Reports from the United States, Australia, and the United Kingdom describe sexual harm reduction practices like serosorting among men who have sex with men (MSM).1-7 In serosorting, potential partners discuss their HIV status and engage in risky behavior only with those believed to have concordant HIV status.5,8 Serosorting might explain why the HIV incidence in San Francisco and other cities in the United States is still stable,9,10 despite increases in sexual risk behavior and sexually transmitted infections (STIs).11-16
Serosorting has been studied mainly in London, where HIV testing rates are around 75%,17 and in San Francisco and Sydney, where HIV testing rates are >90%.18,19 In Amsterdam, HIV testing rates are lower, at 70%.20,21 We studied sexual behaviors among HIV-negative and HIV-positive MSM attending the Amsterdam STI Outpatient Clinic or participating in the Amsterdam Cohort Studies (ACS) to evaluate if, despite the lower HIV testing rates, serosorting is also practiced by MSM in Amsterdam. We regarded sexual behavior as serosorting when unprotected anal intercourse (UAI) was more frequently practiced with partners of the same HIV status than with partners of different or unknown HIV status (ie, nonconcordant HIV status).
Contrary to the stable HIV incidence found in the United States, the Amsterdam STI Outpatient Clinic has seen a rising HIV incidence among MSM older than 35 years of age from 1999 to 2005.22,23 Among MSM participating in the ACS, HIV incidence has remained relatively low and stable since 1991.14,23 We thus hypothesized that serosorting could explain the difference in HIV incidence trend between the STI Outpatient Clinic and ACS. We therefore tested whether MSM participating in the ACS more frequently reported UAI with concordant traceable partners than MSM attending the STI Outpatient Clinic. Additionally, we tested whether MSM younger than 35 years of age more frequently reported UAI with concordant traceable partners than MSM older than 35 years of age to verify whether serosorting explains the rising HIV incidence among MSM older than 35 years of age at the STI Outpatient Clinic.
Study Population and Setting
We included 2 groups of MSM: MSM attending the STI Outpatient Clinic of the Health Service of Amsterdam and MSM participating in the ACS. In total, 513 MSM were included (Fig. 1).
We defined MSM attending the STI Outpatient Clinic as being at “high risk” for HIV. The HIV incidence increased in this population from 1999 to 2005 and was estimated at 3.75 per 100 person-years (PYs) during that period.22,23 The Amsterdam STI Outpatient Clinic offers free-of-charge examination and treatment for STIs, and every client is routinely screened for Chlamydia trachomatis, including lymphogranuloma venereum (LGV); Neisseriagonorrhea; and Treponema pallidum. HIV antibody testing is not routinely performed but is offered to all clients. For this study, we included 232 MSM (defined as reporting at least 1 sexual contact with someone of the same gender in the prior 6 months) who had traceable partners, including self-defined main partners, from January 2004 to June 2006. Indicative of high risk, each was diagnosed with gonorrhea, syphilis, LGV, Chlamydia, or nonspecific proctitis (defined as inflammatory signs on anoscopy and/or >10 white blood cells per high-power field in a Gram-stained anal mucosal smear). MSM were known to be HIV-negative (ie, they tested HIV-negative at least twice, with their most recent HIV-negative test result at or within 6 months before study participation) or HIV-positive (ie, they received an HIV diagnosis before study participation).
We defined MSM participating in the ACS as having lower risk for HIV. Their HIV incidence was stable at 1.24 per 100 PYs from 1999 to 2005,14,23 which was considerably lower than that among MSM attending the STI Outpatient Clinic. The ACS, started in 1984, is an ongoing open cohort that includes mainly HIV-negative but also HIV-positive MSM aged 18 to 65 years with at least 1 male sexual partner in the 6 months before intake.24,25 Every 6 months, ACS participants complete a self-administered questionnaire on personal health, sexual risk behavior, and STIs at the Health Service of Amsterdam; blood is taken for HIV testing. For this study, we included all (n = 281) MSM in the ACS reporting traceable partners from January 2004 to June 2006.
After giving consent, all participants completed our study questionnaire on the background characteristics (eg, nationality, age, HIV status, sexual risk behaviors) of a maximum of 4 traceable partners (ie, partners who can be contacted by telephone, e-mail, and/or home address) in the prior 6 months. If a participant reported >4 partners, only information on the last 4 partners was collected. In addition, information was collected on sexual risk behaviors with anonymous partners over the same period. At the ACS and the STI Outpatient Clinic, highly similar questionnaires were used, only differing with respect to the information collected on anonymous partners; for MSM in the ACS, information was collected for anonymous partners with HIV-negative, HIV-positive, or unknown HIV status separately, whereas for MSM attending the STI Outpatient Clinic, only pooled information on anonymous partners was collected. Additionally, our questionnaire was self-administered by MSM in the ACS but was administrated by a public health nurse in the clinic group.
For lower risk and high-risk MSM separately, we calculated the proportion of HIV-negative and HIV-positive MSM reporting anal intercourse and UAI with partners who were concordant (ie, perceived to have the same HIV status), partners who were discordant (ie, perceived to have different HIV status), and partners of unknown HIV status, including 95% confidence intervals (95% CIs). For the remainder of this article, we refer to partners with discordant or unknown HIV status combined as “nonconcordant” partners. We performed these calculations separately for traceable (lower risk and high-risk MSM) and anonymous partners (only lower risk HIV-negative MSM). To assess whether sexual behaviors (eg, anal intercourse, UAI) differed significantly across the HIV status of partners, we used univariate logistic regression analysis, with partner HIV status as the independent variable and sexual risk behavior as the dependent variable. Because MSM could report 1 to 4 traceable partners, >1 traceable partner was allowed per HIV status category. For example, if men reported 3 traceable partners of concordant HIV status, all 3 partners were included. We therefore corrected estimates and standard errors for intraindividual correlation using generalized estimation equations (GEEs), assuming an exchangeable correlation matrix.
Finally, we studied predictors for having concordant traceable partners among men reporting UAI, using logistic regression. We were mainly interested in risk group and age to evaluate whether differences in UAI with concordant or nonconcordant partners could explain differences in HIV incidence trend between MSM of lower risk and high-risk groups or between MSM older and younger than 35 years of age. In addition, we analyzed the potential association of the following characteristics of the study participants: ethnicity, HIV status, LGV, syphilis or gonorrhea infection, having a main/nonmain traceable partner, meeting a traceable partner through the Internet, having sex with traceable partners of the same ethnicity or age, having anonymous partners, and having UAI with anonymous partners. Because men could report up to 4 traceable partners, we used GEE, assuming an exchangeable correlation matrix, and allowed for the inclusion of multiple concordant or nonconcordant partners. All variables were first analyzed univariately. Thereafter, we constructed a multivariate model that included risk groups and age. First, we assessed interaction effects of variables of interest with risk group or age, considering only those effects with a P value <0.05. Thereafter, we included variables by forward selection, including only variables with a P value <0.05 based on the Wald test.
Of the 232 high-risk MSM, 54% were HIV-negative and 46% were HIV-positive. For the 281 lower risk MSM, these numbers were 93% and 7%, respectively. Lower risk MSM were younger (mean age = 35 years, SD = 9 years) and more often of Dutch background (84%) than high-risk MSM (mean age = 38 years, SD = 9 years, and 74%, respectively). A total of 1286 traceable partners were reported, with 19% being main partners; 29% of the MSM reported only 1 partner, whereas 32% reported 4 partners.
Evidence of Serosorting With Traceable Partners?
Most of the lower risk and high-risk HIV-negative MSM reported partners of unknown HIV status (83% and 58%, respectively). Forty percent of the lower risk and 56% of the high-risk HIV-negative MSM reported partners of concordant HIV status, whereas these numbers were only 11% and 11% for discordant partners, respectively (Table 1). Importantly, lower risk and high-risk HIV-negative MSM reported partners of >1 HIV status category. To test whether HIV-negative MSM more frequently practiced sexual behaviors with concordant partners than with nonconcordant partners, comparisons were made across the HIV status of partners (see Table 1). Lower risk and high-risk HIV-negative MSM more often had concordant anal intercourse and UAI than nonconcordant anal intercourse and UAI. Fifty percent of the lower risk and 72% of the high-risk HIV-negative MSM with concordant partners practiced UAI. For MSM with nonconcordant partners, levels of UAI were significantly lower but still quite high, especially among high-risk MSM; 38% of the high-risk MSM with discordant partners and 39% of the high-risk MSM with partners of unknown HIV status had UAI (see Table 1).
A large proportion of the lower risk and high-risk HIV-positive MSM reported partners of concordant HIV status (43% and 60%, respectively). These numbers were 33% and 27% for discordant partners and 57% and 40%, respectively, for partners with unknown HIV status. Similar to HIV-negative MSM, lower risk and high-risk HIV-positive MSM reported partners of >1 HIV status category. Lower risk and high-risk HIV-positive MSM were more likely to practice concordant than nonconcordant anal intercourse and UAI. The number of lower risk HIV-positive MSM was too small to reach statistical significance, however (see Table 1). Compared with HIV-negative MSM, HIV-positive MSM reported higher levels of concordant UAI; of the lower risk and high-risk HIV-positive MSM with concordant partners, UAI was practiced by 72% and 82%, respectively. Similar to HIV-negative MSM, the level of UAI with nonconcordant partners was lower but still substantial; 22% of the lower risk and 36% of the high-risk MSM with discordant partners and 36% of the lower risk and 40% of the high-risk MSM with partners of unknown HIV status had UAI.
We also assessed whether anal intercourse and UAI varied among partner HIV status for main and nonmain traceable partners separately. Although levels of UAI with concordant partners were higher with main partners than with nonmain traceable partners, patterns of UAI across partner HIV status were consistent with serosorting for both types of partners; UAI was significantly higher for those with concordant partners than for those with nonconcordant partners (data not shown).
Evidence of Serosorting With Anonymous Partners?
Most of the MSM reported anonymous partners: 62% of the lower risk HIV-negative MSM, 67% of the lower risk HIV-positive MSM, 70% of the high-risk HIV-negative MSM, and 76% of the high-risk HIV-positive MSM. The median number of anonymous partners in the preceding 6 months differed substantially across the lower risk and high-risk MSM, being 5 in the lower risk HIV-negative MSM (interquartile range [IQR]: 2 to 11), 10 in the lower risk HIV-positive MSM (IQR: 5 to 20), 10 in the high-risk HIV-negative MSM (IQR: 4 to 25), and 12 in the high-risk HIV-positive MSM (IQR: 5 to 28). Table 2 shows the HIV status and UAI pertaining to anonymous partners tabulated against the HIV status of traceable partners for lower risk and high-risk HIV-negative and HIV-positive MSM separately. Most of the anonymous partners were of unknown HIV status. Additionally, Table 2 shows that levels of UAI with anonymous partners were high, particularly among high-risk MSM, including those having concordant traceable partners. Of the 64 high-risk HIV-positive MSM with concordant traceable partners, 60% reported UAI with anonymous partners. More importantly, for the 70 high-risk HIV-negative MSM having concordant traceable partners, 29% reported UAI with anonymous partners. This behavior poses a risk for HIV transmission, because most anonymous partners were of unknown HIV status. For lower risk HIV-negative and HIV-positive MSM with concordant traceable partners, levels of UAI with anonymous partners were substantially lower, being 10% and 22%, respectively.
To test whether MSM with concordant anonymous partners had risky sex more often than MSM with nonconcordant anonymous partners, comparisons were made across the HIV status of the partners. We performed this analysis only for the lower risk HIV-negative MSM, because numbers were too small for the lower risk HIV-positive MSM and data were unavailable for the high-risk MSM. Table 3 shows that sexual risk behavior did not vary by the HIV status of anonymous partners in lower risk HIV-negative MSM.
Predictors for Having Concordant Traceable Partners Among Men Who Have Sex With Men Reporting Unprotected Anal Intercourse
We were especially interested in risk group and age to test if their association with concordant UAI would explain why HIV incidence is stable in the ACS and increasing among MSM older than 35 years of age in the STI Outpatient Clinic group. For this subanalysis, we included all 235 MSM reporting UAI. These MSM reported UAI with a total of 372 partners, of whom 60% were of concordant HIV status and 40% were of nonconcordant HIV status. Univariately, being older than 35 years of age and being in the high-risk group were borderline significantly associated with having concordant UAI partners (odds ratio [OR] = 1.61, 95% CI: 0.97 to 2.70; OR = 1.55, 95% CI: 0.93 to 2.60, respectively). Other factors univariately significantly associated with having concordant UAI partners were being HIV-positive (P = 0.03), the partner being a main partner (P < 0.001), and having a syphilis infection (P = 0.03). In multivariate analysis, including age and risk group, the UAI partner was more likely to be concordant in HIV-positive MSM than in HIV-negative MSM (OR = 2.1, 95% CI: 1.1 to 4.1), in MSM having UAI with main partners than in MSM with nonmain partners (OR = 3.6, 95% CI: 2.2 to 5.9), or in MSM who met their partner by way of the Internet (OR = 1.6, 95% CI: 1.1 to 2.5). Having sex with someone of the same age category was borderline significantly associated with concordant partners (OR = 1.5, 95% CI: 0.97 to 2.3). In multivariate analysis, there was no significant association between risk group and age and having concordant partners (P = 0.9 and P = 0.4, respectively). There was a significant interaction effect between risk group and having UAI with anonymous partners, however (P = 0.02). In lower risk MSM having UAI with anonymous partners, the UAI traceable partner was less likely to be concordant (OR = 0.3, 95% CI: 0.1 to 0.8). This association was not found for high-risk MSM.
We found that lower risk and high-risk MSM with concordant traceable partners more often engaged in UAI than MSM with nonconcordant traceable partners. This suggests that both groups of MSM practice serosorting. There was no statistical significant difference in UAI between lower risk HIV-negative MSM with concordant and nonconcordant anonymous partners, however, suggesting that among lower risk HIV-negative MSM, serosorting takes place mainly with traceable partners rather than with anonymous partners. Unfortunately, information on serosorting with anonymous partners was unavailable for high-risk MSM. Nevertheless, also among high-risk MSM, serosorting with anonymous partners seems unlikely; few high-risk HIV-negative MSM reported concordant anonymous partners, and although HIV-positive MSM reported a substantial number of HIV-positive anonymous partners, most reported partners of unknown HIV status. Among MSM, neither risk group nor age was associated with having concordant UAI with traceable partners. We therefore conclude that there is no difference in serosorting between lower risk and high-risk MSM or between MSM aged younger and older than 35 years, explaining the stable HIV incidence among MSM in the ACS and the increasing HIV incidence among STI Outpatient Clinic attendees older than 35 years of age.
A more plausible explanation for the difference in HIV incidence trend between MSM in the ACS and MSM attending the STI Outpatient Clinic is the higher levels of risk behavior among the latter. High-risk MSM with nonconcordant traceable partners reported higher levels of UAI and more frequently had UAI with anonymous partners, mostly of unknown HIV status, than MSM in the lower risk group. Additionally, serosorting behaviors seem more consistent among lower risk MSM. In lower risk MSM having UAI with anonymous partners, the UAI traceable partner was more likely to be nonconcordant. In other words, in lower risk MSM not having UAI with anonymous partners, the UAI traceable partner was more likely to be concordant. This interaction effect was not found for high-risk MSM, and a considerable proportion of high-risk MSM with concordant traceable partners reported UAI with anonymous partners. It seems that although sexual behaviors suggest serosorting among MSM, its use is currently inconsistent, especially in the high-risk group. This poses a continuous risk for HIV transmission.
Sexual harm reduction strategies are further hampered by unrecognized HIV infection, difficulties with disclosure, and wrong assumptions of a partner's HIV status.26-28 Much of the assumed concordance is based on reports of partners, which is not a guarantee, especially for HIV-negative men. We found that being HIV-positive, a partner being a main partner, and meeting partners through the Internet were significantly associated with concordance in UAI partners. This shows that serosorting is more likely among HIV-positive MSM and in situations where the communication level is high and the exchange of information on HIV status is possible. The Internet may act as a tool for finding concordant partners easily, because users can filter on-line on serostatus profiles29,30 and disclosure of positive HIV status on-line avoids embarrassment, stigmatizing, or face-to-face rejection.16,29,30 Main partners are more likely to disclose HIV status when having first-time sex than nonmain partners.31
From this and other studies based on a description of sexual behaviors, sexual harm reduction practices can only be inferred.1-4 Are sexual associations found by chance, or are they conscious decisions based on the discussion of HIV status with potential partners? We propose the latter, because additional data analysis in the ACS among HIV-negative MSM showed that most who reported UAI with main partners knew that this partner was HIV-negative before having UAI for the first time (108 [65%] of 166 men). This supports our proposed hypothesis of serosorting with traceable, including main, partners. Of the MSM reporting UAI with casual partners, only 26% (19 of 72) knew that the partner was HIV-negative before initiating UAI. This correlates with the lack of evidence we found for serosorting with anonymous partners. Additionally, we found that being HIV-positive, having UAI with main partners, and meeting partners through the Internet were all independently associated with seroconcordant UAI, which is consistent with findings from other cities, such as London.7
Our study is limited by that fact that the clinic attendees we studied were not representative of the entire population of MSM who attended the STI Outpatient Clinic. Our group is likely to exhibit high-risk behavior and contribute to the rising HIV incidence, however, as described by Dukers and colleagues.23 Additionally, our study could be biased by the different settings and modes of questionnaire administration. Evidence indicates that measurement error associated with sexual questions decreases when methods that increase anonymity are used.32,33 Because face-to-face interviews are more likely to elicit socially desirable answers than self-administered questionnaires, underreporting of UAI in the STI Outpatient Clinic group might have occurred. If this occurred, however, the underreporting of UAI among high-risk MSM seems to be independent of the HIV status of the traceable partners; among all 3 partner groups, equally high levels of UAI with anonymous partners were reported (see Table 2). Additionally, a cohort effect cannot be excluded for the ACS population, explaining stable HIV incidence and lower levels of UAI. We found the level of UAI with steady and casual partners increasing among the MSM followed in the ACS from 1995 to 2002, however.14 Moreover, the ACS is an open-cohort study with newly recruited MSM entering the cohort each year, making a cohort effect less likely.
Initial reports on serosorting come from cities like London, where HIV testing rates among MSM are around 75%,6,7,17 and San Francisco and Sydney, where >90% of the MSM have been tested.1,3,4,18,19 In Amsterdam, the rate of HIV testing, a requisite for serosorting, is relatively low (70%).20,21 Nevertheless, serosorting seems just as plausible in Amsterdam as in cities with higher HIV testing rates, finding equal evidence for serosorting in a population with stable HIV incidence and in a population with increasing HIV incidence. The higher level of UAI with anonymous or nonconcordant traceable partners in the latter group seems to be the main reason for the increasing HIV incidence seen among STI Outpatient Clinic attendees.
The authors thank the public health nurses of the STI Outpatient Clinic, particularly H. Koops, A. Ferwerda, W. Bolderik, K. de Jong, I. Peters, L. Abma, S. v.d. Kolk, C. de Jong, S. van Elst, and A. Hendriks, for their contribution to the data collection. They also express their gratitude to H. van Bijnen, B. Stappershoef, and J. Bax of the ACS and to A. van Eeden and M. de Groot of the Jan van Goyen Clinic for their contribution to the data collection; to R. B. Geskus and U. Davidovich for critically reading the manuscript; and to L. Phillips for editing the final manuscript. In addition, this study could not have been performed without the help of the study participants.
A. K. Van der Bij analyzed and interpreted the data and drafted the paper. M.-E. Kolader contributed substantially to the conception of the study and to the design and acquisition of the data. H. J. C. de Vries contributed substantially to the acquisition of the data. M. Prins contributed substantially to the conception of the study and to the design and interpretation of the data. R. A. Coutinho contributed substantially to the conception of the study and to the design and interpretation of the data. N. H. T. M. Dukers had the initial idea of the study, supervised the study, and contributed to the interpretation of the data. All authors contributed to the final version and read and approved the final manuscript.
1. Crawford JM, Kippax SC, Mao L, et al. Number of risk acts by relationship status and partner serostatus: findings from the HIM cohort of homosexually active men in Sydney, Australia. AIDS Behav
2. Golden MR, Brewer DD, Kurth A, et al. Importance of sex partner HIV status in HIV risk assessment among men who have sex with men. J Acquir Immune Defic Syndr
3. 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
4. Parsons JT, Schrimshaw EW, Wolitski RJ, et al. Sexual harm reduction practices of HIV-seropositive gay and bisexual men: serosorting, strategic positioning, and withdrawal before ejaculation. AIDS
. 2005;19(Suppl 1):S13-S25.
5. 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
6. Bolding G, Davis M, Hart G, et al. Gay men who look for sex on the Internet: is there more HIV/STI risk with online partners? AIDS
7. Elford J, Bolding G, Sherr L, et al. No evidence of an increase in serosorting among HIV negative gay men in London, 1998-2005. AIDS
8. 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
9. San Francisco serosorting may explain odd HIV data. STDs have risen, but not new HIV infections. AIDS Alert
10. Buchacz K, Greenberg A, Onorato I, et al. Syphilis epidemics and human immunodeficiency virus (HIV) incidence among men who have sex with men in the United States: implications for HIV prevention. Sex Transm Dis
. 2005;32(Suppl 10):S73-S79.
11. Macdonald N, Dougan S, McGarrigle CA, et al. Recent trends in diagnoses of HIV and other sexually transmitted infections in England and Wales among men who have sex with men. Sex Transm Infect
12. van de Laar MJW, de Boer IM, Koedijk FDH, et al. RIVM Report 441100022: HIV and Sexually Transmitted Infections in The Netherlands in 2004. An Update: November 2005
. Bilthoven, The Netherlands: RIVM; 2005.
13. Rietmeijer CA, Patnaik JL, Judson FN, et al. Increases in gonorrhea and sexual risk behaviors among men who have sex with men: a 12-year trend analysis at the Denver Metro Health Clinic. Sex Transm Dis
14. van der Bij AK, Stolte IG, Coutinho RA, et al. Increase of sexually transmitted infections, but not HIV, among young homosexual men in Amsterdam: are STIs still reliable markers for HIV transmission? Sex Transm Infect
15. Van de Ven P, Rawstorne P, Crawford J, et al. Increasing proportions of Australian gay and homosexually active men engage in unprotected anal intercourse with regular and with casual partners. AIDS Care
16. Elford J. Changing patterns of sexual behaviour in the era of highly active antiretroviral therapy. Curr Opin Infect Dis
17. Williamson LM, Dodds JP, Mercey DE, et al. Increases in HIV-related sexual risk behavior among community samples of gay men in London and Glasgow: how do they compare? J Acquir Immune Defic Syndr
18. Jin FY, Prestage G, Law MG, et al. Predictors of recent HIV testing in homosexual men in Australia. HIV Med
19. Chen SY, Gibson S, Weide D, et al. Unprotected anal intercourse between potentially HIV-serodiscordant men who have sex with men, San Francisco. J Acquir Immune Defic Syndr
20. Stolte IG, de Wit JBF, Kolader M, et al. Low HIV testing rates among younger high-risk men who have sex with men in Amsterdam. Sex Transm Infect
. February 21, 2007 [Epub ahead of print].
21. Hospers HJ, Dörfler TT, Zuilhof W. Monitoronderzoek Amsterdam 2004
. Amsterdam, The Netherlands: Schorer; 2004. Available at: http://www.monitoronderzoekonline.nl
. Accessed September 1, 2006.
22. Dukers NH, Spaargaren J, Geskus RB, et al. HIV incidence on the increase among homosexual men attending an Amsterdam sexually transmitted disease clinic: using a novel approach for detecting recent infections. AIDS
23. Dukers NHTM, Fennema HS, Snoek EM, et al. HIV incidence and HIV testing behavior in men who have sex with men: using three incidence sources, The Netherlands, 1984-2005. AIDS
24. van Griensven GJ, Tielman RA, Goudsmit J, et al. Risk factors and prevalence of HIV antibodies in homosexual men in the Netherlands. Am J Epidemiol
25. Stolte IG, Dukers NH, Geskus RB, et al. Homosexual men change to risky sex when perceiving less threat of HIV/AIDS since availability of highly active antiretroviral therapy: a longitudinal study. AIDS
26. Parsons JT, Severino J, Nanin J, et al. Positive, negative, unknown: assumptions of HIV status among HIV-positive men who have sex with men. AIDS Educ Prev
27. Ciccarone DH, Kanouse DE, Collins RL, et al. Sex without disclosure of positive HIV serostatus in a US probability sample of persons receiving medical care for HIV infection. Am J Public Health
28. 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
29. Davis M, Hart G, Bolding G, et al. Sex and the Internet: gay men, risk reduction and serostatus. Cult Health Sex
30. Davis M, Hart G, Bolding G, et al. E-dating, identity and HIV prevention: theorising sexualities, risk and network society. Sociol Health Illn
31. Hart TA, Wolitski RJ, Purcell DW, et al. Partner awareness of the serostatus of HIV-seropositive men who have sex with men: impact on unprotected sexual behavior. AIDS Behav
32. Catania JA, Gibson DR, Chitwood DD, et al. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull
33. Weinhardt LS, Forsyth AD, Carey MP, et al. Reliability and validity of self-report measures of HIV-related sexual behavior: progress since 1990 and recommendations for research and practice. Arch Sex Behav