Prevalence, type, and correlates of trauma exposure among adolescent men and women in Soweto, South Africa: implications for HIV prevention

We used cross-sectional survey data from adolescents (aged 14–19 years) enrolled in the Botsha Bophelo Adolescent Health Study (BBAHS) in Soweto, South Africa. Soweto is a large township southwest of Johannesburg with a population of approximately 1.3 million predominantly (98.5%) black inhabitants residing in informal and formal settlements [32]. While there are no population-level statistics on HIV prevalence among adolescents in Soweto, a recent study of 11,552 adolescents and young adults (14–25 years) residing in Soweto, reported that 4% of those who accessed HIV testing services at a local youth-centered clinic tested positive for HIV, including 2% of young men 4% of young women [33].

BBAHS was conducted at the Perinatal Health Research Unit (PHRU) and the Kganya Motsha Adolescent Centre (KMAC) in Soweto, South Africa. KMAC was opened in 2008 with a local mandate to address HIV and sexual and reproductive health priorities of adolescents (ages 14–19 years). Earlier pilot studies on adolescent health identified the urgent need for such youth-centred services, and informed the development and implementation of BBAHS [33‐36].

Adolescents aged 14–19 years residing in Soweto were eligible to participate in BBAHS. Participants were recruited from across 41 townships to be representative of adolescents living in formal and informal communities within Soweto. Participant recruitment occurred around local malls, schools, neighbourhood hangouts, through peer-word-of-mouth, and staff outreach. We used a targeted stratified sampling and recruitment approach, based on geographic location, age, and gender. In order to reflect the gendered dimensions of HIV risk in South Africa, we aimed for a sample comprised of 60% young women and 40% young men. The research team approached interested adolescents for participation, and if eligible, were enrolled in the study. A total of 956 interviews were completed between March 2010 and March 2012. This amount of recruitment time was required to meet stratified sampling targets, and to ensure inclusion of youth from more remotely located townships with Soweto and harder-to-reach youth sub-populations. Of 956 completed interviews, n = 126 were excluded as they were determined to be outside of the targeted age criteria or had incomplete data, yielding a final sample of 830 adolescent participants. Additional information about the study procedures of the BBAHS can be found elsewhere [37].

Adolescents under 18 years signed an informed assent form and provided a signed informed consent form from a parent or legal guardian. Adolescents aged 18 or 19 signed an informed consent form. Age was verified using birth certificates or other identity documents.

Ethical approval for the study was granted by the ethics committees of the University of the Witwatersrand (Johannesburg, South Africa) and Simon Fraser University (Burnaby, Canada).

An interviewer-administered, structured, online questionnaire was delivered to participants (supported by SurveyMonkey^TM software) via iPad or desktop computer. Interviewers received extensive training in good clinical practice guidelines, participant recruitment, administering questionnaires, and participant referral in cases where additional support was required after the study visit. Interviews were conducted in either English or isiZulu at the PHRU, the KMAC, or at a private location selected by the participant. Questionnaires took an average of 60 min to complete, and participants received 50 Rand (approximately 7 USD at the time) as compensation for their time and transportation costs. An international team of experts in adolescent health and HIV, including an adolescent Community Advisory Board (CAB), contributed to the development of the BBAHS questionnaire [37].

All analyses were conducted using SAS 9.4, stratified by self-identified gender. Descriptive statistics (median, 1^st quartile [Q1] and 3^rd quartile [Q3] for continuous variables and n, % for categorical variables) were used to characterize baseline distributions of study variables. Differences in baseline variables and trauma scores by gender were compared using Wilcoxon rank sum test for continuous variables and Pearson χ² or Fisher’s exact test for categorical variables.

Univariable and multivariable logistic regression were used to identify variables associated with high PTE score, separately for adolescent men and women. Variables of interest with univariable p-values <0.20 were included in multivariable model selections. After testing for collinearity, only the sexual behaviour variable ‘inconsistent condom use (yes vs. no vs. never had sex)’ was considered for inclusion in the final model. For all other variables, model selections were performed using backward selection based on Type III p-values to reach the optimal (minimized) AIC. All statistical tests were considered statistically significant at α < 0.05.

Of 830 participants, 767 answered all 19 TESI-C items and were included in this analysis of whom 442 (58%) were adolescent women and 325 (42%) were adolescent men (Table 1). Median age was 17 years [Q1-Q3: 16-18], 45% were Zulu, 85% were currently enrolled in school, and 6% had ever been incarcerated. A majority had lived in Soweto since birth (77%), lived in brick house/flat owned by the family (71%), reported high food insecurity (52%), and lived in a household which had received a social grant in the last 12 months (57%).

Overall, 56% of participants reported having ever had sex, including 64% of adolescent men and 50% of adolescent women (p < 0.001 for gender difference). Of those reporting sexual activity in the six months prior to the interview, 35% reported having more than one sexual partner in the previous 6 months (including 56% of adolescent men and 18% of adolescent women [p < 0.001]). Among those who had ever had sex, 54% reported inconsistent condom use (including 53% of adolescent men and 55% of adolescent women [p = 0.729]) and 23% reported ever having been diagnosed with an STI or experienced STI symptoms (including 17% of adolescent men and 28% of adolescent women [p = 0.009]). Overall, 1.4% reported being HIV-positive (1.5% of adolescent men and 1.4% of women, p = 0.19).

Nearly all participants (99.7%) reported experiencing at least 1 PTE. Median number of PTEs experienced was 7 [Q1-Q3: 5-9], with no significant difference by gender (p = 0.19). Overall, 47% of adolescent men and 45% of adolescent women experienced a high PTE score (≥7 events (p = 0.603)).

Table 2 shows the proportion of adolescents who reported experiencing each of the 19 PTE items included in the adapted TESI-C scale by gender. Nearly three-quarters (74%) of adolescent men and women reported experiencing the death of a family member or someone close to them. Over two-thirds (68%) had witnessed a close family member or friend deal with a serious illness or injury. Nearly half reported that their parents were separated or divorced (48%) or that their family struggled with money (46%). In general, adolescent men were more likely to have experienced or perpetuated violent forms of traumatic experiences (e.g. forcing someone to have sex with them [7%], deliberately inflicting harm on another [51%], witnessed an act of violence in the community [76%]). Adolescent women were more likely to experience psychological and emotional experiences of potentially traumatic events (e.g. having a family member have [46%] or die from [41%] HIV/AIDS).

Overall, 14% of adolescent women and 11% of adolescent men reported experiencing forced sex (p = 0.153) while 1.4% and 7.4% reported ever forcing someone to have sex with them (p < 0.001).

In unadjusted models among adolescent men (see Table 3), high PTE score was associated with older age, living in Soweto for <5 years, self-reported Tswana ethnicity, high food insecurity, drug use in the past six months, sexual experience, and inconsistent condom use. In the adjusted model (see Table 3), adolescent men with high PTE scores had significantly higher adjusted odds of being older (aOR = 1.40/year, 95%CI = 1.21-1.63); recently moving to Soweto (<5 years) vs. living in Soweto ‘since birth’ (aOR = 2.78, 95%CI = 1.14-6.76); and high vs. low food insecurity (aOR = 2.63 95%CI = 1.36-5.09).

In the unadjusted models among adolescent women (see Table 4), high PTE score was associated with, high food insecurity, incarceration history, received a household social grant in the last year, probable depression, sexual experience and inconsistent condom use. In the adjusted model (see Table 4), adolescent women with high PTE scores had significantly higher adjusted odds of high food insecurity (aOR = 2.57, 95%CI = 1.55-4.26); probable depression (aOR = 2.00, 95%CI = 1.31-3.03); and inconsistent condom use vs. no sexual experience (aOR = 2.69, 95%CI = 1.66-4.37).

Our results align with previous research indicating that co-occuring multiple PTEs experienced by women influence heightened depression symptomology [8], and compound syndemic risks of HIV transmission through increased HIV risk behaviour such as inconsistent condom use [10, 30]. The synergistic effect of multiple experiences of PTEs and increased HIV acquisition risk may be exacerbated among women living in vulnerable urban environments, such as Soweto, facing economic hardships and high levels of food insecurity [23, 30]. These compounding experiences of structural vulnerability influence economic dependence - placing women in inferior roles in their relationships - in turn increasing experiences of gender-based violence, inability to negotiate condom use, and ultimately HIV transmission risk [3, 23].

Our results indicate that high-PTE scores were more commonly found among older adolescent men who have recently moved to Soweto, and who face high levels of food insecurity. Experiences of trauma can accumulate over the lifecourse, [47], as such older age was a hypothesized finding for higher number of PTEs among men in our study. The exposure to multiple experiences of PTEs at a young age have been found to perpetuate aggressive behaviour and negative views towards women in adulthood [48, 49]. The development of negative views towards women may perpetuate harmful gender norms and inequitable power dynamics in relationships, which has shown to have significant implications for the HIV epidemic in South Africa [24, 50‐52]. Furthermore, young men living in South Africa face extremely high rates of interpersonal violence. A study assessing hospital data on injuries within the Mthatha Hospital Complex in South Africa, found that the majority of injuries occurred among men, with 60% of all cases being for acts of interpersonal violence [27]. Despite extremely high levels of PTEs within men participating in our study, we found that this was not significantly associated with increased depression symptomology or inconsistent condom use. Previous research has explored the relationship between high levels of trauma and post-traumatic growth [53]. Resilience to HIV risk among adolescent men living in HIV hyper-endemic nations experiencing concurrent poverty and high-levels of PTEs should be further explored.

Reducing syndemic risks to traumatic experiences in both adolescent men and women is likely to have a positive impact on HIV transmission through multiple pathways. The scale-up of community and structural level interventions, as well as increased focus on trauma-informed models of care for adolescents in South Africa is critical for addressing the HIV epidemic [21, 54]. For adolescent women, intervention strategies aimed at increasing economic independence, reducing gender-based violence, reducing inequities in relationship power and control, and challenging gender norms, are critical to increase sustained and widespread uptake of HIV prevention options, including male and female condoms and, in more recent years, pre-exposure prophylaxis (PrEP), [48, 55‐57]. Among adolescent women, high rates of sexual violence and inequities in relationship power [50, 58, 59] intersect to compromise opportunities to negotiate condom use [30, 60‐62]. Given demonstrated links between trauma, poor mental health, and sexual behaviours, mediated through pathways of gender and power inequity, central to the efforts to reduce HIV incidence among adolescent women is a clear need to scale-up access to youth-centred, trauma-informed, and women-controlled HIV prevention strategies, inclusive of PrEP [4].

Trauma-focused cognitive behavioural therapy (TF-CBT) has been shown to be highly beneficial in reducing sexual health risk. Hien and colleagues [63] implemented a skill-based TF-CBT program focusing on various domains including: personal self-management, coping, communication, boundary setting, HIV risk reduction and reducing unsafe behaviour in general. Women in the trauma-focused intervention were almost half as likely to report unprotected sex compared to women in the control group [63]. Given the high number of PTEs experienced by young people in South Africa, it is imperative to scale-up such trauma-informed mental health services for adolescents [21].

Community-level interventions addressing harmful gender norms, such as Stepping Stones, have been successful at reducing the perpetuation of intimate partner violence, a significant step forward in reducing HIV transmission and experiences of trauma for adolescent women [48]. For both adolescent men and women, interventions aimed at addressing food insecurities may help to mediate the compounding affects of PTEs on HIV transmission within vulnerable urban environments such as Soweto. This relationship merits further examination. Future interventions should consider the importance of resilience and post-traumatic growth within settings where experiences of traumatic events and HIV risk are extremely high [64].

In conducting a gender-stratified analysis of PTE occurrence, we demonstrated the multitude of implications that PTEs have on both SRH programs and HIV intervention — informing a gendered approach to addressing PTE and HIV risk. However, we did not include measurements within our survey to assess PTSD symptomology which is a known outcome of experiencing trauma [8, 10, 21], thus we acknowledge this is a limitation of our study which should be further examined within future South African adolescent health studies. Further, we are unable to assess causation within this cross-sectional study. Additional limitations include recall and social desirability bias due to self-reported measures of sexual behaviour and other sensitive topics. In addition, we used a modified variation of the TESI-C; therefore, caution should be used in comparing these findings with other studies using the original version of the TESI-C and other scales similarly measuring experiences of trauma.