Effects of Digital Health Interventions to Promote Safer Sex Behaviors Among Youth: Systematic Review and Bayesian Network Meta-Analysis

TL;DR

This study compares digital health interventions for promoting safer sex among youth and finds that their effectiveness varies by type and outcome.

Contribution

The first Bayesian network meta-analysis comparing digital health interventions for HIV/STI prevention in youth.

Findings

Telecommunication-based interventions improved condom use at last sex compared to non-digital interventions.

Interactive online and static web-based interventions outperformed telecommunication-based interventions for consistent condom use.

Non-digital interventions were associated with fewer STIs than static web-based interventions.

Abstract

Youth aged 15‐24 years carry a disproportionate HIV/sexually transmitted infections (STIs) burden. In recent years, different modalities of digital health interventions (DHIs) have been explored to promote safer sex behaviors among youth, but their comparative effectiveness across modalities and relative to nondigital interventions (NDIs) remains unclear.

This study aimed to compare DHI modalities on safer sex behaviors and HIV/STI incidence, rank modalities using Bayesian network meta-analysis (NMA), and position their effectiveness relative to NDIs.

A systematic review and Bayesian NMA of randomized controlled trials were conducted by comprehensively searching PubMed, EMBASE, Web of Science, and Cochrane Library (inception to November 2025). Eligible studies were those that enrolled youth aged 15‐24 years and evaluated mobile app-based intervention, telecommunication-based intervention (TCI), static web-based intervention (SWI), or interactive online-based intervention (IOI)—with an NDI or another DHI. Primary outcomes were condom use at last sexual contact, consistent condom use, and proportion of condom use. Secondary outcomes included condom use self-efficacy, number of sexual partners, and STI incidence (including HIV). Risk of bias was assessed with the Cochrane Risk of Bias 2 tool, and certainty of evidence with GRADE/CINeMA (Confidence in NMA). Bayesian random-effects NMAs estimated odds ratios (ORs) with 95% credible intervals (CrIs), and complementary frequentist NMAs provided 95% CIs and 95% prediction intervals.

Twenty-four randomized controlled trials (20,134 participants) were included, forming treatment networks across 5 intervention types. TCI was the only intervention that significantly improved condom use at last sex compared with NDI (OR 1.13, 95% CrI 1.02‐1.26). For consistent condom use, SWI and IOI outperformed TCI (SWI vs TCI: OR 1.77, 95% CrI 1.03‐3.06; IOI vs TCI: OR 1.68, 95% CrI 1.02‐2.76). For the proportion of condom use, IOI outperformed SWI (OR 1.34, 95% CrI 1.01‐1.80), and mobile app-based intervention ranked highest in probability rankings, though estimates lacked precision. For STI incidence, NDI was associated with fewer STIs than SWI (OR 0.61, 95% CrI 0.46‐0.82).

This is the first NMA to compare the effectiveness of DHIs on condom use and HIV/STI outcomes among youth populations. It demonstrates that the impact of DHIs on HIV prevention varies substantially by intervention modality and outcome type. While TCI demonstrates the most consistent improvement in condom use at last sex, SWI and IOI may be more effective for promoting consistent condom use, though estimates remain imprecise. However, wide prediction intervals and low-certainty evidence suggest that self-reported behavioral changes may not translate into reductions in HIV/STI incidents without integration with offline services and broader structural support. Future trials might consider including standardized outcome indicators and longer follow-up to generate more precise estimates of the effectiveness of DHIs and guide generalization of youth-centered digital HIV/STIs prevention.

Figures