Digital health interventions are a promising approach for improving adolescent family planning and reproductive health (FP/RH) outcomes in low-resource settings. They offer the potential to increase access to accurate health information and comprehensive health products and services through direct-to-consumer channels, mitigating health misinformation and extending quality healthcare coverage. Studies in sub-Saharan Africa (SSA) suggest that such interventions can improve FP/RH knowledge, increase contraceptive use, and reduce adolescent pregnancy. The United Nations advocates for equitable, rights-based digital health technologies to reduce information disparities and increase sexual health literacy, aligning with Sustainable Development Goals. Digital interventions may be particularly suitable for adolescents due to their preference for private information channels and early technology adoption. In SSA, limited youth-friendly FP/RH information and services contribute to low contraceptive knowledge and use, high rates of unintended pregnancy, and persistently high HIV incidence among young people, who account for ~30% of new HIV infections, despite progress in the general population. These detrimental FP/RH outcomes are linked to school dropout, prevalent in SSA. Harmful gender norms and lack of access to quality FP/RH information, services, and products further exacerbate these challenges, preventing many young people from completing school HIV-free and avoiding unintended pregnancy. However, rigorous impact evaluations of digital FP/RH interventions for adolescents, particularly comparing different implementation models using community-led approaches, remain scarce. CyberRwanda is an innovative digital FP/RH intervention developed using a multi-year, participatory, human-centered design (HCD) process to address gaps in FP/RH care in Rwanda. It's an online platform with three main features: STORIES (webcomics on relationships, sex, consent, contraceptive use, adolescent pregnancy, and HIV/STIs); LEARN (informational guides, videos, audio, FAQs on menstruation, puberty, gender equality, relationships, consent, education, careers, money, and goal setting); and SHOP (a direct-to-consumer online store for discreet ordering of health products from nearby pharmacies, health posts, and health facilities trained in youth-friendly services). The platform also includes a facility finder. CyberRwanda was implemented in schools and youth centers, but this study focuses solely on school-based implementation, using two models: self-service (self-guided access on provided tablets) and facilitated (self-guided and group access with peer-led weekly clubs). A mixed-methods pilot study demonstrated the acceptability and feasibility of both models and high adolescent demand. This 24-month Hybrid Effectiveness-Implementation trial in 60 secondary schools aimed to evaluate CyberRwanda's effectiveness on individual-level FP/RH outcomes (modern contraceptive use, childbearing, and HIV testing), compare implementation model effectiveness, and assess student engagement and secondary outcomes (FP/RH knowledge, attitudes, beliefs, self-efficacy, and behavior).

Several studies in sub-Saharan Africa (SSA) have explored the effectiveness of digital family planning and reproductive health (FP/RH) interventions for adolescents. Rokicki et al. (2017) demonstrated the positive impact of a text-messaging program on adolescent reproductive health in Ghana. Nuwamanya et al. (2020) showed a mobile phone application's effectiveness in increasing access to sexual and reproductive health information, goods, and services among university students in Uganda. Other research highlights the potential of game-based learning and gamification (Haruna et al., 2018) and user-driven, web-based applications (Sharma et al., 2022) to improve sexual health knowledge and communication. mHealth interventions using SMS reminders have also shown promise in improving postpartum contraceptive use (Harrington et al., 2019). These studies, while showing promise, often lack the rigor and scale of the current study, which contributes significantly to the limited evidence base on effective interventions for adolescent sexual health in SSA.

This study employed a three-arm, parallel-group cluster randomized, Type 2 Hybrid Effectiveness-Implementation trial design across eight Rwandan districts. The study design and analysis plan were preregistered (clinicaltrials.gov NCT04198272). Eligible secondary schools (excluding boarding schools) were those within 4.5 kilometers of a pharmacy, at least 1.5 kilometers from another secondary school, and with at least 150 students. Within schools, eligible students were aged 12–19, in school levels S1 and S2, and willing to provide consent (18–19 years) or assent and parental consent (<18 years). Sixty schools were randomly selected and assigned to one of three arms: control (receiving only the government-approved Comprehensive Sexuality Education), CyberRwanda self-service (self-guided access to the platform with tablets, internet, and student ambassadors), and CyberRwanda facilitated (self-service components plus peer-facilitators leading weekly clubs). The randomization was stratified by district using a two-stage participatory approach, ensuring balance across districts. Schools were randomly assigned using colored balls drawn from an opaque bag in community events. Due to the intervention's nature, blinding was not possible. Data collection involved baseline, midline, and endline surveys (approximately 45 minutes) administered by trained research assistants using tablets and Qualtrics software. Surveys measured sociodemographic and school characteristics at baseline, and FP/RH knowledge, attitudes, beliefs, self-efficacy, sexual behavior, contraceptive use, pregnancy history, and HIV testing at follow-up points. Three primary outcomes were defined: modern contraceptive method uptake among females, initiation of childbearing among females, and HIV testing among all participants. Secondary outcomes were categorized into knowledge, attitudes and beliefs, self-efficacy, and behaviors. Power calculations were conducted using a simulation-based approach, focusing on modern contraceptive use among females, targeting a sample of 50 female participants per school (3000 total). An equal number of male participants were also recruited (total of 6000 participants). Statistical analyses involved descriptive statistics, generalized linear mixed models (log-binomial or log-Poisson with robust standard errors) to estimate prevalence ratios, and inverse probability of censoring weighting to account for attrition. Random intercepts for schools were included to account for clustering. A joint test of no difference was conducted for primary outcomes, comparing CyberRwanda arms to the control arm. Sensitivity analyses were conducted using complete case models for secondary outcomes and a subgroup analysis of sexually active participants.

Of the 6,078 participants enrolled, 5,552 (91.3%) completed the 24-month endline surveys. At endline, 26.6% reported ever having sexual intercourse (34.7% of males, 19.0% of females). The CyberRwanda intervention did not significantly affect the primary outcomes of modern contraceptive use (PR: 1.04, 95% CI: 0.76, 1.42), childbearing (PR: 1.33, 95% CI: 0.71, 2.50), and HIV testing (PR: 1.00, 95% CI: 0.91, 1.11) in the full sample. However, among sexually active participants, the facilitated CyberRwanda arm showed significantly higher modern contraceptive use. Most indicators of school-level fidelity and individual-level engagement were similar across models but slightly higher in the facilitated arm. 75.9% of participants in CyberRwanda schools reported ever using the platform, with STORIES being the most used feature. In secondary outcome analyses, both CyberRwanda arms showed significantly higher knowledge of when emergency contraception could be taken. The facilitated arm demonstrated significantly more favorable beliefs about condom use and greater confidence in getting a partner to use contraceptives. Sensitivity analyses focusing on sexually active participants revealed significantly higher current modern contraceptive use in the combined CyberRwanda arms compared to the control. This was driven by increased condom use among sexually active males and increased use of modern methods other than condoms among sexually active females. The facilitated model showed stronger effects on FP/RH related attitudes and behaviors. Despite high acceptability and access, SHOP orders were infrequent, likely due to low prevalence and frequency of sexual activity within the cohort.

This study provides valuable data on a large-scale digital intervention for adolescent sexual health in Rwanda. While the study did not find significant impacts on primary outcomes in the full sample, the increased modern contraceptive use among sexually active participants in the CyberRwanda arms, particularly the facilitated arm, suggests potential effectiveness. The positive shifts in intermediate outcomes, like emergency contraceptive knowledge, condom beliefs, and confidence in partner communication, align with the hypothesized impact pathway. The facilitated model's stronger impact on intermediate outcomes points towards the potential benefits of combining digital interventions with in-person support and peer education. The infrequent use of the SHOP feature highlights the challenge of demand in a population with low prevalence of sexual activity and suggests that a long-term follow-up is needed to observe the effect after the study population becomes more sexually active. The findings contribute to the limited evidence base on digital interventions for adolescent sexual health in SSA and suggest the potential for CyberRwanda to improve adolescent FP/RH outcomes.

CyberRwanda did not significantly impact primary FP/RH outcomes in the full sample, but demonstrated significantly higher modern contraceptive use among sexually active participants. The intervention showed strong acceptability and engagement, with positive shifts in intermediate outcomes. The facilitated model's stronger effects suggest the value of integrating in-person components. Future research should focus on the cost-effectiveness of different models and longer-term effectiveness as the study population becomes more sexually active.

The study's limitations include the potential for selection bias due to attrition, social desirability bias in self-reported sensitive outcomes, and limited generalizability to out-of-school youth. Technological challenges and high student-to-tablet ratios may have impeded access and engagement, particularly the SHOP feature's use. The limited number of quantitative metrics to assess in-person engagement in the facilitated model was also a limitation. Addressing these limitations can enhance future implementations and evaluations of the intervention.