Secondary Logo

Journal Logo

Epidemiology and Social Science

The HVTN Protocol 903 Vaccine Preparedness Study

Lessons Learned in Preparation for HIV Vaccine Efficacy Trials

Djomand, Gaston MD, MPH*; Metch, Barbara MS*; Zorrilla, Carmen D MD; Donastorg, Yeycy MD; Casapia, Martin MD§; Villafana, Tonya PhD; Pape, Jean MD; Figueroa, Peter MD#; Hansen, Marianne MPH*; Buchbinder, Susan MD**; Beyrer, Chris MD, MPH†† for the 903 Protocol Team

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: May 1, 2008 - Volume 48 - Issue 1 - p 82-89
doi: 10.1097/QAI.0b013e31817236ab
  • Free


The successful conduct of HIV vaccine efficacy trials requires recruitment, enrollment, and long-term retention of informed, willing, and HIV-negative but at-risk participants.1 Such trials generally depend on HIV infection endpoints, and therefore need to be large enough to ensure interpretable outcomes while providing the highest standard of HIV prevention services, including individualized risk reduction counseling, condom promotion and distribution, and diagnosis and treatment of other sexually transmitted infections (STIs). These scientific and operational challenges, and the considerable resources required to meet them, have led many groups to conduct vaccine preparedness studies (VPSs) to prepare research teams and their partner communities in advance of HIV vaccine efficacy trials.2 VPSs have proven to be of considerable importance to the conduct of efficacy vaccine trials.3 From 2003 to 2007, the HIV Vaccine Trial Network (HVTN) conducted VPSs in a number of countries, collectively titled HVTN Protocol 903.

The objectives of HVTN 903 were tailored to the needs of new and established trial sites seeking to investigate cohort development in new populations. Overall, the study sought to assess site-specific capacity to recruit, enroll, and retain high-risk HIV-negative participants in a standardized protocol, thereby building capacity to conduct trials. The protocol sought to assess behavioral risks over time and willingness to participate in actual vaccine trials in the future. A significant focus for 3 of the 903 sites was the need of the HVTN, and the HIV vaccine field generally, to develop cohorts of women at risk for HIV infection in the Americas. This necessity was driven, in part, by the outcome of the first HIV vaccine efficacy trial conducted, the VaxGen trial of the AIDSVAX (rgp120) candidate, which had insufficient statistical power to assess differential effects of immunization, had there been any, by gender.4

Six sites initially participated in HVTN 903, and each developed cohorts in selected at-risk populations to assess the feasibility of conducting HIV vaccine trials in those groups. The study was not powered to assess HIV incidence. Sites and populations included the following: Iquitos, Peru, with cohort development among men who have sex with men (MSM); Port au Prince, Haiti, with male and female STI clinic attendees; Santo Domingo, the Dominican Republic (DR), with female sex workers (FSWs) and male STI clinic attendees; Gaborone and Jwaneng, Botswana, with male and female police officers and mine workers; and Puerto Rico (PR), with FSW and women with injection drug user (IDU) male sexual partners. A new partner site for the HVTN in Kingston, Jamaica joined HVTN 903 in 2005, enrolling cohorts of FSW and male and female STI clinic attendees.


Study Participants

General eligibility criteria for all participants included being HIV-negative; aged 18 to 35 years; willing to provide adequate information for locator purposes; and, for women, not pregnant at enrollment. Women were eligible if they reported exchange of sex for drugs, money, or services within the past 12 months for Jamaica and PR or within the past month for the DR. MSM were eligible if they reported anal sex with another man in the past 12 months and also had one of the following: last sex contact was unprotected, sex with 5 or more partners, a recent STI, an HIV-positive partner, or exchanged sex for money. Heterosexual men and women were eligible if they had recently been diagnosed with an STI, reported engaging in sex with a partner of the opposite sex who had been diagnosed with a recent STI, or had an event of unprotected vaginal or anal sex with a known HIV-positive partner. Additionally, in PR, women who reported a current male partner who was an IDU were eligible. Men who reported more than 1 sexual partner (Botswana, DR, and Haiti) or sex with an FSW (Jamaica) were also eligible. Recruitment strategies varied by region and included advertising, voluntary counseling and testing at STI centers, street outreach, and referrals from community health services or other study participants.

Study Procedures

HIV testing was done at screening and at 6 and 12 months after enrollment. Consenting participants were tested for syphilis at screening and at month 12 using a rapid plasma reagin (RPR) test with positive results confirmed by a microhemagglutination-Treponema pallidum (MHA-TP) test for Botswana and Peru, venereal disease research laboratory (VDRL) findings and/or fluorescent treponemal antibody absorption (FTA-ABS) test for the DR, and T. pallidum hemagglutination assay (TPHA) for Haiti and PR (commercially available kits were used and varied across sites). Questions about history of STIs over the past 6 months and sexual risk behaviors associated with HIV infection were asked at screening at the 6- and 12-month visits using a standardized questionnaire. For FSW cohorts, questions about sexual behaviors were asked about the prior 30 days. For other cohorts, the time frame was the past 6 months. Willingness to participate in future vaccine trials was assessed for all cohorts at enrollment and month 12 with a question set that included general knowledge and understanding of vaccines, attitudes toward research participation, and awareness of the HIV vaccine effort. Site staff performed locator contacts in person or by telephone at 3 and 9 months after enrollment. Participants were considered lost to follow-up after at least 3 unsuccessful attempts at contact.

Statistical Analysis

Completing study follow-up was defined as attending the last study visit (ie, the 12-month post-HIV testing visit) or becoming HIV infected while on study, the endpoint of interest. Visit retention rates were calculated as the number of participants who completed the visit within the required time window divided by the number of participants who were not HIV infected before the visit. Within site-specific risk cohorts that included 5 or more people who did not complete follow-up, we evaluated factors associated with noncompletion of follow-up. These factors included demographic variables and self-reported sexual risk behaviors. Two-sided Fisher exact tests were used to determine statistical significance for dichotomous variables, and Freeman-Halton tests were used for other categoric variables. Because the numbers not completing follow-up within any site-specific cohort were small, multivariate analyses were not possible.

Comparison of behavioral risk data (number of sexual partners; unprotected vaginal and/or anal sex; and exchange of sex for money, drugs, or services) over time was hampered by the small number of participants enrolled in some site-specific risk cohorts and by missing information from some study visits because of early study termination. Percentages of participants reporting a risk behavior, displayed in graphs or in the text, are based on data from all assessments. Because of the correlation in a participant's responses over time, McNemar tests were used to test for changes between baseline and 6 months as well as at 6 and 12 months for participants who had data from both time periods. McNemar tests were also used to test for differences between baseline and month 12 in willingness to participate in future vaccine trials. P values <0.05 were considered statistically significant.


Overall, 4 cohorts were recruited into the study: MSM, FSWs, high-risk heterosexual men (HRHM), and high-risk heterosexual women (HRHW). Enrollment began at the Peru site in April 2003 and was completed at all sites by March 2006 (the Jamaica site joined the study in 2005). All sites met their target enrollment goals within 6 months (Table 1), except for the 2 sites in Botswana. The site in Jwaneng, Botswana closed after 18 participants were enrolled, and these data are not included in the results. Enrollment rates at all sites varied from an average of 13.4 participants per week in Haiti to 3.5 participants per week in Gaborone.

HVTN 903 Enrollment and Retention Characteristics by Study Population and Sites

Characteristics of enrolled participants varied by site (Tables 2, 3). Participants were representative of the ethnic and racial populations of their geographic region (data not shown). Although participants were classified as high-risk for HIV infection by site-specified criteria, most of them (except for women in PR) did not perceive themselves at risk for HIV infection (see Table 2). Individuals who screened positive for syphilis were specifically recruited in Haiti; thus, the Haitian cohorts had the highest rates of confirmed syphilis (see Table 3). In the DR and Jamaica, HRHM cohorts had the highest rates of self-reported Chlamydia, nonspecific urethritis (NSU) or nongonococcal urethritis (NGU), gonorrhea, and STI symptoms (see Table 3).

HVTN 903 Demographic and Behavioral Characteristics at Baseline by Study Population
HIVTN 903 STIs Reported at Baseline

Study completion rates met the protocol-specified goal of 90% at all sites except for PR (87% completed) and Jamaica (89% completed) (see Table 1). Across the sites, 63 participants did not complete follow-up for the following reasons: loss to follow-up for unknown reasons (46.0%); participant relocated (15.9%); participant missed study visit, including attributable to incarceration (15.9%); death (9.5%); and participant withdrawal (12.7%).

Cohorts with 5 or more participants not completing follow-up were HRHW in Gaborone (n = 6), HRHW in Haiti (n = 6), HRHW and FSWs in PR (n = 5 and n = 8, respectively), and MSM in Peru (n = 15). Statistically significant factors associated with not completing follow-up varied by site and cohort and included the following: sex under the influence of alcohol (P = 0.01) among HRHW in Gaborone; having 10 or more sexual partners (P = 0.004), unprotected oral sex (P = 0.046), and an IDU partner (P = 0.03) among FSWs in PR; and being white compared with mixed race (P = 0.046) among MSM in Peru. For the Haiti HRHW cohort, having more than 1 partner and exchange of sex for money were marginally significant factors (both P = 0.057). No significant factors were identified for the PR HRHW cohort. We explored the relation between risk taking and retention to ascertain whether differential loss to follow-up may have explained the low reporting of risks at study completion.

At each site, 90% or more of participants indicated at enrollment that they would be willing to participate in a future HIV vaccine trial (Table 4). At the 12-month visit, lower response rates were observed for all sites (range: 90.7% to 80.6%), with statistically significant decreases for Haiti, Jamaica, PR, and Peru. Items of concern regarding participation in vaccine trials are shown in Table 4.

HVTN 903 Willingness to Participate in a Future HIV Vaccine Trial and Issues of Concern Regarding Participation

The frequency and pattern over time of self-reported sexual behavioral data varied by site and cohort. The baseline median numbers of sexual partners were highest for the FSW cohorts (6 in the past 30 days for Jamaica and PR and 4 for the DR), followed by the MSM cohort (6 in the past 6 months). The HRHM cohort in the DR reported a median of 3 partners in the past 6 months, and the HRHM cohorts in Haiti, Gaborone, and Jamaica each had a median of 2 partners in the past 6 months. The HRHW cohorts reported the fewest number of partners in the past 6 months, with a median of 2 in Gaborone and 1 in Jamaica, Haiti, and PR.

For the FSW and MSM site-specific cohorts, more than 50% of participants reported a decrease in the number of sexual partners between the baseline and month 6 assessments. In general, the percentages of participants reporting cohort-defined high numbers of sexual partners decreased over time (Fig. 1). Statistically significant differences were observed between baseline and 6 months for FSWs in PR, HRHW in Gaborone, HRHM in the DR, and MSM in Peru. Only the Peru MSM cohort had a significant decrease between 6 and 12 months.

HVTN 903: Percentages of self-reported number of sexual partners by assessment time based on all risk assessment data. For female sex worker (FSW), percent ≥10 partners within the last 30 days; for men who have sex with men (MSM) cohorts, percent ≥10 partners within last 6 months; for high-risk heterosexual women (HRHW), percent ≥2 partners within last 6 months; and for high-risk heterosexual men (HRHM), percent ≥4 partners within last 6 months.

Except for the Jamaican FSW cohort, percentages reporting 1 or more episodes of unprotected anal or vaginal sex declined between baseline and the month 6 assessment (Fig. 2). McNemar tests were statistically significant for FSWs in PR and DR, HRHW in Haiti, HRHM in Haiti and the DR, and MSM in Peru. Again, only the MSM cohort had a significant decline between 6 and 12 months.

HVTN 903: Percentages of any self-reported unprotected vaginal or anal sex by assessment time based on all risk assessment data.

Aside from the FSW cohorts, exchange of sex for money, drugs, or services was most frequently reported in the MSM cohort (44%, 36%, and 32% for the 3 assessment times), with a statistically significant change between baseline and 6 months. The 3 Caribbean HRHM cohorts had baseline reporting rates of 26% in the DR, 14% in Haiti, and 20% in Jamaica, with no statistically significant changes over time. For the Caribbean HRHW, baseline rates were 7% in the Haiti cohort, 2% in the Jamaica cohort, and 24% in the PR cohort, with no significant changes over time.

Because of the small size of the cohorts and limited follow-up time of 12 months, the study did not have sufficient power to estimate HIV-1 incidence reliably. Infections occurred in Gaborone (n = 2), the DR (n = 1), and Peru (n = 6). The HIV seroincidence rates per 100 person-years and 95% confidence intervals (CIs) for these sites were 2.3 (95% CI: 0.3 to 8.2) in Gaborone, 0.5 (95% CI: 0 to 2.9) in the DR, and 3.1 (95% CI: 1.1 to 6.8) in Peru.


The study found consistent declines in risk behaviors across cohorts through the 12-month study period: numbers of sexual partners and percentages of unprotected sexual acts declined across almost all cohorts and sites (see Figs. 1, 2), although at 12 months, more than 40% of participants in all cohorts still reported unprotected anal or vaginal sex. This suggests that the HIV counseling and individualized risk reduction focus of our prevention program was effective for many participants over the study period, and it continues to be provided to all HIV vaccine trial participants in our network. Counseling may not be as effective in the context of a vaccine trial, however, which has the potential for behavioral disinhibition among participants who believe they received the vaccine regimen. The study also found a decline in willingness to participate in trials. Stated willingness to participate in trials may be an imperfect surrogate of actual trial participation, however, as indicated in previous reports.5

A key focus for this study was on recruitment and retention of high-risk women in the Americas. Identification of cohorts of women with sufficient risk for HIV infection to support the seroincidence demands of trials but among whom high rates of retention can be achieved has challenged the vaccine, microbicide, and behavioral intervention arenas. We were able to enroll and retain women at risk in a number of the HVTN 903 cohorts, including FSWs in PR, the DR, and Jamaica and HRHW in Haiti and the other sites in the Americas. Across these cohorts of women, however, HIV infection rates were too low to be accurately measured by our relatively small samples. Several findings of this study helped to inform recruitment of women for the recently suspended phase 2b proof-of-concept study of the Merck adenovirus serotype 5 HIV-1 gag/pol/nef vaccine (STEP study). First, we found that a past history of an STI did not predict current risks: 21% of women with confirmed syphilis who were recruited through STI clinic attendance were not sexually active (data not shown) at the time of recruitment. Inclusion criteria for the phase 2b trial specified current, in addition to past, sexual activity risks. Among sex workers, 12-month follow-ups may be too short to measure HIV incidence. Several FSWs recruited into HVTN 903 subsequently were found to have become HIV infected after study completion in routine clinic settings (data not shown); hence, it is possible that for FSWs, 18- to 24-month periods may more accurately capture risk dynamics, although this has not been substantiated in other sex worker cohorts. Finally, it is encouraging to note that HIV infections were uncommon among this sample of adults in Haiti and the DR. Data from Haitian national surveillance and other studies have shown that a decline in HIV prevalence, and presumably in incidence, has been underway in Haiti for the last several years.6,7 Although we did not collect national trend data for this study, it is a truism that any sample from a population in which disease rates are declining is likely to show decline as well.8-10

Retention rate findings from HVTN 903 were also encouraging, because HIV-1 vaccine trials generally have prespecified windows of time in which visits need to be completed and injections given. As shown in Table 1, most sites had >90% retention for both visits. Gaborone and PR had low retention for the 6-month visit; however, after identifying problems with their retention strategies, they made significant improvements in 12-month retention. In both cases, clinic-based strategies seemed to be too passive to ensure high rates of return, and active outreach to participants in person or by telephone was required. For PR, a key change in strategy to improve retention was to increase street outreach efforts to contact participants directly. This required retraining of clinic staff in working with hard-to-reach women and in protecting confidentiality while accessing women in challenging environments, including street-based sex trade venues and informal housing areas for homeless women.

From the vaccine trial feasibility perspective, the cohort established among MSM in Iquitos, Peru was the most successful among those in HVTN 903. Recruitment was rapid, retention at 1 year was high, and the HIV seroincidence of 3.1 per 100 person-years (95% CI: 1.1 to 6.8) among these men was sufficient for measures of vaccine efficacy. The success of this pilot project seems to be a function of several key factors. First, and most obviously, an epidemic of HIV-1 among MSM is clearly underway in this population, and this outbreak has been well documented at other sites in Peru, including Lima.11-13 Second, the research team in Iquitos had a history of successful research studies among the MSM community before conducting HVTN 903, has strong links to the MSM community, has highly motivated and engaged outreach workers from the community, and was truly integrated with the community. This made for effective recruitment of young men who met the recruitment criteria and were actively engaging in unprotected intercourse with multiple partners. The HVTN 903 prevention efforts for MSM in the study were able to show statistically significant declines in numbers of partners and unprotected sex acts; however, despite these declines, at 12 months, the median number of partners was 3 and some 77% of men continued to report unprotected receptive anal intercourse, sustaining the high HIV infection rates in the cohort (see Fig. 2). HIV vaccines and other prevention tools, including pre-exposure prophylaxis (PrEP) approaches and, eventually, rectal microbicides, seem to be an urgent priority for this group of young men. Encouragingly, an HIV PrEP efficacy trial began in 2007 among MSM in Peru.

Finally, HVTN 903 helped to develop clinical trial site capacity and expertise to mount large-scale HIV vaccine efficacy trials, including the current STEP study, which enrolled participants at an average rate of 35 per week across all sites. The implementation of a strong community education program has been essential in achieving community engagement and community mobilization for efficacy trials. The STEP study partnered with the HVTN 903 sites in Haiti, the DR, Jamaica, and PR and with many other HVTN sites not included in HVTN Protocol 903. Lessons learned from this study were instrumental in developing successful strategies, strengthening community engagement and support, for the successful recruitment of the STEP study and, it is hoped, for future vaccine efficacy trials.


The authors thank Joseph Chiu, Alan Fix, and Paul Sato for their contributions as medical officers for the study and Katie Skibinski for editorial assistance.


1. Excler JL, Beyrer C. Human immunodeficiency virus vaccine development in developing countries: are efficacy trials feasible? J Hum Virol. 2000;3:193-214.
2. Seage GR III, Holte SE, Metzger D, et al. Are US populations appropriate for trials of human immunodeficiency virus vaccine? The HIVNET Vaccine Preparedness Study. Am J Epidemiol. 2001;153:619-627.
3. Vanichseni S, Tappero JW, Pitisuttithum P, et al. Recruitment, screening and characteristics of injection drug users participating in the AIDSVAX B/E HIV vaccine trial, Bangkok, Thailand. AIDS. 2004;18:311-316.
4. Francis DP, Heyward WL, Popovic V, et al. Candidate HIV/AIDS vaccines: lessons learned from the world's first phase III efficacy trials. AIDS. 2003;17:147-156.
5. Buchbinder SP, Metch B, Holte SE, et al. Determinants of enrollment in a preventive HIV vaccine trial: hypothetical versus actual willingness and barriers to participation. J Acquir Immune Defic Syndr. 2004;36:604-612.
6. Gaillard EM, Boulos LM, Andre Cayemittes MP, et al. Understanding the reasons for decline of HIV prevalence in Haiti. Sex Transm Infect. 2006;82 (Suppl 1):i14-i20.
7. Hallett TB, Aberle-Grasse J, Bello G, et al. Declines in HIV prevalence can be associated with changing sexual behaviour in Uganda, urban Kenya, Zimbabwe, and urban Haiti. Sex Transm Infect. 2006;82(Suppl 1):i1-i8.
8. Kerrigan D, Moreno L, Rosario S, et al. Environmental-structural interventions to reduce HIV/STI risk among female sex workers in the DR. Am J Public Health. 2006;96:120-125.
9. Kerrigan D, Ellen JM, Moreno L, et al. Environmental-structural factors significantly associated with consistent condom use among female sex workers in the DR. AIDS. 2003;17:415-423.
10. Mishra V, Vaessen M, Boerma JT, et al. HIV testing in national population-based surveys: experience from the Demographic and Health Surveys. Bull World Health Organ. 2006;84:537-545.
11. Lama JR, Lucchetti A, Suarez L, et al. Association of herpes simplex virus type 2 infection and syphilis with human immunodeficiency virus infection among men who have sex with men in Peru. J Infect Dis. 2006;194:1459-1466.
12. Corey L, Coombs RW. The natural history of HIV infection: implications for the assessment of antiretroviral therapy. Clin Infect Dis. 1993;16 (Suppl 1):S2-S6.
13. Sanchez J, Lama JR, Kusunoki L, et al. HIV-1, sexually transmitted infections, and sexual behavior trends among men who have sex with men in Lima, Peru. J Acquir Immune Defic Syndr. 2007;44:578-585.

high-risk sexual behavior; HIV vaccine preparedness

© 2008 Lippincott Williams & Wilkins, Inc.