TRICHOMONAS VAGINALIS (Tv) INFECTION is the most common nonviral sexually transmitted disease (STD); each year worldwide more than 200 million men and women are infected with Tv. 1,2 Long thought to differ from other STDs in its route of transmission as well as its significance in the reproductive system, 3,4 Tv is now being recognized as a sexually transmitted agent able to amplify the transmission of HIV. 5–8 Like other nonulcerative STDs, Tv might increase HIV virus shedding in the genital tract and result in increased HIV infectivity of HIV/Tv-coinfected individuals. 7–13 Tv is especially prevalent among HIV-infected individuals and individuals with high-risk sexual behaviors. 1–3,14,15 Recurrence of Tv infection among HIV-infected women is particularly high; one study demonstrated that as many as 36% of such women with Tv have subsequent episodes. 14 Whether this represents reinfection or persistent infection is not known. Such high prevalence of repeat Tv infection could have public health as well as clinical implications. If Tv infection increases the infectiousness of HIV among coinfected individuals, reductions in Tv could result in concomitant reductions in the spread of HIV. 6
The associations between Tv and both immune status and viral suppression associated with protease inhibitor (PI) use have not been fully examined. Although one cross-sectional study suggested that Tv incidence increased as CD4 counts decreased, 13 2 other cohort studies found no association between Tv and CD4 count. 14–16 If decreased immune status is associated with increased or PI use is associated with decreased Tv incidence, Tv could represent a type of opportunistic infection. If this is the case, improvements in primary HIV care might result in decreases in Tv and, ultimately, in reductions in HIV transmission.
There were 2 objectives of this retrospective cohort study: 1) to evaluate the association between Tv and both immune status and PI use, and 2) to characterize factors associated with Tv among HIV-infected women. An improved understanding of these factors will identify whether Tv is an opportunistic infection, increasing with decreased immune status and viral suppression. It will also provide insight into the care of the Tv/HIV-coinfected woman, as well as assist in developing public health interventions to reduce STDs among this high-risk population.
To be eligible for this study, women had to be aged 13 years or older, enrolled in the Centers for Disease Control Prevention Adult Spectrum of Disease (ASD) database at the HIV Outpatient Program (HOP) at the Medical Center of Louisiana (MCLNO) between November 1, 1990, and September 18, 2000, have at least 1 year of clinical data available, and be HIV-infected as determined by standard methods. The data source for the study was the ASD database. ASD contains data on demographic, clinical, and some behavioral characteristics of HIV-infected men and women aged 13 years and older. 17 All elements of this study were approved by the governing institutional review board and research review committees before implementation.
The outcome of Tv infection was evaluated systematically for all women attending HOP. Women were screened every 6 months for Tv using wet mount/direct observation even in the absence of symptoms. Because without test-of-cure it is unknown whether subsequent infections are actually new infections or persistent infections (as a result of inadequate treatment or resistance), the outcome of subsequent infections is referred to as “subsequent Tv positivity” rather than reinfection or recurrence. Women at HOP are screened every 6 months for Chlamydia trachomatis (ELISA for C. trachomatis antigen or urine screening), Neisseria gonorrhoeae (culture or urine screening), and syphilis (serology) in addition to Tv. Presence of genital warts is evaluated clinically as well. Women positive for Tv were treated with metronidazole (either a 2-g one-time dose or 500 mg twice a day for 7 days), irrespective of symptoms. Patient-delivered partner treatment was not routinely offered, although women were encouraged to refer sex partners for treatment.
Only PI-containing highly active antiretroviral treatment (HAART) regimens were considered in this study as a result of the relatively small proportion of women on non-PI-containing regimens. CD4 counts were considered in the following categories: <200 cells/μL, 200–500 cells/μL, and >500 cells/μL. These categories were selected because of their correspondence with immune staging. Individuals with stable immune systems (>500 cells/μL category) were selected as the referent category.
Univariate and bivariate analyses were performed to describe the study sample and to evaluate potential confounders. Generalized estimating equation (GEE) methods were used to evaluate associations between independent variables and the outcomes initial Tv infection (yes/no) and subsequent Tv positivity (yes/no). GEE assesses associations between predictors and outcomes in a serially cross-sectional fashion, allowing for dependence between observations within a cluster, provided there is independence between clusters. 18–20 Exchangeable correlation matrices and robust estimators of variance were used; parameter estimates were exponentiated to provide odds ratios (OR) and confidence intervals (CI). This method was selected to allow for time-dependent covariates, and provides an OR (as a result of its estimate of prevalence at each point in time, with a variance that incorporates the intraperson correlations over time) rather than a relative risk. To calculate incidence of initial and subsequent Tv, women contributed person-time until their first infection, after which they contributed person-time toward their having a subsequent infection. Only women with Tv at least once were at risk for subsequent positivity.
A bivariate screening procedure was used to develop multivariable models that evaluated the predictors of interest and characterized factors associated with initial and subsequent Tv infection. Variables that were associated (P <0.25) with the outcomes in the bivariate analyses were eligible for inclusion. A stepwise procedure was then used, and variables with significant (P <0.05) associations or those whose addition or removal resulted in a 5% change in other parameter estimates were included in the adjusted model. To ensure that residual confounding was not missed, after the final models were selected, potential confounders were reentered into the model and the model was checked again for alteration in coefficients. Time was included in each of the adjusted models, as well as duration of follow up (as a proxy for access to or secular changes in care), which was dichotomized at the median. Once the final model's main effects were established, all 2-way interactions were assessed for significance. Stata Software, version 7.0 (Stata Corp, College Station, TX) was used for all analyses.
Description of Study Sample
Of the 1655 women with available data during the study period, 1578 (95.3%) were eligible to participate in the study. Table 1 displays the baseline characteristics of subjects. The majority (67.6%) of subjects were 35 years of age or under (mean, 31.47 y; standard deviation [SD], 9.75 y), black (83.2%), urban dwellers (86.6%), had stable immune status (CD4 >200 cells/μL) (57.8%), and identified as being infected through heterosexual exposure or having unidentified risk (72.2%). Nearly 18% (17.7%) of the women were documented intravenous drug users (IDU) or non-IDU drug users at baseline, and 24.5% were pregnant at baseline. Approximately half (51.3%) of the cohort was on antiretroviral treatment at least once during follow up, and 35.4% were on a PI-containing regimen at least once during follow up. Over three fourths (79.2%) of the women had between 1 and 6 years of follow up (mean, 5.47 y; SD, 2.40 y).
Tv was the most common of all STDs (Table 2). No other STD was as prevalent at or within 1 year of baseline as Tv, with 13.1% having Tv, compared with 5.3% for C. trachomatis and 4.9% for N. gonorrhoeae. Almost one fourth (22.4%) had any STD (C. trachomatis, N. gonorrhoeae, primary syphilis, or genital warts) at baseline. Over the study period, 30.2% of the 1578 women studied had Tv at least once. Of 786 episodes of Tv, there were 301 subsequent positive tests among 176 women, indicating that 36.9% of women had at least one subsequent positive test. Women with at least one prior episode of Tv had the bulk of the infections, with 61.7% of all infections among women who had at least one previous positive test. Among women with at least one episode of Tv, the mean number of events was 1.65 (median, 1; SD, 1.08; range, 1–7). Among women with at least 2 episodes of Tv, the mean number of events was 2.76 (median, 2; SD, 1.11; range, 2–7).
The crude incidence of initial Tv was 8.9 per 100 person-years and the incidence of subsequent Tv was 16.4 per 100 person-years. Women under 22 years of age, black women, and substance users had significantly higher incidences of initial positivity than did their counterparts (14.3, 10.3, 13.2, 23.8, and 17.4 per 100 person-years, respectively) (P <0.001). Women who were substance users had significantly higher incidences of subsequent positivity than did nonusers (52.3 and 50.0 per 100 person-years, respectively) (P <0.001).
Characteristics Associated With Initial and Subsequent Tv Positivity
Table 3 displays adjusted ORs and 95% CIs from GEE for the outcome of having an initial and subsequent Tv infection. Before adjustment for confounders, women with initial Tv were significantly more likely than those without to be less than 22 years old (OR, 1.37; 95% CI, 1.37–2.17), black (OR, 2.66; 95% CI, 1.82–3.89), to have become HIV-infected through IDU exposure (OR, 1.27; 95% CI, 1.01–1.59), be substance abusers (OR, 4.00; 95% CI, 3.15–5.08), have incident AIDS-defining diagnosis (OR, 1.42; 95% CI, 1.07–1.88), pregnancy (OR, 2.21; 95% CI, 1.75–2.79), or other STDs (OR, 3.00; 95% CI, 2.50–3.61). They did not differ about immune status, viral load, or antiretroviral treatment. After adjusting for pregnancy and follow-up time in multivariable analysis, age less than 22 years (OR, 1.33; 95% CI, 1.02–1.75), black race (OR, 2.44; 95% CI, 1.63–3.66), substance abuse (OR, 2.65; 95% CI, 1.99–3.53), and have other incident STDs (OR, 1.77; 95% CI, 1.43–2.18) were associated with initial Tv infection, but immune status, PI, and antiretroviral use were not associated.
Before adjustment for confounders, women with subsequent Tv positivity were significantly more likely to be substance abusers (OR, 1.61; 95% CI, 1.10–2.36), have an incident AIDS-defining diagnosis (OR, 1.67; 95% CI, 1.10–2.53), be on antiretroviral treatment (OR, 1.78; 95% CI, 1.36–2.32), and have other incident STDs (OR, 1.99; 95% CI, 1.49–2.67). They were significantly less likely to have a new pregnancy (OR, 0.54; 95% CI, 0.31–0.96). After adjusting for follow-up time in multivariable analysis, having other incident STDs was associated with subsequent positivity (OR, 1.44; 95% CI, 1.06–1.97) and pregnancy (OR, 0.47; 95% CI, 0.24–0.90) was protective of subsequent Tv positivity, but immune status, PI, and antiretroviral use were not associated; neither were age, race, or substance abuse.
This study corroborates the findings of other researchers that Tv is a highly prevalent STD among HIV-infected women, 14–16,21 that pregnancy is associated with decreased repeat Tv infection, 15 and that Tv is likely not an opportunistic infection. 14,16 It also suggests that Tv incidence among HIV-infected women is not associated with PI use.
Like other authors, 14–16 we found no association between CD4 count and Tv infection. This contrasts Ghys et al. 13 who found an association between lowered CD4 count and increased Tv, suggesting that Tv could represent an opportunistic infection. Our retrospective cohort study is a stronger design than the cross-sectional study reported by Ghys et al. Other STDs and gynecologic conditions increase with immunocompromise, and so it would be possible for Tv to behave similarly. However, the current data do not support the hypothesis that Tv is an opportunistic infection. Similarly, PI use was not associated with decreases in Tv infection. In comparing HIV-infected women with high risk for HIV infection, the HIV Epidemiology Research Study investigators 16 found that Tv was not associated with either HIV infection or CD4 cell count.
Our findings were consistent with those of Sorvillo et al. 14 regarding black race as a predictor of Tv as well. They found that Tv infection was significantly associated with black race, and, among non-black women only, trading sex for money or drugs and single marital status. Although they used a Cox proportional hazards approach instead of the GEE approach used in this study, it is interesting to note the similarity in our findings. In our study, caution should be taken with this finding in view of the high proportion of black women in the study sample.
The finding that pregnancy was protective of Tv infection offers insight into methods of Tv control. Women with incident pregnancies were less likely to have a subsequent infection. This finding regarding pregnancy supports that of Niccolai et al. 15 who found, using a Cox proportional hazards method, that pregnant women were at decreased risk for subsequent Tv positivity. Pregnancy's protective association with subsequent Tv infection could result from the more assiduous testing and treating of pregnant women with bacterial vaginosis or the reduction of sexual activity (and thus exposure to the pathogen) during pregnancy. Bacterial vaginosis has been associated with adverse pregnancy outcomes, including premature rupture of membranes. 22–25 As a result, pregnant women are frequently screened for bacterial vaginosis and treated with metronidazole. Asymptomatic Tv infections could thus be treated before diagnosis, explaining the association. That frequent screening and treatment of another vaginal infection results in significant reductions in Tv offers methods for improving clinic protocols to reduce Tv.
Current Centers for Disease Control and Prevention guidelines for the treatment of trichomoniasis 26 do not necessitate follow up for men or women who become asymptomatic after treatment or who are initially asymptomatic. Test-of-cure is not specifically recommended, and the guidelines consider cure of the index patient and partner as when therapy has been completed and both parties are asymptomatic. There are no specific recommendations made for high-risk populations such as HIV-infected individuals. This study's findings suggest that follow up is necessary irrespective of symptoms, and that partner testing and treatment should be aggressively pursued. Especially given the potential adverse outcomes of Tv infection as well as its high prevalence among HIV-infected women, specific guidelines for treatment of HIV/Tv-coinfected women should be developed.
Strengths and Limitations
The primary strength of this study is its ability to evaluate associations between Tv and both PI use and immune status. The size of the cohort provided 89% power to evaluate the association between PI use and Tv infection, and 99% power to evaluate the association between declining immune status and Tv infection. This lends confidence to the conclusion that neither is associated with initial or subsequent Tv infection. Another strength is that the ASD database is an excellent source of data. Abstractors follow a detailed protocol, and the data were checked for quality both locally and within the Centers for Disease Control and Prevention. Although abstractors knew when a subject had Tv or subsequent episodes of Tv, the hypotheses of this study were unknown to them at the time of abstraction, reducing the risk of bias in data collection. The nature of ASD, with over a decade of systematically abstracted data, allowed a comprehensive longitudinal evaluation of factors associated with Tv infection. This is especially important in an HIV-infected population that experiences immune system failure over time. The length of the follow up also allowed a comparison of women taking PIs with those not taking them. One primary reason for the selection of the GEE analysis is that it allowed the covariates, particularly those of immune status and protease inhibitor use, to fluctuate over time and for them to be more easily interpreted than in a Cox model.
There were several limitations to this study. One of the primary limitations stems from the inability to differentiate recurrent from persistent Tv infection. Prospective studies of HIV-infected women are necessary to evaluate these differences and discern what factors are associated with repeat infections versus persistent infections. Selection bias might have been threat to validity. The eligibility criteria required that women had at least 2 clinic visits to join the cohort, making them possibly more adherent than those who did not. Thus, women in the study might represent a more adherent population of women than those not in the clinic. Also, women with more clinic visits might have had Tv diagnosed more frequently. To evaluate this, women who attended the clinic twice or more each 6 months were compared to those attending once or not at all. After adjusting for CD4 count and antiretroviral use, there was no significant association between outpatient visits and initial or subsequent Tv infections. Detection bias was also a concern, because women with Tv symptoms or previous Tv infection might be more likely to be diagnosed than those without. However, we found that HOP's Tv screening protocol was not dependent upon the presence of symptoms or prior infections and that women were screened routinely every 6 months in almost all cases. We were also concerned that women with cervical dysplasia were not diagnosed more frequently with Tv as a result of increased Papanicolaou smears. The overall number of women with cervical dysplasia at baseline was quite low (3.2%) and cervical dysplasia was not significantly associated in any of the outcomes, suggesting that detection bias was not a threat to validity. One crucial limitation is the use of a relatively insensitive diagnostic method of Tv. Wet mount is only 35–80% sensitive, 1–4 and its sensitivity is largely dependent on the experience of the practitioner in visualizing motile protozoa under a microscope. Although the practitioners evaluating the women in the study for Tv were experienced in diagnosing Tv and were the same individuals for the duration of the study, the poor sensitivity of the diagnostic method could have underestimated the incidence of Tv. This problem might also result in either an underestimate or overestimate of the association between other predictors and outcomes of interest.
This study confirms that Tv is the most common nonviral STD among HIV-infected women seeking care at a public HIV outpatient program and that subsequent infection is common. Tv does not appear to be an opportunistic infection and does not appear to reduce with PI use. Efforts to encourage condom use, partner treatment, patient adherence to treatment, and screening protocols are necessary to reduce the burden of subsequent infection. HIV-infected women with Tv might be more infectious than Tv-uninfected as a result of increased HIV shedding in the genital tract. Given the high prevalence of Tv, it is imperative that efforts to prevent and treat infections thoroughly are undertaken. Such efforts might be able to successfully and inexpensively reduce HIV transmission.
1. Krieger JN, Alderete JF. Trichomonas vaginalis
and trichomoniasis. In: Holmes KK, Sparling PF, Lemon SM, et al, eds. Sexually Transmitted Disease. New York: McGraw-Hill; 1999.
2. Sobel JD. Current concepts: vaginitis. N Engl J Med 1997; 337: 1896–1903.
3. Bowden FJ, Garnett GP. Why is Trichomonas vaginalis
ignored? Sex Transm Infect 1999; 75: 372–374.
4. Peterson P. Trichomonas vaginalis
vaginitis. Am J Obstet Gynecol 1938; 35: 1004–1009.
5. Laga M, Manoka A, Kivuvu M, et al. Non-ulcerative sexually transmitted diseases as risk factors for HIV-1 transmission in women: results from a cohort study. AIDS 1993; 7: 95–102.
6. Sorvillo F, Kerndt P. Trichomonas vaginalis
and amplification of HIV-1 transmission. Lancet 1998; 351: 213–214.
7. Fleming DT, Wasserheit JN. From epidemiological synergy to public health policy and practice: the contribution of other sexually transmitted diseases to sexual transmission of HIV infection. Sex Transm Infect 1999; 75: 3–17.
8. Wang CC, McClellland RS, Reilly M, et al. The effect of treatment of vaginal infections on shedding of human immunodeficiency virus type-1. J Infect Dis 2001; 183: 1017–1022.
9. Clemetson DBA, Moss GB, Willerford DA, et al. Detection of HIV DNA in cervical and vaginal secretions. Prevalence and correlates among women in Nairobi, Kenya. JAMA 1993; 269: 2860–2864.
10. Mostad SB, Kreiss JK. Shedding of HIV-1 in the genital tract. AIDS 1996; 10: 1305–1315.
11. Mostad SB. Prevalence and correlates of HIV type-1 shedding in the female genital tract. AIDS Res Hum Retroviruses 1998; 14( suppl 1): S11–S15.
12. Iverson AKN. Genital HIV shedding in women. AIDS Patient Care STDs 1999; 12: 695–701.
13. Ghys PD, Fransen K, Diallo MO, et al. The associations between cervicovaginal HIV shedding, sexually transmitted diseases and immunosuppression in female sex workers in Abidjan, Cote d'Ivoire. AIDS 1997; 11: F85–F93.
14. Sorvillo F, Kovacs A, Kerndt P, Stek A, Muderspach L, Sanchez-Keeland L. Risk factors for trichomoniasis among women with HIV infection at a public clinic in Los Angeles County, California: implications for HIV prevention. Am J Trop Med Hyg 1998; 58: 495–500.
15. Niccolai LM, Kopicko JJ, Kassie A, Petros H, Clark R, Kissinger P. Incidence and predictors of reinfection with Trichomonas vaginalis
in HIV-infected women. Sex Transm Dis 2000; 27: 284–288.
16. Cu-Uvin S, Ko H, Jamieson DJ, et al. Prevalence, incidence, and persistence or recurrence of trichomoniasis among HIV-positive women and among HIV-negative women at high risk for HIV infection. Clin Infect Dis 2002; 34: 1406–1411.
17. Farizo KM, Buehler JW, Chamberland ME, et al. Spectrum of disease in persons with human immunodeficiency virus infection in the United States. JAMA 1992; 267: 1798–1805.
18. Zeger SL, Lang KY. An overview of methods for the analysis of longitudinal data. Stat Med 1992; 11: 1825–1839.
19. Stokes ME, Davis CS, Koch GG. Categorical Data Analysis Using the SAS System, 2nd ed. Cary, NC: SAS; 1998.
20. Zeger SL, Liang K-Y. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986; 42: 121–130.
21. Capps L, Peng G, Doyle M, et al. Sexually transmitted infections in women infected with the human immunodeficiency virus. Sex Transm Dis 1998; 25: 442–447.
22. Meis PJ, Goldenberg RL, Mercer B, et al. The preterm prediction study: significance of vaginal infections. Am J Obstet Gynecol 1995; 173: 1231–1235.
23. Cotch ME, Pastorek JG, Nugent RP, et al. Trichomonas vaginalis
associated with low birth weight and preterm delivery. Sex Transm Dis 1997; 24: 361–362.
24. McGregor JA, French JI, Parker R, et al. Prevention of premature birth by screening and treatment for common genital tract infections: results of a prospective controlled evaluation. Am J Obstet Gynecol 1995; 173: 157–167.
25. Sutton MY, Sternberg M, Nsuami M, Behets S, Nelson A, St. Louis M. Trichomoniasis in pregnant HIV-infected and HIV-uninfected Congolese women: incidence, risk factors, and association with low birth weight. Am J Obstet Gynecol 1999; 181: 656–662.
26. Centers for Disease Control and Prevention. 2002 guidelines for treatment of sexually transmitted diseases. MMWR Morb Mortal Wkly Rep 2002; 51( RR-6): 1–84.