Serosorting is the practice of choosing sexual partners thought to be of concordant HIV serostatus to reduce the risk of HIV transmission during unprotected intercourse.1–3 There is strong evidence to suggest that in various populations of men who have sex with men (MSM) around the world the formation of casual sexual partnerships and behavior within such partnerships has been increasingly influenced by the disclosure of HIV serostatus.1,4–9 But the extent to which the practice of serosorting protects HIV-uninfected MSM from acquiring HIV is unclear. In this study we aim to quantify the relative effectiveness of serosorting across a wide spectrum of contexts related to the prevalence of undiagnosed infections. The effectiveness of serosorting for reducing the risk of HIV acquisition is highly dependent on other population-specific parameters, in particular, the coverage and frequency of HIV testing. Serosorting may in fact lead to increased risk10 if a partner's HIV serostatus is unintentionally misreported because it is unknown. Consequently, the proportion of the sexually active population that is HIV-infected but undiagnosed is a crucial determinant of the effectiveness of serosorting. In the ideal scenario of 100% accurate knowledge and no misreporting of serostatus, serosorting is 100% effective at preventing HIV acquisition by HIV-negative people who practice it. Conversely, serosorting may result in substantially increased risk of HIV acquisition if a large proportion of the HIV-infected population is undiagnosed and serosorting thereby leads to the formation of discordant partnerships and more unprotected sex. Here, we present the results of a mathematical modeling analysis that demonstrate the importance of context, especially the prevalence of undiagnosed HIV infection, in determining the effectiveness of serosorting in reducing the risk of HIV acquisition. We estimate thresholds for the prevalence of undiagnosed HIV infections in a population for which serosorting increases risk.
We developed a mathematical model to investigate the relationship between the proportion of the HIV-infected population that is undiagnosed and the relative risk of serosorting, which refers to the risk of HIV acquisition for MSM who practice serosorting relative to the risk of HIV acquisition for those who do not practice serosorting. A full description of the mathematical formulation of the model is given in the Appendix (available online at http://links.lww.com/OLQ/A2). Briefly, the relative risk of serosorting is determined by calculating the probability that a casual partnership is serodiscordant (a prerequisite for HIV transmission) when serosorting does and does not occur, and the probability that HIV transmission will occur in both cases. The probability of any casual partnership being serodiscordant will be dependent on the proportion of the population who have undiagnosed infection and the prevalence of HIV in the population. The probability of transmission in a serodiscordant partnership is taken to be dependent on the proportion of acts that are insertive, receptive with withdrawal and receptive with ejaculation, and the proportion of partners with diagnosed HIV infection that are on antiretroviral treatment (ART). We assume different rates of transmission from cases on ART than from cases not on ART but we did not consider heterogeneity in transmission rates throughout the natural course of infection. We use detailed behavioral data from the health in men cohort,11,12 a highly studied cohort of gay men, to inform the expected proportion of acts that are insertive, receptive with withdrawal or receptive with ejaculation based on knowledge of the partner's HIV status [see Fig. A1 in the Appendix (available online at http://links.lww.com/OLQ/A3)]. The modeled formulation yields theoretical maximum and minimum relative risks of serosorting of: ∼2.15 when all HIV-infected people are undiagnosed, and 0 when everyone knows their true HIV serostatus, respectively.
To evaluate the relative risk of serosorting in specific settings we chose parameters from 5 representative locations. (1) In Sydney, Australia, the percentage of HIV infections that are undiagnosed has been estimated from a modeling study to be ∼10%13 and although there are no empirical estimates available, this is consistent with data from a community-based study that attempted to recruit HIV-negative gay men.14 In Australia, ∼70% of diagnosed cases are on ART.15 (2) It is widely cited that in the United States ∼48% of all infections are undiagnosed.16 However, there is very large heterogeneity between cities and other demographic variables, particularly age and ethnicity. (3) Other estimates suggest that ∼25% of HIV infections in the United States are undiagnosed.17–19 We assume treatment rates are between 50% and 70% among diagnosed cases in the United States. (4) It has been estimated that ∼44% of HIV infections in London, England are yet to be diagnosed.20 We also assume that treatment rates are around 50% of all diagnosed infections in this population.21 (5) We assume that in many sub-Saharan countries ∼80% to 90% of all HIV infections are undiagnosed and HIV treatment rates are low (20%).
In Figure 1 we present the modeled relationship between the relative risk of HIV acquisition due to serosorting compared with not serosorting, versus the proportion of the HIV-infected MSM population that are undiagnosed, for various treatment rates. As the proportion of undiagnosed infections increases there is a corresponding increase in the risk associated with serosorting. Although it is theoretically possible for serosorting to achieve 100% effectiveness (if everyone knows and discloses their true serostatus), in a worse case scenario the relative risk of practicing serosorting is more than double the risk of not serosorting (Fig. 1); this value is informed by the average proportion of acts that are insertive or receptive with or without withdrawal and the average transmission rates for each type of sex act Appendix (Available online at http://links.lww.com/OLQ/A2). The relative risk of serosorting was not found to be highly sensitive to HIV prevalence but it was moderately sensitive to HIV treatment rates. We estimate that serosorting is only of any benefit at all for reducing HIV risk when the proportion of men with HIV who are undiagnosed is less than ∼20% in the case of high treatment rates (70%) and less than ∼40% for low treatment rates (20%) (Fig. 1), assuming treatment reduces transmission rates by 95%.
In resource-rich countries where there are often high rates of testing for HIV amongst MSM, serosorting may be an effective means of reducing the incidence of HIV infections. For example, we estimate that the relative risk of serosorting in Sydney, Australia (where approximately 10% of HIV infections are undiagnosed) is ∼0.57 (i.e., 43% effectiveness) (Fig. 1). Using commonly cited estimates of the percentage of undiagnosed infections in the United States (of 48%), our model indicates that practicing serosorting in the United States is likely to increase risk by almost 50% (Fig. 1). If just 25% of HIV infections in the United States were undiagnosed then it is unclear whether practicing serosorting would increase or decrease the risk of HIV acquisition (Fig. 1). We estimate that practicing serosorting in London (44% undiagnosed) is likely to increase the risk of acquiring HIV by over 30%. In comparison, in resource-constrained countries where the prevalence of undiagnosed infections is high, serosorting would not be effective even at low treatment rates and could increase HIV risk by ∼90% (Fig. 1). However, these thresholds would be altered if sexual positioning preferences among gay men were considerably different in other settings than they are for gay men in Sydney.
In Figure 2 we present the relative risk of serosorting for different combinations of treatment rates and prevalence of undiagnosed HIV infections. It can be seen that if treatment rates increase then the relative risk associated with serosorting worsens. This finding may seem counter-intuitive but it is important to note that the absolute risk of HIV transmission in the population decreases with increases in treatment rates. But the effectiveness of serosorting is calculated by the ratio of the chance of transmission when serosorting to the chance of transmission when not serosorting in a particular setting. Treatment rates do not influence the risk of acquiring HIV when serosorting and only choosing partners who are thought to be HIV-negative. However, when serosorting does not take place sexual partners of any serostatus could be chosen and could include partners who are HIV-positive and on ART. Consequently, the relative risk of serosorting compared with not serosorting is calculated using a term that is independent of treatment divided by a term that decreases with increasing treatment rates; the result is a relative risk statistic that increases with treatment rates. Therefore, high treatment rates in a population are likely to decrease overall incidence levels, however, serosorting has a reduced relative benefit compared with not serosorting because the risks of transmissions are less. Serosorting can still be of additional value in high-treatment settings if there are significant increases in testing rates which reduce the prevalence of undiagnosed infections. For example, in London the prevalence of undiagnosed HIV infections would need to decrease from 44% to ∼30% for serosorting not to increase risk. But if treatment rates increased from 50% to 70% then the prevalence of undiagnosed infections would need to decrease to ∼20% for the practice of serosorting to have the same risk as not serosorting.
Serosorting is unlikely to be highly beneficial in many populations of MSM, especially when the proportion of undiagnosed HIV infections is relatively high. Indeed serosorting with casual partners will likely increase HIV risk in settings where the percentage of infections that are undiagnosed is greater than approximately 40%. This percentage is likely to be exceeded in most resource-constrained countries and also in many resource-rich settings. However, our calculations suggest that serosorting can lead to effective reduction in risk in those locations where the proportion of undiagnosed HIV infections is relatively low.
Our mathematical modeling analysis of HIV transmission incorporates detailed data of sexual positioning and risk reduction strategies in MSM. A limitation of our analysis is that it is based on behavioral data of gay men in Sydney, Australia and this may not be representative of practices of MSM in other locations. However, similar trends in serosorting and strategic positioning have been observed in MSM throughout the industrialized world.1,5,6,9 Another limitation is that we used average transmission levels across all stages of HIV infection. We assumed that ART reduces transmission by 95%, however, the actual reduction in transmission rates due to ART is not yet well established.
The effectiveness of serosorting depends on accurate disclosure of HIV serostatus. Some HIV-infected people may mistakenly believe that they are not infected and thus disclose as HIV-negative to sexual partners.22 Disclosure of HIV status in sexual partnerships may also lead to other strategies to reduce the risk of HIV transmission, such as strategic positioning and negotiated safety.23–26 But disclosing HIV status may be associated with substantial stigma, so that even if a man knows he is HIV-infected, he may not disclose his HIV status or may even disclose false information (although this is thought to be relatively rare). However, even if truthful disclosure always occurs, serosorting is a potentially risky strategy and its effectiveness for an individual uninfected man also depends on the frequency of testing of his sexual partner and the partner's previous risk exposure. At a population level, the effectiveness of serosorting is specific to each context because it depends on the average HIV testing coverage and frequency in the population, and the rate of treatment. Different testing and treatment patterns exist between different populations.
Serosorting is safer than having unprotected anal intercourse with a serodiscordant partner but our model-based estimates suggest that it is not as safe as consistent condom use. Here, we only considered unprotected sex and not sex where a condom is used. If condoms are only used with partners of known discordant status, then it is imperative to know the true HIV status of partners. Our analysis suggests that unless the surrounding population has a small proportion of undiagnosed infections, serosorting is likely to increase the risk of HIV acquisition in practice. Therefore, it is not appropriate to suggest that serosorting may offer partial protection from HIV. The available evidence suggests that HIV-uninfected men who serosort remain at risk of acquiring HIV infection, and quite possibly at significantly increased risk compared with not serosorting.
Encouraging frequent testing for HIV, particularly among people at high risk, is a very important public health strategy. In addition to greater accuracy in serosorting, the benefits of undiagnosed HIV-infected individuals becoming aware of their serostatus include the tendency for sexual behavior to decrease and treatment can be sought. Both of these lower the risk of HIV transmission. Various behavioral studies provide evidence that risky sexual behavior decreases after HIV diagnosis.27–34 Increasing HIV diagnosis rates requires educating susceptible populations to seek HIV testing on the basis of indicators such as a known exposure to HIV or early recognition of symptoms. The practice of serosorting is increasing among MSM in various settings. In order for this practice to be beneficial, public health campaigns must continue to promote frequent HIV testing among groups at high risk. This message is of even greater importance in regions where current testing rates are relatively low.
1. Mao L, Crawford JM, Hospers HJ, et al. “Serosorting” in casual anal sex of HIV-negative gay men is noteworthy and is increasing in Sydney, Australia. AIDS 2006; 20:1204–1206.
2. Suarez T, Miller J. Negotiating risks in context: a perspective on unprotected anal intercourse and barebacking among men who have sex with men–where do we go from here? Arch Sex Behav 2001; 30:287–300.
3. Suarez TP, Kelly JA, Pinkerton SD, et al. Influence of a partner's HIV serostatus, use of highly active antiretroviral therapy, and viral load on perceptions of sexual risk behavior in a community sample of men who have sex with men. J Acquir Immune Defic Syndr 2001; 28:471–477.
4. Elford J, Bolding G, Sherr L, et al. High-risk sexual behavior among London gay men: No longer increasing. AIDS 2005; 19:2171–2174.
5. Elford J, Ibrahim F, Bukutu C, et al. Sexual behaviour of people living with HIV in London: Implications for HIV transmission. AIDS 2007; 21(suppl 1):S63–S70.
6. Truong HM, Kellogg T, Klausner JD, et al. Increases in sexually transmitted infections and sexual risk behaviour without a concurrent increase in HIV incidence among men who have sex with men in San Francisco: A suggestion of HIV serosorting? Sex Transm Infect 2006; 82:461–466.
7. Golden MR, Wood RW, Buskin SE, et al. Ongoing risk behavior among persons with HIV in medical care. AIDS Behav 2007; 11:726–735.
8. Elford J. Changing patterns of sexual behavior in the era of highly active antiretroviral therapy. Curr Opin Infect Dis 2006; 19:26–32.
9. Xia Q, Molitor F, Osmond DH, et al. Knowledge of sexual partner's HIV serostatus and serosorting practices in a California population-based sample of men who have sex with men. AIDS 2006; 20:2081–2089.
10. Butler DM, Smith DM. Serosorting can potentially increase HIV transmissions. AIDS 2007; 21:1218–1220.
11. Jin F, Prestage GP, Mao L, et al. Transmission of herpes simplex virus types 1 and 2 in a prospective cohort of HIV-negative gay men: The health in men study. J Infect Dis 2006;194:561–570.
12. Jin F, Crawford J, Prestage GP, et al. Unprotected anal intercourse, risk reduction behaviors, and subsequent HIV infection in a cohort of homosexual men. AIDS 2009; 23:243–252.
13. Wilson DP, Hoare A, Regan DG, et al. Mathematical Models to Investigate Recent Trends In HIV Notifications Among Men Who Have Sex With Men in Australia. Sydney, Australia: National Centre in HIV Epidemiology and Clinical Research, 2008. Available at: http://www.nchecr.unsw.edu.au/nchecrweb.nsf/page/BioModInfectDis
14. Jin F, Prestage GP, McDonald A, et al. Trend in HIV incidence in a cohort of homosexual men in Sydney: Data from the health in men study. Sex Health 2008; 5:109–112.
15. Falster K, Gelgor L, Shaik A, et al. Trends in antiretroviral treatment use and treatment response in three Australian states in the first decade of combination antiretroviral treatment. Sex Health 2008; 5:141–154.
16. Glynn MK, Rhodes P. Estimated HIV prevalence in the United States at the end of 2003. Paper presented at: The 2005 National HIV Prevention Conference; June 14, 2005:205 Atlanta, GA.
17. Branson B. Current HIV epidemiology and revised recommendations for HIV testing in health-care settings. J Med Virol 2007; 79(suppl 1):S6–S10.
18. Branson BM, Handsfield HH, Lampe MA, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep 2006; 55(RR-14):1–17; quiz CE1–4.
19. Stekler JD, Golden MR. Learning from the missed opportunities for HIV testing. Sex Transm Infect 2009; 85:2–3.
20. Dodds JP, Johnson AM, Parry JV, et al. A tale of three cities: Persisting high HIV prevalence, risk behaviour and undiagnosed infection in community samples of men who have sex with men. Sex Transm Infect 2007; 83:392–396.
21. MacKellar DA, Valleroy LA, Behel S, et al. Unintentional HIV exposures from young men who have sex with men who disclose being HIV-negative. AIDS 2006; 20:1637–1644.
22. Kippax S, Crawford J, Davis M, et al. Sustaining safe sex: A longitudinal study of a sample of homosexual men. AIDS 1993; 7:257–263.
23. Kippax S, Noble J, Prestage G, et al. Sexual negotiation in the AIDS era: Negotiated safety revisited. AIDS 1997; 11:191–197.
24. Van de Ven P, Kippax S, Crawford J, et al. In a minority of gay men, sexual risk practice indicates strategic positioning for perceived risk reduction rather than unbridled sex. AIDS Care 2002; 14:471–480.
25. Tindall B, Swanson C, Donovan B, et al. Sexual practices and condom usage in a cohort of homosexual men in relation to human immunodeficiency virus status. Med J Aust 1989; 151:318–322.
26. Van de Ven P, Mao L, Fogarty A, et al. Undetectable viral load is associated with sexual risk taking in HIV serodiscordant gay couples in Sydney. AIDS 2005; 19:179–184.
27. National Centre in HIV Social Research. Annual Report of Trends in Behavior. Sydney, Australia: University of New South Wales, 2006.
28. Marks G, Crepaz N, Senterfitt JW, et al. Meta-analysis of high-risk sexual behavior in persons aware and unaware they are infected with HIV in the United States: Implications for HIV prevention programs. J Acquir Immune Defic Syndr 2005; 39:446–453.
29. Cleary PD, Van Devanter N, Rogers TF, et al. Behavior changes after notification of HIV infection. Am J Public Health 1991; 81:1586–1590.
30. Colfax GN, Buchbinder SP, Cornelisse PG, et al. Sexual risk behaviors and implications for secondary HIV transmission during and after HIV seroconversion. AIDS 2002; 16:1529–1535.
31. McCusker J, Stoddard AM, Mayer KH, et al. Effects of HIV antibody test knowledge on subsequent sexual behaviors in a cohort of homosexually active men. Am J Public Health 1988; 78:462–467.
32. Saah AJ, Hoover DR, Weng S, et al. Association of HLA profiles with early plasma viral load, CD4+ cell count and rate of progression to AIDS following acute HIV-1 infection. Multicenter AIDS Cohort Study. AIDS 1998; 12:2107–2113.
33. Smith DK, Warren DL, Vlahov D, et al. Design and baseline participant characteristics of the Human Immunodeficiency Virus Epidemiology Research (HER) Study: A prospective cohort study of human immunodeficiency virus infection in US women. Am J Epidemiol 1997; 146:459–469.
34. Valleroy LA, MacKellar DA, Karon JM, et al. HIV prevalence and associated risks in young men who have sex with men. Young Men's Survey Study Group. JAMA 2000; 284:198–204.