Supplemental Digital Content is Available in the Text.

Interest in sport as a tool for behavioral HIV prevention has grown substantially in the past decade. With dozens of organisations now using sport-based HIV prevention (SBHP) approaches and upcoming randomized controlled trials in South Africa and Zimbabwe, there is a pressing need to synthesize previous evaluation findings and identify gaps in existing research. A systematic review on the effectiveness of SBHP interventions was carried out, identifying both published and unpublished studies on SBHP interventions that measured effectiveness quantitatively. Study quality was scored using an adapted Newcastle-Ottawa Scale. Random-effects meta-analyses were carried out across studies for effects on six categories of HIV-related outcomes. The review identified 952 publications, 21 of which met inclusion criteria. No randomised controlled trials on SBHP interventions and no studies assessing biological outcomes were identified. Mean study quality score was 5.1 (SD 3.1) out of 20 points. Overall strong evidence was observed for positive effects on HIV-related knowledge (RR = 1.26, 95 % CI = 1.15-1.37), stigma (RR = 1.13, 95 % CI = 1.02-1.24), self-efficacy (RR = 1.22, 95 % CI = 1.02-1.41), reported communication (RR = 1.24, 95 % CI = 1.06-1.41), and reported recent condom use (RR = 1.29, 95 % CI = 1.00-1.59). Generally, the review found encouraging evidence for some short-term effects but relied predominantly on low-quality studies. More rigorous research on SBHP is needed to objectively assess effectiveness. Randomised controlled trials could play an important role in guiding policies, strategies, and funding related to SBHP.

South Africa's HIV prevalence among young people remains among the highest in the world. A cross-sectional study was carried out in 2012 to estimate prevalences of sexual risk behavior and hazardous alcohol use (HAU) (via the Alcohol Use Disorder Identification Test) as well as to investigate potential associations between these outcomes and social media use. In all, 4485 students (mean age 15.66 years, SD 1.39) at 46 secondary schools in informal settlements in Cape Town and Port Elizabeth completed mobile-phone-assisted, self-administered baseline questionnaires within a cluster-randomized trial. In all, 312 females (12.5 %) and 468 males (23.5 %) screened positive for HAU (AOR = 1.98, 95 % CI 1.69-2.34). 730 males (39.9 %) and 268 females (11.8 %) reported having had two or more partners in the last year (AOR = 3.46, 95 % CI 2.87-4.16). Among females, having a Facebook account was associated with reported multiple partnerships in the last year (AOR = 1.81, 95 % CI 1.19-2.74), age-disparate sex in the last year (AOR = 1.96, 95 % CI 1.16-3.32) and HAU (AOR = 1.97, 95 % CI 1.41-2.74). Using Mxit-a popular mobile instant messaging application-was associated with higher odds of reported multiple partnerships in the last year among both males (AOR = 1.70, 95 % CI 1.35-2.14) and females (AOR = 1.45, 95 % CI 1.07-1.96) and with HAU among both males (AOR = 1.47, 95 % CI 1.14-1.90) and females (AOR = 1.50, 95 % CI 1.18-1.90). Further longitudinal and qualitative research should explore in more depth the observed links between social media and risk behavior.