Abstract
Objective:
To evaluate the implementation of community-based voluntary HIV counseling and testing (CBVCT) in the Eastern Caribbean.
Methods:
A total of 9782 unique HIV testing events performed through a national program of CBVCT in Antigua and Barbuda (2009-2012) were analyzed. The authors describe testers’ demographic characteristics and assess demographic (education, housing, marital status, nationality, and age) and sexual exposure (partner gender, transactional sex, and multiple partners) factors associated with testing HIV-positivity and with condom use.
Results:
Older men and men having sex with women and women with higher education, of Antiguan nationality, and having sex with men were less likely to test positive for HIV. Younger, educated, and unmarried men and women with multiple partners were more likely to report using condoms.
Conclusion:
The CBVCT model can be successfully implemented in Eastern Caribbean. Demographic differences persist in HIV testing, risk behavior, and infection among vulnerable populations and should be considered in HIV prevention intervention design.
Introduction
Access to community-based voluntary counseling and testing (CBVCT) for HIV is an important step in stemming the tide of the epidemic 1 –3 but has been limited in Eastern Caribbean. 4 In response to the need for CBVCT, the Ministry of Health (MOH) in the island nation of Antigua and Barbuda (A&B) initiated CBVCT through collaboration with local nongovernmental organizations (NGOs) in the mid-2000s. Since its inception, the initiative has aimed to make HIV prevention education and CBVCT more widely available and accessible in communities most at risk of HIV infection. 5 Community-based VCT was chosen because of stigma and discrimination faced by groups disproportionately impacted by HIV in the region—namely, sex workers and men who have sex with men—particularly when they attempt to access health care services.
The model employed in A&B uses peer and lay outreach workers to provide voluntary counseling and testing (VCT) and, when needed, referrals for other services. 3 Evidenced-based reports have shown that peer-based programs can be highly effective in reaching key populations such as sex workers and men who have sex with men. 6 Peer counselors who are identified as leaders in their communities can act as role models and can also be very effective in encouraging behavior change and increasing access to services. 7,8 Training peers as counselors also provides concrete opportunities for meaningful involvement of key populations in response to HIV and AIDS, in line with internationally endorsed principles of community engagement. 9
In this article, we briefly describe the design and implementation of the CBVCT program in A&B. We provide demographics of men and women receiving HIV tests during the first 4 years of CBVCT implementation, between 2009 and 2012. We also assess and describe associations between testers’ demographic and risk characteristics and the receipt of a positive HIV test result. Finally, we assess and describe associations between these same characteristics and consistent condom use, a proven-effective prevention strategy. 10
Description of the Program
The CBVCT model implemented in A&B aimed to increase access to HIV testing among key populations at higher risk for HIV infection and to examine its feasibility and acceptability as the first peer-based CBVCT program implemented in the Caribbean region. The CBVCT model was implemented under the supervision of the MOH through its National AIDS Program (NAP) and the Mount St John’s Medical Centre National Laboratory.
Six key strategies were used in the development and implementation of the model. The first was training peer outreach workers from key populations as well as lay outreach workers from different organizations as community health workers and counselors to talk to their peers and/or people at high risk of infection about the risks of HIV infection. In A&B, it was emphasized to provide these services in community settings and through partner organizations, as opposed to health care delivery sites. By building on existing trust and on the effectiveness of existing peer outreach strategies, peer and other outreach workers were uniquely positioned to discuss sexuality, safer sex, HIV risk behaviors, harm reduction, the impact of stigma and discrimination, informed consent and confidentiality, and the need for people to seek HIV counseling and testing. Furthermore, because they were already trusted members of the community in which they worked, they were able to leverage existing relationships and encourage testing in nontraditional testing venues.
The second strategy of the NAP/NGO collaboration involved training employees at private laboratories and key local organizations to provide counseling as a linked component with the rapid HIV tests. The third critical strategy was to employ a community-based approach that involved community members as full partners and incorporate their feedback from the design phases of the program throughout the program implementation. Fourth, the approach was decentralized so that more people could be reached including those at high risk of infection by training and employing lay and community health workers not previously engaged in this work and launching mobile testing units in urban and rural settings.
Fifth, connections were established between peer and lay outreach workers and other services and support structures. These community connections ensured access to a full complement of services from peer-based ones to existing preventive and clinical services, including those available at government clinics. These personal relationships and the bidirectional communication they support ensure access to sensitive care and support services for key populations. Finally, the sixth strategy was for outreach workers to use rapid HIV tests, which provide results in 30 minutes from a finger stick. Despite some potential technical challenges, this allows same-day diagnosis and eliminates concerns about people not returning for their test results. These 6 strategies contributed to the successful implementation of CBVCT in A&B.
During initial rollout in 2009, 4 outreach workers were hired and trained by the NAP and were certified to provide rapid testing and counseling, congruent with the certification standards established by the Mount St John’s Medical Centre National Laboratory and in accordance with regional and international recommendations. The outreach workers, who were hired because they were key population members, delivered rapid testing and counseling in the NAP office and in collaborating health clinics. This was an important step in terms of building capacity and skills of the outreach workers and supporting a peer- and community-based rapid testing model to reach key populations, a first in the Eastern Caribbean sub-region. This increased the capacity of community outreach workers to provide a broader range of services including community-based counseling and rapid testing services as well as prevention information and referrals using a one-stop shopping approach. After the initial rollout of testing in the NAP office and clinics, community health workers initiated mobile testing activities undertaken in venues where key populations congregated, for example, at the beach, at sex clubs, and in other high-risk neighborhoods. In 2010, access to testing and counseling was increased when The Caribbean HIV/AIDS Alliance (CHAA) provided a small grant to Antigua Planned Parenthood Association to roll out a peer- and community-based rapid testing and counseling program targeting key populations.
Methods
To examine the uptake and outcomes of the CBVCT program in A&B, we used client intake forms from each test performed. Peer outreach workers used standardized forms to document test takers’ demographic characteristics, HIV-risk behavior, and test results. Data entry of paper form was performed at the MOH using Epi Info (US Centers for Disease Control and Prevention, Atlanta, Georgia, USA). 11 Test takers were uniquely identified by anonymous codes, allowing us to exclude data on repeat test takers in the analysis.
Descriptive Analysis
We employed chi-square and t tests to compare men and women test takers by the following characteristics: education, housing status, condom use, marital status, nationality, age, sexual identity, report of multiple sex partners in the last year, and characteristics of sex partners including their gender, HIV status, and whether they were a sex worker. We also assessed differences in self-reported HIV risk behavior among test takers.
Logistic Regression
We employed logistic regression analyses to assess the strength and direction of associations between demographic factors and HIV risk behaviors and the likelihood of receiving a positive rapid test result. We performed the same analyses to assess the likelihood of any condom use. Because of the demographic differences between men and women, we ran separate models for each. The equations included the same variables in both models: binary variables indicating an age above the overall mean (30 years old), marriage status, Antiguan nationality, report of multiple partners, characteristics of sexual partners (gender, HIV status, and transactional), HIV risk factors (occupational exposure, other sexually transmitted infection, and multiple sex partners) and categorical variables indicating the level of education (primary, secondary, and any college) and housing status (own, rent, shared, and none).
Results
Descriptive Analysis
A total of 9782 unique individuals were tested for HIV through the CBVCT program between January 2009 and December 2012 (4 years). Thirty-two percent of testers were men and 63% were women (Table 1). Chi-square tests revealed statistically significant differences between men and women in demographic characteristics and reported behaviors. Women who tested were more likely than their male counterparts to have finished secondary school—47.8% compared to 44.6% (P = .003) but less likely to own their own home (17.5% of women compared to 20.9% of men, P < .001). Women were more likely than men to be single (74.2% compared to 70.8%, P = .001). Antiguans comprised the majority of the population of both sexes, although 69.5% of men were Antiguan compared to 63.5% of women (P < .001). Other nationalities representing more than 5% of the population were Dominican, Guyanese, and Jamaican. More women than men were from countries outside of A&B, more often holding Jamaican nationality (13% of women versus 7% of men, P < .001). The women in the sample were, on average, 2 years younger than the men (29.3 years old compared to 31.7 years old, P < .001).
Demographics of Individuals Receiving Tests between January 2009 and December 2012.a
Abbreviations: MSP, multiple sex partner; SD, standard deviation; STI, sexually transmitted infections.
aN = 9782.
bNot included in regression models due to issues of collinearity and increased precision.
cNot included in one or both regression models due to small numbers
Men who tested were more likely than women who tested to identify as heterosexual or homosexual (92.0% compared to 86.8%, P < .001, and 1.3% compared to 0.8%, P = .013, respectively), and women were more likely than men to identify as bisexual (6.3% compared to 1.6%, P < .001). Five percent of men who tested reported having had a male sexual partner in the last 12 months, and 7.6% of women testers reported having a sexual partner who was also a woman. Men were more likely than women to report having had a sexual partner who they thought was HIV infected (3.3% compared to 2.0%, P < .001) and who were engaged in sex work (12.3% compared to 1.2%, P < .001).
Men who tested reported a greater number of sexual partners during the 12 months prior to the test: 34.3% reported 1 sex partner, 42.2% reported 2 to 4 partners, 10.5% reported 5 to 9 partners, and 3.7% reported 10 or more partners. Fifty-five percent of women who tested reported a single sex partner in the past 12 months, 32.3% reported 2 to 4 partners, 3.0% reported 5 to 9 partners, and 1.0% reported 10 or more partners. All of these comparisons were statistically significant (P < .001). Men were more likely than women to report “always” using a condom—23.9% compared to 15.2% (P < .001).
Logistic Regression
Separate regressions were used to analyze the predictors of an HIV-positive test and condom use for men and women. Due to issues of collinearity, we dropped 2 variables from our regression analyses. Instead of “sexual orientation,” we used “gender of sex partner” in order to assess associations between testers’ sexual behavior, regardless of their sexual identity and HIV status and condom use. We also chose to include “multiple sex partners” rather than “number of sex partners” because it increased the precision of the model with regard to having had sex with sex workers. We also dropped variables with too few observations to permit robust analysis but describe them in the demographics table for reference because characteristics of populations testing for HIV in the Eastern Caribbean are thus far not well described.
The first logistic regression explores predictors of testing positive for HIV (Table 2). Among men, there were no significant differences in education, housing, condom use, nationality, or HIV risk behavior. However, the odds of testing positive were significantly lower in 2 groups of men: those older than 30 years of age had significantly lower odds of testing positive (odds ratio [OR]: 0.15; 95% confidence interval [CI]: 0.02-0.82, P = .035) as did men who reported sex with women (OR: 0.14; 95% CI: 0.02-0.82, P = .03).
Predictors of Receipt of an HIV-Positive Test Result—Multivariable Analysis.a
Abbreviations: CI, confidence interval; OR, odds ratio.
an = 2799.
b P < .05 in boldface.
Although there were no differences in the odds of testing positive among women with a primary school education and those who had completed secondary school, women who reported having any college were less likely to have received a positive test (OR: 0.11; 95% CI: 0.01-1.04, P = .05). Women who reported living in rental and shared housing were significantly less likely to have received a positive test than women who reported owning their own home (OR: 0.15; 95% CI: 0.04-0.64; P = .01 and OR: 0.16; 95% CI: 0.03-0.74; P = .02, respectively). There were no significant differences in testing positive among married versus unmarried women or among women aged older than 30 relative to their younger counterparts, all else being equal. Antiguan women were less likely than their non-Antiguan counterparts to test positive (OR: 0.14; 95% CI: 0.04-0.46, P < .001).
Women who reported sex with men were significantly less likely than women who reported sex with women to test positive (OR: 0.13; 95% CI: 0.04-0.43, P < .01). There were no differences in positive test results depending on whether women reported having multiple sex partners themselves or having had sex with someone with multiple partners.
The second analysis (Table 3) explored the predictors of reporting any condom use (reporting that condoms are used sometimes, usually, or always). Among men, the odds of reporting any condom use increased as the level of education increased: the OR for any use was 1.62 (95% CI: 1.12-2.36, P < .013) among men who completed secondary school compared to those who had completed only primary school. This rose to 1.94 (95% CI: 1.25-3.00, P < .003) among men who had reported any college. Males who reported living in shared housing were significantly more likely to report any condom use compared to males who reported owning their own home (OR: 2.89; 95% CI: 1.91-4.38, P < .001) as were men who reported renting a house or apartment (OR: 1.56; 95% CI: 1.05-2.30; P < .026). The odds of any condom use were significantly lower in 2 groups of men: married men (OR: 0.52; 95% CI: 0.35-0.77; P < .001) and men older than 30 (OR: 0.57; 95% CI: 0.41, 0.84, P = .003). There were no differences in condom use among men who were of Antiguan nationality and those who were not.
Predictors of any Condom Use—Multivariable Analysis.a
Abbreviations: CI, confidence interval; OR, odds ratio.
an = 3703.
b P < .05 in boldface.
All else being equal, the odds of any condom use was higher among men who reported having sex with men (OR: 3.07; 95% CI: 1.16-8.12, P = .024) and women (OR: 2.14; 95% CI: 1.34-3.42; P = .00) and among men who reported having sex with sex workers (OR: 2.19; 95% CI: 1.06-4.53, P = .034) and men who reported multiple sex partners (OR: 2.76; 95% CI: 1.73-4.40, P < .001).
Among women, there were no differences in the odds of any condom use between women with a primary school education and those who had completed secondary school. However, women who reported having any college were more likely to report any condom use (OR: 1.59; 95% CI: 1.19-2.15, P < .002). Women who reported living in shared housing were significantly more likely to report any condom use compared to women who reported owning their own home (OR: 1.70; 95% CI: 1.31-2.22, P < .001) as were women who reported renting a house or apartment (OR: 1.59; 95% CI: 1.22-2.07, P < .001). Women who reported having multiple sex partners were significantly more likely to report any condom use (OR: 1.54; 95% CI: 1.09-2.17, P < .015). The odds of any condom use were significantly lower in 2 groups of women: married women (OR: 0.37; 95% CI: 0.27-0.49, P < .001) and women older than 30 (OR: 0.61; 95% CI: 0.49-0.76, P < .001).
The odds of any condom use was higher among women who reported having sex with men (OR: 2.60; 95% CI: 1.97-3.44, P < .00) and among women who reported multiple sex partners (OR: 1.54; 95% CI: 1.09-2.17, P < .02). There was no difference in condom use among women reporting transactional sex and those not reporting it, all else being equal.
Discussion
The national program of CBVCT rolled out in A&B has been successful in providing access to most at-risk populations. It has also yielded new information about demographics and risk characteristics of key populations receiving tests for public health providers. These data are important for helping the national program to strategically plan future testing efforts. The example provided by the implementation of the model is also helpful for other countries seeking to undertake similar efforts. While these data do not explicitly reflect key populations at higher risk for HIV infection in A&B, such as commercial sex workers or men who have sex with men, these data show demographic and behavioral characteristics of men and women testing through CBVCT that mirror other descriptions of the populations at risk in the region. For example, there are demographic differences in men and women testing through CBVCT that mirror other descriptions of the populations at risk in the region. Women who tested were more often younger migrants who may be involved in domestic work or the sex work industry, as other studies have indicated. 12 Men who tested were more often married and reported sex with other men also described previously. 13,14
Regardless of the differences in the populations of men and women testing, few of the demographic and sex risk variables remained significant as predictors of either testing positive or using condoms, which may indicate success in educating A&B about reducing risk. Educated men who are sexually active in particular may be getting the message about condom use. The picture is less clear about how demographics relate to HIV positivity, at least in this data set. We remain cautious about conclusions related to HIV positivity among men who have sex with men. In general, there is a dearth of data on this at-risk community, and the difficulty in reaching them due to stigma and discrimination limits conclusions about our data. There are prevalence studies in the field currently, which may shed additional light on prevalence among these men.
The findings presented in this article are not without limitations. First and most importantly, the analyses presented are limited to the data available in the client intake forms used in the CBVCT program and the quality of data generated during client intake sessions. Data quality was emphasized throughout the CBVCT program, in part through training of counselors and testers regarding the importance and practice of recording data during client intake sessions. Second, because of the small number of positives, it is difficult to determine whether testers were more or less likely to be finding out for the first time that they were infected with HIV or whether they were receiving a duplicate positive test. Of the 50 testers testing positive, only 5 indicated that they had been tested before, and we have no way of knowing the results of previous tests. With these small numbers, we are unable to determine the impact of repeat testing.
Nevertheless, we present the findings from the first national data set of CBVCT in the Eastern Caribbean. The data represent a rich and useful source of information about the CBVCT program in A&B and has utility for helping guide future directions for CBVCT. These data provide useful information to program managers and service providers as to whom the CBVCT program has reached thus far. Notably, more testing may be needed in lower socioeconomic population groups. The program definitively demonstrates that testing using community peers and/or lay people can be an effective model for reaching the general population. The model employed in A&B required collaboration by multiple agencies and organizations from the NAP to community-based clinics to civil sector prevention providers. Together, these efforts led to the effective delivery of HIV testing in community settings, a first at this scale in the region. In addition, the data collected as testing were provided to offer valuable information about HIV infection patterns and safer sexual behaviors, which should help guide messaging and targeting of community-based HIV prevention efforts in Eastern Caribbean.
Footnotes
Acknowledgments
The authors thank Conrad Otterness at Community Partners International and Karen Blyth at Intrahealth International Inc. for their support of the work. The authors would like to express special thanks to Mr Oswald Hannays from the Antigua and Barbuda Ministry of Health, AIDS Secretariat, for his role in supervising and assisting the data entry process. Finally, the authors are grateful to the outreach workers at the Caribbean HIV/AIDS Alliance and from the AIDS Secretariat of the Ministry of Health in Antigua and Barbuda for their work in the field and for the staff at the Health Information Division and AIDS Secretariat of the Ministry of Health for data entry. The views expressed in this publication do not necessarily reflect the views of USAID, the CDC, the United States Government, or the Government of Antigua and Barbuda.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors thank the US Agency for International Development (USAID), Cooperative Agreement No. 538-A-00-07-00100-00 and the US Centers for Disease Control and Prevention (CDC), Cooperative Agreement No. 5U2GPS001468, for providing funding for this project under the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR).