HIV infection and other sexually transmitted infections (STIs) continue to disproportionately affect men who have sex with men (MSM) in the United States [1,2]. A study  conducted in five major US cities between June 2004 and April 2005 found that 25% of MSM were HIV seropositive, and that 48% of those men were unaware of their infection. In 2007, 72% of estimated HIV/AIDS cases from 38 US areas were attributed to male-to-male sexual contact, and there was a 26% increase in the estimated HIV/AIDS cases among MSM from 2004 to 2007 . Similar increasing trends in primary and secondary syphilis among MSM have been observed since 2000 [5,6]. The increase in the rates of HIV and other STIs among MSM has coincided with wider availability of HAART [1,7].
As the number of MSM diagnosed with HIV has grown in recent years, it is increasingly important to understand the sexual practices of MSM living with HIV. Many HIV-diagnosed MSM believe that they have a responsibility to protect their sex partners from infection [8,9], and many eliminate or reduce HIV transmission behaviors after their HIV diagnosis [10,11]. However, some engage in unprotected sexual behavior, thus placing others at risk for HIV infection and themselves at risk for other STIs [12–14], and possibly re-infection with other strains of HIV [15,16].
To reduce the possibility of transmitting HIV to others, HIV-diagnosed MSM may practice ‘harm reduction’ approaches. Some engage in unprotected sex only with seropositive partners (i.e., serosorting) . Others are less likely to engage in unprotected sex when they perceive that their partners are HIV negative than when they are uncertain about the serostatus of their partners [18,19]. Many prior studies have combined these two types of partners and called the group ‘at-risk partners.’ This approach clouds the understanding of the type of partner (e.g., a known HIV-seronegative partner vs. a serostatus unknown partner) most likely to be placed at risk for infection. Additionally, some HIV-diagnosed MSM selectively engage in unprotected receptive rather than unprotected insertive anal intercourse (i.e., strategic positioning)  because the per-act risk of transmitting HIV is lower when the infected person engages in the receptive role [21,22]. HIV-diagnosed MSM may feel protected from transmitting HIV sexually when they receive HAART or have an undetectable viral load, as evidence suggests that a low viral load may reduce the probability of infecting a sex partner .
Several studies have examined the prevalence of unprotected anal intercourse (UAI) among HIV-diagnosed MSM. An integrated understanding of the empirical findings will help guide prevention efforts. In this meta-analysis, we expanded the scope of previous qualitative reviews [12–14,24] by quantitatively synthesizing the US-based literature. Specifically, we examined the prevalence of UAI in relation to the perceived serostatus of partners, type of UAI (insertive vs. receptive), and HIV clinical variables [on antiretroviral therapy (ART), undetectable viral load, and medication adherence]. We also examined the UAI prevalence stratified by participant characteristics (ethnicity/race) and study characteristics (recruitment time and setting, sampling, and data collection method).
As part of the Centers for Disease Control and Prevention's (CDC) HIV/AIDS Prevention Research Synthesis (PRS) project , we developed multiple search strategies to identify relevant reports published or in press between 2000 and 2007 . A subject expert librarian conducted systematic searches of three electronic databases (MEDLINE, EMBASE, and PsycINFO) by cross-referencing multiple search terms (i.e., keywords and each database's index terms) in two domains: sexual risk behavior descriptors (risk reduction behavior, risk-taking, sexual partners, safe sex, unsafe sex, condoms, abstinence, concordant, discordant, unprotected sex, sex behavior, risk behavior, or intercourse) and HIV-positive status descriptors (HIV seropositivity, HIV infections, HIV status, or various terms for individuals with HIV or AIDS). To fill the indexing gaps, we checked reference lists of pertinent reports to identify relevant citations.
Studies were included if they met the following criteria: conducted in the United States; specifically targeted or provided stratified data for HIV-seropositive MSM; examined sex risk behaviors among MSM who were aware of their seropositive status during the behavioral recall window; reported at least one behavioral outcome with male partners: UAI, insertive UAI, or receptive UAI with any partners, HIV-seropositive partners, HIV-seronegative partners, or HIV serostatus unknown partners; and conducted after HAART became available (late 1995 and early 1996).
To avoid overestimating the UAI prevalence, we excluded studies that recruited MSM exclusively with behaviors that may inflate UAI prevalence (e.g., sex work, methamphetamine use, and clinically diagnosed alcohol dependence) or that recruited MSM entirely from high-risk settings devoted to seeking sex partners (e.g., gay bathhouses, sex parties, and barebacking websites). Studies that only enrolled MSM who had recently engaged in UAI were also excluded. Conference abstracts or dissertations were screened but not included, as they provided too few details of the study findings.
Information from eligible articles was independently extracted by pairs of trained reviewers. Using a standardized coding form, each study was coded for the following variables: authors, year of publication, date of enrollment, sampling method, recruitment settings, sample size, race/ethnicity composition of study sample, sexual orientation (homosexual and bisexual), time since HIV diagnosis, medical status (CD4 cell count, viral load, in HIV primary care, receiving ART, and medication adherence), perceived serostatus of sex partner, type of UAI with male partners (insertive and receptive), behavioral recall window, and data collection methods [computer-assisted self-interviewing (CASI), audio CASI (ACASI), self-administered questionnaires, and interviewer-administered questionnaires]. There was 92% agreement between reviewers across all variables coded. Discrepancies were reconciled by a third independent reviewer.
To obtain the most accurate information about study findings and optimize the comparison of behavioral outcomes across studies, we contacted authors of all eligible studies and requested data on standardized outcome measures and supplemental information not reported in the published articles. Eighty-seven percent of the authors we contacted provided additional information to assist in our coding and analyses.
The following rules guided the data abstraction:
- If multiple publications reported data from the same study, findings from the most recent or most comprehensive report were used to avoid overlap.
- To examine the extent to which the UAI prevalence varied by serostatus of partners and type of UAI, we calculated multiple estimates of UAI if a study provided more than one relevant behavioral outcome. For example, if a study reported the prevalence of UAI with any male partners, UAI with HIV-seropositive partners, and UAI with HIV-seronegative partners, we calculated three estimates for this study and examined these data separately through stratified analyses (described at the end of Methods). The data on the prevalence of UAI with at-risk partners (i.e., HIV-negative or serostatus known partners) were directly obtained from authors or from the published report(s).
- We used data provided directly from the authors of the eligible studies in the analyses when available. If authors did not provide data, we used the information published in the original reports.
Standard meta-analytic approaches were used in calculating prevalence estimates for individual studies and for aggregating estimates across studies . For each relevant behavioral outcome, we calculated the prevalence (i.e., dividing the number of HIV-diagnosed MSM who engaged in that behavior by the total number of HIV-diagnosed MSM in the study) and its corresponding weight (i.e., inverse variance). In estimating the overall prevalence across studies, we multiplied each prevalence estimate by its weight, summed the weighted prevalence estimates across studies, and then divided by the sum of the weights. To meet the normal distribution assumption, we first converted the individual prevalence estimates to logit because there was substantial variation in findings across studies and the logit distribution is approximately normal . All analyses were conducted on the logit, and the final results were then converted back into prevalence for interpretation.
We also tested the heterogeneity of the findings across studies by using the Q statistic. The aggregated findings reported in the Results section are based on a random-effects model, which provides a more conservative estimate of variance and generates more accurate inferences about a population of studies beyond those included in this review .
Stratified analyses were conducted to examine whether the prevalence of UAI with any male partner differed by study recruitment period (prior to year 2000 vs. after), sampling method (random/systematic vs. convenience), recruitment setting (medical settings vs. gay venues), sample characteristics (majority of white vs. MSM of color; being sexually active as eligibility criterion vs. not), and data collection method [methods that increase confidentiality (CASI/ACASI, self-administered) vs. other methods (interviewer-administered)]. We tested the difference in the UAI prevalence by subgroups using the mixed-effects model in which between-study differences were systematically tested using QB tests with the estimated random effects variance component added to the standard error variance associated with each prevalence estimate .
Additionally, we conducted a univariate meta-regression analysis to examine the association between the UAI prevalence with any male partner and each of the three continuous variables: percentage of the sample that was receiving ART, percentage reporting an undetectable viral load, and percentage adhering to at least 90% of HIV medication. Variables significantly (P < 0.05) associated with the prevalence of UAI in the stratified analyses described above or the univariate meta-regression analyses were then entered together in a multivariate meta-regression model with the maximum likelihood method.
In the sensitivity analyses, we compared the overall prevalence estimate with estimates obtained after iterations using k − 1 findings (k = number of independent samples). We removed a finding and calculated the overall prevalence estimate. Then, we replaced that finding, removed another, and repeated the process.
We assessed publication bias by visual inspection of funnel plots. There was no evidence of publication bias. All analyses were conducted using the Comprehensive Meta-Analysis software version 2 (Biostat, Englewood, New Jersey, USA)  and STATA version 9 (StataCorp LP, College Station, Texas, USA) .
Description of studies in the meta-analysis
Thirty English-language studies [31–60], with a total of 18 121 HIV-diagnosed MSM, met the inclusion criteria (Fig. 1). Table 1 provides a descriptive summary of each study. Approximately, half of the studies collected data from multiple US cities, and the most commonly surveyed cities were San Francisco, Los Angeles, New York City, and Seattle. Participants were mainly recruited from medical settings (primarily HIV clinics) and multiple gay venues and events. Twenty-one studies used convenience samples, and nine studies used random sampling methods (e.g., list all patients in a clinic and select participants based on a random number table) or recruitment methods that were more systematic (e.g., every third person who came into the clinic) than convenience sampling. Sixteen studies used self-administered questionnaires with or without CASI or ACASI, and 14 studies used interviewers to administer the questionnaires.
The median age of study participants was 38 years. In 15 studies, MSM of color (primarily black and Latino) comprised the majority of the sample. Many studies did not collect or report information on medical variables. Thirteen studies reported the length of time living with HIV; among these, the median was 70 months. Only one study specifically focused on newly HIV-diagnosed MSM. Fifteen studies reported that 23–89% of their study participants were taking ART (median = 77%). Seven of those 15 studies reported the percentage of HIV-seropositive MSM who adhered to at least 90% of their medication with a range from 26 to 76% (median = 67%). Of 10 studies that reported data on viral load, the percentage of participants with undetectable virus ranged from 39 to 57% (median = 42%). Fourteen studies did not require participants to be sexually active at the enrollment. The behavioral recall window ranged from last sex to the past 12 months with a median of 3 months.
Estimated prevalence of unprotected anal intercourse with male partners
The aggregated findings from 30 studies indicate that the prevalence of having engaged in UAI with any male partner was 43% [95% confidence interval (CI) 37–48, Table 2]. The UAI prevalence did not differ significantly by the length of behavioral recall window (i.e., <3, 3, 6, and 12 months, QB = 3.24, DF = 3, P = 0.36). Figure 2 shows the wide heterogeneity of results across studies (Q29 = 1314, P < 0.0001). However, sensitivity tests did not reveal any individual study that exerted influence on the overall prevalence estimate.
As seen in Table 2, the UAI prevalence was considerably higher with HIV-seropositive partners (30%) than with HIV serostatus unknown partners (16%) or HIV-seronegative partners (13%). The prevalence of UAI with either a serostatus unknown or HIV-seronegative partner was 26%. Considering the type of UAI, the prevalence of receptive UAI exceeded the prevalence of insertive UAI when the partners were HIV seronegative (9 vs. 5%) or of unknown serostatus (12 vs. 8%), but the difference between the receptive and insertive UAI estimates was smaller when partners were HIV seropositive (22 vs. 21%). The lowest prevalence of UAI was observed for insertive UAI with HIV-seronegative partners (5%).
Table 3 summarizes the findings of stratified analyses and meta-regression of factors potentially associated with the prevalence of UAI with any male partner. The UAI prevalence was not associated with: whether studies required participants to be sexually active prior to enrollment; the percentage of study participants who were receiving ART; reporting more than 90% medication adherence; or reporting an undetectable viral load. Five significant covariates were identified in unadjusted analyses. The UAI prevalence was significantly lower in studies that recruited participants prior to 2000 (vs. after), had MSM of color (vs. white) as the majority of the sample, recruited participants from medical settings (vs. multiple gay venues or events), used random or systematic sampling methods (vs. convenience sampling), and had interviewers administering the questionnaire (vs. self-administered questionnaires with or without CASI/ACASI). These five significant covariates were entered into a multivariate meta-regression analysis, and two covariates remained statistically significant: type of sampling method and how the questionnaire was administered.
This meta-analysis provides the first quantitatively synthesized estimates of the prevalence of UAI with male sex partners among HIV-diagnosed MSM during the HAART era. The overall and stratified estimates are useful for targeted prevention efforts, for epidemiologic modeling of HIV transmission among MSM, and for providing parameters for resource allocation models.
We found that a sizeable percentage of HIV-diagnosed MSM had engaged in UAI with male partners. This is an important public health concern, given the high prevalence of HIV and STIs among MSM in the United States. There is also evidence, however, that some HIV-diagnosed MSM engage in behaviors that may lessen the probability of infecting others. First, UAI varied according to the perceived HIV status of male partners. The UAI prevalence was higher with HIV-seropositive partners than HIV-seronegative or serostatus unknown partners. Although we do not know the percentage of HIV-diagnosed MSM who engaged in UAI exclusively with HIV-seropositive partners, this serosorting pattern may be protective for new HIV transmissions. Conversely, it still leaves seropositive persons open to other STIs or possibly acquiring a drug-resistant strain of HIV [16,61]. Second, we also observed a pattern of strategic positioning [19,20] during UAI with serodiscordant partners. The risk of transmitting HIV is highest when HIV-seropositive MSM engage in insertive UAI with HIV-negative partners [21,22]. Our meta-analysis revealed that HIV-diagnosed MSM reported a higher prevalence of receptive UAI than insertive UAI with HIV-seronegative partners. This pattern of strategic positioning was also found when HIV-seropositive MSM were uncertain about the serostatus of their partners but not when partners were perceived to be HIV positive. The differences in the patterns of sexual behaviors with different serostatus partners support the contention of a previous study  that strategic positioning is an intentional and deliberate HIV-related harm-reduction practice rather than merely a reflection of sexual position preference. However, more research is needed on the extent to which MSM intentionally use strategic positioning and serosorting as harm-reduction strategies and the relative safety of those practices.
Consistent with an earlier meta-analysis , we did not find medical factors, such as receiving ART, having an undetectable viral load, and reporting more than 90% medication adherence, to be associated with the prevalence of having engaged in UAI. Our findings on these medical factors were based on a small subset of studies that provided relevant information. The proportion of HIV-diagnosed MSM with an undetectable viral load was relatively low (median of 42%) with a narrow range across studies (39–57%), which might have lessened the opportunity to detect an association with UAI prevalence.
Our meta-analytic findings must be viewed within the context of the methodological limitations of the primary studies. Our findings are based on cross-sectional data, which provide snap shots of HIV-diagnosed MSM's behaviors at the time of data collection. Changes in transmission behaviors over time cannot be investigated using these data. However, some evidence from published cross-sectional data indicates that the prevalence of UAI with at-risk partners among the HIV-diagnosed MSM did not differ significantly by the length of time they had known they were HIV positive [11,62]; the prevalence of unprotected sex with at-risk partners (with a 3-month behavioral recall window) appeared to remain at a relatively stable level (20–30%) over 6–10 years after HIV diagnosis. Additional longitudinal data are needed to better describe behavioral trends in the years following an HIV diagnosis.
There is limited information from the original studies regarding how participants determined the serostatus of their male sex partner (based on either actual information received from a partner or a guess) and whether the participants had disclosed their own HIV-seropositive status to their sex partners. This has important implications for understanding whether serosorting and strategic positioning are effective harm-reduction strategies because both approaches rely on accurate assessment of partner's serostatus and mutual disclosure. Additionally, the number of unprotected sex acts and the number of sex partners were not included as outcomes in our meta-analysis because few studies provided the information. Future studies should collect and report the information to facilitate more precise estimates of HIV transmission.
The UAI prevalence was significantly lower in studies that used random or systematic sampling methods compared with studies that used convenience samples. Convenience samples might have selected more high-risk MSM; however, we did not find a significant difference in the UAI prevalence between the studies that required participants to be sexually active prior to assessment and the studies that did not have this requirement. A close examination suggests that the sampling method is highly correlated with the recruitment setting; seven of nine studies with random or systematic sampling methods were conducted in medical settings, which tended to have a lower prevalence of UAI. Further research is needed to untangle the complex relationship between level of risk, sampling methods, and recruitment settings.
Sexual behaviors were obtained with self-reports and thus open to recall and social desirability biases. The fact that the UAI prevalence was significantly higher in studies with self-administered questionnaires than in studies with interviewer-administered questionnaires suggests that HIV-seropositive MSM may have underreported socially undesirable behaviors when the questionnaire was administered by an interviewer. Thus, if any bias exists in the aggregated findings, it would be an underestimation of UAI prevalence.
Since the release of the Serostatus Approach to Fighting the HIV Epidemic in 2001, there has been an increased nationwide effort in promoting HIV testing, so that infected persons can become aware of their status [63,64]. With increases in the proportion of infected persons who become aware of their seropositive status, there will be a corresponding need for increases in prevention efforts targeting HIV-diagnosed persons. Evidence suggests that integrating prevention into settings in which HIV-diagnosed people receive medical care or other services and addressing an array of health, behavior, and well being issues are most likely to achieve success . Clinicians can reduce patients' risk for transmitting HIV to others by briefly discussing sexual behaviors, communicating prevention messages (protecting others and protecting one's health), identifying and treating STIs during clinic visits, and referring patients to more intensive risk reduction counseling and psychosocial services as needed [39,66,67]. The frequency with which HIV patients receive safer sex counseling from HIV medical providers has been significantly associated with decreased prevalence of UAI , but too many providers do not give this counseling [68,69]. In 2003, CDC, the Health Resources and Services Administration, National Institutes of Health, and the HIV Medicine Association of the Infectious Diseases Society of America developed evidence-based recommendations on incorporating HIV prevention into the medical care of persons living with HIV . As the recommendations become more widely adopted by medical providers nationwide, we may expect a larger reduction in HIV sexual behaviors among MSM who are in care.
Furthermore, prevention with positives programs in community settings that serve MSM (e.g., community-based organizations, pride events, and gay venues) continue to be needed for MSM in the United States, given a higher UAI prevalence in studies that recruited participants from community settings. A priority of these programs should be to address the safety of serosorting and strategic positioning. A concerted effort to establish and maintain these programs in medical as well as community settings is needed to reduce HIV transmission among MSM in the United States.
This work was supported by the Prevention Research Branch, Division of HIV/AIDS Prevention, U.S. CDC and was not funded by any other organization.
We sincerely thank the following authors for providing additional data to assist our coding and analyses: Angela Aidala; David S. Bimbi; Cherilyn R. Bingman; Michael Campsmith; Sanny Y. Chen; Shonda M. Craft; Paul Denning; Ralph DiClemente; Helen Ding; Theresa Exner; Ellen Funkhouser; Lytt Gardner; Christian Grov; Perry N. Halkitis; Ben Hadsock; David Holtgrave; Charlotte Kent; Andrea Y. Kim; Robert Klitzman; Gordon Mansergh; Gary Marks; Willi McFarland; Stephen Morin; Joanne Mullen; Dennis H. Osmond; David E. Ostrow; Jeffrey T. Parsons; Lance M. Pollack; Paul J. Poppen; Michael Reece; Jean L. Richardson; Starley Shade; Peter Theodore; Peter Vanable; Lance S. Weinhardt; William L. H. Whittington; Richard Wolitski. No compensation was received for any contributions made by these individuals.
N.C. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. N.C., G.M., and R.J.W. conceptualized and designed the study. N.C., A.L., M.M.M., L.W.A., E.D.J., R.J.W., and G.M. did acquisition of data. N.C. and G.M. analyzed and interpreted the data and drafted the manuscript. N.C., G.M., R.J.W., A.L., M.M.M., L.W.A., K.J.M., and E.D.J. did critical revision of the manuscript for important intellectual content. N.C. provided statistical expertise. N.C., G.M., R.J.W., A.L., M.M.M., L.W.A., K.J.M., and E.D.J. provided administrative, technical, or material support.
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US CDC.
There are no conflicts of interest.
1. Jaffe HW, Valdiserri RO, De Cock KM. The reemerging HIV/AIDS epidemic in men who have sex with men. JAMA 2007; 298:2412–2414.
2. Wolitski RJ, Stall R, Valdiserri RO, editors. Unequal Opportunity: health disparities affecting gay and bisexual men in the United States
. New York: Oxford University Press; 2008.
3. CDC. HIV prevalence, unrecognized infection, and HIV testing among men who have sex with men: five U.S. cities, June 2004-April 2005.MMWR Morb Mortal Wkly Rep
4. Centers for Disease Control and Prevention. Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention; 2009. HIV/AIDS Surveill Rep 2007; 19. http://www.cdc.gov/hiv/topics/surveillance/resources/reports/
. [Accessed 30 March 2009].
5. CDC. Primary and secondary syphilis: United States, 2002.MMWR Morb Mortal Wkly Rep
6. Heffelfinger JD, Swint EB, Berman SM, Weinstock HS. Trends in primary and secondary syphilis among men who have sex with men in the United States. Am J Public Health 2007; 97:1076–1083.
7. Crepaz N, Hart TA, Marks G. Highly active antiretroviral therapy and sexual risk behavior: a meta-analytic review. JAMA 2004; 292:224–236.
8. Wolitski RJ, Bailey CJ, O'Leary A, Gomez CA, Parsons JT. Self-perceived responsibility of HIV-seropositive men who have sex with men for preventing HIV transmission. AIDS Behav 2003; 7:363–372.
9. Wolitski RJ, Flores SA, O'Leary A, Bimbi DS, Gomez CA. Beliefs about personal and partner responsibility among HIV-seropositive men who have sex with men: measurement and association with transmission risk behavior. AIDS Behav 2007; 11:676–686.
10. Marks G, Crepaz N, Senterfitt JW, Janssen RS. 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.
11. Weinhardt LS. HIV diagnosis and risk behavior. In: Kalichman SC, editor. Positive prevention: reducing HIV transmission among people living with HIV/AIDS. New York, New York: Kluwer Academic/Plenum Publishers; 2005. pp. 29–63.
12. Crepaz N, Marks G. Towards an understanding of sexual risk behavior in people living with HIV: a review of social, psychological, and medical findings. AIDS 2002; 16:135–149.
13. Kalichman SC. HIV transmission risk behaviors of men and women living with HIV-AIDS: prevalence, predictors, and emerging clinical interventions. Clin Psychol 2000; 7:32–47.
14. Van Kesteren NM, Hospers HJ, Kok G. Sexual risk behavior among HIV-positive men who have sex with men: a literature review. Patient Educ Couns 2006; 65:5–20.
15. Blackard JT, Cohen DE, Mayer KH. Human immunodeficiency virus superinfection and recombination: current state of knowledge and potential clinical consequences. Clin Infect Dis 2002; 34:1108–1114.
16. Smith DM, Richman DD, Little SJ. HIV superinfection. J Infect Dis 2005; 192:438–444.
17. Kippax S, Noble J, Prestage G, Crawford JM, Campbell D, Baxter D, et al
. Sexual negotiation in the AIDS era: negotiated safety revisited. AIDS 1997; 11:191–197.
18. Jin F, Prestage GP, Ellard J, Kippax SC, Kaldor JM, Grulich AE. How homosexual men believe they became infected with HIV: the role of risk-reduction behaviors. J Acquir Immune Defic Syndr 2007; 46:245–247.
19. Parsons JT, Schrimshaw EW, Wolitski RJ, Halkitis PN, Purcell DW, Hoff CC, 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.
20. Van de Ven P, Kippax S, Crawford J, Rawstorne P, Prestage G, Grulich A, 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.
21. Varghese B, Maher JE, Peterman TA, Branson BM, Steketee RW. Reducing the risk of sexual HIV transmission: quantifying the per-act risk for HIV on the basis of choice of partner, sex act, and condom use. Sex Transm Dis 2002; 29:38–43.
22. Vittinghoff E, Douglas J, Judson F, McKirnan D, MacQueen K, Buchbinder SP. Per-contact risk of human immunodeficiency virus transmission between male sexual partners. Am J Epidemiol 1999; 150:306–311.
23. Quinn TC, Wawer MJ, Sewankambo N, Serwadda D, Li C, Wabwire-Mangen F, et al
. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group. N Engl J Med 2000; 342:921–929.
24. Elford J. Changing patterns of sexual behaviour in the era of highly active antiretroviral therapy. Curr Opin Infect Dis 2006; 19:26–32.
25. Lyles CM, Crepaz N, Herbst JH, Kay L, for the HIV/AIDS Research Synthesis (PRS) Project. Evidence-based HIV behavioral prevention from the perspective of CDC's HIV/AIDS Prevention Research Synthesis Team. AIDS Educ Prev 2006; 18(4 Suppl A):21–31.
26. DeLuca JB, Mullins MM, Lyles CM, Crepaz N, Kay L, Thadiparthi S. Developing a comprehensive search strategy for evidence-based systematic review. Evid Based Libr Inf Pract 2008; 3:3–32.
27. Lipsey MW, Wilson DB. Practical meta-analysis. Thousand Oaks, California: Sage Publications; 2001.
28. Hedges L, Vevea JL. Fixed and random effects models in meta-analysis. Psychol Meth 1998; 3:486–504.
29. Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive meta analysis. Englewood, New Jersey: Biostat; 2005.
30. StataCorp. STATA Statistical Software
. College Station, Texas: StataCorp LP; 2005.
31. Aidala AA, Lee G, Howard JM, Caban M, Abramson D, Messeri P. HIV-positive men sexually active with women: sexual behaviors and sexual risks. J Urban Health 2006; 83:637–655.
32. Bingman CR, Marks G, Crepaz N. Attributions about one's HIV infection and unsafe sex in seropositive men who have sex with men. AIDS Behav 2001; 5:283–289.
33. Campsmith ML, Nakashima AK, Jones JL. Association between crack cocaine use and high-risk sexual behaviors after HIV diagnosis. J Acquir Immune Defic Syndr 2000; 25:192–198.
34. CDC. High-risk sexual behavior by HIV-positive men who have sex with men: 16 sites, United States, 2000–2002.MMWR Morb Mortal Wkly Rep
35. Chen SY, Gibson S, Weide D, McFarland W. Unprotected anal intercourse between potentially HIV-serodiscordant men who have sex with men, San Francisco. J Acquir Immune Defic Syndr 2003; 33:166–170.
36. Craft SM, Smith SA, Serovich JM, Bautista DT. Need fulfillment in the sexual relationships of HIV-infected men who have sex with men. AIDS Educ Prev 2005; 17:217–226.
37. Denning PH, Campsmith ML. Unprotected anal intercourse among HIV-positive men who have a steady male sex partner with negative or unknown HIV serostatus. Am J Public Health 2005; 95:152–158.
38. DiClemente RJ, Funkhouser E, Wingood G, Fawal H, Holmberg SD, Vermund SH. Protease inhibitor combination therapy and decreased condom use among gay men. South Med J 2002; 95:421–425.
39. Gardner L, Marks G, O'Daniels CM, Wilson T, Golin C, Wright J, et al
. Implementation and evaluation of a clinic-based behavioral intervention: positive Steps for HIV patients. AIDS Patient Care STDS 2008; 22:627–635.
40. Golden MR, Brewer DD, Kurth A, Holmes KK, Handsfield HH. Importance of sex partner HIV status in HIV risk assessment among men who have sex with men. J Acquir Immune Defic Syndr 2004; 36:734–742.
41. Gorbach PM, Drumright LN, Daar ES, Little SJ. Transmission behaviors of recently HIV-infected men who have sex with men. J Acquir Immune Defic Syndr 2006; 42:80–85.
42. Grov C, DeBusk JA, Bimbi DS, Golub SA, Nanin JE, Parsons JT. Barebacking, the Internet, and harm reduction: an intercept survey with gay and bisexual men in Los Angeles and New York City. AIDS Behav 2007; 11:527–536.
43. Halkitis PN, Parsons JT, Wilton L. Barebacking among gay and bisexual men in New York City: explanations for the emergence of intentional unsafe behavior. Arch Sex Behav 2003; 32:351–357.
44. Holtgrave DR, Crosby R, Shouse RL. Correlates of unprotected anal sex with casual partners: a study of gay men living in the southern United States. AIDS Behav 2006; 10:575–578.
45. Kim AA, Kent CK, Klausner JD. Risk factors for rectal gonococcal infection amidst resurgence in HIV transmission. Sex Transm Dis 2003; 30:813–817.
46. Mansergh G, Marks G, Colfax GN, Guzman R, Rader M, Buchbinder S. ‘Barebacking’ in a diverse sample of men who have sex with men. AIDS 2002; 16:653–659.
47. Morin SF, Myers JJ, Shade SB, Koester K, Maiorana A, Rose CD. Predicting HIV transmission risk among HIV-infected patients seen in clinical settings. AIDS Behav 2007; 11(Suppl 5):S6–S16.
48. Osmond DH, Pollack LM, Paul JP, Catania JA. Changes in prevalence of HIV infection and sexual risk behavior in men who have sex with men in San Francisco: 1997–2002. Am J Public Health 2007; 97:1677–1683.
49. Ostrow DE, Fox KJ, Chmiel JS, Silvestre A, Visscher BR, Vanable PA, et al
. Attitudes towards highly active antiretroviral therapy are associated with sexual risk taking among HIV-infected and uninfected homosexual men. AIDS 2002; 16:775–780.
50. Parsons JT, Halkitis PN, Wolitski RJ, Gomez CA. Correlates of sexual risk behaviors among HIV-positive men who have sex with men. AIDS Educ Prev 2003; 15:383–400.
51. Parsons JT, Bimbi DS. Intentional unprotected anal intercourse among men who have sex with men: barebacking – from behavior to identity. AIDS Behav 2007; 11:277–287.
52. Poppen PJ, Reisen CA, Zea MC, Bianchi FT, Echeverry JJ. Serostatus disclosure, seroconcordance, partner relationship, and unprotected anal intercourse among HIV-positive Latino men who have sex with men. AIDS Educ Prev 2005; 17:227–237.
53. Reece M. Sexual compulsivity and HIV serostatus disclosure among men who have sex with men. Sex Addict Compulsivity 2003; 10:1–11.
54. Richardson JL, Milam J, Stoyanoff S, Kemper C, Bolan R, Larsen RA, et al
. Using patient risk indicators to plan prevention strategies in the clinical care setting. J Acquir Immune Defic Syndr 2004; 37(Suppl 2):S88–S94.
55. Theodore PS, Duran RE, Antoni MH, Fernandez MI. Intimacy and sexual behavior among HIV-positive men-who-have-sex-with-men in primary relationships. AIDS Behav 2004; 8:321–331.
56. Vanable PA, Ostrow DG, McKirnan DJ. Viral load and HIV treatment attitudes as correlates of sexual risk behavior among HIV-positive gay men. J Psychosom Res 2003; 54:263–269.
57. Weinhardt LS, Kelly JA, Brondino MJ, Rotheram-Borus MJ, Kirshenbaum SB, Chesney MA, et al
. HIV transmission risk behavior among men and women living with HIV in 4 cities in the United States. J Acquir Immune Defic Syndr 2004; 36:1057–1066.
58. Whittington WL, Collis T, Dithmer-Schreck D, Handsfield HH, Shalit P, Wood RW, et al
. Sexually transmitted diseases and human immunodeficiency virus-discordant partnerships among men who have sex with men. Clin Infect Dis 2002; 35:1010–1017.
59. Wolitski RJ, Parsons JT, Gomez CA, Purcell DW, Hoff CC, Halkitis PN. Prevention with gay and bisexual men living with HIV: rationale and methods of the Seropositive Urban Men's Intervention Trial (SUMIT). AIDS 2005; 19(Suppl 1):S1–S11.
60. Xia Q, Molitor F, Osmond DH, Tholandi M, Pollack LM, Ruiz JD, 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.
61. Blick G, Kagan RM, Coakley E, Petropoulos C, Maroldo L, Greiger-Zanlungo P, et al
. The probable source of both the primary multidrug-resistant (MDR) HIV-1 strain found in a patient with rapid progression to AIDS and a second recombinant MDR strain found in a chronically HIV-1-infected patient. J Infect Dis 2007; 195:1250–1259.
62. Marks G, Millett GA, Bingham T, Bond L, Lauby J, Liau A, et al
. Understanding differences in HIV sexual transmission among Latino and black men who have sex with men: The Brothers y Hermanos study.AIDS Behav
2008 [Epub ahead of print].
63. Janssen RS, Holtgrave DR, Valdiserri RO, Shepherd M, Gayle HD, De Cock KM. The serostatus approach to fighting the HIV epidemic: prevention strategies for infected individuals. Am J Public Health 2001; 91:1019–1024.
64. Janssen RS. HIV testing: rationale for changing recommendations. Top HIV Med 2007; 15:6–10.
65. Crepaz N, Lyles CM, Wolitski RJ, Passin WF, Rama SM, Herbst JH, et al
. Do prevention interventions reduce HIV risk behaviours among people living with HIV? A meta-analytic review of controlled trials. AIDS 2006; 20:143–157.
66. Richardson JL, Milam J, McCutchan A, Stoyanoff S, Bolan R, Weiss J, et al
. Effect of brief safer-sex counseling by medical providers to HIV-1 seropositive patients: a multi-clinic assessment. AIDS 2004; 18:1179–1186.
67. CDC. Incorporating HIV prevention into the medical care of persons living with HIV. Recommendations of CDC, the Health Resources and Services Administration, the National Institutes of Health, and the HIV Medicine Association of the Infectious Diseases Society of America.MMWR Recomm Rep
2003; 52 (RR-12)
68. Margolis AD, Wolitski RJ, Parsons JT, Gómez CA. Are healthcare providers talking to HIV-seropositive patients about safer sex? AIDS 2001; 15:2337–2339.
69. Marks G, Richardson JL, Crepaz N, Stoyanoff S, Milam J, Kemper C, et al
. Are HIV care providers talking with patients about safer sex and disclosure? A multi-clinic assessment. AIDS 2002; 16:1953–1957.