Who is sexually active? Using a multi-component sexual activity profile (MSAP) to explore, identify and describe sexually-active high-school students in rural KwaZulu-Natal, South Africa

The analysis uses baseline data collected during a cluster-randomised control trial (CAPRISA 007) that was conducted in a rural part of the uMgungundlovu district of KwaZulu-Natal, South Africa. A secondary data analysis from students attending 14 secondary schools in the district was included in this analysis. Data were analysed from all students in grades 9 and 10 who successfully enrolled. Details of the cohort selection and inclusion criteria are described elsewhere [10, 30].

The CAPRISA 007 study was a two-arm, matched pair, cash incentivised cluster randomised controlled trial that took place in the study schools between 2010 and 2013. The study has been presented and explained in more detail elsewhere [10, 30].

All participants provided informed consent prior to being enrolled. Students ≥18 years provided first-person consent following a literacy and comprehension assessment, whilst students < 18 years, provided assent, while consent was obtained from the parent/guardian. In the event that a parent/guardian was not available, proxy parental consent was obtained from a member of the School Research Support Group (SRSG). Behavioural and demographic data were collected using self-report, privately self-completed, structured questionnaires available in isiZulu and English. Biological measures included students’ HIV test results, HSV-2 results, and urine pregnancy test result. The details of the measures used in the parent study are discussed elsewhere [10]. All participants received referrals as required. All students consented that their data may be used in additional analyses. This study explores how we estimate sexual activity, and students are not personally identified about incongruencies in reported sexual activity. All ethical approvals were granted by the University of KwaZulu-Natal Biomedical Ethics Committee (BF105/010 and BE 523/14).

We aimed to understand how many students were sexually active using self-reported sexual activity versus a combination of variables including biological and self-reported behavioural markers to create a Multicomponent Sexual Activity Profile (MSAP). For the purposes of this study, self-reported sexual activity was defined as any learner who reported to have ever had vaginal, anal or oral sex. For the MSAP, we included biological variables including: HIV test results, HSV-2 test results, pregnancy test results. MSAP also include self-reported behavioural variables: reported STI symptoms (those symptomatic screening questions used in the South African public health care system, including asking about experience of a vaginal/penile/urethral discharge, a genital sore or ulcer or pain on urination), previous pregnancy, and those who had made someone pregnant before. MSAP variables all suggest possible unprotected previous sexual encounters. The purpose was to identify the maximum number of young people who may have had a sexual encounter, however, it is possible that some students were HIV positive by vertical transmission, other modes of transmission (i.e. injecting drug-use) or reported not sexually active if they had engaged in sex through force. We include HIV positive in the definition as an individual may still pose a risk for infecting others if not virally suppressed. The MSAP variables identified students who self-reported as not sexually active, but whose profile suggested possible sexual activity. The MSAP variables can be easily collected and measured within a research setting [31].

Using a staged analysis of the data collected, we constructed a descriptive flow diagram that identified those students who disclosed as sexually-active and those that reported being not sexually active. In order to construct the flow diagram, several steps were followed to ensure we could identify the prevalence of key variables included in the MSAP (See Fig. 1).

The first step was to identify the individuals in the total cohort who had self-reported as either sexually active or not sexually active. The two groups were then split by sex as we expect to see variation in the MSAP by sex, with young women expected to bear a higher burden of each outcome. Descriptive summaries were generated to obtain the overall prevalence of each variable of the MSAP within each. HIV, HSV-2 and pregnancy prevalences were obtained using the outcomes of biological tests for each group, while prevalence data for experiencing an STI symptom and previous pregnancy/previously making someone pregnant were generated from self-report data collected through self-administered questionnaires. The prevalence of the biological markers (HIV, HSV-2, pregnancy) and the self-reported behavioural markers (previous/previously making someone pregnant and STI symptoms) were then calculated for each sub-group, reported as percentages. The prevalence of key MSAP variables within the group identifying as not sexually active was used to indicate sexual activity. The overall flow diagram was used to look at the difference of MSAP variables within each group, as well as highlighting differences between male and female students.

Summary statistics of the basic demographic data were generated for the overall cohort of students enrolled in the study as well as the groups identifying as sexually active or not. These variables were reported using medians for continuous variables and as frequency distributions for categorical data. As the data was collected from a cluster randomised control trial it was necessary to adjust for any cluster effects that may arise from the school-based sampling. The unadjusted analysis did not account for the clustering, calculating prevalence by combining prevalence of all schools (clusters). Using the adjusted prevalence estimates from each of the 14 clusters, a t-test for two independent samples was performed to compare by sex, reported sexual activity, and ecological variable [21, 32] differences between those newly identified vs self-reported as sexually active. As there were few clusters (< 20) in the main study, regression methods were not completed due to concerns over robustness [33].

Overall, more male students reported having had some type of sex before (p = 0.001), within those self-reporting as sexually active more males reported having been sexually active (58.1%) than females (41.9%). The overall prevalence of key variables in those reporting to be sexually active was, HIV prevalence 5.7%, HSV-2 prevalence 11.1%, positive pregnancy test 6.3%, self–reported ever made someone pregnant or previous pregnancy 79.8% and self-reported STI symptoms 14%. The gender difference in the prevalence of MSAP variables persisted, consistent with the gender disparity seen nationally and regionally [6]. HIV prevalence amongst sexually active young women was approximately seven times greater than their male counterparts, with young women’s prevalence 11.4% to the males 1.6% (p < 0.001). HSV-2 prevalence amongst self-reporting sexually active women was significantly higher compared to young men, reporting a prevalence of 21.0% compared to 3.9% in young men (p < 0.001). Amongst the key self-reported behavioural markers of sexual activity, the gender disparity continued, significantly more females reported having had a previous pregnancy, compared to young men who made someone pregnant before (p < 0.001). Previous pregnancy prevalence for those reporting sexually active should be interpreted cautiously as the available data had high missing values. For those reporting symptoms of an STI, the gender disparity was not significantly different, 13.5% of females reporting to have been sexually active had a symptom of an STI as opposed to 14.5% of young men self-reporting STI symptoms. For young women, it appeared that they were engaging in unprotected sex, indicated by the high pregnancy prevalence (6.3%).

Overall, fewer young men reported that they had never had sex before (p = 0.001). For those self-reporting never having engaged in any type of sex before, fewer males (39.3%) reported not having had sex before compared to females (60.7%). For those reporting to have never been sexually active, the overall prevalence of MSAP variables were HIV prevalence 2.8%, HSV-2 prevalence 4.3%, positive on pregnancy test 2.2%, ever having made someone pregnant or previous pregnancy prevalence was 1% and positive for a self-reported STI symptom was 6.5%. The gender disparity amongst the prevalence of the key indicators of sexual activity remained skewed towards females (Fig. 1) in almost all the variables. HIV prevalence amongst females reporting to have never been sexually active was significantly higher (3.9%) than amongst males (1.3%) (p = 0.001). Young women had four times higher HIV prevalence when compared to their male counterparts. Amongst those reporting not being sexually active the difference in HSV-2 prevalence was less pronounced between the genders but still significantly different. In females, the prevalence of HSV-2 was 5.6% compared to 2.3% in males (p = 0.001), with females having two times greater prevalence than males. Evidence of unprotected sexual encounters was again evident amongst young females reporting no sexual activity where there was a 2.2% prevalence of pregnancy. For the key self-reported behavioural markers that could indicate sexual activity, there was no significant difference between the genders. Firstly, the prevalence of ever been/ever made someone pregnant (1.2% of females and 0.5% of males (p = 0.068)), suggest possible evidence of unsafe sexual practices, despite reporting to have never had sex. Secondly, amongst those reporting never having had sex, 5.3% of females reported to having a positive STI screening symptom. Here the gender disparity was reversed, young men had more reported positive STI screening symptoms than females, with 9.3% of young men experiencing at least one STI symptom by self-report, however, the difference was not significant (p = 0.125).

In total, 59.7% of those students that enrolled self-reported as having never had sex before. We identified 14.1% of these students who appeared to have been sexually active because of the presence of at least one sexual health outcome. Of the 223 sexually active individuals newly identified using the MSAP variables, female students were more likely to under-report sexual activity (p < 0.001) (Table 3). In the 223 individuals, 20.2% were identified by looking at HIV in those reporting to as not sexually active, an additional 27.8% were identified by adding HSV-2, 6.7% were added by adding pregnancy results, 3.6% were added by adding ever pregnant/ever made someone pregnant and the final 41.7% were added by adding self-reported STI symptom. Some variables overlapped, and the breakdown of the variables for those newly identified sexually active individuals are presented in Table 3.

In total, MSAP increased the number identified as sexually active from 40.3 to 48.7%. The basic age demographic for the 223 newly identified sexually active students is presented in Table 3.

Characterising those newly identified as sexually-active compared to those who self-reported as sexually active on various ecological level factors highlighted some important differences (Table 4 in the appendix section). At an individual level, those newly-identified as sexually active were more likely to be female (p = < 0.000), 15 years old or younger (p = 0.008), 20-year-old or older (0.040), less likely to perceive being at risk (p = 0.037) or have ever used alcohol (p = < 0.000). At a relational level, newly-identified as sexually active were less likely to report having ever had a boyfriend/girlfriend (p = 0.000) or ever having had more than one boyfriend/girlfriend at the same time (p = 0.000). They were also less likely to have felt pressured to have sex by their peers (p = < 0.000) or partners (p = 0.008), or have friends that were mostly boys (p = 0.000). At a familial level, those newly identified were more likely to be of medium socioeconomic (SES) status (p = 0.037) and less likely to be low SES (p = 0.029). At a school level, those newly-identified as sexually active were less likely to have repeated a grade (p = 0.024) and were more likely to be engaged in social activities (p = 0.032) at a community level.

To further understand whether those newly identified as sexually active perceived that they were at risk of negative health outcomes we used their self-reported perception of future HIV infection as a proxy marker. For those reporting that they felt they were at ‘some risk’ of future HIV infection, significantly more male students reported to feel at some risk than female students (Table 5). This subset of students, while being part of a group that perceived themselves to be a lower risk (Table 4), appeared to correctly assess that they may still be at some risk. When assessing the risk perception between those who experienced a negative health outcome in the self-reported sexually active and newly identified as sexually active (Table 6), the data showed that amongst those newly identified, those with HSV-2, those who had been pregnant before or those who self-reported an STI symptom were more likely to perceive themselves as being at lower risk.