Web-Based Service Provision of HIV, Viral Hepatitis, and Sexually Transmitted Infection Prevention, Testing, Linkage, and Treatment for Key Populations: Systematic Review and Meta-analysis

Background Despite the growth of web-based interventions for HIV, viral hepatitis (VH), and sexually transmitted infections (STIs) for key populations, the evidence for the effectiveness of these interventions has not been reported. Objective This study aimed to inform the World Health Organization guidelines for HIV, VH, and STI prevention, diagnosis, and treatment services for key populations by systematically reviewing the effectiveness, values and preferences, and costs of web-based outreach, web-based case management, and targeted web-based health information for key populations (men who have sex with men, sex workers, people who inject drugs, trans and gender-diverse people, and people in prisons and other closed settings). Methods We searched CINAHL, PsycINFO, PubMed, and Embase in May 2021 for peer-reviewed studies; screened abstracts; and extracted data in duplicate. The effectiveness review included randomized controlled trials (RCTs) and observational studies. We assessed the risk of bias using the Cochrane Collaboration tool for RCTs and the Evidence Project and Risk of Bias in Non-randomized Studies of Interventions tools for non-RCTs. Values and preferences and cost data were summarized descriptively. Results Of 2711 records identified, we included 13 (0.48%) articles in the effectiveness review (3/13, 23% for web-based outreach; 7/13, 54% for web-based case management; and 3/13, 23% for targeted web-based health information), 15 (0.55%) articles in the values and preferences review, and 1 (0.04%) article in the costs review. Nearly all studies were conducted among men who have sex with men in the United States. These articles provided evidence that web-based approaches are as effective as face-to-face services in terms of reaching new people, use of HIV, VH, and STI prevention services, and linkage to and retention in HIV care. A meta-analysis of 2 RCTs among men who have sex with men in China found increased HIV testing after web-based outreach (relative risk 1.39, 95% CI 1.21-1.60). Among men who have sex with men in the United States, such interventions were considered feasible and acceptable. One cost study among Canadian men who have sex with men found that syphilis testing campaign advertisements had the lowest cost-per-click ratio on hookup platforms compared with more traditional social media platforms. Conclusions Web-based services for HIV, VH, and STIs may be a feasible and acceptable approach to expanding services to key populations with similar outcomes as standard of care, but more research is needed in low-resource settings, among key populations other than men who have sex with men, and for infections other than HIV (ie, VH and STIs).

1 Although authors do not report specific details of the randomisation and/or allocation process are not documented, the study protocol states that 8 cities were randomized using SAS software. Because cities were randomized (not individuals), we did not judge there to be issues with blinding.
2 Loss to follow up between baseline and final follow-up assessment was > 90% but similar rates across arms.
3 Authors used a stratified block randomization scheme.
4 All participants in the intervention arm had the app downloaded onto their phones; levels of engagement with the app varied (i.e. viewing/opening WeTest messages).
5 Given the intervention of interest (online outreach), blinding was not possible for participants and personnel. However, deviations from the intended intervention due to lack of blinding were not documented, and any potential deviations were unlikely to have affected the outcome. However, the outcomes (use of prevention services, use of testing services for HIV/VH/STIs) were judged as potentially influenced by lack of blinding since consistent condom use and HIV testing outcomes were self-reported.
6 While baseline confounding was not adjusted or, demographics and time-varying confounding were somewhat adjusted for (Grindr outreach between October 2012 and March 2013 was compared with the number of contacts initiated through traditional outreach activities during the same 6-month period from October 2011-March 2012; race and age were adjusted for in analysis). However, two years separated the two data collection time periods, and through use of Grindr for outreach was the main intervention, the public health dept may have adjusted other methods of reaching out to potential HIV/STI service users.
7 The outcome data not available for nearly all participants. Because the intervention (outreach via Grindr) was not obvious as research, retention rate after the outreach was identified as a health worker instead of a potential sexual partner was less than 80%, but engagement a secondary outcome after the primary outcome of outreach (# contacts made). There was no denominator (population) to calculate rates; only number of contacts -which may have included duplicates. For the uptake of prevention/testing services outcome, no comparisons to pre-Grindr outreach were reported.
8 Given the intervention of interest, blinding was not possible for personnel, whether blinding for participants was not mentioned in this study. Personnel are well-trained. The outcome measure was unlikely to be influenced by knowledge of the intervention received, the methods of outcome assessment comparable across intervention groups. Among the 120 participants enrolled in the intervention, 60 were lost to follow up and did not complete the intervention. The 60 who completed the intervention were considered the intervention group, and the 60 who were lost to follow up (67% for unknown reasons, 10% for phone loss, 15% for moving out of jurisdiction, and other reasons) were considered the control group. However, characteristics of the control group were not different from the overall sample (or the intervention group).

Online case management
8 Selection of participants into the study was not based on participant characteristics observed after the start of the intervention, but the start of follow-up and start of intervention did not coincide for most participants (intervention group was those who successfully received the case management video conference in time for their release from prison; control group was those who did not receive the video conference in time). 1 Details of the randomisation and/or allocation process are not documented. The study says: "Next, participants were randomized into either the tailored experimental condition or the non-tailored control condition." with no additional information on concealment or any baseline differences between groups.

Targeted online health information
2 Given the intervention of interest (targeted health information), blinding was not possible for participants and personnel. However, deviations from the intended intervention due to lack of blinding were not documented, and any potential deviations were unlikely to have affected the outcome. However, the outcomes (uptake of prevention services, uptake of testing services for HIV/VH/STIs) were judged as potentially influenced by lack of blinding. For uptake of prevention services, the outcome was measured by self-reported vaccination for hepatitis A or B, HPV, or meningococcal meningitis. For uptake of testing services for HIV/VH/STIs, the outcome was measured through self-reported testing by disease.