Background

Across many countries in sub-Saharan Africa (SSA), adolescent girls live in a context of vulnerability and are exposed to a combination of intersecting systemic barriers based on their age, gender, education, ethnicity, socioeconomic status, and place of residence [1,2,3]. Every day, an estimated 1000 adolescent girls and young women aged 15–24 years are newly infected with human immunodeficiency virus (HIV) [4]. Globally, there are now 19.1 million adolescent girls and women living with HIV, of which 80% reside in sub-Saharan Africa [4]. Data from the Joint United Nations Program on HIV and AIDS (UNAIDS) estimates that three out of four new HIV infections in SSA among 15–19 years olds are among young women, and 7 out of 10 young women do not have comprehensive knowledge about HIV [5]. Additionally, the interactive effects of youth poverty and disease are particularly severe in SSA [6]. Decades of economic crisis across SSA have left millions of youth that are currently out of school unemployed [7,8,9]. These youth, particularly young girls, who miss out on education are more likely to engage in risk-taking behavior such as unprotected sex, transactional sex, and age-disparate sex [10,11,12]. Simultaneously, the population of adolescent girls and young women in SSA is expected to double from 100 million in 1990 to 200 million by 2020 [13]. This suggests a potential for new infections, and consequently a need to address the growing education and employment gap already faced by this population [14]. While there may be some challenges posed by the growing youth population in SSA, there are opportunities to optimize the demographic dividends from the “youth bulge”, where more than half of the population is younger than 20 years, to foster youth employment and economic empowerment [15]. The potential for elevated infection rates among young females in SSA demonstrates an urgent need for sustainable programs that leverage on the capabilities of young people to avert new HIV infections in adolescent girls in high HIV risk settings [16]. If not properly addressed, the mutually reinforcing crisis of poverty and disease may threaten fragile development gains. The result of which is a devastating downward spiral in human development over the next generation for millions of adolescent girls and young women in the region.

Recognizing the urgency of the crisis, considerable research has been devoted over the past two decades to developing effective strategies to prevent HIV among adolescents and young people globally [17,18,19,20]. A number of theory-based prevention approaches targeting individual-level, group, community, and structural barriers to HIV have been implemented, with some targeting girls in schools [21,22,23,24,25,26,27,28] or within their communities [29,30,31] and some showing evidence of efficacy or effectiveness. HIV prevention interventions also led to the development of effective approaches to combat a spectrum of other health and behavioral problems, including depression, risky sexual behaviors, pregnancy intentions, and intimate partner violence [17, 32,33,34,35]. Examples of economic empowerment interventions include microfinance, vocational skills training, business development training, micro-enterprise development, cash transfers, and savings-led asset-based programs that work to alleviate girls’ household economic hardships through the infusion of financial assets and resources [36,37,38]. Available evidence suggests that when implemented in conjunction with financial literacy curricula, such economic empowerment programs increase school attendance and personal savings among girls [39, 40]. Additionally, when these programs are combined with other social empowerment programs such as safe spaces, peer-support, and mentoring on female-specific issues related to health and well-being, they can increase girls’ bargaining power, decrease their financial dependence on others, and reduce engagement in sexual risk-taking behaviors [41]. Combination HIV prevention interventions that include economic empowerment activities are particularly beneficial in low-resource settings such as SSA, where adolescent girls and young women are at increased risk to engage in transactional and cross-generational sex due to limited economic assets [42,43,44]. Numerous studies continue to show that women who lack economic independence are less able to negotiate safe sex with partners, less able to leave an abusive relationship, and are more likely to engage in transactional sex as means of survival [45, 46]. These in turn increases their risk for HIV. Such evidence shows a strong link between economic instability and risky sexual behaviors that increase HIV risk among adolescent girls and young women in the region [43, 47].

Nevertheless, despite the increase in the number of these interventions targeting adolescent girls and young women in the region, it can take up to 17 years for these interventions to make their way to other adolescent girls underrepresented in scientific trials or in settings where its delivery could reasonably produce benefit [48, 49]. Additionally, a sizable gap remains between what is known about what works and how to effectively translate these interventions into practice [50, 51]. One potential solution is the use of implementation science, and by this, we mean the scientific inquiry into what, why, and how interventions work in “real world” settings and to test approaches to improve them [52,53,54]. As described by Peters and colleagues, “implementation research seeks to understand and work within real-world conditions, rather than trying to control for these conditions” [55]. It also implies working with populations that will be affected by an intervention (i.e., adolescent girls themselves serving as an advisory board), rather than selecting beneficiaries who may not represent the target population of an intervention (such as studying only in-school girls or excluding girls who have comorbidities) [54, 56, 57]. One goal of implementation science is to appropriately expand the use of interventions that have been found efficacious and as broadly as feasible in order to foster the greatest public health impact [51, 58]. The reach, effectiveness, adoption, implementation, and maintenance (RE-AIM) model is an implementation science framework for expanding interventions that have been found to be effective in research settings [58, 59]. The model focuses on the reach of the intervention to a representative proportion of the target population, the effectiveness of a program on specific outcomes, adoption of the program in a specified setting, and details of program implementation and maintenance [58, 59]. To date, there are no published studies using the RE-AIM framework to evaluate the public health impact of economic empowerment HIV prevention programs for girls, and none of the published RE-AIM studies have looked at adolescent girls and young women populations in SSA.

The present study seeks to bridge this gap between research and practice in SSA. Our objectives are twofold: (1) to review the extent to which EE HIV prevention interventions for AGYW in SSA report on implementation outcomes, as conceptualized in the RE-AIM (reach, efficacy/effectiveness, adoption, implementation, and maintenance) framework; and (2) to make recommendations for using the RE-AIM framework to advance the implementation of these interventions for girls and young women in the region. Through highlighting the reach, effectiveness, adoption, implementation, and maintenance of economic empowerment HIV interventions for adolescent girls and young women in SSA, we aim to assist researchers, practitioners, and policymakers in scaling up and evaluating new and existing economic empowerment interventions aimed at reducing the rate of new HIV infections.

Methods

A multi-step process was used to identify, review, and analyze existing economic empowerment HIV prevention interventions targeting adolescent girls and young women in SSA using the RE-AIM framework. For the purposes of this review, economic empowerment intervention was defined as a set of economic-related actions (i.e., microfinance, cash transfers, financial literacy, savings, and asset-based programs) [60, 61] with a coherent objective to bring about change or produce identifiable HIV prevention outcomes in three broad sectors: health (i.e., girls’ overall sexual and reproductive health, HIV and sexually transmitted infections (STIs), sexual risk-taking behaviors, pregnancy, and gender-related violence), social (i.e., education-related outcomes such as school attendance, employment, mental health, future outlook, etc.), and economic (savings, asset accumulation, small business, etc.).

Search strategy

A systematic search of the literature was executed from October 2018 to July 2019 to locate studies published in academic journals. Figure 1 outlines the search strategy, which was reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (see Additional file 1). Two reviewers (JI and UN) independently searched PubMed, Ovid/MEDLINE, Science Direct, Ebscohost, PsycINFO, Scopus, and Web of Science databases with the following approximate search terms: (girls or young women) AND (HIV or AIDS) AND (prevention or intervention or program) AND (economic empowerment or microfinance or cash transfers or savings-led programs or asset-based programs) AND (sub-Saharan Africa or country-specific terms for each SSA country). The search teams were modified for each database. A detailed search strategy for the PubMed database is provided in Additional file 2. In addition, published systematic reviews focused on economic strengthening for HIV prevention, as well as reference lists from the included studies, were searched to augment the database literature search. The titles and abstracts of potentially relevant articles were independently screened by two reviewers (JI and UN) for eligibility. The full texts of articles that met the eligibility criteria were obtained and assessed by the two reviewers (JI and UN) independently for inclusion in the review. Discrepancies in the screening process and study eligibility were discussed and addressed based on consensus between the two reviewers (JI and UN).

Fig. 1
figure 1

Flow diagram of the search strategy. A total of 25 unique interventions reported in 45 articles were included in the review

Full size image

Eligibility criteria

Inclusion and exclusion criteria were developed to identify original research that empirically evaluated or tested economic empowerment strategies to prevent HIV among adolescent girls and young women in SSA. Articles were eligible for inclusion if they were (a) conducted in sub-Saharan Africa, (b) described an economic empowerment intervention with outcomes related to HIV prevention, (c) targeted adolescent girls and young women aged 10–24 or interventions that were not specific to AGYW but reported separately on AGYW, (d) written in English, and (e) published between 2000 and 2019. We included studies that used intervention designs ranging from randomized control trials to quasi- and non-experimental evaluations of the interventions. Non-empirical studies (e.g., reviews, commentaries, editorials, and dissertations) and studies that did not explicitly assess the effect of economic empowerment on HIV prevention were excluded from the review.

Data extraction

For studies meeting the inclusion criteria, we extracted the following data: (1) title, author, country, study objective, and design; (2) information on the intervention being evaluated, including type of economic empowerment HIV prevention intervention, and target AGYW populations; (3) components of the intervention; and (4) RE-AIM framework implementation outcomes that included (a) reach (absolute number, proportion, and representativeness of AGYW in the economic empowerment HIV prevention interventions); (b) efficacy/effectiveness (impact of the intervention on AGYW HIV prevention behaviors, including overall sexual health factors, social factors, and economic outcomes); (c) adoption (absolute number, proportion, and settings participating in the intervention, and the extent to which the settings selected are representative of settings that the target population use or visit); (d) implementation (consistency of delivery as intended, time, and cost of implementation); and (e) maintenance (extent to which a program has become part of routine practice at the organizational level or the long-term effects of a program on outcomes at the individual level) [59].

Data analysis

Data from the articles included in this review were analyzed using narrative synthesis [62], with details on the reporting of the RE-AIM components synthesized. The articles included in this review are heterogeneous in terms of study design and measured outcomes; therefore, it was not practical to conduct a meta-analysis. To evaluate the included interventions within each dimension of the RE-AIM framework, two authors coded and scored each article independently using an adapted RE-AIM data extraction form that included a series of yes or no questions used to identify components within each of the RE-AIM dimension outcomes [63,64,65]. The adapted RE-AIM data extraction form is presented in Additional file 3. The form was used for calculating percentages of interventions meeting the criteria for the five RE-AIM dimensions (reach, efficacy/effectiveness, adoption, implementation, and maintenance). We summarized RE-AIM components using frequencies, proportions, and means. First, the frequencies and proportion of reported 26 components for each RE-AIM dimension were calculated separately for each study included in the review. Secondly, the average proportion of components within each RE-AIM dimension across the 25 unique interventions included in the review was calculated. The percentage and number of interventions reporting each RE-AIM dimension were reported to provide a comparable summary score across interventions.

Risk of bias

To systematically compare the interventions, we evaluated the rigor of each intervention using the Cochrane Collaboration risk-of-bias tool [66, 67]. The tool consists of six domains: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias [66, 67]. The risk of bias was independently rated as low, high, or unclear by two authors using the guideline for each domain. The raters discussed each domain of the assessment tool to apply consistent judgment. If ratings differed, the rationale for the rating was discussed, and the study was re-reviewed to reach consensus. The Cochrane Collaboration risk of bias assessment tool was only used to evaluate the internal validity of the interventions included in the review; no study was excluded from the review based on the risk-of-bias score.

Results

Study selection

The initial database search yielded 2205 potentially relevant citations based on publication titles and abstracts (Fig. 1). A total of 542 papers were retrieved for full-text review for eligibility, and 443 were excluded. The most common reasons for exclusions were the absence of an economic empowerment HIV prevention intervention targeting adolescent girls and young women aged 10–24 (n=46), study location outside of sub-Saharan Africa (n=3), or not meeting study design criteria (n=15). 

Characteristics of included studies

Forty-five articles were retained in the final group of articles. Most of the articles were pulled from the electronic data searches, except for 13 articles that were identified from the manual search of reference lists. The characteristics of the final included articles are presented in Table 1. We reported on 25 unique interventions that were highlighted across the forty-five articles included in the review. All 25 interventions reported in the articles were published between 2006 and 2018, thirteen interventions were conducted in Southern Africa, nine in East Africa, and one in Central Africa, West Africa, and North Africa respectively. The majority of AGYW targeted were between the ages of 15 and 24 years. Seventeen (68%) interventions were randomized controlled trials, three (12%) were cross-sectional interventions, three (12%) were quasi-experimental designs, one (4%) observational study, and one (4%) time-series design study. Six types of economic empowerment (EE) interventions were in the included in the review: cash transfers (conditional or unconditional) [18, 41, 68, 73, 80, 86, 88, 91, 96, 97, 107], job skills or business development [70, 84, 86, 99, 108], matched savings account [71, 102], vocational skills training [75, 77, 82, 101, 105], payment of school fees and school supplies [23, 78], and loan credit [92]. Several HIV prevention outcomes were targeted, including reduction in HIV incidence [18, 22, 68, 80, 82, 97], increase in condom use [70, 73, 75, 77, 78, 83, 84, 86, 90, 92, 96, 101, 107], reduction in the number of sexual partners [70, 73, 78, 86, 101, 107], reduction in transactional sex [41, 71, 78, 82, 88, 90, 91], increase in sexual and reproductive health knowledge [71],  self-efficacy [71, 84], and delay in sexual debut [22, 73, 78, 88, 101].

Table 1 General characteristics of 25 interventions reported in 45 articles included in the review
Full size table

Quality of evidence

The quality assessment of the selected articles is reported in Table 2. The level of bias varied widely, with a range of 0.0% to 71.4% risk among the interventions. Among the interventions using quantitative methods, one of the interventions [18] was found to have a 0.0% (low) risk of bias. The risk of bias for quantitative methods ranged from 0.0% (low) [18] to 71.4% (high) [77]. Among interventions using mixed methods, one of the interventions [92] was also found to have a 0.0% (low) risk of bias. The risk of bias for mixed methods interventions also ranged from 0.0% (low) [92] to 71.4% (high) [99]. The only qualitative study in the review had a high risk of bias (71.4%) [91].

Table 2 Reporting on quality of included interventions (25 interventions reported in 45 papers included in the review)
Full size table

The most common strengths of the interventions that utilized quantitative methods were: the ability to conduct a longitudinal follow-up of study participants over time, the random selection and assignment of participants, and the reporting of descriptive intervention details. However, one of the common weaknesses was the limited use of intent-to-treat analysis, although attrition was acknowledged by the majority of the interventions [18, 22, 41, 70, 71, 75, 78,79,80, 83, 84, 88, 92, 96, 97, 101, 102]. For the interventions that utilized mixed-methods [71, 73, 90, 92, 99, 105], common strengths were the ability to triangulate data obtained from qualitative and quantitative methods and providing additional explanation for the quantitative data using qualitative data. The strength of the qualitative studies were the use of detailed quotes and narratives to explain study findings.

Reporting of RE-AIM dimensions

The reporting of RE-AIM dimensions was assessed using a previously developed and validated data extraction tool that included implementation outcome components based on the RE-AIM framework [59, 63]. Across all the interventions, average reporting rates (defined here as the overall percent of components) were highest for efficacy/effectiveness ≈19(74.4%) and adoption ≈17(67.2%), followed by reach 16(64.0%), and lowest for implementation ≈9(37.3%) and maintenance ≈7(26.4%). Table 3 provides details on each of the components assessed across the RE-AIM framework and a summary of the overall percentage of interventions reporting on each of the RE-AIM dimensions. The reporting status for the 26 components for the RE-AIM dimensions per study is provided in Additional file 2.

Table 3 Proportion of interventions reporting RE-AIM dimensions and components
Full size table

Reach

The average proportion reporting on the reach components was 16(64.0%). The sample size and participants’ characteristics were the most frequently reported item 25 (100%). All interventions reported on sample size, defined as the number of participants who consented to participate in the study/intervention. Of the interventions that recruited only AGYW (n=12) [18, 22, 41, 70, 72, 73, 75, 84, 87, 92, 99], sample size ranged from n=40 in the O’Neill Berry and colleagues study [99] to n=4800 in the study by Bandiera and colleagues [75]. In interventions that recruited other populations in addition to AGYW, the sample size ranged from n=46 to n=6576 (with the AGYW sample size being between 122 and 1705). Five (20%) interventions [77, 91, 96, 97, 101] did not explicitly report the sample size for AGYW, although it was mentioned that AGYW were included in the study. Participant characteristics included; reports on age, gender (for interventions that included other population), employment status, education attainment, and socioeconomic status (measured as household income in some interventions). The next commonly reported reach component was the method for identifying the target population for the study and this was assessed in 24(96.0%) of the studies reviewed [18, 22, 39, 41, 68, 70, 71, 73, 75, 77, 78, 80, 82, 84, 86, 88, 90,91,92, 96, 97, 99, 101, 105]. Most of the interventions were conducted in Southern Africa, specifically in South Africa where 7 [18, 68, 79, 90,91,92, 105] out of the 25(28%) interventions were located. The description of methods utilized to identify the target population varied across interventions from single-sentence descriptors to detailed reporting of the protocol used. Strategies utilized to identify the target population included using schools, youth centers, and community stakeholders. Regarding factors that foster or hinder the ability to reach the target audience, stakeholders’ engagement and school recruitment were emphasized as beneficial strategies to enhance reach.

Sixty-four percent(16) of the interventions [18, 26, 41, 70, 73, 75, 80, 83, 84, 86, 88, 90,91,92, 96, 97] reported study participants’ inclusion criteria. Only 3(12.0%) studies [18, 73, 80] explicitly stated participants’ exclusion criteria. Participant inclusion criteria were typically related to participants’ age, place of residence, membership (e.g., being part of the school), parental status (being an orphan), and gender. Individuals were mainly excluded if they did not meet the inclusion criteria for the interventions. Participation rate was reported in nine(36.0%) [18, 68, 70, 75, 79, 83, 90, 92, 96] of the included interventions. The participation rate ranged from 21% to 97.50%. Eighteen(72.0%) of the interventions [18, 41, 68, 70, 73, 75, 78, 80, 83, 84, 86, 88, 90,91,92, 97, 102, 107] reported on the representativeness of recruited study participants’ relative to the target population. This was determined based on comparing demographic characteristics (e.g., age, education level) of study participants to those of the target population. Reporting on this component allowed the researchers to assess the extent to which the intervention could be generalizable across the target population and setting. The rigor of the study design was reported as an indicator of representativeness. Interventions that utilized randomized controlled trials reported representativeness as one of the strengths of their studies. According to the RE-AIM framework, studies should describe the characteristics of participants of the target population in comparison with non-participants. Eight(32%%) of the interventions provide some form of information on the characteristics of individuals who did not participate in their study. Some of the reasons for non-participation included unavailability of individuals (e.g., going back to school and having full-time jobs), inability to complete study procedures (e.g., not wanting to test for HIV, not returning for study procedure, and not obtaining consents from parents), limited access to the study location (e.g., distance from the individuals’ residence to study site was a barrier to participating and geographic relocations) and lack of interest in the study.

Efficacy/effectiveness

Efficacy/effectiveness was the most consistently reported RE-AIM dimension across all interventions (74.4%). Twenty four(96.0%) interventions reported on at least one post-intervention effect; 5(20.0%) interventions used intent-to-treat analyses and the remainder analyzing only data from participants who completed the intervention. All interventions included in the review included HIV prevention measures as primary outcomes. HIV prevention measures included; reduction in HIV incidence [22, 68, 79, 80], reduction in number of sexual partners [70, 73, 79, 86], condom use [70, 73, 75, 77, 78, 83, 84, 86, 90, 92, 101], decrease in transactional sex [41, 71, 78, 83, 89,90,91], and sexual debut [23, 73, 78, 83, 87, 88, 97]. Of the 25 interventions that measured HIV prevention outcomes, 20(80%) reported that the economic empowerment HIV prevention intervention resulted in statistically significant positive changes in HIV prevention outcomes.

Sixteen(64.0%) interventions reported their percent attrition [22, 70, 71, 75, 78,79,80, 83, 84, 88, 92, 96, 97, 100,101,102], which ranged from 5% to 40%. Attrition rates were examined in relation to participants’ loss to follow-up and non-use of the intervention [75]. Reasons for attrition included; participants’ relocation, death, change of phone number, and logistics challenges. In terms of logistics challenges, Erulkar and Chong [84] reported some delays in participants receiving their loans and accessing their savings account as a result of limited human resources which accounted for some of the attritions they faced. Some participants were concerned that they may not have access to their savings account or loans; therefore, they dropped out of the study. In addition, Bandiera and colleagues [75] examined how participants’ characteristics influence attrition between the intervention and control groups and found that married AGYW in the intervention were less likely to be tracked at follow-up. A high proportion of the interventions 23(92.0%) reported on participants’ quality of life [18, 22, 39, 41, 70, 71, 73, 75, 77, 78, 80, 82, 84, 88, 90,91,92, 96, 97, 99, 101, 105] and found that economic empowerment HIV prevention interventions generally improved quality of participants lives and did not have any significant negative outcomes.

Adoption

The average proportion reporting on adoption components was 17(67.2%). Twenty-two of the interventions [22, 39, 41, 68, 70, 71, 73, 75, 78, 80, 82, 84, 86, 88, 90, 92, 96, 97, 99, 101, 105] provided some description of the staff who delivered the intervention. Interventions were delivered by a range of staff with different levels of expertise and included research assistants, community leaders, and organization staff. Staff responsibilities included delivering parts of the interventions that consisted of moderating the discussion and intervention meeting groups, distributing conditional cash incentives, training participants’ income-generating skills, and educating participants on intervention curriculum (e.g., sexual and reproductive health training, financial training, and income-generating skills and crafts). Seventy-six percent (19) of the interventions explicitly stated implementing staff level of expertise [22, 39, 41, 68, 70, 71, 73, 75, 80, 82, 84, 86, 88, 90, 92, 99, 101, 105], but for those that were not stated, it could be inferred from their job titles or their organization’s focus. Sixteen(64.0%) interventions reported on the methods used to identify staff who delivered the intervention [22, 70, 71, 73, 75, 78, 80, 82, 86, 88, 90, 92, 97, 99, 101, 105]. Intervention staff were mainly identified through their participation in the research project or collaborating organization.

The most commonly reported adoption component was the description of intervention location, reported by 24(96.0%) studies [18, 22, 39, 41, 68, 70, 71, 73, 75, 78, 80, 82, 84, 86, 88, 90,91,92, 96, 97, 99, 101, 105]. Intervention locations included schools, community centers, and refugee camps. These locations were identified as typical settings that the target population visit or use. Also, most of the interventions were restricted to a specific geographical area. Most of the interventions were implemented in one site. The least reported adoption component was the adoption rate. Only 3 (12.0%) interventions reported on intervention adoption rate among participants [86, 99, 101]. There were no reports on setting level adoption rates.

Implementation

The average proportion reporting on implementation components was about 9(37.3%). All 25(100.0%) interventions reported on the format of the intervention; specifically, they provided information on intervention duration and frequency [18, 22, 39, 41, 68, 70, 71, 73, 75, 77, 78, 80, 82, 84, 86, 88, 90,91,92, 96, 97, 99, 101, 105]. Intervention ranged in duration from a single session to two or more (up to 14) sessions. None of the interventions explicitly reported on fidelity or the extent to which the intervention protocol was delivered as intended.

The cost of delivering the intervention was mentioned in only three (12.0%) interventions [70, 73, 75]. Implementation cost items included skills training cost [70, 75], administrative cost [73, 75], and cost of monetary incentive [73, 75]. Two interventions [70, 75] further conducted cost-benefit analyses to determine if the benefits/returns from the interventions for the participants outweighed the cost of implementing the interventions. These two interventions assessed intervention benefit based on the number of participants who participated in the income-generating component of the intervention. The authors highlighted that equipping AGYW with skills to generate sustainable income, which would in return reduce their chance of engaging in risky sexual behaviors [70, 75]. Specifically, Adoho and colleagues [70] found that the value provided by the program was equivalent to a 3 year increase in income among EE program participants. The study by Bandiera and colleagues [75] reported gains/benefits to the economic empowerment intervention in the form of delaying early marriage and childbirth and improving HIV and pregnancy-related knowledge.

Maintenance

The average proportion reporting on maintenance components was about 7(26.4%). Among the maintenance components, individual-level indicators were reported more frequently than program-level indicators. Twenty-three (92.0%) interventions reported at least one follow-up measure, particularly the primary outcomes at 6 months [18, 22, 39, 41, 68, 70, 71, 73, 75, 77, 78, 80, 82, 84, 86, 88, 91, 92, 96, 97, 99, 101]. The longest follow-up period reported was 24 months after baseline assessment [71]. The majority of the post-intervention assessments were conducted within 12 to 24 months after completion of the intervention. There were a few interventions that had follow-up assessments beyond 24 months after intervention completion; 6 for 36 months [18, 22, 78, 84, 86] follow-up, 2 for 48 months [88, 89] follow-up, and 1 for 60 months follow-up [23].

In terms of program-level maintenance, six interventions reported [22, 75, 77, 80, 86, 92] on indicators of program level maintenance or sustainability. Only two(8%) interventions explicitly stated that the interventions were sustained beyond the study period [75, 86] For one study, the intervention was adapted to fit the context by including an additional component [75]. Two(8%) interventions were discontinued before the study period end date [77, 80], and another two(8%) ended at the completion of the study period [22, 92]. For the two interventions that were completed at the end of the study period, it was not clear if they were sustained beyond the study period.

Discussion

The primary aim of this review was to systematically assess the implementation of economic empowerment HIV prevention programs for AGYW in SSA. This review goes beyond an assessment of intervention effectiveness to report implementation outcomes as conceptualized by the RE-AIM framework. The RE-AIM framework was used as a guideline to determine the impact of EE HIV prevention interventions for AGYW. We evaluated five key components important for the translation of research findings to practice: reach, effectiveness, adoption, implementation, and maintenance [59, 109]. These components are important in understanding the factors that influence, not only adoption, but the cost and sustainability of economic empowerment interventions as a strategy for HIV prevention among AGYW in SSA.

A total of 25 (reported in 45 papers) economic empowerment interventions among AGYW were identified, described, and evaluated based on the five RE-AIM dimensions. On average, the included interventions reported on 14(53.86%) of the 26 components that constitute the RE-AIM dimensions. Major knowledge gaps exist relating to reporting of implementation and maintenance (least reported RE-AIM dimensions) of economic empowerment HIV interventions for AGYW in SSA. Specifically, the interventions in the review mainly focused on reporting intervention-specific components (e.g., sample size, intervention location, and effectiveness), with minimal reporting of broad or system-level components such as implementation costs, program-level sustainability, and intervention fidelity. Although concerning, the underreporting of broad or system-level elements is consistent with reports from other systematic reviews using the RE-AIM framework [110,111,112,113,114,115] that also found limited reporting of these dimensions. This further confirms the previous report on the predominant focus on intervention effectiveness, with limited attention to external factors that may impact the translation of effective interventions to real-world settings. Researchers need to also focus on reporting broad or system-level measures as well as intervention-specific measures. Broad or system-level factors are critical with understanding how findings from interventions apply to local settings, population, and available resources [116]. It informs the overall relevance and appropriateness of these interventions in real-world settings, and the potential for health gains by reducing HIV incidence among AGYW in SSA. 

Reporting on intervention reach is important to inform future dissemination of interventions that have been found to be effective or efficacious towards behavior change. To scale-up economic empowerment HIV prevention intervention, there is the need to understand how to reach target populations. In this review, participants’ characteristics and sample sizes were consistently reported across interventions. This is congruent with previous reviews on HIV prevention interventions [32, 117] that reported frequent reporting of participants characteristics such as their demographics. Some of the interventions in the review specified the degree to which the  target samples were representative of the larger population. Information on the external population from which a study sample is drawn from helps to inform the generalizability of the findings to a larger population [111]. However, the characteristics of non-participants and participants, as well as the reasons for non-participation, were rarely reported in the interventions. This limits the understanding of contextual factors that may influence AGYW participation in such interventions. With scant information on characteristics of non-participants, researchers may be missing individuals who are most in need of these interventions, such as AGYW residing in remote areas, rural areas, and those with low literacy. To enhance the translation of intervention to a wider population, researchers should improve on the reporting of the characteristics of non-participants as this may extend program reach and inclusivity.

Consistent with past reviews, intervention effectiveness was the most commonly reported RE-AIM element across all interventions, with baseline activity measures reported for all included interventions [110, 111]. The outcome measures included; HIV incidence, number of sexual partners, condom use, transactional sex, and sexual debut. Findings from this systematic review highlight the impact on economic empowerment intervention on HIV reduction among AGYW, with about 19(74.7%) of the interventions reporting statistically significant improvements on HIV risk reduction measures among intervention participants compared with controls. For effectiveness analyses, only 5(20%) interventions reported using intent-to-treat analyses; this in turn may have impacted the positive effect of the intervention across the interventions. The positive effect found in these interventions were only limited to participants who were present for follow-up assessments and did not account for attrition. There were variations in the reporting of attrition rates across the interventions, and few studies provided information on reasons for attrition. Information on the reasons for attrition may help to highlight barriers or challenges that influence AGYW participation in the interventions. For instance, one of the interventions encountered some logistics challenges in the form of delays in providing loans to participants [84]. Such logistical challenges are critical information that may influence AGYW attrition and participation in HIV prevention interventions. Thus, efforts to account for factors influencing attrition are necessary for identifying barriers and challenges to AGYW continued participation in interventions.

For adoption, the description of the intervention location, staff delivering the intervention, and level of staff expertise were well documented in the reviewed interventions. However, there was minimal reporting on the methods used to enhance staff and intervention settings adoption. This is consistent with other reviews using the RE-AIM framework, where there is consistent under-reporting on methods used to enhance adoption by intervention delivery agents [111, 118]. This makes it challenging to determine what types of delivery agents may be appropriate for the optimal implementation of the intervention [111]. Furthermore, only 3(12%) interventions reported on the intervention adoption rate. Reporting of adoption rate and characteristics of participating intervention locations versus non-participating locations may help highlight components of intervention design that either hinder or foster adoption across various settings [111].

The cost of intervention implementation is an important factor in determining the translation of research findings to real-world settings. Three (12%) of the 25 interventions in the review reported on the cost of intervention delivery. The findings of the review reveal a paucity of data on the cost and cost-effectiveness of implementing economic empowerment HIV prevention interventions among AGYW. Report on delivery cost allows for effective planning to optimize the yield and reach of economic empowerment HIV prevention for AGYW [119,120,121]. Likewise, documenting cost-effectiveness is crucial for sustainability and large-scale dissemination of HIV prevention interventions in SSA [120]. Cost information also helps to allocate resources efficiently particularly in settings were resources are scare. This in turn may help maximize the impact of positive health outcomes among AGYW [119].

In terms of implementation, intervention duration and frequency were consistently measured across the interventions. However, none of the interventions reported on the fidelity of the study, although it is a critical measure of the internal validity of the interventions. Therefore, it is unclear if the reported intervention impact were attributed to the fidelity of the intervention or to the actual intervention components [110]. Considering the critical role of these components in enhancing the impact and scale-up of such intervention, the scarcity of evidence in this area is a concern. Future interventions should clearly specify implementation components such as fidelity to enhance the translation of these interventions to other settings and populations.

Regarding maintenance, about 7(26%) reported on this RE-AIM dimension. This is a favorable result, compared to other reviews that have reported between 0.0% and 11.0% maintenance [110, 122]. This RE-AIM dimension helps to understand the long-term maintenance of behavior change among intervention participants and the sustainability of the interventions at implementing locations. Most of the interventions measured maintenance of individual behavior at least 6 months following the completion of the intervention, with only 4 of the interventions measuring behavior at 48 months after intervention completion. While individual-level maintenance components were frequently reported, little attention was paid to the assessment of setting- and program-level maintenance components. Intervention maintenance also known as sustainability is influenced by an interplay of individual-, program-level factors and broader socio-cultural- and community-level factors, which collectively determine long-term intervention impact. Therefore, future research should address critical gaps in the assessment of intervention maintenance, and apply a more comprehensive approach in the evaluation of this implementation outcome dimension.

Limitations

Our review has some limitations. First, our conclusions are based on the degree to which the included interventions reported on specific RE-AIM dimensions. It is possible that some of the RE-AIM dimensions were measured, but not reported in the interventions due to editorial restrictions. To address this limitation, we included all available articles on a specific intervention. Second, we did not conduct a meta-analysis. While this was not the focus of this systematic review, the heterogeneity of the included interventions and variations in HIV prevention outcomes would not have supported a meta-analysis. Third, our search strategy was limited to published articles and those available in English; this is potentially subject to selection bias. Fourth, it is worth noting that the Cochrane Collaboration risk of bias assessment tool used in assessing study quality is biased towards purely quantitative study designs and quite limited in appraising mixed and qualitative study designs. Given the limitation of this tool, it was only used to evaluate the internal validity of the interventions included in the review and not to select articles included in the review.

Nonetheless, this study has a number of strengths. First, this review was conducted with a well-constructed search strategy, created with the help of the college librarian, and was supplemented by a manual search of the reference list of included articles. Second, to the best of our knowledge, this is the first study to collate and examine the measurement of implementation outcomes among economic empowerment interventions HIV prevention for adolescent girls and young women in sub-Saharan Africa using the RE-AIM framework as a guide.

Conclusion

Emerging evidence suggests that economic strengthening interventions can be effective in reducing adolescent girls’ and young women’s risks for HIV. RE-AIM assessment showed that economic empowerment intervention provides AGYW with skills to reduce their risk of HIV. Our findings further show that although researchers frequently reported on intervention-specific implementation science outcome components, broad or system-level implementation outcome indicators of these interventions are scarce. Considering the critical role of these implementation factors in enhancing the ultimate impact of combination economic strenghtening intervention on HIV prevention among AGYW in SSA, the scarcity of evidence is a concern. We recommend the use of RE-AIM components in future EE HIV interventions targeting AGYW, with special consideration given to factors relevant to the adoption, implementation (such as implementation cost, adoption rate, and intervention fidelity) and long-term sustainability of these interventions in SSA. We further suggest the measurement of other implementation science outcomes beyond RE-AIM indicators to provide a holistic indicator of factors and measures to promote intervention scale-up, dissemination and sustainability. Overall, the findings of this systematic review and the use of the RE-AIM framework, have the potential to accelerate the tempo of implementation and dissemination of evidence-based interventions for addressing HIV prevention among at-risk AGYW in SSA.