV and TV, few data that would help clarify the temporality of the relationship between BV and TV infection are available.
Understanding the relationship between BV and TV has been problematic because most studies have been cross-sectional in nature.13–17 Nevertheless, results of a number of studies suggest that TV colonization has increased in the presence of BV-defining phenomena such as elevated amine production, loss of facultative lactobacilli, and increased pH.18–22 This study describes the results of a secondary analysis of data originally collected in a prospective cohort study examining the relationship of abnormal vaginal flora and incident HSV-2 infections among young women of reproductive age in Mysore, India.
MATERIALS AND METHODS
Participant recruitment and laboratory methods have been described in detail elsewhere.23–25 In brief, 15- to 30-year-old, nonpregnant, sexually active women were recruited in 2005–2006 through health education camps offered in the rural and peri-urban communities around Mysore City in south India. At baseline, and at 3- and 6-month visit, participants underwent an interviewer-administered questionnaire and a physical examination, during which biologic specimens, including vaginal, high cervical swabs, and venous blood, were collected. Of the 898 women who completed the baseline visit, 853 provided data from at least 2 consecutive study visits. These 853 participants comprise the study sample for this analysis.
TV infection was diagnosed on the basis of a positive result from either wet-mount microscopy for detection of motile trichomonad and/or culture (InPouch Culture Kit, BioMed Diagnostics, White City, OR). All women had specimens collected for both TV tests. Women with TV infection were treated with a single dose of 2 grams of oral metronidazole, and the same treatment was given to the participants to give to their sex partners. BV was diagnosed clinically using Amsel's criteria and treated with 400 mg oral metronidazole twice daily for 1 week.26 For analytic purposes, BV was diagnosed by Nugent scoring of Gram-stained vaginal smears.27 ELISA testing was done for HSV-2 IgG antibodies (Focus Technologies, Cypress, CA). Institutional review boards at the University of California, Berkeley, and the Asha Kirana Hospital in Mysore, India, approved the study protocol. Informed consent was obtained from all study participants.
In a cross-sectional analysis of baseline data, increased prevalence of TV was found among those testing positive for BV (Nugent scores, 7–10) and among those with intermediate scores (Nugent scores, 4–6), both of which indicate the presence of abnormal vaginal flora.24 Thus, for this analysis Nugent scores were categorized into a dichotomous measure of vaginal flora status, with 0 to 3 considered “normal” and 4 to 10 “abnormal.”27 To estimate the relationship between changing vaginal flora status and TV infection risk, log-linear generalized estimating equations (GEE) with an exchangeable covariance structure, and bootstrapped results with 5000 repetitions for robust standard errors, were used.28 Four exposure patterns were defined based on participant vaginal flora status on a given visit (abnormal or normal) and status at the previous visit (abnormal or normal). The four patterns represent different exposure periods of abnormal vaginal flora between visits. With 3 study visits, participants could contribute 2 exposure patterns (for 0–3 months, and 3–6 months) to the GEE model, to estimate the risk of TV at 3 and 6 months, respectively. The GEE model provides covariate-adjusted estimates of risk ratios (aRR) for TV by different vaginal flora exposure patterns between visits, compared with those whose Nugent tests scores were normal at both visits (Fig. 1). In all, 820 participants contributed at least one observation to the GEE model; unreadable vaginal smears accounted for all but 3 of the 33 participants not included in the analysis.
The selection of possible confounders and cut points for categorization of continuous variables were based on findings from the baseline cross-sectional analyses for BV and TV infections12,24 and previously published literature. For the multivariable GEE model, variables with small strata (<10% of sample) were not controlled for. Strata from ordinal categorical measures were combined if the baseline TV analysis showed similar prevalences of TV across adjacent groups. For the religious identification measure, Hindu and Christian women were combined into a single “Non-Muslim” category. In three observations where a participant's marital status was not recorded, her marital status at the previous visit was used, as there was minimal change in marital status over time in this cohort. For HSV-2 testing, 21 missing results were recoded to negative, as HSV-2 incidence was low in this cohort. A socioeconomic status (SES) index was created using the first factor generated from a principal components analysis of assets owned, financial instruments used, and dummy-coded cooking stove type.29 SES score was made into a dichotomous categorical measure to create balanced categories of low and high SES. Two participants with missing asset data were coded into the Low SES category.
Two additional GEE models were run to examine the sensitivity of results due to the clinical definition of abnormal flora. In the first case, abnormal flora was defined by a Nugent score of ≥7. In the second case, abnormal flora was defined by an Amsel score of ≥3. A third GEE model was run using our original definition of abnormal flora (Nugent score of ≥4), but using only incident TV infections. All three additional GEE models used the same covariate and bootstrapping specifications as used by the original model. Data were analyzed using Stata 11.1 (StataCorp, College Station, TX).
A description of the full cohort with baseline TV prevalence has previously been reported elsewhere12; additional baseline measures are described here for the analyzed cohort. A majority of women had been with their partners for 7 to 19 years (78.1%), and these women were at higher risk of a TV diagnosis (9.4% vs. 6.5% for women who had partners for <7 years). Women were equally divided across SES categories, with a higher prevalence of TV found in the lower SES category (10.6% vs. 6.4% with higher SES). Although a large majority (69.1%) of women reported at least 13 sexual acts in the 3 months before the baseline visit, prevalence of TV did not vary across sexual activity categories. Very few women reported having a partner who had other sex partners, or using an intrauterine device for contraception, leading to unstable estimates of the prevalence of TV in these strata. No participants reported smoking, using hormonal contraception, or vaginal douching. These latter measures were thus not included as potential confounders in the multivariable GEE model.
Table 1 shows TV risk by change in vaginal flora status between the baseline and 3-month visit. For ease of reporting, only the baseline- to 3-month visit results are tabulated; results from the 3- to 6-month visit (data not shown) demonstrate a similar risk gradient. Most of the women had normal vaginal flora at both visits (54.9%), and few of these women (1.4%) tested positive for TV at 3 months. Fewer women had normal flora at baseline and abnormal flora at the 3-month visit (12.4%), though a higher proportion tested positive for TV at 3 months (6.2%). Of the women who had abnormal flora at baseline and normal flora at the 3-month visit (13.2%), a similar proportion (7.8%) tested positive for TV at 3 months. Approximately, 1 in 5 women had abnormal vaginal flora at both the baseline and at 3-month visit (20.6%), and these women were at greatest risk for testing positive for TV at the 3-month visit (13.0%).
Prevalence and Incidence of TV Infection and Abnormal Vaginal Flora
The prevalence of TV declined by study visit, from 8.5% at baseline to 5.5% at 3 months and 3.0% at 6 months. Of the 775 women who were TV negative at baseline, 23 (3.0%) had TV diagnosed at their 3-month follow-up visit and 8 (1.1%) at their 6-month visit; of those 8 women, 5 were diagnosed with TV for the first time. In comparison, of the 73 women who had TV at baseline, 24 (32.9%) were TV infected again at the 3-month visit, and 10 had TV at all 3 visits, despite having received repeated metronidazole treatment. The prevalence of abnormal vaginal flora also decreased by visit, from 33.7% at baseline to 32.4% at 3 months and 28.7% at 6 months. Of the 526 women with normal vaginal flora at baseline, 35 (6.6%) were Amsel positive for BV, and 97 (18.4%) had abnormal flora at 3 months. In comparison, of the 264 women with abnormal flora at baseline, 60 (22.7%) were Amsel positive, and 161 (61.0%) had abnormal flora again at 3 months, and 94 participants had abnormal flora at all 3 visits. For all visits, Amsel clinical diagnosis resulted in treatment of 4.1% of those with normal flora (Nugent scores, 0–3), 18.3% with intermediate flora (Nugent scores, 4–6), and 34.7% with BV (Nugent scores, 7–10).
TV Risk by Multivariable GEE Analysis
The participants with abnormal vaginal flora across any 2 consecutive visits had 9 times higher risk (95% CI: 4.1, 20.0) of TV infection at the latter visit relative to those women with persistently normal flora, adjusted for covariates (Table 2). Women whose flora became abnormal after being normal at the previous visit had a higher risk of TV (aRR: 7.11, 95% CI: 2.7, 18.2), as did those whose flora became normal after being abnormal at the previous visit (aRR: 4.50, 95% CI: 1.7, 11.7). Of the covariates examined, Muslim religion (aRR: 0.44, 95% CI: 0.2, 1.1), having a partner for at least 7 years (aRR: 1.94, 95% CI: 0.9, 4.0), and HSV-2 infection (aRR: 1.68, 95% CI: 0.9, 3.0) each had weaker evidence for an association with TV infection.
Table 3 shows results from the alternate GEE models. The GEE models using different definitions for abnormal flora provided results that were all consistent in direction, though with reduced magnitude relative to the original GEE model. For the 3 longitudinal exposure categories normal-abnormal, abnormal-normal, and abnormal-abnormal, the model using Nugent scores of ≥7 to indicate abnormal flora resulted in aRR of 4.12 (95% CI: 1.9, 8.9), 4.44 (95% CI: 2.0, 10.1), and 3.15 (95% CI: 1.4, 7.0), respectively, and the model using Amsel score of ≥3 resulted in aRRs of 2.81 (95% CI: 1.5, 5.3), 1.94 (95% CI: 0.7, 5.8; P = 0.215), and 4.89 (95% CI: 2.3, 10.5), respectively. The GEE model using our original abnormal flora definition of Nugent scores of ≥4 but only including incident TV infection resulted in aRR of 8.59 (95% CI: 3.2, 22.7), 3.15 (95% CI: 0.9, 10.5; P = 0.043), and 4.75 (95% CI: 1.8, 12.6), respectively. Unless otherwise indicated, all P values were less than 0.005.
We found evidence for a 4- to 9-fold increased risk of TV infection among women who had abnormal vaginal flora within a 3-month span, with the highest risk among those women found to have abnormal flora at consecutive visits. Our alternate GEE models indicate the relationships are generally maintained when other widely accepted abnormal flora measurement criteria are used, and for incident TV infection. This finding is consistent with the findings of cross-sectional studies showing that a disturbed vaginal flora is associated with an increased risk of sexually transmitted infections, including HIV and TV.13,30–33 It also confirms the findings from longitudinal studies in Kenya and the United States,20,22 which show that abnormal vaginal flora is associated with subsequent acquisition of TV infection, and thus strengthens the evidence for a causal role. Thurman and Doncel and others have suggested several mechanisms whereby disturbed vaginal flora might increase the risk of HIV and other STI infections, including “initiation of a clinical or subclinical mucosal inflammatory response, alteration of innate mucosal immunity, alteration of normal vaginal microflora and pH, and weakening or breach of the cervicovaginal mucosa.”22,34–38 As other studies have suggested, all of these mechanisms could also plausibly pertain to the relationship between abnormal vaginal flora and acquisition of TV infection.
The declining prevalence of TV infection and abnormal flora status over time in this cohort is unsurprising, given that the women were treated with metronidazole, which has been shown to be effective in treating both conditions.39,40 As TV infection is typically self-limiting in men,41 and reported extramarital partnerships were low among participants, the reduction in prevalence from 8.5% to 3.0% after 2 rounds of diagnosis and treatment was expected, and demonstrates that control of this infection with metronidazole in resource-limited settings is possible. In contrast, the small reduction in the prevalence of abnormal flora is of concern, as is the high rate of repeat diagnosis. Treatment based on the Amsel diagnostic criteria identified only 35% of women who were found to have BV based on Nugent test (and 18% of women with Intermediate Nugent scores). In this setting, the Amsel criteria did not appear to be a sufficiently sensitive method for diagnosis of BV. This is also a notable finding because BV has been found to increase the risk for STIs and adverse birth outcomes, which are commonly found in resource-constrained settings.42
Consistent with other studies that found Muslim religion to be associated with a lower risk of HIV and STI, our study found that the risk of TV was lower among Muslim women, which could be due to cultural practices, such as male circumcision among their sex partners.43,44 We also found some evidence of an increased risk of TV infection associated with HSV-2 infection, consistent with the findings of other studies of TV.16,45 HSV-2 infection has been shown to increase the risk of acquisition of STIs because of genital ulcers,46 and may be a proxy of past sexual risk behavior.47
This study has a number of strengths, including a large sample size, standardized laboratory testing for TV and BV, minimal loss to follow-up visits, and the use of longitudinal measures. On the other hand, there were also several limitations: First, we did not use a population-based sample, so our findings may not be generalizable. Second, as vaginal flora are highly dynamic,48 it is possible that we underestimated new abnormal flora episodes between visits. Third, because we did not do a “test of cure” for women treated in the study, we do not know whether TV infections were new or persistent. Because metronidazole treatment is thought to be 95% effective,49 it is unlikely that this limitation would significantly affect the findings. Furthermore, although all women diagnosed with TV were provided with treatment for their sex partners, we could not ascertain whether partners were offered or accepted the treatment. What appears to be repeated TV infection may be prevalent infections from the perspective of the sexual dyad, and as such the results from this analysis cannot fully rule out reverse causation. However, the final GEE model, using only incident TV infection, does provide additional evidence of abnormal flora preceding TV infection. Finally, given that a minority of participants with abnormal vaginal flora was treated based on Amsel criteria, the aRR estimates are likely biased toward the null.
In conclusion, the findings of this study emphasize the need to screen for TV when women are found to have abnormal vaginal flora. More research is needed to help ascertain the mechanisms or cofactors involved in the relationship between BV and acquisition of TV including common exogenous factors, such as other sexually transmitted infections and host immune factors. It will be important to understand the mechanisms by which abnormal vaginal flora enhances susceptibility to TV.
1. Atashili J, Poole C, Ndumbe P, et al. Bacterial vaginosis and HIV acquisition: a meta-analysis of published studies. AIDS (London, England) 2008; 22:1493.
2. Feigin R. Textbook of Pediatric Infectious Diseases. 5th ed. Philadelphia, PA: Saunders; 2004.
3. Mitchell H. Vaginal discharge—causes, diagnosis, and treatment. BMJ 2004; 328:1306.
4. Schwebke JR, Burgess D. Trichomoniasis. Clin Microbiol Rev 2004; 17:794–803.
5. Paige M, David M, Augustyn M, et al. Bacterial vaginosis and preterm birth: a comprehensive review of the literature. J Nurse Midwifery 1998; 43:83–89.
6. Ness R, Kip K, Hillier S, et al. A cluster analysis of bacterial vaginosis–associated microflora and pelvic inflammatory disease. Am J Epidemiol 2005; 162:585.
7. Cherpes TL, Wiesenfeld HC, Melan MA, et al. The associations between pelvic inflammatory disease, Trichomonas vaginalis
infection, and positive herpes simplex virus type 2 serology. Sex Transm Dis 2006; 33:747–752. Doi: 10.1097/01.olq.0000218869.52753.c7.
8. Grodstein F, Goldman M, Cramer D. Relation of tubal infertility to history of sexually transmitted diseases. Am J Epidemiol 1993; 137:577.
9. Pellati D, Mylonakis I, Bertoloni G, et al. Genital tract infections and infertility. Eur J Obstet Gynecol Reprod Biol 2008; 140:3–11.
10. McClelland R, Sangaré L, Hassan W, et al. Infection with Trichomonas vaginalis
increases the risk of HIV-1 acquisition. J Infect Dis 2007; 195:698.
11. Nagot N, Ouedraogo A, Defer M, et al. Association between bacterial vaginosis and Herpes simplex virus type-2 infection: implications for HIV acquisition studies. Sex Transm Infect 2007; 83:365.
12. Madhivanan P, Bartman MT, Pasutti L, et al. Prevalence of Trichomonas vaginalis
infection among young reproductive age women in India: implications for treatment and prevention. Sex Health 2009; 6:339–344.
13. Brotman R, Erbelding E, Jamshidi R, et al. Findings associated with recurrence of bacterial vaginosis among adolescents attending sexually transmitted diseases clinics. J Pediatr Adolesc Gynecol 2007; 20:225–231.
14. Moodley P, Connolly C, Sturm A. Interrelationships among human immunodeficiency virus type 1 infection, bacterial vaginosis, trichomoniasis, and the presence of yeasts. J Infect Dis 2002; 185:69–73.
15. Demirezen S, Korkmaz E, Beksac M. Association between trichomoniasis and bacterial vaginosis: examination of 600 cervicovaginal smears. Cent Eur J Public Health 2005; 13:96–98.
16. Kaul R, Nagelkerke N, Kimani J, et al. Prevalent herpes simplex virus type 2 infection is associated with altered vaginal flora and an increased susceptibility to multiple sexually transmitted infections. J Infect Dis 2007; 196:1692–1697.
17. Gatski M, Martin DH, Clark RA, et al. Co-Occurrence of Trichomonas vaginalis
and bacterial vaginosis among HIV-positive women. Sex Transm Dis 2011; 38:163–166.
18. Josey W, Schwebke J. The polymicrobial hypothesis of bacterial vaginosis causation: a reassessment. Int J STD AIDS 2008; 19:152.
19. Schwebke J. Bacterial vaginosis. Curr Infect Dis Rep 2000; 2:14–17.
20. Martin H Jr, Richardson B, Nyange P, et al. Vaginal lactobacilli, microbial flora, and risk of human immunodeficiency virus type 1 and sexually transmitted disease acquisition. J Infect Dis 1999; 180:1863–1868.
21. Livengood C III. Bacterial vaginosis: an overview for 2009. Rev Obstet Gynecol 2009; 2:28.
22. Brotman R, Klebanoff M, Nansel T, et al. Bacterial vaginosis assessed by gram stain and diminished colonization resistance to incident gonococcal, chlamydial, and trichomonal genital infection. J Infect Dis 2010; 202:1907.
23. Krupp K, Madhivanan P, Karat C, et al. Novel recruitment strategies to increase participation of women in reproductive health research in India. Glob Public Health 2007; 2:395–403.
24. Madhivanan P, Krupp K, Chandrasekaran V, et al. Prevalence and correlates of bacterial vaginosis among young women of reproductive age in Mysore, India. Indian J Med Microbiol 2008; 26:132.
25. Madhivanan P, Krupp K, Chandrasekaran V, et al. The epidemiology of herpes simplex virus type-2 infection among married women in Mysore, India. Sex Transm Dis 2007; 34:935.
26. Amsel R, Totten P, Spiegel C, et al. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
27. Nugent R, Krohn M, Hillier S. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991; 29:297.
28. Hubbard AE, Ahern J, Fleischer NL, et al. To GEE or not to GEE: comparing population average and mixed models for estimating the associations between neighborhood risk factors and health. Epidemiology 2010; 21:467.
29. Vyas S, Kumaranayake L. Constructing socio-economic status indices: how to use principal components analysis. Health Policy Plan 2006; 21:459.
30. Cohen C, Duerr A, Pruithithada N, et al. Bacterial vaginosis and HIV seroprevalence among female commercial sex workers in Chiang Mai, Thailand. AIDS 1995; 9:1093.
31. Sewankambo N, Gray R, Wawer M, et al. HIV-1 infection associated with abnormal vaginal flora morphology and bacterial vaginosis. Lancet 1997; 350:546–550.
32. Taha T, Gray R, Kumwenda N, et al. HIV infection and disturbances of vaginal flora during pregnancy. J Acquir Immune Defic Syndr 1999; 20:52.
33. Hillier S, Krohn M, Klebanoff S, Eschenbach D. The relationship of hydrogen peroxide-producing lactobacilli to bacterial vaginosis and genital microflora in pregnant women. Obstet Gynecol 1992; 79:369.
34. Thurman A, Doncel G. Innate immunity and inflammatory response to Trichomonas vaginalis
and bacterial vaginosis: relationship to HIV acquisition. Am J Reprod Immunol (New York, NY: 1989) 2011;65:89–98.
35. Alderete JF, Garza GE. Specific nature of Trichomonas vaginalis
parasitism of host cell surfaces. Infect Immun 1985; 50:701.
36. Diamond LS. In vitro cultivation of the Trichomonadidae: a state of the art review. Acta Univ Carol-Biol 1986; 30:221–228.
37. Kostara I, Carageorgiou H, Varonos D, et al. Growth and survival of Trichomonas vaginalis
. J Med Microbiol 1998; 47:555.
38. Rein M, Chapel T. Trichomoniasis, candidiasis, and the minor venereal diseases. Clinical Obstet Gynecol 1975; 18:73.
39. Mitchell C, Hitti J, Agnew K, et al. Comparison of oral and vaginal metronidazole for treatment of bacterial vaginosis in pregnancy: impact on fastidious bacteria. BMC Infect Dis 2009; 9:89.
40. Lossick J. Single-dose metronidazole treatment for vaginal trichomoniasis. Obstet Gynecol 1980; 56:508.
41. Bowden F, Garnett G. Trichomonas vaginalis
epidemiology: parameterising and analysing a model of treatment interventions. Sex Transm Infect 2000; 76:248.
42. Schmid G, Steen R, N′Dowa F. Control of bacterial sexually transmitted diseases in the developing world is possible. Clin Infect Dis 2005; 41:1313–1315.
43. Gray P. HIV and Islam: is HIV prevalence lower among Muslims? Soc Sci Med 2004; 58:1751–1756.
44. Weiss H, Thomas S, Munabi S, Hayes R. Male circumcision and risk of syphilis, chancroid, and genital herpes: a systematic review and meta-analysis. BMJ 2006; 82:101.
45. Klinger E, Kapiga S, Sam N, et al. A community-based study of risk factors for Trichomonas vaginalis
infection among women and their male partners in Moshi urban district, northern Tanzania. Sex Transm Dis 2006; 33:712.
46. Brown J, Wald A, Hubbard A, et al. Incident and prevalent herpes simplex virus type 2 infection increases risk of HIV acquisition among women in Uganda and Zimbabwe. AIDS 2007; 21:1515.
47. Centers for Disease Control. Seroprevalence of herpes simplex virus type 2 among persons aged 14–49 Years: United States, 2005–2008. MMWR 2010; 59:456–459.
48. Brotman R, Ravel J, Cone R, et al. Rapid fluctuation of the vaginal microbiota measured by Gram stain analysis. Sex Transm Infect 2010; 86:297.
49. Gülmezoglu A, Garner P. Trichomoniasis treatment in women: a systematic review. Trop Med Int Health 1998; 3:553–558.