Huebner, David M PhD, MPH; Binson, Diane PhD; Woods, William J PhD; Dilworth, Samantha E MS; Neilands, Torsten B PhD; Grinstead, Olga PhD
Studies of the efficacy of voluntary HIV counseling and testing (VCT) suggest that traditional VCT may reduce the risk behaviors of individuals diagnosed as HIV-positive and of serodiscordant couples.1 Although risk reduction among HIV-negative individuals is less certain after standard VCT, enhanced interactive models of brief risk reduction counseling for HIV-negative individuals have been shown to reduce risk behaviors and the incidence of sexually transmitted infections (STIs).2,3
In 2003, the US Centers for Disease Control and Prevention (CDC) announced that diagnosing unidentified HIV infections would become a cornerstone of its new domestic approach to HIV prevention.4 To this end, one of the CDC's 4 primary priorities is to implement new models for VCT outside medical settings. Settings likely to be most effective include those where large numbers of high-risk individuals are willing to test.
Among men who have sex with men (MSM), individuals who engage in unprotected anal intercourse (UAI) and who have multiple casual sexual partners are at greatest risk for HIV infection.5,6 In a probability sample of MSM residing in 4 major urban areas, 76% of the men who reported having UAI with a nonprimary partner in the past year also reported that they had sought out sex at a public venue during that time.7 Thus, these venues may be a promising place to target MSM with appreciable risk for HIV infection. Some evidence also indicates that these venues are a good place to identify men who are unaware of their HIV status. In the same sample described previously, more than half of the men had not been tested for HIV in the last year, and almost a fifth of those men reported engaging in UAI with a casual or secondary partner. Among the high-risk men who had not recently been tested for HIV, more than half reported that they went to bathhouses or sex clubs.8
Anecdotal evidence suggests that testing for STIs in bathhouses has been occurring since the 1970s.9 More recently, studies have shown that HIV VCT is possible to conduct in bathhouses and that it is being implemented in selected sites.10 Only limited documentation exists regarding the feasibility of this approach in reaching high-risk men and men who otherwise test infrequently, however.11 Furthermore, no study that we are aware of has ever studied the impact of VCT in changing risk and protective behaviors among bathhouse patrons. Thus, the present study had 2 primary goals: first, to document the feasibility of recruiting high-risk men from bathhouses into on-site VCT and, second, to provide preliminary evidence for the effectiveness of VCT in this setting. Specifically, we hypothesized the following:
1. When VCT is offered in a bathhouse, MSM with histories of recent sexual risk behaviors would be willing to be tested for HIV in this setting and to return for their results.
2. Three months after participation in an interactive VCT intervention in a bathhouse, MSM would report fewer HIV-related risk behaviors (ie, UAI, sex under the influence of alcohol or drugs) and more precautionary behaviors (ie, discussing both HIV-status and condom use with partners).
BACKGROUND OF THE STUDY
The California State Office on AIDS “High Risk Initiative” funded the Center for AIDS Prevention Studies to provide technical assistance and to conduct process and outcome evaluation of high-risk initiative interventions in 3 California health jurisdictions. This report focuses on one of these health jurisdictions, a city health department, which was funded to provide VCT in a gay bathhouse setting. In collaboration with the city health department, from October 2001 to November 2002, we evaluated the health department's counseling and testing intervention that served MSM in this setting.
Research Design Overview
Given the community-collaborative nature of this study and limitations in the scope of the project budget, it was not feasible to conduct a full randomized controlled trial of the effectiveness of VCT in the bathhouse. Thus, we opted for a simpler design that was agreeable to all collaborating parties. This design involved a within-subjects pre- to postintervention comparison of risk and precautionary behaviors. Participants were interviewed before receiving HIV counseling and testing and again 3 months later. Men were paid $20 in cash for the baseline survey and $25 for the follow-up survey.
At the beginning of the study period, testing was offered at the venue on Saturday evenings only. Later in the evaluation period, testing hours were expanded to include Wednesday afternoons. Testing was offered 5 nights per week during June 2001 (“Gay Pride” month). In total, testing was offered during 72 shifts over the 13-month evaluation period. Given staffing resources, approximately 12 men could be tested during any given shift.
The testing program that operated on-site at the bathhouse followed California State outreach HIV counseling and testing protocols for testing outside a medical clinic. All testing was done anonymously. An outreach worker actively recruited patrons to test as they passed through the area of the club where the program was set up, fliers announcing the testing were posted around the club, and periodic announcements were made over the public address system. Those interested in being tested were handled on a first-come, first-served basis. Initially, pretest counseling was provided in a private room in the bathhouse (ie, otherwise rented to customers) that was set aside for the shift for testing purposes, and, later, counseling occurred in rooms built specifically for outreach at the club.
In the pretest counseling session, health department counselors were trained to use harm-reduction and client-centered approaches in addressing issues thought to be of particular relevance to the bathhouse population, namely, increasing discussion of HIV status and negotiating condom use. Some attention was also given to encouraging men to seek out social support about their sexuality, particularly for those men who used the bathhouse as a sexual outlet but who do not otherwise identify as gay. After pretest counseling, a client was escorted to the testing room, where oral secretions were collected for an HIV test (another private room or the office of the club's health director was usually used for this purpose).
Test results were given in the context of a posttest counseling session a week or more later. Because of concern that men might not necessarily want to return to the bathhouse to obtain their results during the scheduled testing times, the posttest session was offered at one of several venues. The primary location was the health department's clinic located roughly 2 blocks from the club. Clients also could receive a posttest counseling session in a testing van parked in a private parking lot next to the club during hours when the testing staff was at the club administering the on-site outreach program. This allowed men to return for their results without having to enter the club. On several occasions, exceptions were made to allow posttest counseling inside the club during the on-site outreach program.
Sample for Evaluation of On-Site Program
As soon as a man expressed interest in getting an HIV test at the on-site program, the testing clerk told him about the evaluation survey and invited him to participate. Not all men receiving tests were invited to participate, because test counselors outnumbered evaluation interviewers. Participation in the evaluation study was offered on a first-come, first-served basis whenever an interviewer was available. Those who agreed to participate were immediately introduced to the survey interviewer, who led the individual to a private room to complete the informed consent form, take contact information (to be used to schedule the 3-month follow-up interview), and conduct the baseline interview. Because VCT was conducted anonymously, data collected in the course of the VCT session, including HIV test results, were not linkable to data from the evaluation study.
Inclusion criteria for participation in the evaluation survey included any man at the bathhouse who requested HIV testing through the on-site HIV/sexually transmitted disease (STD) testing program. Because the initial questionnaire was designed to assess attitudes and behaviors before VCT, men were excluded if they had already tested that day or had received HIV test counseling in the preceding 3 months. Men were also excluded if they were not able to speak English or Spanish. Club patrons were required to be at least 18 years of age; thus, no participants were minors. Participation in the evaluation survey was completely voluntary, and participants could withdraw from the study at any point without affecting their options to test at the on-site testing program. After completing the evaluation interview, the participant was led back to the testing clerk, who introduced him to the next available test counselor, who would then conduct his pretest counseling and collect saliva for the test.
Three-month follow-up interviews were conducted at one of 3 possible sites: the research team offices, the health department offices, or the bathhouse. Interviews were scheduled via telephone or e-mail correspondence initiated by the study interviewer approximately 2 weeks before the 3-month follow-up. Participants who failed to make 2 scheduled appointments were considered lost to follow-up, as were men for whom the contact information was no longer accurate. All study procedures and instruments were approved by the Committee on Human Research at the University of California, San Francisco.
Questionnaires were developed in collaboration with community partners and built on data obtained from 3 focus groups: 1 each with bathhouse experts, bathhouse patrons, and test counselors. The purpose of the focus groups was to gather information about how to word questions about men's sexual activities generally, and, specifically, in bathhouse settings. In addition to helping us arrive at suitable question wording, the focus group discussions prompted us to use an innovative survey technique, “conversational interviewing.”12,13 Conversational interviewing is a survey research technique that engages the respondent and interviewer in a type of conversation rather than the “question-and-response” exchange that is typical in standard survey practice. This technique was used in all the sexual behavior sections of the questionnaire (but not in other sections) to ensure that the participant had a clear understanding of the researcher's intent in asking each question related to sexual behavior. Before implementing the assessment instrument and procedure, it was piloted extensively over a 3-month period.
The items used in this analysis were the same at the baseline assessment and at the 3-month follow-up. Assessment interviews were conducted by men and lasted approximately 35 minutes.
Sexual Behavior Overview
Participants were first asked if they had had any anal, vaginal, or oral sex partners in the previous 3 months. For participants who reported any anal sex with a man, detailed questions were asked about their 2 most recent anal sex partners. Men with only 1 anal sex partner in the last 3 months were asked questions only about this 1 partner. For participants who reported no anal sex partners, detailed questions were asked about the 2 most recent oral sex partners. Men were asked whether they knew the HIV status of their partners and what relationship they had with each partner.
Unprotected Anal Intercourse
UAI was calculated as the total number of unprotected insertive or receptive anal sex acts in the past 3 months with the 2 most recent anal sex partners,* regardless of partner type or HIV status of the partner. Men with no male sex partners or only oral sex partners were coded as 0. This variable was dichotomized so that we could also determine the proportion of men engaging in UAI, in addition to the number of acts of UAI.
Communication About Condoms
Respondents who reported any anal sex with a man in the past 3 months were asked whether they “ever talk[ed] about using condoms” with their 2 most recent anal sex partners. This variable was dichotomized so that men were coded as communicating with 100% of their 2 most recent anal sex partners or less than 100% of those partners. Men with no male anal sex partners in the past 3 months were not included in condom communication analyses.
Communication About HIV
Respondents who reported any anal or oral sex partners in the past 3 months were asked whether, with their 2 most recent anal or oral sex partners, they had “ever discuss[ed] whether [one of them had] HIV or not.” This variable was dichotomized so that men were coded as communicating with 100% of their most recent 2 partners or less than 100% of those partners. Men with no oral or anal sex partners in the past 3 months were not included in HIV communication analyses.
Sex While Drunk or High
Respondents who reported any anal or oral sex partners in the past 3 months were asked the number of times with their 2 most recent partners that they had “either insertive or receptive anal/oral sex…while high on drugs or alcohol.” Sex while under the influence was calculated as the total number of times they reported being drunk or high when having sex with their 2 most recent partners.
At baseline, participants self-reported their age, education, income, employment status (full-time or less than full-time), race/ethnicity, and relationship status. Because of the ethnic breakdown of the sample, participants were coded as white, Latino, or other ethnicity for purposes of the analyses. Relationship status was coded as involved in or not involved in some type of committed relationship (eg, marriage, domestic partnership).
Data analyses were performed in SAS 9.1 and Stata 8.2 for personal computers.14,15 To maximize statistical power and accuracy of inferences, multiple imputation was used to address incomplete data before performing the inferential analyses.16 Data were imputed for a total of 25 cases in which no follow-up data were available. The SAS MI and MIANALYZE procedures were used to generate Markov Chain Monte Carlo-based multivariable multiple imputations and to combine inferential results, respectively.14 Fifty imputed data sets were generated for each of the bivariate and multivariable inferential analyses. To improve the accuracy of imputed values, auxiliary variables were included in the multiple imputation generation process.17 These auxiliary variables included the following: number of visits to the public sex environment in the past year, region of residence, number of places lived in the past 6 months, and gender of sex partners at both time points.
Bivariate and Multivariable Modeling
For each count outcome (number of unprotected anal sex acts and number of occasions having sex while drunk or high), we used negative binomial regression. For dichotomous outcomes (any UAI, communication about HIV, and communication about condoms), we used logistic regression. Because each respondent contributes 2 observations (1 at each time point), observations are nested within respondents; therefore, all inferential analyses reported here make use of generalized estimating equations (GEEs), with robust standard errors to account for the clustered nature of the data.18
We began inferential inquiry by evaluating the statistical significance of each covariate with each outcome. First, we fit a typical repeated-measures model containing the explanatory covariate score, the time predictor, and, for continuous predictors, a time by covariate interaction. The bivariate analyses were then extended to include multiple explanatory variables to complete our final multivariable models. Because of the formative nature of this research, the study team decided a priori to follow the recommendations of Hosmer and Lemeshow19 by using a bivariate P value of 0.25 or lower as the cutoff for inclusion of explanatory variables in multivariable models. Once variables for the initial multivariable model were selected, additional models were successively estimated, removing nonsignificant variables one at a time in order of their P values, until all remaining parameter estimates were significant with P values less than 0.05. This process resulted in the final multivariable models reported.20
During the 13-month evaluation period, a total of 492 men received HIV VCT at the collaborating bathhouse. Of those men who were tested, 133 were recruited into the evaluation sample and completed the baseline (ie, pre-VCT) assessment. Of the 133 men who were interviewed at baseline, 108 (81.2%) completed the follow-up interview 3 months later. Table 1 displays the demographic characteristics and HIV-related risk behaviors of the evaluation sample at baseline and the portion of the evaluation sample retained to follow-up. Among men in the evaluation sample, those who were lost to follow-up did not differ from men who were interviewed at follow-up in terms of their age, ethnicity, education, income, sexual orientation, or history of risk or precautionary behaviors at baseline (all P > 0.23). Data from a recent probability sample of patrons from the same bathhouse suggest that the sample of men testing in the bathhouse is similar to the overall population of bathhouse patrons in terms of age and ethnicity but may be slightly less educated (64% of testers had college degrees relative to 73% of men at the bathhouse) and less likely to identify as gay (68% of testers identified as gay relative to 75% of men in the bathhouse).21
We hypothesized that when VCT was offered in a sex club, MSM with histories of recent sexual risk behaviors would be willing to be tested for HIV in the venue and would return for their results. Data from the evaluation sample indicate that 38.4% of men reported engaging in UAI with at least 1 of their 2 most recent anal sex partners during the 3 months before testing, and 48% of these men with recent risk had not been tested for HIV in the previous 12 months. Among men who did not report UAI, 53% had not tested in the past year. On average, men in the sample reported engaging in 3.2 unprotected acts across their 2 most recent partners.
Among all 492 men who tested during the evaluation period, 347 (70.5%) returned for their results. Eighteen men tested positive for HIV (3.7% of the men tested), and 9 of these men returned for their results (50% of the men who tested positive).
We also hypothesized that bathhouse-based VCT would show preliminary evidence of effectiveness in reducing risk and increasing precautionary behaviors. Table 2 presents the results of the bivariate and multivariable (ie, adjusted) logistic and negative binomial regression models predicting communication with sex partners about HIV status and about condoms. Table 3 presents models predicting any UAI, the number of UAI acts, and number of sex acts while drunk or high. Variables with significant coefficients for the “time” variable changed from before testing to the 3-month follow-up. At follow-up, men were more likely to report communicating with their sex partners about HIV but not about condoms. In addition, the proportion of the sample engaging in UAI decreased from baseline to follow-up, as did the reported number of sex acts while drunk or high. The raw number of acts of UAI did decrease from baseline to follow-up, but this decrease was attenuated when control was added for demographic characteristics.
Our data indicate that when VCT is offered in a gay bathhouse setting, men who have recent histories of HIV-related risk behavior and who have not otherwise been tested in the past year are willing to get tested. Thirty-eight percent of the men using VCT in this study reported UAI with at least 1 of their 2 most recent anal sex partners in the 3 months before testing, and, on average, men in the sample engaged in 3.2 unprotected acts across these 2 partners. Nearly half of the men reporting UAI had not been tested for HIV in the past year. In a separate study using a probability sample of all men in the same bathhouse, 27.5% of men reported having UAI in the 3 months before assessment (Woods and Binson, unpublished manuscript), suggesting that the men accessing VCT in the bathhouse are at least as risky as, if not more risky than, the general population of bathhouse patrons.
Additionally, most men tested in the bathhouse returned for their results (70.5%). Data from the local county health department indicated that during the same period as our evaluation, 88.4% of MSM tested outside the bathhouse setting returned for their results.21 Thus, the return rate in the bathhouse might be lower; however, our return rate was similar to that observed in a separate study of VCT offered in a bathhouse in Seattle, Washington, where 73.7% of men tested returned for their results.11 Half of the men in the bathhouse who tested HIV-positive returned for their results, although given the small numbers of men testing positive, this estimate has a wide confidence interval. Seventy-three percent of MSM testing positive outside the bathhouse setting returned for their results.21 Although the return rates for HIV-positive and HIV-negative MSM tested in the bathhouse seem to be lower than in other test sites, a greater proportion of HIV-positive men were identified in the bathhouse. Among MSM testing at the bathhouse, 3.7% were HIV-positive relative to only 1.6% of MSM testing at other sites.21
We also found initial evidence for the effectiveness of VCT offered in bathhouses. Three months after participation in bathhouse-based VCT, participating MSM were less likely to engage in UAI and reported having less sex under the influence of alcohol or drugs and more communication with sexual partners about HIV status. Decreases in the raw number of unprotected anal sex acts were more modest; the reductions we observed in the number of unprotected anal sex acts attenuated once we added control for demographic variables. Communication about condoms also did not increase after VCT.
Presently, it is unclear why we observed significant changes in the proportion of men who engaged in some UAI but not in the mean frequency of the behavior. Because risk behavior is often found to be more common with steady than with casual partners,22,23 it might be that some men eliminated their occasional UAI with casual partners, whereas others continued to engage in more frequent UAI with steady partners. This dynamic could have obscured our ability to detect a decrease in overall frequency of UAI while still revealing a decrease in the proportion of men engaging in the behavior. It is also unclear why we observed a change in the proportion of men consistently communicating about HIV status but not about condoms. One explanation could result from differences in how certain topics are raised or discussed. Although communication about HIV status requires a conversation, use of condoms can be accomplished with no communication or only nonverbal communication, in which case, it is possible that change in communication about condoms might be harder to detect.
Despite the promise of these findings, they must be qualified by a number of limitations. First, data on the efficacy of VCT were gathered in the context of an uncontrolled nonexperimental design. Thus, we cannot rule out the possibility that other factors (eg, social desirability, regression toward the mean) might have contributed to the positive effects of VCT that we observed. Additionally, we obtained detailed information only on participants' 2 most recent anal or oral sex partners during a 3-month period. An assessment that allowed for reporting on more partners would have provided a more complete picture of the risk in this population. Even assessing only the 2 most recent partners, our sample of men testing in this venue nevertheless reported significant HIV-related risk behavior at baseline, and a more detailed assessment could only have increased the numbers of risk behaviors observed, thereby making an even stronger case for offering VCT in these venues. Finally, given constraints on space at the venue and on our interviewing resources, we were only able to interview a relatively small convenience sample of the men who requested testing in this venue. Thus, our statistical power to detect effects was low, and our findings share the limitations inherent in all research with nonprobability samples.
Despite these limitations, our data do provide evidence that bathhouse-based VCT is a feasible approach to identifying new HIV cases, at least as efficient as VCT conducted in other settings. It also provides preliminary evidence that VCT in this setting might change some risk and precautionary behaviors over a 3-month period. These findings are consistent with those of studies from other nontraditional VCT settings that suggest VCT offered outside clinic settings can reach high-risk individuals, many of whom are at greater risk than those who access HIV testing in clinics.24,25 Our preliminary effectiveness data are also consistent with some studies showing that VCT can lead to modest reductions in risk behaviors, particularly when counseling sessions are tailored to the specific target population or otherwise enhanced relative to “standard” VCT.3 The counseling model used in the present study was tailored to issues likely important to bathhouse patrons, which could explain the promising results. Another explanation could be the unique setting in which counseling occurred. Important cognitive or affective states might be more highly activated when men receive counseling in a sexualized setting relative to a “sterile” clinic setting.
This study documented not only the feasibility of conducting VCT in a bathhouse setting but of conducting research in this context. We were able to recruit a sample of bathhouse patrons participating in VCT for a baseline assessment and were successful in follow-up 3 months later with more than 80% of those men. Thus, the prospect of conducting future research in this setting is promising. One priority in this area is to replicate our preliminary effectiveness data with a randomized controlled trial. Additionally, the emergence of rapid HIV testing could be an important development for offering VCT in bathhouse settings,11,26 and future research should explore how this approach differs from traditional VCT in terms of feasibility and effectiveness. Although rapid testing should certainly improve the return rate for receiving test results, it is possible that the higher risk men would be less willing to test under rapid testing conditions. Both of these possibilities should be examined in future work. Another promising area for future research is to explore the venue-level impact of having VCT on the premises of a bathhouse. Presently, it is unclear how men react, behaviorally and in terms of their patronage, when a bathhouse commits itself to conducting VCT or other HIV prevention efforts on the premises.
Given recent calls by the CDC and other prevention scientists to begin implementing VCT outside traditional clinic and medical settings, the results of the present study are especially timely. VCT offered in a bathhouse is feasible, reaches appreciable numbers of high-risk individuals, and shows preliminary evidence of effectiveness. As further research accumulates that corroborates these findings, disseminating this VCT model to other bathhouse settings should become a priority.
The authors acknowledge Steamworks, the collaborating venue in this research, along with the club's owners and the general manager, without whom conducting the program and evaluation would have been impossible. They thank the club customers as well, especially those who took the time to participate in the program and evaluation. They also recognize the efforts and support of several people at the local health department: the AIDS Director, LeRoy Blea, the coordinator of HIV testing, Amity Balbutin-Burnham, and the testing program manager at the club, Librado Nunez, and his outreach testing team: Tom Tseng, Greg Melhauf, Patrick Bork, Mike Tachet, Jose Luis Tello, Rafael Rodriguez, and Megel McCoy. The evaluation could not have been successful without the exceptional efforts of Bob Siedle-Khan, who was the primary research interviewer and man-on-the scene for the study team. The evaluation at the club was part of a larger evaluation of HIV testing programs in multiple counties, and the authors therefore want to acknowledge the assistance of the other study interviewer, Charles Pearson, the study project director, Thomas Riess, the project assistant, Sherry Fung, and our coinvestigators Karen Vernon and Moher Downing.
1. Weinhardt LS, Carey MP, Johnson BT, et al. Effects of HIV counseling and testing on sexual risk behavior: a meta-analytic review of published research, 1985-1997. Am J Public Health. 2000;90:1152-1153.
2. Dilley JW, Woods WJ, Sabatino J, et al. Changing sexual behavior among gay male repeat testers for HIV: a randomized, controlled trial of a single session intervention. J Acquir Immune Defic Syndr. 2002;30:177-186.
3. Kamb ML, Fishbein M, Douglas JM, et al. Efficacy of risk-reduction counseling to prevent human immunodeficiency virus and sexually transmitted diseases: a randomized controlled trial. Project RESPECT Study Group. JAMA. 1998;280:1161-1167.
4. Centers for Disease Control and Prevention. Advancing HIV prevention: new strategies for a changing epidemic-United States, 2003. MMWR Morb Mortal Wkly Rep. 2003;52:329-332.
5. Buchbinder SP, Vittinghoff E, Heagerty PJ, 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.
6. Valleroy L, MacKellar D, Karon J, et al. HIV prevalence and associated risks in young men who have sex with men. JAMA. 2000;284:198-204.
7. Binson D, Woods WJ, Pollack L, et al. Differential HIV risk in bathhouses and public cruising areas. Am J Public Health. 2001;91:1482-1486.
8. Binson D, Woods WJ, Pollack L, et al. Bridging HIV/STI testing programmes to high risk men. Int J STD AIDS. 2006;16:600-604.
9. Judson FN, Miller KG, Schaffnit TR. Screening for gonorrhea and syphilis in the gay baths-Denver, Colorado. Am J Public Health. 1977;67:740-742.
10. Woods WJ, Binson D, Mayne TJ, et al. HIV/sexually transmitted disease education and prevention in US bathhouse and sex club environments. AIDS. 2000;14:625-626.
11. Spielberg F, Branson BM, Goldbaum GM, et al. Choosing HIV counseling and testing strategies for outreach settings: a randomized trial. J Acquir Immune Defic Syndr. 2005;38:348-355.
12. Conrad FG, Schober MF. Clarifying question meaning in a household telephone survey. Public Opin Q. 2000;64:1-28.
13. Schober MF, Conrad FG. Does conversational interviewing reduce survey measurement error? Public Opin Q. 1997;61:567-602.
14. SAS Institute. SAS/Stat User's Guide, version 9.1. Cary, NC: SAS Institute; 2004.
15. Stata Corporation. Stata Version 8.0 Reference Guide. College Station, TX: Stata Press; 2004.
16. Schafer JL. Analysis of Incomplete Multivariate Data. London, UK: Chapman & Hall; 1997.
17. Collins LM, Schafer JL, Kam C. A comparison of inclusive and restrictive strategies in modern missing data. Psychol Methods. 2001;6:330-351.
18. Hardin JW, Hilbe JM. Generalized Estimating Equations. Boca Raton, FL: Chapman and Hall/CRC; 2003.
19. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; 1989.
20. Vittinghoff E, Glidden DV, Shiboski SC, et al. Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models. New York, NY: Springer; 2005.
21. Berkeley City Health Department. HIV Counseling and Testing Surveillance Data. 2001, Berkely, CA, USA.
22. 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. 2006;10:325-331.
23. Davidovich U, de Wit JB, Stroebe W. Assessing sexual risk behaviour of young gay men in primary relationships: the incorporation of negotiated safety and negotiated safety compliance. AIDS. 2000;14:701-706.
24. DiFranceisco W, Holtgrave DR, Hoxie N, et al. HIV seropositivity rates in outreach-based counseling and testing services: program evaluation. J Acquir Immune Defic Syndr Hum Retrovirol. 1998;19:282-288.
25. Varghese B, Peterman TA. Cost-effectiveness of HIV counseling and testing in US prisons. J Urban Health. 2001;78:304-312.
26. Summers T, Spielberg F, Collins C, et al. Voluntary counseling, testing, and referral for HIV: new technologies, research findings create dynamic opportunities. J Acquir Immune Defic Syndr. 2000;25(Suppl 2):S128-S135.
*Our data revealed that at baseline, only 9 men (6.8% of the sample) reported sex with primary partners (eg, boyfriends, husbands, lovers) of more than 3 months' duration and who the respondent knew to be HIV-negative. Thus, only a small percentage of the sample was engaging in what has been termed negotiated safety. Excluding sex acts with these men did not change the results of the analysis. Cited Here...
© 2006 Lippincott Williams & Wilkins, Inc.