Four interventions (14.3%) incorporated randomized controlled trial procedures using a team (n = 1), a school (n = 1), and participants (n = 2) as the unit of randomization; of these, one blinded investigators to participant status during data collection. The remaining studies were characterized by the authors as quasi-experimental (n = 9; 32%), pretest–posttest designs that included nonrandomized comparison conditions (n = 7; 25%), prospective cohort studies (n = 3; 10.7%), and mixed methods (n = 3; 10.7%). The authors of two articles did not mention a specific design, but a quasi-experimental approach could be deduced based on their description of the methods.

The authors of 10 studies (35.7%) discussed a specific theoretical framework that guided their research, including social learning theory (n = 3), the social-ecological model (n = 1), self-determination theory (n = 1), the theory of planned behavior (n = 1), and psychosocial theory (n = 1). However, among those that mentioned a theoretical framework, the authors of only five studies described how that framework specifically informed their intervention. For example, Tian et al. (2017) explained how self-determination theory guided decisions to provide participants with extrinsic rewards. Other authors described how their use of interactive soccer-based activities, led by slightly older peer coaches or role models, aligned with social learning theory (Clark et al., 2006; Hershow et al., 2015; Kaufman et al., 2016) and the social influences encompassed in the theory of planned behavior (Awotidebe et al., 2014).

The average sample size across studies was 704 participants; a mean of 199 participants (28.3%) dropped out, leaving a mean of 505 participants who completed the interventions (n = 280 males, n = 225 females). Sample size was highly variable across studies, ranging from 27 to 3,814 participants. The average age of participants across studies was 14.08 years, ranging from 7.8 to 30.8 years.

Measured outcomes included those related to physical fitness (n = 9; 32.1%), HIV (n = 8; 28.6%), physical activity levels (n = 6; 21.4%), life skills (n = 3; 10.7%), physical health (n = 3; 10.7%), injury prevention (n = 2; 7.1%), and mental health (n = 3; 10.7%). Three studies examined a compilation of essential health behaviors (i.e., physical activity, clean water, sanitation, substance abuse, nutrition, malaria prevention, vaccinations, prescription medications, and HIV awareness) as well as gender equality and social support (n = 3; 10.7%). In 82.1% (n = 23) of included studies, at least one of the measured outcomes yielded statistically significant improvements (see Table 1).

Most studies (78.6%; 22/28) cited a funding source, including Elimu, Michezo Na Mazoezi Kicking AIDS Out Program (n = 3); Fédération Internationale de Football Association (n = 3); KwaZulu-Natal Department of Health and Education (n = 2); National Research Foundation (n = 2); Bill and Melinda Gates Foundation (n = 1); Dphil Scholarship, University of Oxford (n = 1); International Development Research Center, Ottawa, Canada (n = 1); Medical Research Council of South Africa (n = 1); Nelson Mandela Metropolitan University (n = 1); South African National Research Foundation (n = 1); 3ie (n = 1); University of Cape Town Research Committee (n = 1); Universidad Politécnica de Madrid (n = 1); VLIR-UOS (n = 1); and World Diabetes Foundation (n = 1). The authors of one study mentioned receiving four sources of funding, including Imago Dei, Elton John AIDS Foundation, MAC AIDS Fund, and USAID–New Partnerships Initiative. The authors of another study reported using their own research funding to finance their intervention but did not cite a specific source.

Few studies in our review used randomized controlled trials, which are often considered the gold standard for intervention effectiveness research. Without randomization, researchers cannot control for preexisting differences between experimental and control conditions, which limits both internal and external validity (West & Spring, 2017). Although the authors of several studies acknowledged this as a limitation, they explained that randomizing participants was not always logistically possible and adjusted their designs accordingly. For example, when participants could not be randomized across classrooms due to school structure policy and scheduling conflicts, the authors instead assigned clusters of classrooms to an experimental or control condition (Awotidebe et al., 2014; Fuller et al., 2010, 2011, 2015). This flexibility in study design is critical given the nature of field research. In the future, researchers may also consider alternative randomization procedures, such as using wait-list comparison groups or cluster randomizing by region in larger trials to maximize the validity of their findings. Critically, researchers must also weigh the benefits of improving the scientific quality of their research against the unique concerns of the communities they intend to serve. For example, at-risk youth may benefit immediately from receiving essential health information, which poses ethical dilemmas relative to including a control group.

Using empirical theory to inform study design and methodology allows researchers to examine social and psychological phenomena systematically (Anfara & Mertz, 2014). A considerable body of literature suggests that interventions guided by theory are more effective (Glanz & Rimer, 2005) because they provide researchers with a framework, or lens, from which they can interpret findings and draw conclusions (Rapport et al., 2018). Fewer than half the studies included in our review cited a theoretical framework, and even fewer linked a theory specifically to their intervention design, content, and delivery. In a review of peer-led HIV prevention initiatives, Maticka-Tyndale and Barnett (2010) reported similar findings and hypothesized that many interventions were based on researchers’ or practitioners’ personal experience alone. Integrating theoretical foundations derived from the literature with valuable experiential knowledge is likely the most ideal approach to developing, planning, and evaluating sport-based health promotion interventions. This balance may be best achieved by intentionally collaborating with stakeholders in existing nonprofit organizations whose mission and goals are often already grounded in a theoretical framework. For example, Grassroot Soccer has increasingly used social learning theory to inform peer-led HIV prevention initiatives in Africa (Maticka-Tyndale & Barnett, 2010; Simoni et al., 2011). Eight studies included in our review were similarly administered in collaboration with Grassroot Soccer, several of which involved experienced staff members training new facilitators or delivering the intervention themselves (e.g., Fuller et al., 2010, 2011, 2015; Hershow et al., 2015). The knowledge, experience, and resources of existing organizations can and should be used to enhance the development, implementation, and evaluation of sport-based health promotion interventions.

Given the variability of sample sizes across interventions, our findings suggest that sport-based health promotion interventions can be implemented on both small and large scales. Aligned with recommendations from the sport for development literature (Schulenkorf et al., 2016), it is advantageous for researchers to continue using the popularity of sport in order to attract larger groups that increase sample size and maximize participants’ exposure to the benefits of sport-based interventions. It has also been suggested that the scholastic setting, often a safe space for students to engage in organized activities, is ideal for conducting interventions with youth in Africa due to the high number of youth that can be reached (Struthers, 2011; World Health Organization, 1997). Although our findings support that schools can be an effective setting for sport-based health promotion interventions, we found that attrition is a significant barrier. Absenteeism in schools throughout Africa is one plausible explanation (Plummer et al., 2007), but drawing conclusions is difficult when reasons for attrition were rarely reported. Collecting data on participant attrition could help researchers identify and address barriers that may inhibit participants from completing the intervention. Furthermore, considering that many youth in Africa do not attend school at all, researchers should also consider conducting sport-based interventions in settings where youth spend time performing activities they value, such as popular community centers or sport fields (Maro et al., 2009). Importantly, recommendations for increasing sample size should not negate efforts to work with special populations, such as incarcerated youth or youth with disabilities (Bloemhoff, 2006, 2012; Chetty & Edwards, 2007) or small schools (Ferguson et al., 2015; Kemp & Pienaar, 2009) where a critical need exists but larger sample sizes are not feasible.

Due to the high variability in participant age observed across interventions, our results suggest that sport-based health promotion interventions can be effective with young children, adolescents, and young adults. It was also promising that an approximately equal number of males and females were examined across studies. However, it was unclear how the interventions were specifically aligned with participants’ developmental level to maximize teaching and learning, especially in a mixed-age sample. Visek et al. (2013) highlighted the unique developmental challenges that accompany working with youth. The researchers suggested that scientists and practitioners who interact with youth participants should have a comprehensive understanding of child and adolescent development. For example, adjusting interventions to be more concrete (e.g., hands-on and movement-based) when working with children in the mid-childhood stage of development, versus abstract (e.g., incorporation of greater discussion) when working with adolescents, would be appropriate as cognitive abilities change over the lifespan (Visek et al., 2013).

Our results suggest that sport-based health promotion interventions can be effective in addressing multiple aspects of a single health concern. For instance, eight interventions assessed the knowledge, attitudes, communication skills, awareness, and uptake of preventative treatments specifically for HIV; all these interventions identified statistically significant improvements. Our results also suggest that sport-based health promotion interventions can effectively target multiple health concerns concurrently. Fuller et al. (2010, 2011, 2015) conducted an intervention to improve participants’ knowledge and awareness of nine essential health practices as well as outcomes related to social cohesion and gender equality; participants in these studies demonstrated statistically significant improvements across most outcomes. These findings suggest that researchers should conduct thorough needs assessments to gain a comprehensive understanding of the health concerns of communities and cater their intervention to address the most relevant health topics and concerns.

The most common specific health concern addressed by the interventions in our review were related to HIV. In a meta-analysis examining the effectiveness of sport-based HIV prevention with both youth and adults globally, Kaufman et al. (2013) found that the interventions were linked only to attitudinal changes among participants, and none measured behavioral outcomes. However, research published since this meta-analysis, and subsequently included in our review, suggests that sport-based interventions can significantly improve the uptake of HIV-related behavioral and mental health treatment services (Hershow et al., 2015; Kaufman et al., 2016). Sport-based health promotion initiatives are thus a promising avenue for attitudinal and behavioral changes that may prevent HIV.

Other targeted concerns were related to physical activity and mental health. It has been well documented that physical activity is associated with improved fitness outcomes in youth (Armstrong et al., 2011) and can reduce disease and promote wellness (Booth et al., 2000). A strong positive relationship also exists between physical activity and improved mental health with children and adolescents, including decreased symptoms of depression and anxiety and improved self-esteem and cognitive functioning (Biddle & Asare, 2011). The improvements observed in various aspects of physical (e.g., improved fitness) and mental health (e.g., improvements in depressive symptoms) across studies included in our review suggest that the physical activity component of sport-based interventions may have inherent physical and mental health benefits for youth. Furthermore, participants across several studies reported positive experiences with sport-based health promotion interventions (Fuller et al., 2010, 2011, 2015), which suggests that youth enjoy participating in sport while learning about essential health practices.