Secondary Logo

Journal Logo

Epidemiology and Social Science

Brief Cognitive Counseling With HIV Testing To Reduce Sexual Risk Among Men Who Have Sex With Men

Results From a Randomized Controlled Trial Using Paraprofessional Counselors

Dilley, James W MD*; Woods, William J PhD; Loeb, Lisa MPH*; Nelson, Kimberly BA*; Sheon, Nicolas PhD; Mullan, Joseph PhD; Adler, Barbara LMFT*; Chen, Sanny MHS§; McFarland, Willi MD, PhD§

JAIDS Journal of Acquired Immune Deficiency Syndromes: April 15, 2007 - Volume 44 - Issue 5 - p 569-577
doi: 10.1097/QAI.0b013e318033ffbd
  • Free


The ability of counseling associated with voluntary HIV testing to effect behavior change has been the subject of numerous studies. A review article from 1991 suggested that a positive overall reduction in high-risk sexual or drug use behavior was found after counseling and testing but concluded that because of a range of methodologic difficulties, the question could not be fully answered.1 An ambitious meta-analysis in 1999 of 27 different studies concluded that “HIV counseling and testing appears to provide an effective means of secondary prevention for HIV positive individuals, [though it] is not an effective primary prevention strategy for uninfected participants.”2 The authors focused on reductions of high-risk sexual behavior (reduced frequency of unprotected intercourse and increased condom use) as the measure of success and suggested the need to develop and examine further the effectiveness of specific counseling approaches. It should also be noted that 23 of the 27 studies reviewed by Weinhardt et al2 were published before 1993, the year that Centers for Disease Control and Prevention (CDC) published guidelines for client-centered counseling; thus, further studies of the effectiveness of the client-centered model are warranted.3

In 1997, we developed a new approach to counseling associated with testing to address the needs of high-risk men who have sex with men (MSM) who had tested multiple times for HIV yet persistently reported unprotected anal intercourse (UAI) with nonprimary partners. The new intervention, based on cognitive theory and earlier work by Gold and colleagues,4-6 targeted the “self-justifications” (thoughts, attitudes, or beliefs) that persons employ when deciding to engage in high-risk behavior. This “personalized cognitive counseling” (PCC) intervention was conducted by trained mental health professionals in a single counseling session during the week between standard pre- and posttest HIV test counseling sessions.7 The goal was to identify the specific thoughts, attitudes, or beliefs used by the client “in the heat of the moment” when he decided to engage in UAI, re-evaluate those ideas in the “cold light of day” in the company of an empathic counselor, and discuss alternative ways to think about and approach future sexual situations. Our clinical trial found that the PCC intervention was effective: at the 6- and 12-month follow-up visits, participants who received PCC reported significantly fewer episodes of UAI with nonprimary non-HIV-seroconcordant partners than controls who received CDC-defined client-centered counseling or usual care (UC).7 Controls also demonstrated a small decline in high-risk behavior at 6 months, but this effect dissipated by 12 months.

Although these results were encouraging, the projected cost of employing mental health professionals to provide the PCC intervention was thought to be unsustainable in the “real world.” As such, we conducted and report here on a randomized controlled trial designed to assess the effectiveness of the PCC intervention when conducted by trained paraprofessional antibody test counselors during the pretest counseling session compared with UC (HIV risk reduction counseling as described by the CDC8).


Overall Study Design

We conducted a randomized controlled trial between October 2002 and September 2005 at publicly funded HIV counseling and testing venues in San Francisco. The primary study aims were to determine (1) whether paraprofessional counselors could learn and conduct the PCC intervention according to protocol, (2) whether this cognitively focused intervention would remain effective in reducing future episodes of UAI when provided by paraprofessionals and be more effective than UC, and (3) whether client satisfaction would be at least as high in the experimental condition versus the control condition. Study participation consisted of 3 visits. At visit 1, baseline risk behavioral assessment, counseling intervention, an HIV test, and multiple tests for sexually transmitted infections were conducted; visit 2 occurred 6 months later and consisted of the assessment with an optional HIV test. Visit 3 occurred 12 months after baseline and included the assessment and exit HIV and sexually transmitted disease (STD) tests. To assess the participants' experience of the intervention, a standardized client satisfaction questionnaire with a client's study code was mailed to the client 1 week after visit 1 with a self-addressed stamped envelope for return.


Study participants were MSM who presented for anonymous HIV counseling and testing between October 2002 and September 2004. Each was screened for eligibility when he telephoned to schedule an appointment or presented at a drop-in clinic, and eligible testers were scheduled to receive study enrollment and data collection at the same visit as their HIV test. Eligible participants were men 18 years of age or older with a history of 1 or more previous HIV-negative antibody tests at least 6 months before enrollment and at least 1 episode of UAI (receptive or insertive) in the prior 12 months with a nonconcordant (unknown or known positive HIV serostatus) and nonprimary (neither a husband, domestic partner, nor boyfriend for more than 3 months) male partner. Persons who reported nonprescription intravenous drug use during the prior 12 months were excluded. The study interviewer obtained the participant's written informed consent when he presented for testing, administered the baseline interview, and retrieved the random assignment to one of 2 study arms. Participants were paid $5 for completing the client satisfaction questionnaire, $40 for completing the baseline interview, $45 for completing the 6-month interview, and $50 for completing the 12-month interview.


Participants completed the baseline study interviews at the HIV testing site just before their pretest counseling session, with follow-up study interviews at 6 months and 12 months after baseline. All 3 interviews included assessments of HIV testing history, history of sexual behavior, history of STDs, and drug/alcohol history (Addiction Severity Index)9,10 using audio computer-assisted self-interview (ACASI) technology.11 Unless otherwise specified, the time frame for retrospective behavioral questions was the preceding 90 days.

For the 2 most recent sex partners, participants reported the number of anal sex episodes, number of condom-protected anal sex episodes, partner's HIV testing history, partner's HIV serostatus, method of establishing the partner's HIV serostatus (direct [eg, “he told me”] vs. indirect [eg, “he looked healthy, so he must be negative”]), and relationship type (“committed relationship” vs. various other nonprimary relationship types). From these measures, we calculated the number of episodes of UAI in the prior 90 days with nonprimary HIV-nonseroconcordant (discordant or of unknown HIV serostatus) male partners. This calculated measure was selected as the primary outcome because it most closely corresponded to the focus of the intervention (ie, preventing events in which the subject knowingly places himself at elevated risk for HIV acquisition). The primary outcome for the efficacy hypothesis was defined a priori as a change in the number of episodes of UAI with nonprimary nonconcordant male partners in the prior 90 days as measured via self-report in the ACASI interview. Episodes of UAI with partners from a “committed relationship” were excluded, because we found in our previous study that the intervention was ineffective in addressing these behaviors.7

To compare the acceptability and utility of the 2 interventions from the clients' perspective, we used the client satisfaction survey, a standardized instrument with excellent psychometric properties.12 This self-administered questionnaire asks the participant to rate his satisfaction with the session and his attitude about the utility of the session.


Randomization was conducted using a “blocked randomization” scheme. Four opaque envelopes were prepared: 2 with the UC assignment inside and 2 with the PCC assignment inside. The envelopes were shuffled and placed in a green box. Group assignment was retrieved by study staff at the site while the participant was completing the ACASI interview by selecting the first envelope in the box, and the group assignment was recorded in the chart. The used envelope was retired into a red “discard” box. Once all 4 envelopes had been retired, they were shuffled and placed back into the green box. Blinding was used for participants (ie, they were not informed of their group assignment at any time) and data analysts. On-site study staff (recruiters and counselors) were masked until group assignment was made and were not masked after that time (required for scheduling), but they were not involved in data collection or outcome assessment.


All participants received standard counseling associated with HIV testing according to the usual practices of the testing center and in accord with CDC guidelines.13 The CDC recommends use of “client-centered counseling,” which involves active listening, respect for the client's concerns, and an assessment of the participant's “general” sexual activities and willingness to change based on the stages of change theory to help clients identify the behaviors and circumstances that put them at increased risk for HIV transmission.14 HIV knowledge, risk behavior, and substance use are assessed, and a realistic and incremental plan for reducing risk is negotiated. Persons in the UC group received no further intervention. The UC intervention lasted approximately 30 minutes.

All study participants received their HIV test results and posttest counseling in the same manner, regardless of study arm (ie, by standard clinic protocol). They were scheduled to return 7 days after the pretest counseling session, when they were assigned by order of arrival at the clinic to the next available posttest counselor. Posttest sessions at our clinics last 15 to 20 minutes for HIV-negative results and 1 hour for HIV-positive results (on average). For HIV-negative test results (persons with HIV-positive test results were not included in the present analysis), the posttest counseling protocol includes an assessment of readiness to receive the test result, disclosure of the test result, assessment of risk during the 1-week waiting period, and negotiation of an achievable risk reduction goal for the client.

Persons in the PCC group, in addition to the standard counseling, received the experimental intervention for an average of 50 total minutes of counselor contact. On joining the participant, the counselor asked the participant to complete a revised “self-justifications” questionnaire (SJQ-R).15 To do so, the participant was instructed to recall a recent episode of UAI with a male partner of unknown or known HIV-positive serostatus. After bringing a specific episode to mind, the participant completed the SJQ-R, noting whether and to what extent any of the items from the questionnaire were present in his mind just before or during the sexual encounter. The revised instrument, derived from our experience in our earlier study,15 presented a list of 33 possible self-justifications for having anal intercourse without a condom. For each self-justification, the participant rated how strongly the thought had occurred to him using a Likert scale, where 1 indicates “I had this thought strongly (in the forefront of my mind),” 2 indicates “I had this thought to a moderate degree,” 3 indicates “I had this thought slightly (in the back of my mind),” and 4 indicates “I didn't have this thought at all.” The response “can't remember at all” was also offered for each self-justification. Examples of self-justifications include “I want to have unprotected sex because it feels good”; “he looks healthy, so he must be HIV-negative”; and “a condom will mean (I) can't get an erection, and the sex will be spoiled.”

After completing the SJQ-R, the counselor asked the participant to narrate as fully as possible the events that led up to his episode of high-risk sexual activity using the following guidelines:

  • Before start of the episode: describe events, activities, mood states, and other descriptors that characterized his state of mind before the high-risk episode.
  • At the start of the episode: describe where the participant met his partner, his wishes about specific sexual activity, and how strong his sexual attraction was for this particular partner, for example.
  • At the start of sexual activity: describe details about his mood, desires, and how sexually aroused he was and what influence alcohol or drugs may have had in the moment. Communication between the partners about HIV or specific sexual activities was assessed.
  • During sexual activity: describe communication, specifically about condom use, body language, the use of drugs/alcohol, and to estimate the level of risk involved in the episode.

On completing the narrative, the counselor and participant discussed any identified self-justification(s). The participant was asked to reflect on his view of the self-justifications he employed during the episode and how these thoughts may have played a role in the participant's decision to engage in UAI. The counselor helped the participant to confront these ideas and work toward a plan to address these in the future.

The paraprofessional counselors who conducted the PCC intervention were bachelor's level-trained and California-certified HIV test counselors with a minimum of 1 year of HIV test counseling experience. Before the start of the study, they received 4 hours of didactic training on the principles of cognitive behavioral interventions and instruction on implementing the PCC, completed 4 supervised role plays of PCC sessions, and reviewed audiotapes of those role plays with the investigators to refine their technique. During the study, they received regular supervision by one of the investigators, and audiotapes of the sessions were reviewed for adherence to protocol.

Data Management and Statistical Methods

A sample size of 300 subjects was selected a priori to detect a 25% decrease in the number of episodes of UAI based on a data completion rate of 80%, power of 80%, and α = 0.05.

Data were collected using Questionnaire Development System 2.0 (NOVA Research Systems, Bethesda, MD)11 and were analyzed using STATA 8.0 (StataCorp, College Station, TX).16 All electronic data were password protected and stored on a secure server. The impact of the counseling interventions was assessed as “intent to treat” by comparing the episodes of UAI with a nonprimary HIV-nonseroconcordant partner between men randomized to PCC and those randomized to UC using random-intercept Poisson regression. Random-intercept Poisson regression models assess the effect of group assignment (PCC vs. UC) while accounting for the dependence of repeated measures on individuals over time. This analytic approach assesses the individual's change in risk behavior rather than the difference in average risk between the 2 arms at follow-up. In other words, the effect of the intervention is measured by how much a person reduces his risk rather than by how much the control group reduced overall risk on average.

We compared the difference in the outcome between PCC and UC at baseline, 6 months, and 12 months. We also compared the change from baseline to 6 months and from 6 to 12 months within each study arm. To adjust for incomplete data, we used the “offset” option for the generalized linear latent and mixed model (gllamm) command in STATA, which specifies that a variable to be added to the linear predictor without estimating a corresponding regression coefficient. A likelihood ratio test was used to assess the fit of the model. Comparisons of baseline characteristics between PCC and UC participants were done using the χ2 test for proportions or the t test for means.

Human Subjects Protections

All participants gave written informed consent before enrollment. The study protocol was approved and monitored by the Institutional Review Board of the University of California at San Francisco and by a Data and Safety Monitoring Board consisting of 4 volunteers (an ethicist, an epidemiologist, a person living with HIV, and a physician specializing in HIV). The trial was registered on the Web site ( [NCT00218699]) after data collection began but within the first week of the availability of the trial registration system and before data analysis was begun.


Of the 7087 telephone calls during the study period, 587 (8%) were from persons initially determined to be eligible for the study (Fig. 1; diagram of flow of participants through each stage). The remaining 6500 persons were ineligible because they failed to meet 1 or more of the study criteria (most [92%] because they did not self-report UAI in the prior 12 months). Of the 587 eligible men, 336 (57.2%) consented to participate. Of the 336 enrollees, 31 were later excluded (10 who reported UAI at screening but denied UAI at interview, 19 who reported HIV-negative serostatus at screening but tested HIV-positive at baseline, 1 who was deemed unable to give informed consent, and 1 who participated in the entire protocol twice [his second course was excluded]), yielding a final sample of 305 men (147 were randomized to PCC and 158 to UC). Those who were eligible but declined study participation did not differ significantly from those who agreed to participate in terms of race, age, or reported HIV serostatus.

Trial participant recruitment, randomization, and retention.

There were no significant differences between groups on any demographic characteristics (Table 1) at baseline. Overall, mean age was 35.5 years, 35.7% of the cohort were men of color (including 7.5% African American and 11.8% Latino), and the modal annual household income was between $30,000 and $59,999. Although 21.7% had no more than a high school degree, 76.0% had at least 1 advanced degree. Risk for HIV acquisition was high: the mean number of male anal sex partners in the previous 90 days was 4.6 (range: 0-70), and 22.6% of the sample had been diagnosed with an STD in the previous 12 months (Table 2 shows the history of STDs). Risk for acquiring HIV (Table 3) was acknowledged by 79.3%, who agreed or strongly agreed with the statement “given my behavior, I could get infected with HIV,” and desire to change this risk was also evident, because 54.1% reported considerable or extreme desire to change their risk. These men were highly experienced testers; the mean number of prior HIV tests was 9.6 (range 1-40).

Demographic Characteristics of Trial Participants by Group Assignment
Self-Reported History of STDs by Group Assignment
Behavioral Characteristics of Trial Participants by Group Assignment

Complete follow-up achieved for UC participants (91.8%) was higher than for PCC participants (83.1%) (P = 0.033). There were no differences in age (P = 0.86), income (P = 0.93), and education (P = 0.55) between intervention and control subjects lost to follow-up. We also compared mean episodes of UAI at baseline for those who were ultimately lost to follow-up by 12 months with those who were retained. For controls, there was no difference in mean episodes of UAI (4.46 for those lost to follow-up vs. 4.82 for those retained; P = 0.87). For intervention, there was no significant difference, but it was borderline higher for those lost to follow-up (7.29 for those lost to follow-up vs. 3.61 for those retained; P = 0.068).

PCC and UC participants reported comparable levels of high-risk sex at baseline (mean number of episodes: 4.2 vs. 4.8, respectively; P = 0.151; Fig. 2). At 6 months, the mean number of episodes of high-risk sex decreased to 1.9 (P < 0.001) for the PCC intervention participants, a level significantly lower than for the UC controls (mean number of episodes = 4.3 at 6 months; P = 0.029 for difference to intervention; P = 0.069 for change in controls over baseline). The reduction in mean number of episodes of high-risk sex among the PCC intervention participants was sustained at 1.9 from 6 to 12 months (P = 0.181), whereas the mean number of episodes significantly decreased to 2.2 among the UC control participants (P < 0.001) during the same interval, a level not significantly different from the 12-month PCC intervention level (P = 0.756).

Mean episodes of UAI with a nonprimary partner of unknown HIV serostatus or known discordant serostatus in the preceding 90 days, intervention versus control counseling, in MSM in San Francisco, 2002 to 2004.

Regarding risk behavior variables (Table 3), the randomization produced comparable groups with the notable exceptions of amphetamine and cocaine use at baseline. The differences persisted at 6 and 12 months of follow-up. To assess the potential impact of these imbalances on the effectiveness of the intervention, we repeated the primary analysis removing men reporting use of these drugs at baseline. In doing so, the prevention effect of the intervention was slightly but not substantially lower (by 9% with respect to amphetamine use and 4% with respect to cocaine use).

Client Satisfaction Questionnaire Results

The standardized client satisfaction questionnaire12 was mailed to all 305 participants and was returned by 230 (75.4%). Results are presented in Table 4. The return rate did not differ by study group (74.8% for PCC participants and 75.9% for UC participants; P = 0.821). PCC participants were significantly more likely than UC participants to rate the quality of service they received as “excellent” (69.1% vs. 54.2%, respectively; P = 0.022), to rate their perception of their counselor's competence as “high” (58.2% vs. 39.2%; P = 0.005), or to rate their overall satisfaction as “very satisfied” (78.2% vs. 59.2%; P = 0.003). More than one quarter (25.8%) of the UC group reported that the problems that led them to take an HIV test (presumably risk for HIV infection, suspected exposure to HIV, or other sexual risk issues) remained “unchanged” after receiving the UC counseling, whereas only 9.1% of the PCC group did so (P = 0.001). From the clients' perspective, the PCC counseling seemed to be more useful and effective in addressing the participants' presenting concerns. Included in the standardized client satisfaction questionnaire is an assessment of stress experienced by the participant as a result of the counseling session. None of the participants in either group reported “considerable stress” (0 of 110 PCC participants and 0 of 120 UC participants), and reports of any stress at all (some or moderate) were comparable for the 2 groups (41 [37.3%] of 110 PCC participants and 47 [39.2%] of 120 UC participants; P = 0.136).

Client Satisfaction Survey Results by Group Assignment


This study demonstrated that our PCC approach could be taught to and successfully executed by experienced paraprofessional HIV antibody test counselors. Further, when compared in a randomized controlled trial with usual client-centered risk reduction counseling, the approach had a stronger and more immediate effect at reducing the incidence of UAI among high-risk repeat testing MSM. Additionally, the effect was persistent: the sharp decrease in risk behavior among the PCC group from baseline to 6 months was sustained at 12 months after the intervention, a finding consistent with the long-term effects of the same intervention when conducted by licensed mental health professionals.7 Moreover, the approach seems highly acceptable to this key behavioral risk population.

The finding of decreased risk among control subjects from 6 to 12 months is puzzling. We considered 5 possible explanations. First, we were aware that a new counseling approach entailing review of a narrative of a recent risk episode (a key component of the PCC intervention) began to be applied by some counselors providing UC at the study sites based on the success of the well-known Project RESPECT.17 We were aware that some clients received nonscheduled HIV tests during the interval between visit 1 and visit 3, and we hypothesized that this new counseling approach, if received by control subjects at these nonscheduled HIV tests, may have caused “cross-contamination” in our study. To test this hypothesis, we identified all control subjects who reported receiving any HIV counseling and testing between their baseline and 12-month visits (n = 45) and conservatively reclassified any such controls as having received the PCC intervention. We then reanalyzed the main outcome. The reduction in risk among the control group from 6 to 12 months did not substantially change, suggesting that this type of cross-contamination was not responsible for the effect.

Second, we wondered if the number of times that a participant tested for HIV between study visits may have affected the outcome. We hypothesized that repeatedly interviewing participants on the details of their sexual behavior may have resulted in their reporting fewer episodes of risky sex over time because of social desirability response bias or a real effect. To test this hypothesis, we assigned a score for each participant representing the total number of counseling and testing sessions received during study participation (range: 1-4) and tested for a dose-response type of effect on risk behavior. Results failed to support this hypothesis: in fact, we found that as the number of counseling and testing sessions increased, risk did as well.

Third, we considered whether outliers could account for the apparent intervention effect or the drop in risk among controls from 6 to 12 months. Repeating the primary analysis while removing men with extremely high levels of risk (ie, >20 episodes of potentially discordant UAI) did not alter the magnitude or significance of the principal findings.

Fourth, we wondered whether the differential loss to follow-up for intervention versus control might have affected the drop in risk among controls from 6 to 12 months or affected the principal outcome. Assessing the impact of loss to follow-up on interpretation of the outcome, we compared mean episodes of UAI at baseline for those ultimately lost to follow-up by 12 months with mean episodes of UAI of those who were retained. For controls, there was no difference in mean episodes of UAI (4.46 for those lost to follow-up vs. 4.82 for those retained; P = 0.87). For the intervention group, there was no significant difference, but mean episodes of UAI at baseline was borderline higher for those lost to follow-up (7.29 for those lost to follow-up vs. 3.61 for those retained; P = 0.068). This does raise the issue that some of the decrease in episodes of UAI in the intervention arm may be attributable to loss of the highest risk subjects. This does not necessarily undermine the evidence of an intervention effect, however, for 2 reasons: (1) it could be that those who are at highest risk are also those who respond best to the intervention, and the data could thus underestimate the effect, and (2) the analysis adjusts for the individual person's level of risk, so it is the person's change in risk (not just the group differences) that also matters.

A fifth possibility is that the finding may be explained by “regression to the mean.” In other words, our study may have recruited men in a period of higher than usual risk behavior. To the extent that when men entered the study, they were a engaging in more risk than usual (eg, the reasons for seeking testing), it might be expected that they would return to a lower risk baseline in the long term. Evidence that this is the case may be in the finding that men in the intervention arm reduced their risk in the period immediately after the session, whereas those in the control arm did so in a longer time frame. Evidence against this is that we did not see this effect in our previous trial.7

It is important to note that the primary outcome variable in the present study (number of episodes of UAI in the past 90 days with a nonprimary partner) is more specific than that in the first study (any UAI in the past 90 days with a nonprimary partner) and is limited to up to 2 nonprimary partners during the recall period. This was done as a result of the evidence collected during the first study: the median number of sex partners in the prior 90 days for those study participants was 2.7 For the current study, we wanted to collect in-depth data regarding sex partners and encounters (eg, insertive vs. receptive position, serostatus and race/ethnicity, use of alcohol or other drugs during sex) and decided to limit the number of partners reported on to minimize the burden on the participants and potential recall bias. Ultimately, limiting the recall and reporting to the past 2 partners should result in a valid representation of the past 90 days, because few participants reported more than 2 nonprimary partners in that period. On balance, we believe the specificity we gained with this limitation was preferable.

Several limitations of the study narrow the generalizability of our findings. First, to date, this approach has only been tested by our group. Ideally, other research teams at other sites would demonstrate the effectiveness of the intervention. Second, the participants in the PCC group received more counselor contact time (50 vs. 30 minutes), and we have no way of telling if this time difference in attention may have played a role in the eventual outcome. Third, persons who reported nonprescription injection drug use during the prior 12 months were excluded based on the assumption that the physical imperatives associated with drug use may lead to cognitive dynamics that might not be susceptible to the intervention. Fourth, the study was limited to MSM only; it is possible that the self-justifications that contribute to sexual risk taking among non-MSM populations would also be amenable to this approach, but we elected to limit our study population to that which represents most new HIV cases in San Francisco.

The imbalance between study arms with respect to use of amphetamines and cocaine in the previous 30 days could introduce bias but only if the intervention works differently among drug users versus nonusers (because our analysis is based on the individual's change in behavior rather than on the group differences). The sensitivity analysis we conducted suggests that this was not the case. Finally, some other unknown bias may have been present in our sample that could have influenced the outcome; still, our use of randomization may have balanced such biases across study arms, such that the likelihood of compromise of our primary results should be minimal.

Despite these limitations and the puzzling results of the control group results from 6 to 12 months, we believe that when compared with standard client-centered counseling, this new approach (PCC) has shown itself to be superior when working with repeat testing MSM who continue to report high-risk sexual activity. The rapid and steep decline in unprotected anal sex from baseline to 6 months and the persistent reduced numbers of such episodes at 12 months argue strongly that the intervention was having some real effect. Further, the client satisfaction data suggest that the participants actually preferred this approach and that significantly more of those who received the new intervention thought it had been helpful in addressing the problems that led to their testing episode.

What is it about this novel single-session intervention that produced future behavior change? One possibility is that for these men who had repeatedly tested for HIV (mean number of previous tests was 9.6), an important aspect was that the intervention itself was different and, with its emphasis on a recent specific high-risk exposure, perhaps was experienced as more “personally relevant” than standard counseling associated with testing.18 It may also be that this approach, which focuses on the specific thoughts, attitudes, and beliefs employed by these men at the time they decided to have unsafe sex, offered them an opportunity to discover something about their own decision-making process that was previously unknown to them. Finally, by recounting an experience in great detail and “reliving” this experience in the company of an empathic counselor, the participant may have had an affective experience that was memorable, and thus more readily retrievable, when faced with future potential high-risk exposures. The memory of the counseling experience may have provided him an opportunity to recall the discomfort associated with a high-risk experience and eventually to exercise greater control over his decision making.

This study and several recent studies conducted in STD clinic populations have demonstrated the usefulness of counseling in achieving sexual risk reduction,19-23 despite initial review papers that showed equivocal support for counseling associated with testing.1,2 This newer information is particularly important, given the recent decision by the CDC that counseling no longer be a required component of HIV testing,24 and is data that should be considered in future discussions about the merits of this approach. Repeat testers remain at particularly high risk for seroconversion, and our intervention has now been shown in 2 different studies to reduce their risk behavior favorably.25 Considering the limited resources available for prevention efforts with this high-risk population, a cost-effectiveness analysis comparing this effective intervention with other interventions is the next logical extension in this area.

By its nature, counseling is a fluid interaction, and it is easy for counseling protocols to “slip.” To be done well, counseling requires thought and intention on the part of the counselor. Supervision is necessary to help counselors remain on task and on target. Taken together, these studies suggest that if applied thoughtfully and under a supervised protocol, even short-term counseling delivered by paraprofessionals can achieve meaningful behavior change in high-risk repeat testing populations, and our new counseling approach seems to offer an effective and viable alternative to standard counseling associated with testing.


The authors sincerely thank all the clients of the AIDS Health Project who volunteered their time and shared personal information with us during this study. They are also indebted for the professional contributions of Joanna Rinaldi, Nancy Bomse, Shannon Casey, Kerry Gentile, Meghan Geary, Michelle Yu, and Sandy Schwarcz.


1. Higgins DL, Galavotti C, O'Reilly KR, et al. Evidence for the effects of HIV antibody counseling and testing on risk behaviors. JAMA. 1991;266:2419-2429.
2. 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. 1999;89:1397-1405.
3. Kamb ML, Peterman TA, Wolitski RJ. Prevention counseling for HIV-negative persons. Am J Public Health. 2000;90:1152-1153.
4. Gold R, Skinner M, Grant P, et al. Situational factors and thought processes associated with unprotected intercourse in gay men. Psychol Health. 1991;5:259-278.
5. Gold R, Skinner M, Grant P, et al. Situational factors and thought processes associated with unprotected intercourse in young gay men. AIDS. 1992;6:1021-1030.
6. Gold R, Skinner M, Ross M. Unprotected anal intercourse in HIV-infected and non-HIV-infected gay men. J Sex Res. 1994;1:59-77.
7. 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.
8. Centers for Disease Control and Prevention. Revised guidelines for HIV counseling, testing, and referral. MMWR Recomm Rep. 2001;50(RR-19):1-57 [quiz: CE51-19a51-CE56-19a51].
9. McLellan AT, Luborsky L, Cacciola J, et al. Guide to the Addiction Severity Index: Background, Administration, and Field Testing Results. Washington, DC: Government Printing Office; 1985.
10. McLellan AT, Luborsky L, Woody GE, et al. An improved diagnostic evaluation instrument for substance abuse patients. The Addiction Severity Index. J Nerv Ment Dis. 1980;168:26-33.
11. Questionnaire Development System [computer program]. Version 2.0. Bethesda, MD: NOVA Research Systems, 1998.
12. Attkisson CC, Zwick R. The client satisfaction questionnaire. Psychometric properties and correlations with service utilization and psychotherapy outcome. Eval Program Plann. 1982;5:233-237.
13. Centers for Disease Control and Prevention. Technical guidance on HIV counseling. MMWR Recomm Rep. 1993;42(RR-2):11-17.
14. UCSF AIDS Health Project. Building Quality HIV Prevention Counseling Skills: The Basic I Training, A Training Curriculum for Counselors Working in the Context of HIV Counseling and Testing. 4th ed. San Francisco, CA: University of California at San Francisco; 2000.
15. Dilley J, McFarland W, Woods W, et al. Thoughts associated with unprotected anal intercourse among men at high risk in San Francisco 1997-1999. Psychol Health. 2002;17:235-246.
16. STATA [computer program]. Version 8.0. College Station, TX: StataCorp, 2003.
17. Kamb ML, Fishbein M, Douglas JM Jr, 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.
18. Gold RS, Rosenthal DA. Examining self-justifications for unsafe sex as a technique of AIDS education: the importance of personal relevance. Int J STD AIDS. 1998;9:208-213.
19. Boyer CB, Barrett DC, Peterman TA, et al. Sexually transmitted disease (STD) and HIV risk in heterosexual adults attending a public STD clinic: evaluation of a randomized controlled behavioral risk-reduction intervention trial. AIDS. 1997;11:359-367.
20. DiFranceisco W, Pinkerton SD, Dyatlov RV, et al. Evidence of a brief surge in safer sex practices after HIV testing among a sample of high-risk men and women. J Acquir Immune Defic Syndr. 2005;39:606-612.
21. Huebner DM, Binson D, Woods WJ, et al. Bathhouse-based voluntary counseling and testing is feasible and shows preliminary evidence of effectiveness. J Acquir Immune Defic Syndr. 2006;43:239-246.
22. O'Donnell CR, O'Donnell L, San Doval A, et al. Reductions in STD infections subsequent to an STD clinic visit. Using video-based patient education to supplement provider interactions. Sex Transm Dis. 1998;25:161-168.
23. Shain RN, Piper JM, Holden AE, et al. Prevention of gonorrhea and Chlamydia through behavioral intervention: results of a two-year controlled randomized trial in minority women. Sex Transm Dis. 2004;31:401-408.
24. Branson BM, Handsfield HH, Lampe MA, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep. 2006;55(RR-14):1-17 [quiz: CE11-CE14].
25. MacKellar DA, Valleroy LA, Secura GM, et al. Repeat HIV testing, risk behaviors, and HIV seroconversion among young men who have sex with men: a call to monitor and improve the practice of prevention. J Acquir Immune Defic Syndr. 2002;29:76-85.

cognitive behavioral therapy; CONSORT; HIV prevention counseling; men who have sex with men; paraprofessional; personalized cognitive counseling; voluntary counseling and testing

© 2007 Lippincott Williams & Wilkins, Inc.