Secondary Logo

Journal Logo

Article

Estimates of the Annual Number and Cost of New HIV Infections Among Women Attributable to Trichomoniasis in the United States

Chesson, Harrell W. PhD*; Blandford, John M. PhD*; Pinkerton, Steven D. PhD

Author Information
doi: 10.1097/01.olq.0000137900.63660.98
  • Free

CERTAIN SEXUALLY TRANSMITTED DISEASES (STDs) such as syphilis, gonorrhea, chlamydia, and genital herpes can facilitate the sexual transmission and acquisition of HIV.1–4 Recently, attention has been directed to the possible role of trichomoniasis as a cofactor for sexual transmission of HIV.3,5–10 Trichomoniasis is a common STD caused by the single-celled protozoan parasite Trichomonas vaginalis (Tv).11 An estimated 7.4 million new cases of trichomoniasis occurred in the United States in 2000,12 and it is the most common curable STD among young, sexually active women in the United States.11

Because infection with Tv typically elicits an inflammatory response of vaginal epithelium and exocervix in women and of the urethra in men, trichomoniasis could facilitate both transmission and acquisition of HIV.9 Studies demonstrating an association between HIV-1 and Tv in women13–15 suggest that trichomoniasis could increase women’s risk of HIV acquisition by a factor of 2 to 3.7,9,16 In addition, trichomoniasis-associated urethritis in men can increase HIV shedding in semen by a factor of 8, similar to the effects of gonococcal urethritis.5,8 Researchers have suggested that because trichomoniasis is so common—and because most of those infected manifest mild or no symptoms and are likely to remain sexually active while untreated—the number of HIV acquisitions attributable to trichomoniasis could be substantial.7,9,10

In this study, we used a simple model of HIV transmission to estimate the annual number of new HIV infections among women in the United States attributable to trichomoniasis. We also estimated the cost of the trichomoniasis-attributable HIV infections in women using a published estimate of the discounted, lifetime medical care costs of HIV. We focused exclusively on male-to-female transmission of HIV because published data do not allow reliable estimation of the effect of Tv on female-to-male or male-to-male HIV transmission.

Methods

We adapted a mathematical model of HIV transmission that was previously developed as a general tool for estimating the number of STD-attributable HIV infections and associated costs for any given STD.17 We applied Tv-specific inputs and used the modified model to estimate the probability that a given case of trichomoniasis would facilitate transmission of HIV from an HIV-infected male to a previously HIV-uninfected female sex partner.

The presence of Tv in 1 or both sex partners can facilitate male-to-female HIV transmission resulting from: 1) increased HIV infectiousness resulting from increased shedding in a Tv-infected male’s semen, or 2) enhanced susceptibility in a Tv-infected female. Specifically, let p denote the per-act probability of male-to-female HIV transmission in the absence of Tv infection in either partner. The presence of Tv in the male partner increases this per-act probability by a factor of θ1 (as a result of increased HIV infectiousness), and the presence of Tv in the female partner increases it by a factor of θ2 (as a result of increased HIV susceptibility).

Our model focused on women who acquire Tv and included 2 stages of their sexual partnerships: first when they are exposed to a partner infected with Tv (and possibly HIV) but have not yet acquired Tv, and second after the women have acquired Tv and are potentially exposed to HIV in the same or subsequent sexual relationship(s).

First consider the partnership in which the woman initially acquired Tv and assume that Tv was transmitted on the mth sex act but not before. If the Tv-source partner was infected with HIV as well as with Tv, then, taking into account the transmission-facilitating effect of Tv, the probability that the woman would acquire HIV as a consequence of these m acts is Ω = 1 - (1-θ1p)m.

We assumed that some proportion, T, of Tv-infected women would be symptomatic and that all of these women would receive curative treatment immediately after acquiring Tv (ie, just after the mth sex act with the Tv-source partner and before engaging in any additional sex acts). Women who had not already acquired HIV as a consequence of the m sex acts with the Tv-source partner (either because he was not HIV-infected or because the virus was not transmitted) were assumed to engage in n additional sex acts with the Tv-source partner or with 1 or more new partners. For simplicity and to keep our estimates conservative, we assumed that the additional sex acts were, in fact, with the Tv-source partner rather than with different partners (which would increase the likelihood of HIV transmission), but that there would be no increase in HIV transmissibility as a result of the male partner’s Tv infection. Instead, we assumed that the woman (if untreated for Tv) would be at increased susceptibility to HIV. If the male partner was HIV-infected, the cumulative HIV transmission probability over these n additional sex acts is Φ = 1 - (1-θ2p)n if the woman was not treated for her Tv infection and Φ* = 1 - (1-p)n if she was.

Thus, the combined probability that the woman acquired HIV during the m initial acts or the n additional acts equals Hm(1-ΔHw) [Ω + (1-Ω){(1-T)Φ + TΦ*}]. In this equation, the Hm(1-ΔHw) term represents the probability that a particular relationship would involve an HIV-infected man and an HIV-uninfected woman. This probability would equal Hm(1-Hw) if the partners were chosen randomly from a population in which the fraction of men who were HIV-infected was Hm and the fraction of women who were HIV-infected was Hw. In reality, partners are not chosen randomly by HIV status.18 To account for nonrandom partner selection, the model includes an assortative mixing variable Δ to indicate that an HIV-infected man is Δ times more likely to select a partner who is HIV-infected than would be expected by chance alone.

Notice that if Tv did not increase the likelihood of HIV transmission and acquisition, then the woman’s risk of HIV infection would have equaled Hm(1-ΔHw) [Ω* + (1-Ω*)Φ*], where Ω* = 1 - (1-p)m. Therefore, the total number of Tv-attributable HIV infections in women, per year, equals

where C is the number of women who acquire Tv each year.

Model Inputs

Model inputs are summarized in Table 1. Approximately 7.4 million new cases of trichomoniasis occur each year in the United States,12 and we assumed that half of these cases occur in women (C = 3.7 million). Base case values of the probability of HIV infection in men and women (Hm and Hw) were based on HIV prevalence rates for the general population, which suggest that approximately 0.78% of men and 0.16% of women in the 18- to 59-year-old age group are infected with HIV.19 We assumed that because of assortative mixing, HIV-infected persons were 3 times more likely to select an HIV-infected partner than would be suggested by chance alone (Δ = 3).18,20

T1-6
TABLE 1:
Model Inputs: Base Case Values and Ranges

After acquiring Tv, the woman might experience symptoms of Tv within 5 to 28 days.11 We conservatively assumed that all women with symptoms would promptly seek and obtain appropriate treatment and would abstain from sex until treatment was complete. Thus, only women with asymptomatic infections would be at subsequent enhanced risk of HIV acquisition as a result of increased susceptibility. Approximately 20% to 50% of Tv infections in women are symptomatic.21 In the base case analysis, we assumed 35% of women with Tv would experience symptoms and that all these women would receive treatment (T = 0.35).

The per-act probability of male-to-female HIV transmission (p) was based on published epidemiologic studies.22,23 Existing estimates of the cofactor effect (θ1) of increased infectiousness of a male sex partner coinfected with HIV and Tv are limited. However, the effect of trichomoniasis-associated urethritis in men on HIV shedding in semen is similar to that of gonococcal urethritis.5,8 Previous mathematical models of HIV transmission17,24 have applied a cofactor effect of 10 for gonorrhea. We applied a base case value of 5 for the cofactor effect of Tv on HIV infectivity, assuming that the effect of trichomoniasis on HIV transmission would be at least half that of gonorrhea.

Infection with Tv in women has been estimated to increase HIV susceptibility by a factor (θ2) of 1.8 to 3.0.7,9,13–15 Although the data on the magnitude of the θ2 is limited, a prospective cohort study from Africa provides a reasonable base case value of 1.9.13

In the base case analysis, we assumed that the number of sex acts (m) with the Tv-source partner (before the woman acquired Tv) was equal to the average number of sex acts required for Tv transmission to occur. If the per-act probability of Tv transmission is 0.5, then acquisition of Tv would be expected, on average, on the second sex act with an Tv-infected partner. This base case value could underestimate the number of sex acts with the Tv-source partner, because the per-act probability of Tv transmission could be less than 0.5.25

We assumed each woman not seeking treatment for Tv would have 33 unprotected sex acts (n) while at increased susceptibility to HIV. This estimate was calculated as the product of 180 days of asymptomatic infection and 0.186 unprotected sex acts per day.26 Although Tv infection could be of indefinite duration,25 we assumed that women with untreated Tv infection would be at increased susceptibility for 180 days. The number of unprotected sex acts per day (0.186) was based on data from the National Health and Social Life Survey.26 In estimating the number of unprotected sex acts per day for women at risk of acquiring an STD, we 1) calculated the mean rate of sexual activity, 2) adjusted for the likelihood that the activity was vaginal intercourse and that intercourse occurred without a condom, and 3) adjusted the population-weighted rates of unprotected sexual activity, assuming that Tv infection was equally distributed across age groups.12,27

Cost of HIV Infections Attributable to Trichomonas vaginalis

To estimate the cost of HIV infections attributable to Tv, we multiplied the estimated number of HIV infections attributable to Tv by a published estimate of the discounted, lifetime medical cost per case of HIV (approximately $224,000 in 2002 U.S. dollars).28

Sensitivity Analyses

We performed univariate and multivariate sensitivity analyses. In the univariate sensitivity analyses, we varied 1 model input at a time, holding other model inputs at their base case values. In the multivariate analysis, we performed a Monte Carlo simulation29 in which we estimated the number of HIV infections attributable to Tv 10,000 times, each time choosing random values for each model input from a uniform distribution between that input’s lower and upper bounds.

Results

Under base case assumptions, an estimated 746 new HIV cases among women can be attributed to Tv each year. Among the 3.7 million women who acquire Tv annually, an estimated 1734 also would acquire HIV. In the absence of Tv infection, only 988 of these women would have been expected to acquire HIV. The lifetime medical care costs of the 746 Tv-attributable HIV infections was estimated to be $167 million.

In the univariate sensitivity analyses (Table 2), the estimated HIV cases attributable to Tv ranged from 343 to 1149. The results were particularly sensitive to the cofactor effect (θ), the probability of HIV transmission (p), the number of sex acts with the Tv-source partner and with subsequent partners (n and m), the probability that the woman’s sex partner is HIV-infected (Hm), and the number of Tv infections in women (C). In the multivariate sensitivity analyses (Monte Carlo simulations), the mean number of HIV infections attributable to Tv was 739 (cost: $166 million) and ranged from 207 to 1703 (cost: $46–382 million) in 90% of the simulations.

T2-6
TABLE 2:
Estimated Number of New HIV Cases Among Women Attributable to Trichomoniasis, After Applying Upper and Lower Bound Values of Model Inputs: Univariate Sensitivity Analysis

Discussion

Our results indicate that the number and cost of new HIV infections attributable to Tv are substantial. Because these results depend on a number of assumptions, it is useful to examine the reasonableness of our findings in light of other available estimates. Our estimate of the number of trichomoniasis-attributable HIV infections (746) among women represents approximately 6.2% of the estimated 12,000 new HIV infections that occur among women in the United States each year.30 Previously published calculations suggest that the percentage of HIV infections attributable to Tv in a given population could range from 0% to more than 30% (depending on the prevalence of Tv in that population), assuming that Tv causes a 2- or 3-fold increase in the probability of HIV transmission.9 Another modeling exercise suggested that STD treatment for persons with HIV attending STD clinics could reduce HIV transmission from dually infected persons by 27%.31 In light of these previous studies, and because trichomoniasis is so common, it would not be unreasonable to expect trichomoniasis to account for 6.2% of new HIV infections among women.

Our study was intended to provide a “ballpark” estimate of the number and cost of Tv-attributable HIV cases among women. Our estimates are subject to several limitations, most importantly, the uncertainty in our model inputs and the sensitivity of our results to changes in some of these inputs. The most important input is the cofactor effect of Tv on HIV transmission and acquisition (θ1 and θ2). We based our estimates of the cofactor effect on odds ratios reported in studies that examined HIV acquisition over a series of sexual encounters (such as over the course of a partnership or a given period of time). Because Tv might not have been present over the entire series of sexual encounters, these odds ratios could understate the true per-act cofactor effect.32 Conversely, although there is strong evidence that Tv can facilitate HIV transmission and acquisition, this evidence is not conclusive and the possibility that Tv might have no effect on HIV transmission and acquisition cannot be ruled out definitively.

We assumed that women with asymptomatic Tv would be at increased susceptibility to HIV. If symptomatic Tv were required to facilitate HIV acquisition, then, under the conservative assumption that all symptomatic women receive prompt treatment, the only HIV transmission-facilitating effect of Tv would be the result of the increased infectiousness of the male Tv-source partner. If we assumed that each Tv-uninfected women had only 2 sex acts while her TV-infected partner was at increased risk of transmitting HIV (like in the base case analysis), the number of Tv-attributable infections would drop to only 222.

Another limitation is that we do not address the possibility that a woman or her sex partner has another non-HIV STD (such as syphilis or chlamydia) at a given time. Because we applied a low per-act cofactor effect of Tv on the probability of HIV transmission, however, it is possible that such a small increase in the probability of HIV transmission might still be attributable to Tv even in the presence of another STD. We would need a much more complex transmission model to incorporate the effect of multiple STD infections, secondary HIV transmissions, and other dynamic factors. Additional strengths and limitations of our approach and of Bernoulli models of HIV transmission in general are discussed more thoroughly elsewhere.33–35

Despite these limitations, our model provides a useful estimate of the number of HIV infections in women attributable to Tv. Our results show that even under conservative assumptions, the impact of Tv on HIV can be considerable. Under less conservative assumptions, the estimated impact of Tv on HIV would be even greater, as shown in the multivariate sensitivity analysis in which the number of Tv-attributable HIV infections exceeded 1703 in 5% of the simulations. Although more research is needed on the role of Tv in HIV transmission, current research suggests that Tv infection might play an important role in HIV transmission and acquisition.

Efforts to prevent Tv could help prevent HIV transmission and could help reduce the estimated $167 million in future medical treatment costs associated with the Tv-attributable HIV cases that occur each year. These Tv-attributable HIV costs can be used in evaluating the cost-effectiveness of STD prevention programs.17 Because Tv is so common, however, a substantial number of men and women with Tv would have to be treated to have a discernible impact on HIV. Future research is needed to examine the cost-effectiveness of Tv prevention as a tool for HIV prevention.

References

1. Wasserheit JN. Epidemiological synergy. Interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis 1992; 19:61–77.
2. 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.
3. Centers for Disease Control and Prevention. HIV prevention through early detection and treatment of other sexually transmitted diseases—United States. MMWR Recomm Rep 1998; 47(RR-12):1–24.
4. Cohen MS. Sexually transmitted diseases enhance HIV transmission: No longer a hypothesis. Lancet 1998; 351(suppl 3):5–7.
5. Cohen MS, Hoffman IF, Royce RA, et al. Reduction of concentration of HIV-1 in semen after treatment of urethritis: Implications for prevention of sexual transmission of HIV-1. Lancet 1997; 349:1868–1873.
6. Jackson DJ, Rakwar JP, Bwayo JJ, et al. Urethral Trichomonas vaginalis infection and HIV-1 transmission. Lancet 1997; 350:1076.
7. Sorvillo F, Kerndt P. Trichomonas vaginalis and amplification of HIV-1 transmission. Lancet 1998; 351:213–214.
8. Hook EW III. Trichomonas vaginalis—No longer a minor STD. Sex Transm Dis 1999; 26:388–389.
9. Sorvillo F, Smith L, Kerndt P, et al. Trichomonas vaginalis, HIV, and African-Americans. Emerg Infect Dis 2001; 7:927–932.
10. Schwebke JR. Update of trichomoniasis. Sex Transm Infect 2002; 78:378–379.
11. Division of STD Prevention. Trichomoniasis. Atlanta: Division of STD Prevention, Centers for Disease Control and Prevention, 2002.
12. Weinstock H, Berman S, Cates W Jr. Sexually transmitted diseases among American youth: incidence and prevalence estimates, 2000. Perspect Sex Reprod Health 2004; 36:6–10.
13. 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.
14. Ghys PD, Diallo MO, Ettiegne-Traore V, et al. Genital ulcers associated with human immunodeficiency virus-related immunosuppression in female sex workers in Abidjan, Ivory Coast. J Infect Dis 1995; 172:1371–1374.
15. ter Meulen J, Mgaya HN, Chang-Claude J, et al. Risk factors for HIV infection in gynaecological inpatients in Dar es Salaam, Tanzania, 1988–1990. East Afr Med J 1992; 69:688–692.
16. Laga M, Alary M, Nzila N, et al. Condom promotion, sexually transmitted diseases treatment, and declining incidence of HIV-1 infection in female Zairian sex workers. Lancet 1994; 344:246–48.
17. Chesson HW, Pinkerton SD. Sexually transmitted diseases and the increased risk for HIV transmission: Implications for cost-effectiveness analyses of sexually transmitted disease prevention interventions. J Acquir Immune Defic Syndr 2000; 24:48–56.
18. Dow WH, Philipson T. An empirical examination of the implications of assortative matching on the incidence of HIV. J Health Econ 1996; 15:735–749.
19. Rosenberg PS. Scope of the AIDS epidemic in the United States. Science 1995; 270:1372–1375.
20. Chesson HW, Pinkerton SD, Voigt R, et al. HIV infections and associated costs attributable to syphilis coinfection among African Americans. Am J Public Health 2003; 93:943–948.
21. Krieger JN, Alderete JF. Trichomonas vaginalis and trichomoniasis. In: Holmes KK, Sparling PF, Mardh P, et al., eds. Sexually Transmitted Diseases. New York: McGraw-Hill, 1999:587–604.
22. Mastro TD, de Vincenzi I. Probabilities of sexual HIV-1 transmission. AIDS 1996; 10(suppl A):S75–82.
23. Royce RA, Sena A, Cates W Jr, et al. Sexual transmission of HIV. N Engl J Med 1997; 336:1072–1078.
24. Rehle TM, Saidel TJ, Hassig SE, et al. AVERT: A user-friendly model to estimate the impact of HIV/sexually transmitted disease prevention interventions on HIV transmission. AIDS 1998; 12(suppl 2):S27–S35.
25. Bowden FJ, Garnett GP. Trichomonas vaginalis epidemiology: Parameterising and analysing a model of treatment interventions. Sex Transm Infect 2000; 76:248–256.
26. Laumann EO, Gagnon JH, Michael RT, et al. The Social Organization of Sexuality. Chicago: The University of Chicago Press, 1994.
27. Cotch MF, Pastorek JG, Nugent RP, et al. Demographic and behavioral predictors of Trichomonas vaginalis infection among pregnant women. Obstet Gynecol 1991; 78:1087–1092.
28. Holtgrave DR, Pinkerton SD. Updates of cost of illness and quality of life estimates for use in economic evaluations of HIV prevention programs. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 16:54–62.
29. Doubilet P, Begg CB, Weinstein MC, et al. Probabilistic sensitivity analysis using Monte Carlo simulation. A practical approach. Med Decis Making 1985; 5:157–177.
30. Division of HIV Prevention. A Glance at the HIV Epidemic. Atlanta: Centers for Disease Control and Prevention, 2001.
31. Rothenberg RB, Wasserheit JN, St. Louis ME, et al. The effect of treating sexually transmitted diseases on the transmission of HIV in dually infected persons: A clinic-based estimate. Sex Transm Dis 2000; 27:411–416.
32. Hayes RJ, Schulz KF, Plummer FA. The cofactor effect of genital ulcers on the per-exposure risk of HIV transmission in sub-Saharan Africa. J Trop Med Hyg 1995; 98:1–8.
33. Chesson HW, Pinkerton SD, Irwin KL, et al. New HIV cases attributable to syphilis in the USA: Estimates from a simplified transmission model. AIDS 1999; 13:1387–1396.
34. Pinkerton SD, Abramson PR. Evaluating the risks: A Bernoulli process model of HIV infection and risk reduction. Eval Rev 1993; 18:504–528.
35. Holtgrave DR, Leviton LC, Wagstaff DA, et al. Cumulative probability of HIV infection: A summary risk measure for HIV prevention intervention studies. AIDS Behav 1997; 1:169–172.
© Copyright 2004 American Sexually Transmitted Diseases Association