HIV Prevention in High-Risk Women in South Africa: Condom Use and the Need for Change

Introduction Young women are at disproportionate risk of HIV infection in South Africa. Understanding risk behaviors and factors associated with ability to negotiate safe sex and condom use is likely to be key in curbing the spread of HIV. Traditionally prevention efforts have focused on creating behavioral changes by increasing knowledge about HIV/AIDS. Methods This was a cross-sectional analysis from a prospective observational cohort study of 245 women at a high-risk of HIV infection in KwaZulu-Natal, South Africa. Results Participants demonstrated a high level of HIV/AIDS knowledge. Overall, 60.3% of participants reported condom use. Reported condom use at last sexual encounter varied slightly by partner type (57.0% with steady versus 64.4% with casual partners), and self-perceived ability to choose to use a condom was significantly lower with steady partners compared to casual partners (p<0.01). In multivariate analysis, women who had high school education were more likely to use condoms at their last sex encounter compared to those with only primary school education (RR of 1.36 (95% Confidence Interval (CI) 1.06–1.75) and 1.46 (95% CI 1.13–1.88) for grades 8–10 and 11–12, respectively). Those who used condoms as a contraceptive method were twice as likely to use condoms compared to women who did not report using them as a contraceptive method. Greater perceived ability to choose to use condoms was associated with higher self-reported condom use at last encounter, irrespective of partner type (RR = 2.65 (95% CI 2.15–32.5). Discussion Self-perceived ability to use condoms, level of formal education and condom use as a contraceptive were all significantly associated with self-reported condom use at last sexual encounter. These findings suggest that that gender inequality and access to formal education, as opposed to lack of HIV/AIDS knowledge, prevent safer sexual practices in South Africa.


Introduction
In the South African province of KwaZulu-Natal, an estimated 25.8% of all persons aged 15 to 49 years are HIV-positive. This prevalence rate is one of the highest in South Africa, a country that carries the highest number of HIV-infected persons in the world [1,2]. HIV is predominantly spread via heterosexual sex, with young women being the country's most vulnerable population [1,3].
Traditionally prevention efforts have focused on creating behavioral changes (i.e. safer sexual practices) within the general population by increasing knowledge about HIV/AIDS. Studies in the United States and other developed countries have demonstrated successful reduction in HIV infection rates with educa-tional campaigns, while South Africa has continued to experience a rise in the prevalence of HIV despite increased public knowledge and awareness of HIV/AIDS [1,4,5]. Recent reductions in HIV incidence in South Africa have instead been attributed to increased access to HIV treatment [6]. Additionally, several studies have found that improved access to information about HIV risk is not translating into an effective behavioral change in women at high-risk of infection [1,7,8,9,10,11,12,13,14]. Researchers have investigated the possible causes of this disconnect and one theory is that behavioral changed is impeded by the 'social context' of this community [15].
Reported level of condom use has been used as way of identifying and understanding barriers to behavioral change.
This investigation, as part of a larger behavioral study, focused on women at high-risk of contracting HIV/AIDS. It aimed to examine the impact of HIV/AIDS knowledge, as well as related sexual practices on self-reported condom use, and the influence these factors have on HIV acquisition in a high-risk population.

Methods
This study was a cross-sectional analysis of baseline behavioral data gathered from women enrolled into a prospective observational cohort study performed at the Centre for the AIDS Programme of Research in South Africa (CAPRISA), in Durban, KwaZulu-Natal Province, South Africa. This investigation, known as the CAPRISA 002 Acute Infection Study, began in 2004 to research the natural history of acute HIV infection and associated risk behaviors in a cohort of high-risk women, 78.8% of who were self-identified sex workers.
CAPRISA 002 study enrolled women . = 18 years of age who self-reported sex with more than three different partners in the previous three months, and who were HIV antibody negative on screening with rapid tests (first with the Determine test (Abbott Laboratories, Tokyo, Japan) and, if positive, confirmed with a Capillus test (Trinity Biotech, Jamestown, NY, USA)). Discordant rapid HIV tests were resolved with a confirmatory HIV enzyme immunoassay (BEP 2000; Dade Behring, Marburg, Germany). More detailed information regarding CAPRISA 002 study methodology, including recruitment methods and eligibility criteria, is described elsewhere [29].
Prior to study initiation, the protocol and informed consent forms were reviewed and approved by the research ethics committees of the University of KwaZulu-Natal (E013/04), the University of Cape Town (025/2004), and the University of the Witwatersrand (MM040202). Consent forms and information to patients were available in English and were translated into isiZulu, the local language. Written informed consent was obtained at each phase of the study, including collecting the behavioral risk data and to store the study specimens.
Participants completed a structured, in-person interview in either English or isiZulu (per participant preference) to identify their HIV risk behaviors. Upon enrollment, demographic, locator information and laboratory samples were obtained from each participant who received HIV pre-and post-test counseling, HIV and sexually transmitted infection (STI) risk reduction counseling, a behavioral risk assessment as well as a routine clinical evaluation. During the administration of the HIV/AIDS knowledge questions, participants were not prompted, but were asked to state their responses to the questionnaire administrator who ticked off the appropriate category according to the participant's response. Responses were then labeled with a participant identification number and entered into a data collection system. Each participant was paid for her participation in the study as a means of reimbursement for their time and travel expenses.
Descriptive and statistical analysis was performed using SAS software version 9.2 (SAS Institute Inc., Cary). Unadjusted and adjusted risk ratios, measuring the associations between the factors of interest, were calculated using generalized estimating equation regression models with the log link function. Variables in the unadjusted model with a p-value less than 0.20, or variables deemed as important to adjust for, were fitted to the multivariate model. The McNemar test was used to assess differences in perceived choice of condom use between participants' casual and steady partners.

Results
A total of 245 participants were enrolled in the study, the basic characteristics of whom are presented in Table 1. This was a cohort of high-risk women; 85.4% reported having 2-5 casual partners in the 3 months prior to enrollment and 78.8% selfidentified as sex workers. Overall, 58.8% of women reported condom use at their last sexual encounter.
All participants reported that they had heard of HIV/AIDS at the time of study enrollment and, overall, participants demonstrated a good level of knowledge of HIV disease. These data are presented in Table 2.
Participants were knowledgeable about how HIV/AIDS is spread; almost all participants listed vaginal sex as a risk factor, but considerably smaller proportion listed anal or oral sex. When asked about the risk of contracting HIV via anal and oral versus penovaginal sex, slightly more than a third of participants answered that anal and oral sex carried more risk compared to vaginal sex. Similarly, there was high level of awareness how to prevent acquisition or transmission of HIV. Almost all of the participants listed condoms as an effective method, while only 39% and 30% indicated abstinence and staying faithful to one partner as an effective HIV prevention method. Even though treatment was not widely available at the time of the interviews, the majority of participants knew that HIV could not be cured, while a similar proportion agreed that it could be treated.
Sexual behavior data is presented in Table 3, and indicate that participants were much more likely to have had sex with a steady partner than a casual partner (defined as one seen only occasionally, or even once, that was not a sex work client) at last encounter (75.9% steady versus 24.1% casual). In the three months prior to enrollment, over 85% of participants reported two to five casual partners. With respect to total lifetime partners, 54.1% reported more than five steady partners and 76.1% reported more than five casual partners. Participants reported that they engaged in penovaginal sex an average of 10 times a month, with low levels of oral and anal sex. Approximately 80% of participants reported that they used contraceptives, with 60.3% listing condoms as a method of contraception. Importantly, none of these baseline characteristics or sexual practices were found to be statistically associated with subsequent HIV acquisition. While frequency of contraceptive use was not assessed, condom use at the most recent sexual encounter varied depending on partner type (see Table 4). For example, 57.0% of participants     Several factors were associated with condom use at last sexual encounter. For example, self-perceived ability to choose to use condoms was associated with self-reported use both in univariate and multivariate analyses (see Table 5). In the multivariate model, women who believed they had a high degree of control over condom use (able to choose always or over half the time whether or not to use condoms) were more than twice as likely to have used a condom than those who reported never having the ability to choose or who felt they could only insist on using a condom less than half the time (RR 2.21, 95% CI 1.79-2.72).
Education level was significantly associated with condom use. In multivariate analysis, women with Grade 8 to 10 education were approximately 40.0% more likely to report using condoms compared to women with primary school education (p = 0.02); and those with Grade 11 to 12 education were about 50.0% more likely to report using condoms (p,0.01). Women who reported having no partner (single, divorced, widowed or separated) were 1.39 times more likely (95% CI 1.01-1.90) to have used a condom at their last sex act compared to women who were in a stable relationship or married (p = 0.04).
In univariate analysis, age was significantly associated with condom use, with older women being less likely to use condoms. It was no longer significant in the multivariate model, however, most likely because schooling was closely associated with age, with older women being less educated. Age was also associated with perceived choice or insistence on condom use; younger women reported being able to insist on condom use more often than older women. For every five year decrease in age, women were 1.03 times more likely (95% CI 1.01-1.05) to insist on condom use (p = 0.01), adjusting for education level.
Women who reported using condoms as a method of contraception were twice as likely to have used a condom at their last sex act compared to women who did not use condoms as contraception (RR 2.02, 95%CI 1.53-2.66, p,0.01). However, women who said they were using hormonal methods as a contraception method were no more likely to have used a condom at their last sex act compared to women who were not on hormonal contraception (p = 0.51).

Discussion
This study provides information about the behavioral characteristics, level of HIV knowledge and predictors of condom use in a population of high-risk women in KwaZulu-Natal, South Africa. Specifically, this study illustrates that knowledge of HIV and HIV transmission does not correlate with high condom use in this population. These behavioral risk data are also supported by high STI and HIV infection rates in this cohort over time [29]. Previous studies also illustrated that lack of HIV specific knowledge does not explain low rates of condom use [7,8,11,12]. Of note, there was lower awareness of STI and HIV transmission risk with oral and anal sex, suggesting that this would be an important opportunity for more targeted education.
In our cohort, an important finding is the higher reported efficacy of condom use choice at last sexual encounter with casual rather than steady partners. Similar observations were made in other studies from South Africa and the proposed explanations include concern about perceived lack of trust and threat to intimacy and commitment between steady partners [12,13,14,15,30]. This highlights the fact that negotiating condom use with steady partners is challenging as it may represent infidelity or, conversely, that nonuse of condoms may signal trust. Ironically this means that steady sexual partnerships may represent greater risk of HIV infection, for either partner.
In our cohort, several factors were significantly associated with reported condom use during last sexual encounter. We found an association between education level and increased frequency of reported condom use, a result consistent with other studies that show an association between schooling and reduced levels of HIV risk, particularly in young women [31]. Our results also suggests that using condom for contraceptive purpose seems to be a more acceptable motivation, and women are more likely to use condoms in their sex acts if this is being done as the primary means of contraception. It remains to be seen whether or not use of hormonal contraception in this social context would undermine condom use.
Importantly, we noted that women who perceived that they are able to make a choice about using a condom were twice as likely to report using one. It is therefore important to focus on female empowerment in sexual relationships. Previous literature suggests that encouraging self-agency and endorsing egalitarian gender roles in sexual relationships may have a positive impact on rates of condom use [22]. Similarly, a recent preliminary analysis from a randomized trial in South Africa has shown that interventions to empower a woman's ability to negotiate condom use can lead to an increase in self-reported condom use at the last sex act [32,33]. Negotiating condom use for women in our cohort is further complicated by the social and cultural context in which these relationships are developed. For although, it has been shown that women have positive attitudes toward condom use [11], South African men are traditionally the decision makers, especially regarding issues of sexual practice [12,15,24,26,27,34,35]. Therefore, despite high levels of knowledge about HIV transmission and prevention, some women are not able to protect themselves from infection.
A strength of this study is the prospective design, with a relatively large cohort, and the high response rate with almost all participants responding to each question. These findings are thus important as they represent unique data for a large cohort of HIV negative women at high risk for HIV infection, who had a broad range of risk factors, including both sex work and non-sex work partners. We believe these data contribute to greater understanding of the sexual networks and risk behaviors in this specific context. Furthermore, this level of data is particularly important in designing culturally relevant risk reduction counseling messages in the context of, for example, HIV prevention studies. A limitation of the study is that the HIV knowledge and behavioral risk questionnaires were only administered at baseline in the HIV negative cohort, limiting the analysis to cross-sectional data. For example, since data about factors associated with condom use are collected at a single time point, it is not possible to determine causality; prospective study would be needed to determine whether factors such as greater access to formal education would empower the woman to negotiate for condom use. Another notable weakness, shared by much behavioral research, is that the data are self-reported and therefore subject to reporting and recall bias. Attempts were made to minimize reporting bias by making the questions neutral and assuring participants of confidentiality.

Conclusions
Women at high-risk of HIV infection were found to have had a high level of knowledge about HIV acquisition and prevention of transmission. This indicates that education campaigns have reached women in need of this information. However, importantly, knowledge about the high risk of transmission during anal sex was lacking. Future HIV/AIDS educational campaigns therefore need to include messages regarding the risk of transmission via anal sex.
The self-perceived ability to choose to use condoms is significantly associated with reported condom use at last encounter, and furthers the argument that social barriers, particularly gender power imbalances in relation to sexual decision-making, as opposed to lack of knowledge, may be influential factors preventing safer sexual practices in South Africa. This information highlights the importance of female empowerment in prevention efforts. In particular, and supporting this empowerment hypothesis, our data show that women who attained higher levels of formal education, and those who choose to use condoms for contraception, are more likely to report that they use condoms during sex. Our findings, from a unique cohort of women at high risk for HIV infection with a range of partner types, suggest that focusing on ensuring higher levels of formal education in women and de-stigmatizing condom use, for example by emphasizing condom use as contraception, particularly among men who may view condom use for STI prevention as an indication of infidelity or a lack of trust, are essential components of HIV prevention efforts and promoting safer sexual practices in this part of the world.