Background of Uganda VACS and objectives

In 2015, the Uganda Ministry of Gender, Labour, and Social Development, the Uganda Bureau of Statistics, the U.S. Centers for Disease Control and Prevention (CDC), U.S. Agency for International Development (USAID), UNICEF Uganda, Makerere University School of Public Health, and The ChildFund International Consortium, including The AfriChild Centre for Excellence, worked in partnership to implement the Uganda Violence Against Children and Youth Survey (VACS). The purpose of the Uganda VACS was to produce nationally representative and regional estimates on the burden and consequences of sexual, physical, and emotional violence. VACS are national, cross-sectional surveys of children and youth age 13–24 years [32].

Sampling and DREAMS settings

A geographically clustered, multi-stage sample design was used to produce nationally representative estimates of violence. A split sample approach selected separate enumeration areas for males and females in order to protect privacy, confidentiality, and safety of participants; however, only females were included in this analysis. During the first sampling stage, 368 female enumeration areas were selected using probability proportional to size. During the second stage, 25 households were systematically sampled in each enumeration area, and in the third stage, one eligible female was randomly selected per household and included in the sample, yielding a sample size of 3,159 completed female interviews. A description of the sample size and oversampling in each of the regions can be found in the Uganda VACS Report [9]. The overall response rate for females was 76.6%, with 3,970 females selected, 56 refusals, 620 not available, 7 incomplete interviews excluded from analysis, and 128 who were incapacitated/had language issues. Oversampling for AGYW in two regions, DREAMS Central 1 and DREAMS Central 2, yielded sub-national estimates for these regions. DREAMS Central 1 included Bukomansimbi, Ssembabule, and Rakai districts and DREAMS Central 2 included Mubende, Mityana, Gomba, and Mukono districts (Fig 1). A third DREAMS region was oversampled but is not presented separately due to a small number of females reporting TS in this area. All remaining enumeration areas (including the third DREAMS region and those regions not included in the DREAMS 1 and DREAMS 2 regions) were combined into Other Districts for the sub-national analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Uganda PEPFAR DREAMS clusters.

https://doi.org/10.1371/journal.pone.0257030.g001

Eligibility criteria

Eligibility criteria included 13–24 years of age at the time of the survey and speaking one of the survey languages: English, Ateso-Karamajong, Luganda, Lugbara, Luo, Swahili, Runyankole-Rukiga, or Runyoro-Rutoro. AGYW who lived in institutions rather than households or who had a physical or cognitive disability that would interfere with participating in the survey were ineligible and not included in the study.

Response plan and IRB review/ethics

The study protocol was independently reviewed and approved by the Makerere University College of Health Sciences ethics review committee, the Uganda National Council for Science and Technology (UNCST) review board, and the CDC Institutional Review Board. Prior to participation in the study, informed consent was obtained for 18-24-year-old participants. Informed consent by a parent or guardian and participant informed assent were obtained for the 13-17-year-old participants. Informed consent and assent were obtained verbally and documented through electronic signature in the netbooks. The oral consent/assent process was approved by each ethics institution during the independent review process. Ugandan survey teams trained by CDC and Uganda partners administered household face-to-face interviews and recorded participant responses on netbook computers [33]. Counseling and referral services were provided to participants who met specific criteria or requested help and participants who self-reported a positive HIV status were referred to the nearest government clinic for services. The study followed World Health Organization recommendations on ethics and safety in studies of violence against women [34].

Weighted data for representative results

Survey weights were applied to the Uganda VACS data to yield nationally representative estimates and sub-national estimates for DREAMS 1, DREAMS 2, and Other Districts.

Questionnaire

The VACS questionnaire was developed by a group of violence subject matter experts by drawing from the literature and including validated measures when available. The questionnaire was then adapted to the Ugandan context by Ugandan stakeholders. Prior to the survey in Uganda, the VACS questionnaire was cognitively tested to ensure that the VACS questions capture the intended content. Prior to Uganda VACS data collection, a pilot test of all data collection instruments and protocols was conducted. The Uganda VACS questionnaire can be accessed at the Together for Girls website (togetherforgirls.org).

Variables

The outcome variable, recent TS, was asked among participants who answered yes to “Have you had sex in the past 12 months?” (n = 1,593) and assessed through the question, “In the last 12 months, have you had sex with this person because they provided you with material support or help in any other way? Material support means helping you to pay for things, or giving you gifts or things such as food, school fees or money.” This question was asked for the three most recent partners in the past 12 months. Sex was defined as vaginal, oral, or anal sex. Because the field and this study defines TS as being between non-married partners [10], where partner data were not missing, females who only engaged in TS with spouses were removed from analysis (n = 21 removed). Females who had a TS partner other than a spouse and those who had multiple TS partners where one or more partners was a non-spouse, were included in the analysis (n = 203). Females who engaged in TS in the past 12 months were also asked their age at first experience of TS, their relationship to their three most recent TS partners within the past 12 months (current or ex-romantic partner, current or ex-spouse, friend, and other), age of most recent TS partner (less than five years older and five or more years older), material goods provided by the most recent TS partner (money, food, school fees or supplies, favors or gifts, shelter or protection, and other), and HIV testing history and result.

Sociodemographic factors in these analyses include age, ever married or lived with someone as if married, child marriage, mean age at first marriage or cohabitation, family wealth, ever pregnant, highest level of school completion, orphanhood, food insecurity in the household, having a close or very close relationship with mother, having talked to friends about important things, early sexual debut, and ever tested for HIV. Violence and violence-related experiences included childhood witnessing of violence in the home or community and lifetime experiences of sexual, physical, and emotional violence. Sociodemographic factors, violence, and violence-related experiences are defined in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Definitions of sociodemographic factors and violence and violence-related variables.

https://doi.org/10.1371/journal.pone.0257030.t001

Statistical analysis

Descriptive analyses among sexually active AGYW ages 13–24 years assessed prevalence of sociodemographic factors, recent TS, and individual characteristics of AGYW who engaged in recent TS. To identify associations between recent TS and risk factors, univariate and multivariable logistic regressions were conducted. Variables that were significant at the p-value < 0.2 level in the univariate models were entered into multivariable logistic regression models. Control variables: sexual debut, ever married, family wealth, education, and ever pregnant, were chosen to be included in the multivariable models with a combination of identification as predictors of TS in prior studies [13,15] and had a p-value of less than.05 in one or more regions. Covariables in each model were limited to six variables. The cut-off was determined as a function of sample size using the rule of ten events, as not to over-fit the models. Interaction terms between region and types of violence assessed the associations between violence and recent TS within geographic regions (DREAMS Central 1, DREAMS Central 2, and Other Districts) and were included in separate unadjusted and adjusted models. Unweighted n’s, weighted percentages, weighted means, 95% confidence intervals (CIs), unadjusted odds ratios (ORs), adjusted ORs, and p-values were calculated using SAS 9.4 (SAS Institute Inc., Cary, North Carolina, USA). All calculations accounted for the complex survey design using the SURVEYFREQ procedure which includes STRATA, CLUSTER, and WEIGHT statements to account for design strata, clusters, and sampling weights.

Socio-demographic factors associated with TS

In the Other Districts (not DREAMS Central 1 or 2), orphanhood was associated with higher odds of recent TS, aligning with previous research in which TS was more common among youth raised without one or both parents [27]. Low socioeconomic status in Other Districts and marriage/cohabitation in DREAMS Central 1 and Other Districts were negatively associated with TS, as was ever experiencing a pregnancy in Other Districts. The protective nature of marriage aligns with previous studies in which TS was less common among married AGYW and aligns with anthropological analyses suggesting that demonstrations of male provision and offering sexual access in return may be a part of the process toward establishing a union for some young people [39]. The negative association between family wealth and TS in Other Districts counters common narratives of TS among the most vulnerable AGYW. Low family wealth was not significantly associated with TS in DREAMS Central 1 or 2, potentially reflecting differences in access to TS exposure for the poorest members of communities in mostly rural Uganda. The proportion of AGYW in the lowest wealth tertile was higher in Other Districts (28.8%) than in DREAMS Central 1 (13.8) or 2 (12.8%). These findings further support the idea that TS among AGYW in Uganda is not necessarily a means to meet basic survival needs. Rather, these findings indicate that the underlying vulnerabilities leading to TS are more socially and psychologically complex than simple economics. In further considering potential economic motivation for TS, we note that most common items received in exchange for sex were money, favors or gifts, or food, suggesting that AGYW in Uganda do participate in TS to meet certain perceived needs. However, these findings indicate that even as a sub-set of AGYW may engage in TS toward meeting their basic needs (e.g. sex in exchange for food), most are not engaging in TS to survive. Rather, these findings indicate that the underlying vulnerabilities leading to TS are more socially and psychologically complex than simple economics. These findings may be better understood in light of Stoebenau et al’s [10] paradigms for transactional sex: ‘sex for material gain’, ‘sex for improved social status’, and ‘sex as an expression of love’. This summary of findings related to the motivations of AGYW to engage in TS provides insight into the perceptions of AGYW on why they engage in TS while this research may provide additional understanding of why AGYW have the motivations they do. Specific to Central Uganda, Stoebenau, 2019 found that it is socially acceptable to find a second partner if a primary partner is not providing adequate support and multiple other studies have provided evidence that material gain for increased social status as a motivator for TS.

Critically, our findings indicate that factors associated with TS differ somewhat across geographic regions or contexts. Unique patterns emerged in associations between TS for orphan status, marriage, and family poverty in Other Districts. These factors did not emerge as significantly associated with TS in the DREAMS Central 1 or 2 regions, indicating that factors associated with TS differ based on local context. We hypothesize that this trend would not be limited to the Ugandan context and that this analytic approach to assess potential determinants for TS regionally could provide important information for more targeted programming in other sub-Saharan African countries.

Early sexual debut was associated with recent TS in DREAMS Central 1 and 2, though not in Other Districts. Both early sexual debut and TS are associated with risk for HIV, though limited research has focused on interrelatedness among these two factors. Our findings suggest early sexual debut and TS are associated in areas with high HIV prevalence.

Violence exposures and TS

Exposure to sexual violence and emotional violence by parents were both associated with increased odds of recent TS in all regions. Additional associations between TS and violence emerged in both DREAMS Central 1 and 2, with higher odds of TS for those who reported witnessing violence in the home or community in childhood along with lifetime sexual or emotional violence. In DREAMS Central 1, physical violence was also associated with six-fold increased odds of TS. The associations between violence and TS in this study expand upon previous work establishing links between violence and TS through varying mechanisms [15,16,27].

Although less is understood about specific pathways between exposure to violence and TS, childhood trauma resulting from such adverse childhood events as violence victimization or witnessing violence are well-established risk factors for negative mental and physical health outcomes including increased likelihood of engaging in sexual risk behaviors [40].

Fewer significant associations between violence and TS emerged in the Other Districts, though generally patterns reflected similar directional relationships. The interaction between violence and region further highlights the regionally specific relationship between violence and TS. Although DREAMS Central 1 and 2 are not the only regions with high HIV burden in Uganda, the complex contexts and social environments therein, including fatalism among the fishing communities in DREAMS Central 1, and the transience of populations in DREAMS Central 2, contribute to the HIV epidemic. Data on violence and TS in these focused geographic regions could be reflective of a clustering of many of the same risk factors that both contribute to and emerge as a result of the HIV epidemic. Given the geographic findings on the protective nature of specific social factors, future research could examine the relevance of social connectedness as a factor contributing to associations between violence and TS and could consider emotional dysregulation as a pathway between violence and TS, one that may be exacerbated by other complex factors in Central 1 and 2. These findings indicate that our understanding of the local context is critical to effectively respond to and prevent TS behavior and its role in the HIV epidemic.

TS is associated with a 50% increased risk of HIV among AGYW, making it a top-priority issue in HIV epidemic control [10]. TS prevention efforts have focused largely on relieving economic pressure and changing harmful gender norms [41]. The present findings indicate that factors associated with TS differ by context, suggesting interventions should be tailored to meet the needs of AGYW in their unique social environments. Targeted interventions must take into consideration patterns of violence that may impact AGYW’s decision making processes in addition to or even more than their economic vulnerability.

Considering the association between violence and TS and the protective association with talking to friends about important things and TS, efforts to reduce new HIV infections among AGYW in Uganda may benefit from approaches that prevent and respond to violence and promote social asset building. Evidence-based approaches including cash transfers, group savings and loan programs, and microfinance programs such as Empowerment and Livelihoods for Adolescents [42] could provide much needed support to AGYW facing socio-economic hardship with an aim to prevent TS. Access to such programming, tailored to meet the challenges AGYW face in their communities, can provide beneficiaries, their families, and their communities with the tools to prevent violence, and provide services to victims of violence as secondary prevention to address potential ongoing health issues. The Government of Uganda’s District Action Centers (DAC) hold promise in providing a safe drop-in location for children to seek help for violence and in ensuring the national Child Helpline (Sauti 116) is able to respond to reported cases of violence throughout the country by linking with DAC staff for quality case management.

Recognizing the high prevalence of emotional, sexual and physical violence among AGYW, there is an urgent need for rapid expansion of violence prevention and response programming in Uganda. Programming has increasingly incorporated parenting interventions, such as Parenting for Lifelong Health, implemented as ‘Sinovuyo’ in DREAMS programming in Uganda, which promote nurturing and positive parent-youth relationships and reduce physical and emotional violence by parents [42]. These findings provide evidence that efforts to increase positive parenting can prevent violence and in turn reduce risk for TS and HIV.

Witnessing violence in the home and in the community were associated with TS in both DREAMS Central 1 and 2. Evidence-based strategies that reduce youth exposure to violence in their homes and communities can ensure protective environments for young people and may reduce risk for TS and its negative health consequences.

Recognizing the strong associations between violence and TS, it is critical that interventions aim to decrease HIV risk associated with TS and incorporate early comprehensive violence prevention programming. This research adds evidence that witnessing violence and experiencing physical and emotional violence are important considerations, in addition to the previously recognized role of sexual violence as a driver of TS and HIV [43]. Curbing the HIV epidemic among AGYW may involve a broader approach to violence prevention, taking multiple forms of violence exposure into account. Home-grown evidence-based violence prevention programs endorsed in the INSPIRE technical package, including Raising Voices’ SASA! and Good Schools Toolkit can be scaled up quickly and efficiently with increased commitments from government, partners, and donors.

Limitations

These findings are subject to several limitations. Because this analysis examined TS in the past 12 months and lifetime exposure to violence, the temporality of these experiences is unknown. We expect that most experiences of violence occurred prior to the past 12 months, but it is also likely that many participants experienced violence both recently and historically and that some participants only experienced recent violence. In Uganda, among females who experienced sexual violence prior to age 18, 83% had experienced multiple incidents of sexual violence. Similarly, the majority of those who experienced physical and emotional violence experienced multiple incidents of that type of violence (91% and 82%, respectively) [9]. The relationship between violence and TS is likely bi-directional, with those witnessing and experiencing violence at higher risk of TS and those engaging in TS at increased risk of violence victimization. Previous research among females in Central Uganda has identified high rates of IPV among women engaged in TS and noted the bi-directional nature of the relationship, though definitions of TS appear synonymous with sex work and the focus was on very localized populations [44]. Self-report of HIV status is another limitation of this study to consider, particularly since TS was negatively associated with HIV testing, therefore knowledge of HIV status in this population may be limited. The sample of AGYW who experienced transactional sex in the past 12 months was relatively small. Although we believe the analysis to be powered sufficiently to detect significant differences, or lack thereof, type-2 error may lead to false negatives. Related, an absence of association in some settings and not others may be due to lower prevalence of the factor in that setting. Some point estimates, such as the physical violence*region DREAMS Central 1 OR in Table 5 (AOR = 7.51, 95% CI: 2.10, 27.04), have wide confidence intervals indicating a less precise estimate and should be considered in the interpretation of the result. Finally, there are limitations to the generalizability of the findings. The Uganda VACS did not collect data on some of the most vulnerable AGYW, such as those who live in institutions, on the streets, those with disabilities or those unable to consent.