Herpes Simplex Virus 2 (HSV-2) is a chronic, sexually transmitted infection (STI) which is infectious during both its symptomatic and asymptomatic periods [1]. This asymptomatic nature of HSV-2 infection assists in the spread of the infection in the general population [2]. It has been estimated that around 500 million people are currently infected with HSV-2 worldwide and that approximately 20 million new cases occur each year. In European countries, the prevalence of HSV-2 fluctuates significantly between countries, varying from approximately 5% for England and Spain, approximately 20–30% for countries such as Germany and Sweden up to 40% for Turkey [3]. Considerably higher rates of HSV-2 have been observed in sub-Saharan Africa (SSA), with age‐adjusted prevalence in adults ranging from 10 to 50% in men and 30 to 80% in women [4]. Cross sectional studies in SSA have shown a prevalence of 5–53% in 13–24 year old men and women, respectively [5]. In Uganda, men aged 15–19 years old had 10% prevalence while 20–24 year old men had a 27% prevalence. Ugandan women, showed a 35% prevalence among 15–19 year olds versus a 74% prevalence in women aged 20–24 years old [5]. In South Africa, studies have shown a 31% prevalence of HSV-2 in women aged 15–26 [6], while an 84% prevalence was shown among female commercial sex workers in the province of KwaZulu-Natal (KZN) [7]. The Methods for Improving Reproductive Health in Africa (MIRA) diaphragm study showed a 73% HSV-2 prevalence in Durban, with 41% of these women co-infected with Human Immunodeficiency Virus (HIV) [8].

Two decades of observational and biological research supports the theory that HSV-2 increases the risk of HIV-1 acquisition up to four fold [1, 2, 4, 9–11] and increases the risk of HIV-1 transmission and infectiousness [4, 9, 12]. A meta-analysis demonstrated that HSV-2 seropositivity was associated with a HIV acquisition risk ratio of 2.7 in men and 3.1 in women. HSV-2 infection has also been associated with increased incidence of other STIs, such as Neisseria gonorrhoeae (NG), Treponema pallidum (TP) and Trichomonas vaginalis (TV) [11]. This would emphasize the need to test HSV-2 control strategies as a means to reduce the transmission not only of HIV, but also of other STIs [11].

In the KZN province, HIV prevalence among sexually active women attending three clinics was estimated to be 40% [13]. There are many reasons as to why the epidemic is spreading uncontrollably in this region; socio-economic conditions and specific population dynamics (such as patterns of sexual networking, levels of condom use and STI infections), are known to be important determinants in the spread of HIV infection [14]. Poor socioeconomic conditions often result in women engaging in commercial sex activity. Poverty is also known to result in male migrant labour, resulting in men being away from their primary sexual partners for extended periods of time. This often leads to men having multiple sexual partners [14, 15]. In KZN, a high rate of multiple, concurrent and age-discrepant sexual relationships also contribute to the spread of HIV in the region. In this largely patriarchal society, gender inequality and the inability of women to negotiate condom use further contribute to these factors [14, 15]. Given the relationship between HSV-2 and HIV infection, it is important to ascertain the factors that put HIV uninfected women at risk of HSV-2 infection in regions of high HIV endemicity.

The current study estimates the prevalence and risk factors associated with HSV-2 infection among women who tested HIV-1 negative in a cohort from an HIV prevention trial. We also estimated the incidence of HIV-1 and HSV-2 infection among these women during study follow-up. These data provide an opportunity to gain further insight into the HSV-2 epidemic in the KZN region and to inform the design of future HIV/STI prevention strategies.

The Microbicide Development Programme (MDP) 301 study (2005–2009), was a multicenter, international, double-blind, randomized, placebo-controlled trial assessing the safety and efficacy of 0.5 and 2% PRO 2000/5 gels for the prevention of vaginally acquired HIV infection [16, 17]. Women from six research centers (13 sites) in Africa were enrolled. However, since KwaZulu-Natal is considered the epicenter of the HIV epidemic, only women enrolled at the three Durban sites of the Medical Research Council (MRC) of South Africa research center were considered in this analysis. The protocol was approved by the Biomedical Research Ethics Committee based at the University of KwaZulu-Natal, the Medicines Control Council in South Africa and the US Food and Drug Administration. The eligibility criteria included: willingness and ability to give informed consent; HIV-seronegative status at screening; age 18 years or older; non-pregnant and not planning to become pregnant for the duration of the trial; willing to have regular speculum examinations and urinary pregnancy tests; willing to be tested for HIV infection; willingness to use study gel as advised; likely to be sexually active; and willingness to receive counseling about condoms. Women were excluded if they had any severe clinical, laboratory or gynecological abnormalities; had sex >14 times per week, and/or were unlikely to comply with the protocol. Women were followed up every 4 weeks for a total of 52 weeks. Women were provided with HIV testing and counseling, risk reduction education and diagnosis and treatment of STIs. Participants diagnosed with curable STIs were invited to bring their male partners to the study site for treatment or were provided with referral cards to access STI treatment at local clinics. Women received unrestricted supplies of condoms at all study visits and were counseled to use condoms during all sex acts. The primary efficacy endpoint of the MDP301 trial was HIV-1 infection, using a novel HIV confirmatory testing algorithm [18]. Briefly, the algorithm included parallel HIV-1 rapid tests at the clinics, with positive or discordant tests after enrolment prompting Enzyme Immunoassay testing at local laboratories and confirmation at a central laboratory in South Africa. The central laboratory analysed samples from the study visit at which the first positive rapid test result was obtained, in addition to previous study visits at which samples were obtained. At enrolment and at predefined study visits, additional procedures were carried out according to the study schedule. These included a clinical evaluation with genital examination, and the collection of additional specimens for the confirmation of the diagnosis of HIV (Enrolment, week 12, week 24, week 40 and week 52), HSV-2 (Enrolment, week 40 and week 52), syphilis (Enrolment, week 24 and week 52), and other infections [N. gonorrhoeae (NG), C. trachomatis (CT), Trichomonas vaginalis (TV), bacterial vaginosis and candida] at the Enrolment and week 24 visits. HSV-2 was assessed using the HerpeSelect 2 IgG ELISA (FOCUS Diagnostics, California, USA); Focus index values of ≥3.5 were considered HSV-2 seropositive [sensitivity of 92% (95% confidence interval (CI) 82–98) and specificity of 91% (95% CI 79–98)] [19]. Chlamydial and gonococcal infections were assessed using amplicor swabs collected during the genital examination by using nucleic acid amplification assays (COBAS Amplicor, Roche Molecular Diagnostics, Pleasanton, CA, USA). Baseline and confirmatory syphilis testing was conducted on sera using the rapid plasma reagin (RPR) (BD Macrovue) and Treponema palladium haemagglutination (TPHA) (Fujibrio) methods respectively. Testing for TV was conducted using sterile swabs with the In-Pouch-TV test kit [16].

The primary outcome measure of this analysis was HSV-2 seropositivity at baseline and the risk factors associated with HSV-2 seropositivity. In addition, we assessed the risk factors associated with HSV-2 incidence during study follow-up. The following variables were considered: age (in quartiles), education level (secondary school completed vs not completed), number of sex acts in last week (less than 3 vs more than 3), number of lifetime sexual partners (1 partner vs 2 or more partners), employment (yes/no), religion (Christian vs other), age at first sex act (15 and younger vs 16 and over), contraception use [none (reference), hormonal contraception, condoms, other (tubal ligation, rhythm method)] and being infected with another STI (CT, NG, TV or syphilis). Univariate and multivariate logistic regression models were used to identify the correlates of HSV-2 infection at screening. The multivariate model was constructed using a forward stepwise method. We used data driven cut point methods, such as quartiles, to categorize the continuous age variable. Regarding the number of sex acts per week, the data driven method was used in such a way that each level had enough observations; therefore they will not be over- or under-representative of the respective groups. Central tendency measures, such as the median, and dispersion measures [interquartile ranges (IQRs)] were as follows: the median for overall age was 27 (IQR 22–37), with the median for HSV-2 negative women being 23 (IQR 20–29) and the median for HSV-2 positive women being 31 (IQR 23-39) (p value for rank test is <0.001); the median for number of sex acts per week was three for the study population overall (IQR 2–4), with a median of 3 (IQR 2–4) for both the HSV-2 positive and negative populations (p value for rank test was 0.511). Independent predictors of HSV-2 infection were assessed using Cox proportional hazard regression model. A univariate logistic regression was conducted first; risk factors with a p value of <0.10 were considered as candidates for the multivariate model. We used a forward stepwise technique to build the final model. All the factors with a p value of <0.05 were retained in the final multivariate model. Variables were considered statistically significant if p < 0.05. HIV incidence was defined as the time from enrolment to seroconversion and/or censoring at the end of study follow-up on the basis of a discrete time scale determined by a woman’s study visits at 12, 24, 40 and 52 weeks. The time of seroconversion was defined as the time of the initial positive HIV test result. If one or more visit between the last negative HIV tests and the first positive tests were missed, the time of seroconversion was presumed to be the visit comprising the midpoint between these two time points. Similarly, the time to an HSV-2 positive result after enrolment (tested for at weeks 40 and 52) or censoring was used to calculate the HSV-2 incidence rate during study follow-up. All analyses were conducted using STATA 12.0 (College Station, Texas, USA).

A total of 2236 HIV-negative women were screened and enrolled into the MDP 301 trial at clinical research sites in Durban, South Africa. The HSV-2 prevalence at baseline in this cohort of women was 65%. The HIV incidence among HSV-2 negative women was 5.2 per 100 person years (PY), compared to 8.7 per 100 PY for women that were HSV-2 positive at baseline (p < 0.001, log rank test). The HSV-2 incidence during study follow-up was 22.3 per 100 PY (95% CI 20.4–24.3).

Although the HSV-2 prevalence in this study is high, it is similar to previous studies from developing countries [4, 5] and other South African studies [6, 7, 20, 21]. This study was conducted at three urban clinical research sites in Durban. Interestingly, the MIRA study, which was conducted by the MRC at one rural and one semi-rural clinical research site in the Durban area 2 years prior to the MDP 301 trial, showed a HSV-2 prevalence of 73% (median age was 26; IQR 22–34) [8]. This higher prevalence in rural/semi-rural areas of Durban could be explained by several factors. Since women would have less access to jobs in a rural area, they may engage in more transactional sex; rural areas have poorer access to healthcare, including STI treatment and free condoms; and women in rural areas may have had less opportunity for education than those women from urban areas, which may increase their risk of acquiring HSV-2. The HSV-2 incidence during study follow-up is staggering and is much higher than previously seen in other South African studies. The HSV-2 incidence seen in our study compares to HSV-2 studies in Kenya and Uganda among female sex workers [22]. However, another cohort of Kenyan women, who considered themselves to be at high risk of HIV infection but were not sex workers, had a similar incidence of 22.1 per 100 person-years [23], while a study at an STI clinic in Alabama showed women had a HSV-2 incidence of 20.5 per 100 woman-years [24]. Women participating in these HIV prevention trials usually consider themselves at high risk of HIV infection, which could explain this high incidence of HSV-2 due to risky sexual behavior. The HIV incidence seen in this cohort of women was comparable to the HIV incidence among HSV-2 positive women in the MIRA trial [20].

Previous analyses of HSV-2 risk factors have shown that women whose sexual debut was at an age younger than 16 were at higher risk of being HSV-2 positive, however this was not significant in our cohort [8, 25]. Similar to other studies, our results showed that women who had not attended high school were at greater risk of having acquired HSV-2 [26] and this is possibly due to high school attendance influencing the structure of sexual networks [27]. Unemployment also put a woman at risk of being HSV-2 positive, most likely due to unemployed women being dependent on their partner for income, which may result in them being unable to negotiate safe sex with partners. Additionally, unemployment can drive women to engage in transactional sex. Consistent with other studies, the risk of being HSV-2 positive increased with age [25, 28]. This is most likely due to an increased number of sexual partners and increased years of sexual activity with age (both risk factors of HSV-2 infection [25, 26, 28]). Women presenting for this clinical trial in Durban, who were infected with an STI were significantly more likely to be co-infected with HSV-2 than women in the trial that did not have an STI, although syphilis showed the greatest risk of HSV-2 co-infection. We did note some instances of co-infections with STIs. Thirty three women were co-infected with TV and CT, while 31 women were found to be co-infected with NG and CT (data not shown). The relationships between HSV-2 and other STIs were expected, since it is widely known that there is increased incidence of STIs with HSV-2 infection, specifically NG, TV, and T. pallidum infections [11, 25, 28]. However, the same was not observed when considering HSV-2 incidence during study follow-up. This is not surprising, as women would have been treated for their STIs as soon as they were diagnosed, according to the study protocol requirements, which could explain why having an STI is a risk factor prior to study enrolment, but not during the study. The increased risk of HSV-2 infection we experienced in pregnant women is not surprising, since both are as a result of participating in unprotected sex.

Another study looking at HSV-2 in an HIV prevention trial in rural/semi-rural Durban sites found that women older than 25, having less than high school education, being single or cohabiting, being a Christian, young age of sexual debut, more than four lifetime partners, giving birth to more than one child and co-infection with NG put women at an increased risk of being HSV-2 positive [8]. Together with this study, our analysis provides us with a detailed set of risk factors suited to this region that has high HSV-2 and HIV infection.

The use of hormonal contraceptives has recently come under the spotlight for possibly increasing the risk of HIV-1 acquisition. However, our results showed that a woman on hormonal contraception was less likely to be HSV-2 infected at baseline in addition to being at less risk of HSV-2 infection during study follow-up. The link between hormonal contraception and HIV infection is a highly debated one, with the majority of experts agreeing that hormonal contraception does not increase risk of HIV infection, while the use specifically of depot medroxyprogesterone acetate may pose a risk of infection [29, 30]. In fact, some studies have shown a decreased risk of HIV infection with progesterone only pills and combined oral contraception [adjusted hazard ratio (HR_a)  =  0.86, 95% confidence interval (CI) 0.32, 1.78] [31]. Of note with our data is that hormonal contraception includes injectable contraception and oral contraceptives, which have not shown an increased risk in HIV acquisition. One study evaluating the effect of hormonal contraception on risk of HSV-2 infection showed that injectable contraception did not increase this risk [32]. Further studies need to be conducted to confirm these findings.

HSV-2 has emerged as the most prevalent STI, and since it is most often asymptomatic, unrecognized and, most importantly, incurable, it puts individuals at increased risk of acquiring other STIs in addition to HIV. Given the high prevalence and incidence of HSV-2 infection reported here, screening for symptomatic HSV-2 among high risk populations should be incorporated into HIV and STI screening and treatment packages. Women also need to be encouraged to seek treatment for STIs, including symptomatic HSV-2. Young women need to be educated about the risk of HSV-2 infection, its impact on HIV and STI acquisition, and how to prevent HIV and STI acquisition. Despite condoms being an effective method of HIV prevention, they have only been shown to reduce the transmission of HSV-2 by 30% [33]. Understanding the prevalence and risk factors for common causes of ulcerative genital disease in the general population would inform current STI syndromic management and HIV testing strategies in high HIV prevalence regions. Lastly, it is vital that prevention approaches integrate the management of both bacterial and viral STIs for young women in these high prevalence regions.

Several limitations need to be considered when interpreting our results: this data was taken from the responses of women participating in a large, randomized, controlled, HIV prevention trial, and therefore may have limited representativeness. Women participating in these trials often participate because they perceive themselves to be at high risk of HIV infection and therefore may not represent the general population. The trial protocol also limits the sex and age of participants. In addition to this, we have not considered the effects of unmeasured characteristics, such as cultural differences (e.g. polygamy), poverty, commercial sex and multiple or concurrent sex partners in our findings. As this analysis was limited to data that was collected as part of an HIV prevention trial, the independent variables included are limited. No socioeconomic or behavioral data was collected from the male partners of these women. Due to the fact that we could not determine the characteristics of women who did not present for enrollment into this clinical trial, we must be cautious when extrapolating these results to the population as a whole. However, these results provide valuable information about HIV negative women presenting to participate in a HIV prevention trial. These results could also indicate that there is a much higher prevalence of HSV-2 in the HIV-1 positive population in KZN.