Three randomized, controlled trials to determine the effectiveness of male circumcision in preventing HIV among predominantly heterosexual men in sub-Saharan Africa demonstrated a 50–60% reduction in HIV infection [1–3]. This prompted the World Health Organization and the Joint United Nations Programme on HIV/AIDS to recognize male circumcision as an additional strategy to prevent heterosexual HIV acquisition . It is not clear, however, if this intervention would be an effective prevention strategy in reducing HIV incidence in the US. The majority of HIV transmission occurs among the population of men who have sex with men (MSM) . In this group, circumcision would be presumed to protect only the insertive partner. There are limited data and mixed results addressing the question of a protective effect of male circumcision among MSM. Some studies have found or suggested a protective effect of circumcision on HIV infection among MSM [6–12], whereas others [13–17] have not. Moreover, results from a recent meta-analysis did not suggest a protective effect of male circumcision on HIV infection among MSM . Additional analyses in this area could help inform recommendations and whether or not to conduct a clinical trial to determine the efficacy of male circumcision as an intervention for reducing the risk of HIV infection among MSM.
To further explore this question, data from a completed HIV vaccine clinical trial (VaxGen, VAX004) were reanalyzed. The objectives were to determine the percentage of MSM participants who were circumcised; demographic characteristics and risk factors associated with circumcised and uncircumcised MSM participants; whether circumcision status was associated with incident HIV infection controlling for demographic and risk factors, and whether lack of circumcision among men engaging in unprotected insertive anal sex with an HIV-positive partner was a risk factor for HIV infection.
VaxGen VAX004 was a randomized, double-blind, placebo-controlled efficacy trial of an HIV vaccine (bivalent rgp120 HIV-1 subtype B). It was carried out at 61 sites (US: n = 57, Canada: n = 3; The Netherlands: n = 1). Participants were randomly assigned to the arms of the study, 2/3 to the vaccine arm and 1/3 to the placebo arm. Seven vaccine doses were administered over the 36-month trial (baseline, 1, 6, 12, 18, 24 and 30-month visits). At each trial visit, participants were provided HIV counseling, testing and education about the trial. The trial was conducted according to the Declaration of Helsinki and with the approval of the local Institutional Review Boards. Participants provided informed consent for the study. The vaccine showed no evidence of effectiveness in either preventing HIV infection or in lowering levels of viremia post HIV infection . For more information about the trial, please refer to the study by Harro et al..
Eligible participants for VaxGen VAX004 comprised MSM and women at risk for HIV infection. Enrollment was from June 1998 to October 1999. For the present reanalysis, men in both the vaccine and placebo arms of the trial were included; women were excluded. Men eligible to participate in the trial included MSM who engaged in anal sex with a male sex partner during the previous 12 months. Men not eligible to participate included MSM who were in a monogamous relationship with an HIV-seronegative partner for at least 12 months, anyone who reported a history of injection drug use in the last 3 years, and anyone who was HIV-infected or seroconverting at enrollment.
At baseline and every 6 months throughout the trial, face-to-face interviews were conducted with participants in which questions were asked about HIV risk behaviors during the previous 6 months (sexual risk behavior, drug use, and having had a sexually transmitted infection or STI). Questions pertaining to adverse events associated with vaccination were asked 2 weeks after each vaccination. Circumcision status was self-reported.
HIV infection status and the time of HIV testing on or near (typically within 1 month) the time of the interview comprise the outcome for this study. HIV testing was done with enzyme-linked immunosorbent assay kits approved by the Food and Drug Administration (FDA). Positive tests were confirmed by FDA-approved immunoblot kits. Predictors included demographic characteristics, occurrence of a STI, engaging in specific risk behaviors (sexual risk behaviors and drug use) and treatment arm assignment. Participants were asked about their sexual behavior with partners of positive, negative, and unknown serostatus. All predictors indicating instances of sex refer to instances of sex not using a condom.
We included all time points in the analysis with the exception of the baseline time point. For purposes of analysis, we excluded certain participants and categorized some predictors. Participants who became circumcised during the trial or reported a sex change were excluded from the analysis. Female-related risk predictors were excluded from the analyses because too few MSM participants reported sex with females. Predictors measured as counts (e.g. reported number of times had insertive anal sex with an HIV-positive partner) were categorized based on quantiles when possible. In cases when over 50% of predictor data were zero, we either trichotomized based on the upper quartile or dichotomized.
In preparation for multivariable analyses, we used factor analysis to determine whether any potential covariates possessed high inter-correlation (i.e. loaded highly to a latent factor). Groups of predictors loading highly together (based on a threshold of 0.70) are included in Table 1 along with the representative predictor used in the analyses. Factor analysis found the insertive anal sex and receptive anal sex predictors to be highly correlated. For this reason, two models were run separately, first with the insertive, then with receptive anal sex predictors. All participants, with the exception of those noted above, were included in each of the models.
We assessed associations with circumcision and predictors using bivariate analyses. Chi-squared tests were used to assess differences between circumcised and uncircumcised MSM at the 6-month visit. For multivariable analyses of the main outcome, a generalization of proportional hazards regression called a counting process method  was used. This method allowed us to analyze data from each visit of the trial (excluding baseline) by incorporating each interval of time between visits into the model. Including data from each visit is important as this allows us to more closely associate the reported risk behaviors between visits to the outcome. Thus, instead of aggregating risk behaviors across the trial, we can align a respondent's self-reported, sexual behaviors with the period of time between a participant's HIV tests as recorded in the survey at the visit concluding this time interval. The counting process method allows us to model these multiple visits. As with single-event Cox models , a counting process seeks to determine whether predictors increase or decrease the likelihood of experiencing the outcome and relate that likelihood to a reference via a hazard ratio. The difference between the two methods is the counting process uses multiple visits from a single participant and then adjusts the variances by using the sandwich estimator . More information on counting process models is available . The outcome in this model is whether or not the participant became HIV-infected in the time before the visit in question and the time of the visit. Any visits when a participant was not observed are excluded from analyses. Included in the model were all predictors not loading highly to a latent factor, representatives of latent factors, and circumcision status. Finally, in the insertive anal sex model, we included the interaction of the insertive anal sex with an HIV-positive partner indicator and being uncircumcised. Here, the interaction is included to indicate whether uncircumcised men engaging in insertive anal sex with an HIV-positive partner were at increased risk for HIV infection in this study. Even though HIV infection through receptive anal sex would not be influenced by circumcision status, we similarly included the interaction of the receptive anal sex with an HIV-positive partner indicator and being uncircumcised to be consistent across models. In a recent HIV vaccine trial, men who were uncircumcised and who received the vaccine had a higher risk of HIV infection than uncircumcised men who received the placebo . Thus, we included the interaction of treatment arm assignment and being uncircumcised to rule out the possibility that uncircumcised men receiving the VAX004 vaccine were at greater risk of HIV infection than all others. All statistical tests were two-sided and all confidence intervals were based on a 5% level of significance. SAS version 9.1.3 (SAS Institute Inc., Cary, North Carolina, USA) was used for the counting process analysis performed in PROC PHREG. Weighted Schoenfeld residuals  were checked to validate the proportional hazards assumption. SAS version 9.2 was used to compute the relative risk confidence limits with a Taylor series linearization  in PROC SURVEYFREQ to assess the variability in the relative risk for HIV infection rate in circumcised and uncircumcised men who reported unprotected insertive anal sex with an HIV-positive partner.
A total of 5417 participants enrolled in the study from June 1998 to November 1999. Removed from the analysis were 309 women, 13 men who were indicated by nucleic acid testing to be HIV-infected at baseline, one participant who reported a sex change, and five men whose circumcision status changed during the study. The final number of participants for this analysis, which excluded 200 participants who only attended the baseline visit, was 4889. Of these, 4209 (86.1%) were circumcised. A total of 342 (7.0%) men became HIV-infected during the study and of these, 43 (12.6%) were uncircumcised and 299 (87.4%) were circumcised. The HIV incidence rate was 2.6 infections per 100 person-years of observation. Of participants in our analysis, 80.2% (3921/4889) were recorded as completing the study.
Bivariate analyses: comparison of uncircumcised versus circumcised MSM
Table 2 shows the results of the bivariate analysis. Compared to uncircumcised MSM, a greater proportion of circumcised MSM had a college degree, were non-Hispanic white, attended US study sites, did not report one or more STIs, did not report using amyl nitrite (poppers), reported at least two HIV-negative partners, reported at least one instance of oral sex with HIV-negative partner, and reported at least one instance of insertive anal sex with an HIV-negative partner. A smaller proportion of circumcised than uncircumcised MSM reported at least seven HIV-unknown status partners. The differences in proportions were greatest for race/ethnicity, study site location, poppers usage, and number of HIV-negative male partners.
Multivariable analysis with insertive anal sex predictors: factors associated with HIV infection
Table 3 shows the results of the multivariable analysis that includes the unprotected insertive anal sex predictors. Receptive anal sex predictors were not included due to high correlation with insertive anal sex predictors. Controlling for demographic characteristics and risk behaviors, being uncircumcised was not associated with HIV infection during the 36-month trial [adjusted hazards ratio (AHR) = 0.97, confidence interval (CI) = 0.56–1.68]. Moreover, the interaction between being uncircumcised and reporting insertive anal sex with an HIV-positive partner and HIV infection (AHR = 1.78, CI = 0.90–3.53) was not significant [χ2(1) = 2.76, P < 0.10]. However, there was an association of insertive anal sex with an HIV-positive partner alone with HIV infection (AHR = 2.25, CI = 1.72–2.93). In addition, the following predictors were associated with an increased risk of HIV infection: high school education or less compared to a graduate degree (AHR = 1.45, CI = 1.07–1.96); use of poppers (AHR = 1.75, CI = 1.39–2.19); use of amphetamines/hallucinogens (AHR = 1.50, CI = 1.18–1.90); total number of male partners compared to 0 or 1 [2–4 (AHR = 1.58, CI = 1.05–2.38); 5–10 (AHR = 2.03, CI = 1.34–3.08); >10 (AHR 3.17, CI = 2.09–4.79)]; and number of HIV-negative male partners compared to 0 or 1 [>3 (AHR = 1.38, CI = 1.04–1.84)]. Decreased risk of HIV infection was associated with increasing age (AHR = 0.98, CI = 0.96–0.99) and reported number of times insertive anal sex with an HIV-negative partner (AHR = 0.78, CI = 0.61–0.997). Neither treatment arm assignment nor the interaction of vaccine arm assignment and being uncircumcised was associated with HIV infection.
Multivariable analysis with receptive anal sex predictors: factors associated with HIV infection
Table 4 shows the results of the multivariable analysis model that includes the unprotected receptive anal sex predictors. Insertive anal sex predictors were not included due to correlation with receptive anal sex predictors. Controlling for demographic characteristics and risk behaviors, there was no association of HIV infection during the 36-month trial with being uncircumcised (AHR = 1.04, CI = 0.61–1.80) nor with the interaction of being uncircumcised and reporting receptive anal sex with an HIV-positive partner (AHR = 1.26, CI = 0.62–2.57). In addition, the following predictors were associated with an increased risk of HIV infection: high school education or less compared to a graduate degree (AHR = 1.50, CI = 1.11–2.03); use of poppers (AHR = 1.54 CI = 1.23–1.95); use of amphetamines/hallucinogens (AHR = 1.41, CI = 1.11–1.79); total number of partners compared to 0 or 1 [2–4 (AHR = 1.52, CI = 1.01–2.29); 5–10 (AHR = 1.83, CI = 1.21–2.76); >10 (AHR 2.47, CI = 1.63–3.74)]; number of HIV-negative male partners compared to 0 or 1 [(>3 (AHR = 1.39, CI = 1.06–1.83)], number of times receptive anal sex with an HIV-positive partner (AHR = 3.45, CI = 2.58–4.61), and number of times receptive anal sex with an HIV-unknown status partner (AHR = 2.27, CI = 1.74–2.96). Decreased risk of HIV infection was associated with increasing year of age (AHR = 0.99, CI = 0.97–0.999). As with the insertive anal sex model, neither treatment arm assignment nor the interaction of vaccine arm assignment and being uncircumcised was associated with HIV infection.
Follow-up comparison of HIV infection rate in circumcised and uncircumcised men who reported unprotected insertive anal sex with an HIV-positive partner
Of the study visits when a participant reported unprotected insertive anal sex with an HIV-positive partner, HIV infection among circumcised men was reported in 3.16% of the visits (80/2532) and among uncircumcised men in 3.93% of the visits (14/356) [relative risk (RR) = 0.80, CI = 0.46–1.39]. This analysis did not exclude participants who also had unprotected receptive sex or unprotected insertive anal sex with partners of negative or unknown HIV serostatus.
There was no overall significant association between circumcision status and HIV infection among MSM participating in the VaxGen VAX004 HIV vaccine trial in our models that included reported insertive and receptive anal sex separately. This result differs from the African circumcision trial results when circumcision was partially protective for heterosexual men [1–3]. Moreover, whereas having unprotected insertive and receptive anal sex with an HIV-positive partner was associated with HIV infection, there was no significant interaction of being uncircumcised and reporting unprotected insertive anal sex with an HIV-positive partner. As expected, there was no significant interaction of being uncircumcised and reporting unprotected receptive anal sex with an HIV-positive partner. In a follow-up analysis, no significant difference was found between the rates of HIV infection of circumcised and uncircumcised MSM who reported unprotected insertive anal sex with an HIV-positive partner at study visits.
Similar to other studies, we found a high proportion (86.1%) of circumcision in this primarily US-based group of MSM participating in the VaxGen VAX004 vaccine clinical trial. This percentage is slightly higher than the 79% overall prevalence reported in a National Health and Nutrition Examination Survey study  and the 75% reported in HIV-uninfected gay/bisexual men in San Francisco . The reason for the higher percentage of circumcised men in our study may be due to the small proportion of minorities enrolled, as some minority groups are reported to be less likely to be circumcised than whites .
Circumcision status played a role in HIV infection as an effect modifier in the step HIV vaccine trial . Among men who were circumcised in that trial, the risk of HIV infection was not associated with receipt of vaccine. However, among uncircumcised men, the vaccine/placebo hazard ratio of HIV infection was 3.8 (95% CI 1.5–9.3). This was not the case in this reanalysis of the VAX004 data.
There is not clear evidence for an overall benefit of circumcision in preventing HIV infection among MSM due in part to the multiple factors (e.g. HIV prevalence in the population, use of highly active antiretroviral therapy or HAART) that can impact results of observational studies. In a study of Latino immigrant MSM (including men from Brazil, Colombia, and the Dominican Republic) living in New York City, uncircumcised compared to circumcised men were nearly twice as likely to be HIV-infected. Though in follow-up analyses, results showed that the protective effects of circumcision occurred only in the men from Colombia, perhaps because of the higher HIV prevalence among MSM in that country . Additionally, whereas the recent meta-analysis of 15 observational studies assessing the association between circumcision status and HIV infection among MSM found insufficient evidence to conclude that male circumcision protects against HIV infection , there was a significant protective effect for circumcision in studies conducted before, but not after, the advent of HAART. This result may have been influenced by the lower risk of transmitting the disease for persons taking HAART medication  and/or an increase in risk behaviors among some individuals documented after the advent of HAART  which could outweigh any protective effect of circumcision .
Our results showed that being uncircumcised and engaging in insertive anal sex with a HIV-positive partner was not significantly associated with HIV infection over the period the participants were in the 36-month study. Two Australian studies assessing circumcision status and its effect on HIV infection have also suggested no protective effect of circumcision for the insertive partner [13,14]. In a cross-sectional study comprised of homosexual men recently diagnosed with HIV, there was no association between men being uncircumcised and men reporting insertive anal sex as their highest risk activity . In a longitudinal study of MSM, circumcision status was not associated with HIV infection even when the analysis was restricted to men who reportedly did not engage in receptive anal sex . However, in a reanalysis of the same data among participants with a preference for the insertive role, being circumcised was associated with a reduction in HIV infection . Because many MSM take on both receptive and insertive anal sex roles , this factor should be taken into account in future studies. It is of note that the per-contact risk of HIV transmission for men having unprotected anal sex with an HIV-positive or unknown-status partner is more than four times as high for the receptive than for the insertive partner (0.27 versus 0.06%) . This risk could outweigh any possible protective effect of male circumcision for the insertive partner in men who take on both roles. In fact, a South African cross-sectional study of MSM who reported engaging exclusively in insertive anal sex showed that circumcision was significantly protective for HIV infection, though most of these men had partnerships with women as well .
The study must be interpreted in the context of several potential limitations. First, the purpose of this longitudinal study was to test the efficacy of an HIV vaccine, not to assess the risk of not being circumcised on HIV infection. Second, circumcision was self-reported, not clinically determined, although self-reported circumcision status has been found to be a valid measure among homosexual men . Third, the sample size of minority MSM was small and likely resulted in a smaller proportion of uncircumcised men, given that some minority groups are more likely to be uncircumcised . Fourth, participants in the overseas sites represented a disproportionate percentage of uncircumcised men (29.4%) compared to their overall representation in the trial (8.26%), which could be important if the data are used to make decisions about the US MSM population. Fifth, we could not control for the risk of receptive anal sex in the model because the insertive and receptive anal sex variables were highly correlated. Finally, the number of uncircumcised men in our analysis who became HIV-infected during the trial was small (n = 43), so these results should not be considered definitive. The strength of the study is that it provides a longitudinal assessment of the relationship between circumcision status and HIV infection when risk behaviors, as well as circumcision status were collected every 6 months.
In conclusion, results of this study did not show a significant protective effect of circumcision status on HIV infection among MSM engaging in unprotected insertive anal sex with a HIV-positive partner; however, we could not control for the HIV risk of receptive anal sex and the proportion of uncircumcised men and the number of incident HIV infections was small. The analysis of observational studies was suggested by an expert panel convened by CDC as a first step before considering whether a randomized controlled trial of male circumcision among MSM was warranted . Our finding contributes to the understanding of the potential effect of male circumcision on HIV infection among MSM in the United States. However, additional studies with more incident HIV infections, men who are exclusively insertive partners, or a larger proportion of uncircumcised men will likely provide greater power to determine if circumcision is associated with lower rates of HIV infection among MSM who engage in insertive anal sex with HIV-infected partners. It is of note that a recent mathematical modeling study that looked at the relationship between the proportion of circumcision and long-term HIV prevalence in an MSM community in a developed country found the decrease in HIV prevalence as a result of increasing circumcision was lower than that experienced for heterosexual men. The authors concluded circumcision as an HIV prevention intervention for MSM in developed countries may not substantially reduce HIV prevalence . Relatedly, CDC scientists state that due to the high circumcision rate in the US and other factors such as the fact that most HIV transmission occurs among MSM, the impact of circumcision on HIV infection would be limited [36,37]. In addition to further analyses of observational studies, several issues should be considered in developing recommendations for circumcision as a prevention intervention for HIV infection including acceptability of the procedure among uncircumcised men at high risk for infection, continued emphasis on consistent condom use communications, and healthcare coverage for the procedure .
The authors thank Global Solutions for Infectious Diseases, the exclusive licensee of the VAX003 and VAX004 clinical data.
Author contributions: All authors contributed to the interpretation of the results and in editing and revising the manuscript. R.W. researched and conducted all statistical analyses and wrote the analysis section. K.K. provided input into variable selection for the models. P.K. suggested the follow-up analysis. R.W., B.C., K.K., S.S., P.K. and B.B. provided critical review of the manuscript. D.G. initiated the study, provided input into variable selection for the models, and drafted the manuscript.
1. Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med 2005; 2:e298.
2. Bailey RC, Moses S, Parker CB, Agot K, Maclean I, Krieger JN, et al
. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. Lancet 2007; 369:643–656.
3. Gray RH, Li X, Kigozi G, Serwadda D, Nalugoda F, Watya S, et al
. The impact of male circumcision on HIV incidence and cost per infection prevented: a stochastic simulation model from Rakai, Uganda. AIDS 2007; 21:845–850.
4. Joint United Nations Programme on HIV/AIDS. New data on male circumcision and HIV prevention: policy and programme implications. 2007. Available: http://www.who.int/hiv/pub/malecircumcision/research_implications/en/index.htm
. Accessed 7/9/08.
5. CDC. HIV/AIDS surveillance report: cases of HIV infection and AIDS in the U.S. and dependent areas. Revised edition 2007. Available http://www.cdc.gov/hiv/topics/surveillance/resources/reports/2005report/default.htm
. Accessed 7/20/08 2005; 17.
6. Buchbinder SP, Vittinghoff E, Heagerty PJ, Celum C, Seage GR, Judson FN, et al
. Sexual risk, nitrite inhalant use, and lack of circumcision associated with HIV seroconversion in men who have sex with men
in the United States. J Acquir Immune Defic Syndr 2005; 39:82–89.
7. Kreiss JK, Hopkins SG. The association between circumcision status and human immunodeficiency virus infection among homosexual men. J Infect Dis 1993; 168:1404–1408.
8. Reisen CA, Zea MC, Poppen PJ, Bianchi FT. Male circumcision and HIV status among Latino immigrant MSM in New York City. J LBGT Health Res 2007; 3:29–36.
9. Templeton D, Jin F, Mao L, Prestage GP, Donovan B, Imrie J, et al.Circumcision and risk of HIV infection in Australian homosexual men.AIDS
2009 [Epub ahead of print].
10. Lai S, Hong C, Lan Y, Chen KT, Wong WW, Hu BS, et al.Molecular epidemiology of HIV-1 in men who have sex with men from gay saunas in Taiwan from 2000-2003
[abstract WePeC6097]. International Conference on AIDS
; July 11–16, Bangkok, Thailand. 2004.
11. Sanchez J, Lama JR, Peinado J, Paredes A, Lucchetti A, Russell K, et al
. High HIV and ulcerative sexually transmitted infection incidence estimates among men who have sex with men
in Peru: awaiting for an effective preventive intervention. J Acquir Immune Defic Syndr 2009; 51(Suppl 1):S47–51.
12. Lane T, Raymond HF, Dladla S, Rasethe J, Struthers H, McFarland W, et al.High HIV prevalence among men who have sex with men in Soweto, South Africa: results from the Soweto men's study.AIDS Behav
2009 [Epub ahead of print].
13. Grulich AE, Hendry O, Clark E, Kippax S, Kaldor JM. Circumcision and male-to-male sexual transmission of HIV. AIDS 2001; 15:1188–1189.
14. Templeton D, Jin F, Prestage G, Donovan B, Imrie J, Kippax SC, et al.Circumcision status and risk of HIV seroconversion in the HIM cohort of homosexual men in Sydney.4th International AIDS Conference
; Sydney, Australia. 2007.
15. Millet G, Ding H, Lauby J, Flores S, Stueve A, Bingham T, et al
. Circumcision status and HIV infection among Black and Latino men who have sex with men
in 3 U.S. cities. Epidemiol Soc Sci 2007; 46:643–650.
16. Mor Z, Kent CK, Kohn RP, Klausner JD. Declining rates in male circumcision amidst increasing evidence of its public health benefit. PLoS ONE 2007; 2:e861.
17. Elford J, McKeown E, Nelson S, Anderson J, Low N. Circumcision status and HIV infection in a diverse sample of MSM living in Britain.XVII International AIDS Conference
; Mexico City, August 3–8, 2008.
18. Millet G, Flores S, Marks G, Reed J, Herbst J. Circumcision status and risk of HIV and sexually transmitted infections among men who have sex with men
: a meta-analysis. JAMA 2008; 300:1674–1684.
19. Gilbert PB, Peterson ML, Follmann D, Hudgens MG, Francis DP, Gurwith M, et al
. Correlation between immunologic responses to a recombinant glycoprotein 120 vaccine and incidence of HIV-1 infection in a phase 3 HIV-1 preventive vaccine trial. J Infect Dis 2005; 191:666–677.
20. Harro CD, Judson FN, Gorse GJ, Mayer KH, Kostman JR, Brown SJ, et al
. Recruitment and baseline epidemiologic profile of participants in the first phase 3 HIV vaccine efficacy trial. J Acquir Immune Defic Syndr 2004; 37:1385–1392.
21. Andersen P, Gill R. Cox's regression model for counting processes: a large sample study. Ann Statistics 1982; 10:1100–1120.
22. Cox DR. Regression models and life-tables. J Roy Statist Soc B 1972; 34:187–222.
23. Lin DY, Wei LJ. The robust inference for the proportional hazards model. J Am Statist Assoc 1989; 84:1074–1078.
24. Therneau TM, Hamilton SA. rhDNase as an example of recurrent even analysis. Statist Med 1997; 16:2029–2047.
25. Buchbinder SP, Mehrotra DV, Duerr A, Fitzgerald DW, Mogg R, Li D, et al
. G. Efficacy assessment of a cell-mediated immunity HIV-1 vaccine (the Step Study): a double-blind, randomized, placebo-controlled, test of concept trial. Lancet 2008; 372:1881–1893.
26. Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994; 81:515–526.
27. Lohr SL. Sampling: design and analysis. Pacific Grove, CA: Duxbury Press; 1999.
28. Xu F, Markowitz LE, Sternberg MR, Aral SO. Prevalence of circumcision and herpes simplex virus type 2 infection in men in the United States: the National Health and Nutrition Examination Survey (NHANES), 1999–2004. Sex Transm Dis 2007; 34:479–484.
29. Bunnell R, Ekwaru J, King R, Bechange S, Moore D, Khana K, et al.3-Year Follow-up of Sexual Behavior and HIV Transmission Risk of Persons Taking ART in Rural Uganda [abstract 29].Conference on Retroviruses and Opportunistic Infections
; Boston, MA February 3–6, 2008.
30. Crepaz N, Hart TA, Marks G. Highly active antiretroviral therapy and sexual risk behavior: a meta-analytic review. JAMA 2004; 292:224–236.
31. Moskowitz D, Rieger G, Roloff M. Tops, bottoms, and versatiles. Sex Relation Ther 2008; 23:191–202.
32. 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.
33. Templeton DJ, Mao L, Prestage GP, Jin F, Kaldor JM, Grulich AE. Self-report is a valid measure of circumcision status in homosexual men. Sex Transm Infect 2008; 84:187–188.
34. Smith DK, Taylor A, Kilmarx PH, Sullivan P, Warner L, Kamb M, et al
. Male circumcision in the United States for the prevention of HIV infection and other adverse health outcomes: report from a CDC consultation. Public Health Rep 2010; 125:72–82.
35. Londish G, Templeton D, Regan D, Kaldor J, Murray J. Small population health benefits on HIV by circumcising men who have sex with men.2008 Australasian Sexual Health Conference
; September 17–20, Perth, Australia. 2008.
36. Sullivan PS, Kilmarx PH, Peterman TA, Taylor A, Nakashima M, Kamb M, et al
. Male circumcision for prevention of HIV transmission: what the new data mean for HIV prevention in the United States. PLoS Med 2007; 4:e223.
37. Kilmarx PH, Kretsinger K, Millet GA. Considerations in the role of male circumcision in the prevention of HIV transmission in the United States. HIV Ther 2009; 3:241–254.
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.