Review
Abstract
Background: There remain unmet HIV prevention needs in China, particularly among gay, bisexual, and other men who have sex with men. Technology-based interventions are increasingly used in HIV prevention worldwide.
Objective: We aimed to conduct a systematic review and meta-analysis of studies to assess the effectiveness of technology-based HIV prevention interventions to improve HIV testing and consistent condom use in China.
Methods: We searched English-language (PubMed and MEDLINE, Embase, and Web of Science) and Chinese-language (Wanfang, WEIPU, and China National Knowledge Infrastructure) databases for technology-based HIV prevention intervention studies published between January 1, 2004, and September 30, 2021. Eligible studies were technology-based HIV prevention intervention studies with outcomes of HIV testing or condom use among men who have sex with men or transgender women using randomized controlled or nonrandomized pretest-posttest designs in China. The intervention technologies identified were apps, web pages, and other types of electronic communications (eg, email, SMS text messages, and video messages). A Bayesian meta-analysis was conducted to estimate the pooled effect size and 95% credible interval (CrI). We added study and intervention features as covariates to explore their associations with the study effects. Study quality was assessed using the integrated quality criteria for review of multiple study designs. Publication bias was assessed using funnel plots and robust Bayesian meta-analyses.
Results: We identified 1214 and 1691 records from English-language and Chinese-language databases, respectively. A total of 141 records entered full-text screening, and 24 (17%) studies were eligible for the review. Approximately half (14/24, 58%) of the interventions were delivered through social media platforms, predominantly using message-based communication. The remaining studies used email and web-based platforms. The pooled effect sizes estimated were an absolute increase of 20% (95% CrI 10%-30%) in HIV testing uptake and an absolute increase of 15% (95% CrI 5%-26%) in consistent condom use. The pooled point estimate of the effect of randomized controlled trials was smaller than that of nonrandomized studies for HIV testing uptake (16% vs 23%) and consistent condom use (10% vs 19%), but their CrIs largely overlapped. Interventions lasting >6 months were associated with a 35% greater uptake of HIV testing (95% CrI 19%-51%) compared to those lasting 6 months.
Conclusions: Technology-based HIV prevention interventions are promising strategies to improve HIV testing uptake and consistent condom use among men who have sex with men in China, with significant effects found across a broad array of studies and study designs. However, many studies in this review did not include randomized designs or a control group. More rigorous study designs, such as randomized controlled trials, are needed, with outcome measurements that address the limitation of self-report outcomes to inform the development and implementation of future intervention programs.
Trial Registration: PROSPERO CRD42021270856; https://www.crd.york.ac.uk/PROSPERO/view/CRD42021270856
doi:10.2196/63111
Keywords
Introduction
Background
HIV remains an important health concern and an area of public health efforts in China, with a particular focus on HIV transmission among gay, bisexual, and other men who have sex with men due to high and sustained risk of HIV infection. A cross-sectional survey of 47,000 men who have sex with men from 61 cities in China in 2008 to 2009 found an overall HIV prevalence of 4.9% []. Since 2010, the HIV prevalence among men who have sex with men has emerged as the highest population-specific prevalence in China, increasing from 5.7% to 7.8% by 2014 according to China’s national HIV sentinel surveillance system []. Studies have suggested that transgender women may face an even higher HIV prevalence (>10%) than cisgender men who have sex with men [-].
Applying technology-based approaches to deliver HIV prevention information has been proposed as a critical part of engaging sexual minority populations that are disproportionately missed in HIV prevention and care in China []. These technology-based interventions include social media messaging [], web-based HIV referral services [], chatbots [], and gay social networking apps [-]. Given that >80% of men who have sex with men do not disclose their sexual orientation to health care providers in China [], technology-based approaches both allow for improved targeting of interventions to this hidden population and represent an opportunity to offer men who have sex with men and transgender women easier access to HIV prevention and care services while reducing fears related to stigma and discrimination. A literature review published in 2019 found that multilevel technology-based interventions in China were effective in delivering HIV prevention and care services []. In addition to the various modes of technology, there is a popular all-in-one smartphone app in China, WeChat, providing opportunities to develop innovative interventions. WeChat offers multiple functions, including group chat messaging, voice and video calls, gaming, wallet and payments, and location sharing, in a single platform. WeChat has billions of monthly active users in China []. WeChat offers developers tools to make WeChat-based mini apps that do not have to be downloaded and installed from an app store []. These mini apps can include different functions and features in WeChat and can deliver complex eHealth interventions such as secondary distribution of HIV self-testing kits [].
Several systematic reviews have identified increased use of technology-based HIV interventions over the past 2 decades and have documented mixed evidence regarding the efficacy of such interventions worldwide. After reviewing studies published in 2007 to 2019, a systematic review of technology-based HIV interventions described a trend of increasing eHealth modalities. Interventions were reported to be efficacious in increasing safe sex behaviors in the short term; only a quarter of the interventions reported maintaining the behavior change over a year []. Another systematic review from 2015 to 2020 found more pilot or quasi-experimental study designs than efficacy randomized controlled trials (RCTs). The review found promising feasibility and acceptability outcomes among pilot studies and some efficacious outcomes among RCTs []. In addition to describing the types of technology used to deliver interventions and the reported behavior changes, a few recent meta-analyses have assessed the effect sizes of technology-based interventions. A meta-analysis of eHealth interventions among men who have sex with men found a small but significant effect size for all 3 behavioral outcomes over the reviewed study period: any condomless anal intercourse with nonpaying male partners (Cohen d=−0.21; P<.001), HIV testing (Cohen d=0.38; P<.001), and having multiple sex partners (Cohen d=−0.26; P=.02) []. Another meta-analysis reviewing studies published between 2010 and 2018 estimated that the proportion of HIV testing uptake among study participants who were exposed to technology-based interventions meant to increase testing was 1.5 times higher than that among unexposed study participants []. Across the aforementioned 2 reviews, a number of intervention features were associated with greater impact on HIV testing, including interventions delivered through mainstream social media, providing self-testing kits or HIV testing referral services, placing interventions on interactive platforms, involving target users, and providing longer periods of intervention exposure [,]. Both meta-analyses reported high heterogeneity across their reviewed studies (I2=83.3% [] and I2=65.2% []). So far, all these systematic reviews have included technology-based HIV prevention interventions worldwide, but none have been restricted to China. Given the numerous technology-based interventions and the unique WeChat platform in China, there is a need for a systematic assessment to document these intervention strategies and effects.
Objectives
We aimed to describe technology-based HIV prevention interventions among men who have sex with men and transgender women in China, document the intervention strategies, estimate intervention effects, and explore the relationships between intervention features and effect size. This review is narrower in geographic focus than previous reviews, allowing for an understanding of local intervention effects in China. In previous reviews, small sample sizes and high heterogeneity across studies were often the major concerns given the use of conventional meta-analysis. Therefore, we conducted a Bayesian meta-analysis instead of the conventional frequentist meta-analysis to improve the precision of the pooled effect size estimation and to provide the probability of the intervention effect being >0 [-].
Methods
The protocol for this review was registered in PROSPERO (registration number CRD42021270856). This review was conducted following the guidance of the Cochrane Handbook for Systematic Reviews of Interventions and was reported following the guidance of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist () [].
Eligibility Criteria
The summary of the eligibility criteria for this systematic review is presented in 
Textbox 1. In this review, we used a broad definition of technology to include the internet (eg, web pages, social media platforms, and smartphone apps) and other types of electronic communication (eg, SMS text messages and video messages). The inclusion criteria and search strategy were informed by the population, intervention, comparison, and outcome framework [], including appropriate populations for the review’s goals (eg, HIV-negative men who have sex with men or transgender women residing in China, including Taiwan, Hong Kong, and Macau), use of a technology-based intervention, an eligible study type (eg, RCTs and nonrandomized designs such as 1-group pretest-posttest studies and pretest-posttest studies with a nonequivalent comparison group), eligible study outcomes (eg, HIV seroconversion, HIV testing, condom use, pre-exposure prophylaxis initiation, and pre-exposure prophylaxis awareness and willingness), and eligible data collection time frame (January 1, 2004, to September 30, 2021) and publication time frame (January 1, 2004, to September 30, 2021). Exclusion criteria were studies that exclusively used phone calls or in-person interventions, focused on persons living with HIV, or did not have quantitative HIV prevention outcomes (eg, qualitative outcomes).
Inclusion criteria
- Article or study type:
- Randomized controlled trials
- Nonrandomized trials with a pretest and posttest design
- Study population:
- HIV-negative men who have sex with men
- Transgender women
- Study setting:
- China (including Taiwan, Hong Kong, and Macau)
- Study intervention:
- The internet (eg, web pages, social media platforms, and smartphone apps)
- Electronic communication (eg, SMS text messages and video messages)
- Main outcomes:
- HIV seroconversion
- HIV testing
- Condom use
- Pre-exposure prophylaxis (PrEP) initiation
- PrEP awareness and willingness
- Quality criteria:
- Peer-reviewed journal articles
- Conference abstracts of the International AIDS Society (IAS) Conference and the Conference on Retroviruses and Opportunistic Infections (CROI) or the Chinese National Conference of HIV/AIDS
- Language:
- English or Chinese
- Data collection period:
- January 1, 2004, to September 30, 2021
- Publication period:
- January 1, 2004, to September 30, 2021
Exclusion criteria
- Article or study type:
- Reviews, narratives, commentaries, and editorials
- Qualitative studies
- Study population:
- Other populations
- Study setting:
- Outside China
- Study intervention:
- Interventions that did not use any form of internet or electronic communication
- Main outcomes:
- Studies that exclusively used phone calls or in-person interventions
- Quality criteria:
- Studies that were not published in a peer-reviewed journal
- Conference abstracts that were not presented in the IAS Conference, CROI, or Chinese National Conference of HIV/AIDS
- Language:
- All other non–English or Chinese languages
- Data collection period:
- All periods outside January 1, 2004, to September 30, 2021
- Publication period:
- All periods outside January 1, 2004, to September 30, 2021
Study Identification and Selection
We searched PubMed (including PubMed Central and MEDLINE), Embase, Web of Science, Wanfang, WEIPU (including the Chinese Scientific Journals Database–), the China Academic Journals Full-Text Database and China National Knowledge Infrastructure, and the Chinese Biomedical Literature Database for studies published between January 1, 2004, and September 30, 2021. Conference abstracts were searched from the online archives of the International AIDS Society Conference and the Conference on Retroviruses and Opportunistic Infections to obtain English-language publications, as well as the archives of the Chinese National Conference of HIV/AIDS for Chinese-language publications. Search terms were first developed in English and then translated into academic Chinese terminology for searching Chinese-language databases. A full description of the search strategy can be found in .
Records identified from the keyword search were managed in Covidence (Veritas Health Innovation). The selection procedure followed the PRISMA guidelines. Duplicate records were checked across English- and Chinese-language literature and were removed (n=371). In total, 2 researchers screened titles and abstracts independently to exclude records that clearly did not meet the inclusion criteria or clearly met the exclusion criteria. The eligibility of the remaining records was assessed through full-text articles. During the full-text review, all the articles (n=141) were retrieved and reviewed by 2 researchers independently. Discrepancies were resolved through discussion among the 2 researchers and a senior researcher. After 115 articles were removed according to the exclusion criteria, a total of 24 (20.9%) articles (n=11, 46% articles in English; n=12, 50% articles in Chinese; and n=1, 4% abstracts in English) were included in the final review.
Study Quality Assessment
Study quality was assessed using the integrated quality criteria for review of multiple study designs (ICROMS), a tool developed for a wide range of study designs, including RCTs, pretest-posttest studies, cohort studies, and interrupted time-series analyses []. The ICROMS consists of two components: (1) a scoring system for quality of the study design and (2) a decision matrix that provides mandatory criteria and minimum requirement scores for specific study designs related to the robustness of the study. The ICROMS assessment has 7 dimensions, ranging from study aims to managing bias. Each dimension has 3 to 7 specific criteria for quality assessment. This measure has been used in previous systematic reviews of eHealth interventions [,]. Each study was assessed by 2 researchers independently. Discrepancies were resolved through discussion between the 2 researchers.
Data Extraction
Data extraction for each study was conducted by 2 coders (WH, SSW, KN, and YW) independently. Data extraction for studies published in Chinese was only conducted by bilingual coders (WH, KN, and YW). A standard, Microsoft Excel–based (Microsoft Corp) tool was used to extract data on publication year, language (English or Chinese), study design, sample size, technology mode, behavioral theory applied, intervention development, intervention delivery, outcome measurement (eg, the recalled period, such as in the last time, and in the previous month), and results and analyses (eg, number of consistent condom uses, number of HIV tests, effect size measurement, and reported effect size). Several intervention features were further coded for use as covariates. These features included whether the intervention was theoretically based (ie, developed based on a behavior change theory), whether the intervention had active engagement with users (eg, using an interactive dialogue box, motivational interviewing, and one-on-one counseling sessions), whether the intervention was designed to have more than a 1-time interaction, whether intervention development and delivery involved the target population, the length of the intervention (eg, <6 months, 6 months, and >6 months), and the length of study follow-up (eg, 3 months, 6 months, and 12 months). Studies that did not report these features were coded as no. Extracted data were compared, and discrepancies were discussed with a third coder until consensus was reached.
Data Synthesis
The primary outcomes for this review were HIV testing uptake and consistent condom use. We defined HIV testing uptake as the proportion of participants who reported receiving a facility-based HIV test during the study period or who used an HIV test kit during the study period. Consistent condom use was defined as the proportion of participants with no condomless anal sex with any partners during the study period. When the articles reported multiple follow-up assessments, the longest follow-up interval was used for the outcomes of interest. Effect sizes were calculated as mean differences either between the intervention and control or comparison groups or between the pretest and posttest time points depending on the study design. The uncertainty was assessed using SEs. SEs were extracted from the studies that reported them; for studies that did not report SEs, we manually calculated them by comparing either the intervention and control groups in the case of RCTs or the pretest and posttest outcomes or the intervention and comparison groups in the case of nonrandomized studies. Therefore, 1 HIV testing uptake outcome or 1 consistent condom use outcome were obtained for each study. Studies using noninferiority designs (2/24, 8%) were excluded from the meta-analysis given the goal of the review to estimate overall effect sizes. We decided to use the absolute effect instead of the relative effect because the absolute effect is easier to interpret for clinical practice; in contrast, relative measures sometimes provide misleading estimates [].
Given the expected heterogeneity across the studies, a Bayesian random-effects model was used to estimate the pooled effect size and 95% credible interval (CrI) for behavior changes in HIV testing uptake and consistent condom use associated with the interventions []. This model assumes that reported study estimates (y) follow a normal distribution, with the mean equal to the true study effect and the SD equal to the reported SE. The true study effects (θ) were assumed to arise from a normal distribution, with the mean equal to the overall meta-analytic estimate (μ) and study heterogeneity (τ; []). In a robust version of this hierarchical model, we specified the true study effects as arising from a t-distribution (with estimated df) to increase robustness against potential outliers. We chose vaguely informative default priors for all model parameters and conducted prior sensitivity analyses. [] provides more details. All models were estimated using Hamiltonian Monte Carlo sampling []. We ran 4 Markov chains for 20,000 iterations, dropping the first 10,000 iterations as a burn-in phase. Forest plots were used to display the estimates.
To explore relevant study and intervention features that could be associated with the study effect sizes, we added the study and intervention features as covariates to the Bayesian model. Heterogeneity and potential publication bias were evaluated qualitatively through examining contour-enhanced funnel plots, which included the statistical significance of study estimates, and quantitatively through examination of the Bayes factor with robust Bayesian meta-analysis. We also conducted subgroup analyses to estimate the pooled effect sizes and 95% CrIs by RCT and nonrandomized designs. All analyses were conducted using the brms and RoBMA packages in R (version 4.3.1; R Foundation for Statistical Computing).
Results
Study Characteristics
We identified 1214 records from English-language databases and 1691 records from Chinese-language databases published between January 1, 2004, and September 30, 2021. Of these 2905 records, after removing 371 (12.77%) duplicates, 2534 (87.23%) were screened by title and abstract, and a total of 141 (4.85%) records were included in the full-text screening. After the full-text screening, 24 studies were included in the final review. The study selection process and reasons for exclusion are presented in 
Figure 1. 
Table 1 presents the characteristics of each reviewed study, including study design (with dates and sample size), technology type, intervention, theoretical framework and intervention development, intervention delivery and frequency, and reported results. Half (12/24, 50%) of the reviewed studies were published in English, and half (12/24, 50%) were published in Chinese. Most study sites (19/24, 79%) were in mainland China. Other study sites included Hong Kong (3/24, 12%) [-] and Taiwan (2/24, 8%) [,]. Although transgender women were included in our search strategy, the eligible studies were restricted to men who have sex with men. Over half (13/24, 54%) of the studies were RCTs. Of the 13 RCTs, 8 (62%) were 2-group parallel RCTs, 2 (15%) were 3-group parallel RCTs [,], 2 (15%) were noninferiority RCTs [,], and 1 (8%) was a stepped-wedge cluster RCT []. Among the nonrandomized studies (11/24, 46%), 9% (1/11) used a pretest-posttest design with a nonequivalent comparison group [], 45% (5/11) used 1-group pretest-posttest designs [-], and 45% (5/11) used pretest-posttest designs with repeated cross-sectional surveys [-].
| Study and location | Study design, period, and sample size | Intervention technology | Intervention and control group | Theoretical framework and intervention development | Intervention delivery and frequency | Reported resultsa | |
| Traditional RCTsb with 1 intervention group and 1 control group (n=8; n=6 in English and n=2 in Chinese) | |||||||
| Cheng et al [], 2019; mainland China | Nonblind RCT; September 2010-June 2011; N=1100 | Web pages | Intervention: online interactive scenarios about HIV-related risk behaviors and health messages; control: standard HIV referral service only | TPBc; developed via mixed methods formative research | A total of 4 weeks—1-time delivery of 5 scenarios via web-based interactive dialogue box and weekly health messages sent via email for 3 weeks | Intervention vs control, BLd to 6-month FUe—condomless anal sex in the previous 3 months decreased (difference in proportion=9.3%, 95% CI 1.1%-17.5%) | |
| Chiou et al [], 2020; Taiwan | RCT; August 2015-May 2017; N=300 | App—developed by the research team for men who have sex with men | Intervention: use of a multifunction SBSf app for 6 months; control: no intervention | IMBg model; developed through a qualitative study soliciting men who have sex with men’s needs, men who have sex with men user reviews, and expert review on content validity | A total of 24 weeks—a message reminding the user of the app’s functions was sent every 2 weeks, and a quiz and 2 activities related to HIV testing, safe sex, and recreational drug use were sent every 3 weeks | Intervention vs control, BL to 6-month FU—HIV and syphilis testing frequency increased (mean difference=0.217; SE 0.08); percentage of condom use during anal sex in the previous 3 months increased (difference=20.7%; SE 0.06); other outcomes: disease and safe behavior knowledge increased (difference=2.13; SE 0.21) | |
| Lau et al [], 2008; Hong Kong | RCT; study dates not specified; N=477 | Intervention: health messages in the form of graphics, HIV-related risk behavior electronic monthly form with tailored feedback, and peer counselors; control: no intervention | Did not report that any theory was used; developed by a panel consisting of the authors | A total of 6 months—a health message was sent via email every 2 weeks and an e-form with feedback (“report card”) was sent every month | Intervention vs control, BL to 6-month FU—no change in HIV testing in the previous 6 months; no change in condom use during anal sex in the previous 6 months; other outcomes: no change in HIV- or STDh-related knowledge and perceptions | ||
| Wang et al [], 2018; Hong Kong | Nonblind RCT; study dates not specified; N=430 | App—a live chat app | Intervention: an online video promoting HIV testing, another online video and an MIi promoting HIVSTj with online real-time instructions and pretest and posttest counseling (HIVST-OICk), and a free HIVST kit; control: an online video promoting HIV testing | HBMl; developed through literature review and a stakeholder panel discussion | 1-time—online HIV testing promotion video (3 minutes), online HIVST-OIC promotion video (4 minutes), and a brief MI (15 minutes) | Intervention vs control, BL to 6-month FU—HIV testing rate increased (RRm=1.77, 95% CI 1.54-2.03; ARRn=39.1%, 95% CI 31.3%-46.9%) | |
| Yun et al [], 2021; mainland China | Double-blind RCT; October 2017-June 2018; N=192 | App—WeChat | Intervention: a comprehensive intervention package including 4 components—HIV risk assessment, HIV testing facility recommendation, free condoms and HIVST kit, and health education website; control: a link to an HIV health education web page | AIDS Risk Reduction Model; did not report the process of intervention development | A total of 3 months—information related to the intervention was sent every 4 weeks | Intervention vs control, BL to 3-month FU—no change in HIV testing in the previous 3 months; condom use among casual partners in the previous 3 months increased (ORo 2.81, 95% CI 1.23-6.39) | |
| Zhu et al [], 2019; Hefei | RCT; September 2017-June 2018; N=100 | App—WeChat | Intervention: WeTest program plus watching a brief video demonstrating the use of the HIVST kit and results interpretation and 2 oral HIVST kits with standard information about regular HIV testing; control: only the demonstration video and 2 oral HIVST kits with standard information about regular HIV testing | IMB model; developed through formative in-depth interviews for intervention content, cognitive interviews for refining content, and a pilot test of the beta version of the WeTest program | A total of 6 months—2 weekly health messages about personal stories, data on HIV and STIp infections among men who have sex with men, national policies, and general health concern stories of men who have sex with men and 1-time instructions for oral HIVST kit use | Intervention vs control, BL to 6-month FU—HIV testing in the previous 6 months increased (adjusted RR=1.99, 95% CI 1.03-3.84); oral HIVST in the previous 6 months increased (adjusted RR=2.17, 95% CI 1.08-4.37); no change in consistent condom use in the previous 6 months | |
| Li et al [], 2020; mainland China | RCT; March 2017-April 2019; N=600 | App—WeChat | Intervention: WeChat-based counseling, HIV testing, and test result notification, as well as HIV infection risk evaluation and personal behavior change suggestions; control: on-site intervention, including regular distribution of HIV education materials, in-person class, counseling, and HIV testing | Did not report that any theory was used; did not report the process of intervention development | Did not report the length of the intervention—the HIV infection risk evaluation was sent every week | Intervention vs control, BL to 6-month FU—condomless anal sex in the previous 6 months decreased (from 42% to 16% in the intervention group and from 41% to 31% in the control group; χ21=18.0, P<.05) | |
| Xiao et al [], 2020; Shanghai | Double-blind RCT; July 2017-June 2018; N=200 | App—WeChat | Intervention: men who have sex with men–focused HIV prevention message plus regular HIV prevention message; control: regular HIV prevention message | Did not report that any theory was used; did not report the process of intervention development | A total of 12 months—a weekly regular HIV prevention message and a weekly men who have sex with men–focused HIV prevention message | Intervention vs control at the 12-month FU—no change in consistent condom use in the previous 6 months; other outcomes: knowledge increased in several AIDS knowledge items | |
| RCTs with 2 intervention groups and nontraditional RCTs (n=5; all in English) | |||||||
| Lau et al [], 2016; Hong Kong | Nonblind 3-group parallel RCT; study dates not specified; N=402 | Email and web page | Intervention group 1: STD-related cognitive approach—a video about STD prevention (video 1) and a video about UAIq prevention among men who have sex with men (video 2); intervention group 2: STD-related cognitive plus fear appeal imagery approach—an additional movie with fear appeal and visual imagery (video 3) plus video 1 and video 2; control: HIV-related information–based approach—received factual HIV-related text information but no video | Parallel response model; developed by an interdisciplinary panel with meetings with men who have sex with men | 1-time videos—video 1 and video 2: 5 minutes; video 3: 10 minutes | Intervention group 1 vs intervention group 2, intervention group 1 vs control, and intervention group 2 vs control at the 3-month FU—no change in UAI in the previous 3 months after the intervention across the 3 groups; other outcomes: no significant association between immediate fear-related emotional responses and UAI | |
| Luo et al [], 2021; Beijing, China | 3-group parallel RCT; October 2017-September 2018; N=9280 | App—Blued | Intervention group 1: HIV risk assessment and tailored feedback plus routine HIV education; intervention group 2: HIV risk assessment plus routine HIV education; control: routine HIV education | Did not report that any theory was used; did not report the process of intervention development | Did not report the length of the intervention and delivery frequency | Intervention group 1 vs control, BL to 12-month FU—mean number of HIV tests in the previous 12 months increased (IRRr=1.32, 95% CI 1.09-4.58); intervention group 1 vs intervention group 2, intervention group 1 vs control, and intervention group 2 vs control, BL to 12-month FU—no statistically significant differences in the proportion of UAI among the 3 groups | |
| Tang et al [], 2016; mainland China | Noninferiority RCT; study dates not specified; N=721 | Video message | Intervention group 1: a crowdsourced video promoting HIV testing; intervention group 2: a health marketing video promoting HIV testing; control: no control or comparison group | Did not report that any theory was used; the 1-minute crowdsourced video was developed via a crowdsourcing contest; the health marketing video was developed by a marketing company with public health professional guidance | 1-time videos—both videos were 1 minute long | Intervention group 1 vs intervention group 2 at the 3-week FU—no difference in first-time HIV testing, with a noninferiority margin of −3% | |
| Tang et al [], 2019; mainland China | Noninferiority single-blind RCT; November 2015-February 2016; N=1173 | Video message | Intervention group 1: a crowdsourced video promoting condom use; intervention group 2: a social marketing video promoting condom use | Did not report that any theory was used; the 1-minute crowdsourced video was developed via a crowdsourcing contest; the social marketing video was shot by a marketing company following a script by social marketing experts and approved by young men who have sex with men | 1-time—both videos were 1 minute long | Intervention group 1 vs intervention group 2 at the 3-week FU—no difference in proportion of condomless sex, with a noninferiority margin of +10%, and no difference in HIV testing in the previous 3 weeks, with a noninferiority margin of +10%; intervention group 1 vs intervention group 2 at the 3-month FU—no difference in proportion of condomless sex, with a noninferiority margin of +10% | |
| Tang et al [], 2018; Guangdong and Shandong | Stepped-wedge cluster RCT; July 2016-August 2017; N=1381 | App—WeChat and other social media platforms | Intervention: routine CDCs and CBOt promotional efforts, HIV testing promotional image, a free HIVST kit, and a local CBO-led contest for HIV testing stories; control: routine CDC and CBO promotional efforts | Did not report that any theory was used; developed through a nationwide open contest, a regional strategy designathon contests, and local participatory contests | A total of 3 months—6 HIV testing promotional images were sent via WeChat biweekly, 1-time access to a free HIVST kit, and 1-time local CBO-led contest | Intervention vs historical control, BL to 12-month FU—the proportion of HIV testing increased (difference in proportion=8.9%, 95% CI 2.2%-15.5%); the proportion of HIVST increased (RR=1.89%, 95% CI 1.5%-2.38%); no change in facility-based HIV testing; no change in condom use; no change in syphilis testing or anticipated HIV stigma | |
| Nonrandomized pretest-posttest designs (n=11; n=1 in English and n=10 in Chinese) | |||||||
| Ko et al [], 2013; Taiwan | Pretest-posttest design with nonequivalent groups using a repeated cross-sectional survey; October 2010-November 2011; n=1008 at BL and n=1037 at posttest | Web page—Facebook | Intervention: trained iPOLsu shared and exchanged news, video clips, reports, and opinions and connect with others for advice and support on an online iPOL platform built on Facebook using the Web 2.0 two-way communication format; control: another nonequivalent control website with no intervention | The DOIv theory; did not report the process of intervention development | A total of 6 months—no regular schedule; frequent sharing of information and 2-way conversations; a total of 432 posts, 503 comments, and 804 likes on the iPOL platform; and an estimated 959,088 people viewed the posts on the iPOL platform | Intervention vs control, BL to 6-month FU—HIV tests in the previous 6 months increased (43.89% vs 22.31%; χ21=54.8, P<.01); condom use during anal sex with online sex partners increased (34.15% vs 26.19%; χ21=13.4, P<.01) | |
| Liu et al [], 2012; mainland China | Pretest-posttest design using a repeated cross-sectional survey; study dates not specified; n=1293 at BL and n=1014 at posttest | Web page and email | Intervention: HIV prevention messages about risk of HIV infection, safe sex practice, and HIV tests through a web page and via email; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 2 months—3 sessions of HIV prevention education messages; did not report the intervention frequency | BL to 3-month FU—the proportion of lifetime HIV testing increased (49% to 54%; P<.05); the proportion of last-time condom use in sex with men increased (65% vs 71%; P<.05); other outcomes: knowledge increased in several AIDS knowledge items | |
| Xie et al [], 2018; Dongguan | Pretest-posttest design using a repeated cross-sectional survey; study dates not specified; n=1510 at BL and n=1321 at posttest | App—ZANK | Intervention: HIV prevention intervention (did not report any intervention details); control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 2 months—did not report the intervention delivery and frequency | BL to 3-month FU—the proportion of lifetime HIV testing increased (49% vs 55%; χ21=7.7, P<.05); consistent condom use in anal sex with men in the previous 6 months increased (50% vs 55%; χ21=9.8, P<.05); the proportion of last-time condom use in anal sex with men increased (67% vs 73%; χ21=10.8, P<.05); other outcomes: knowledge increased in several AIDS knowledge items | |
| Liu et al [], 2014; Jining | 1-group pretest-posttest design; July 2012-December 2012; N=213 | App—QQ and WeChat | Intervention: one-on-one HIV prevention intervention session with CBO volunteers via QQ or WeChat and setup of a hotline for HIV and HIV testing counseling; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 6 months—each monthly intervention session lasted 30 minutes | BL to 6-month FU—the proportion of lifetime HIV testing increased (62% vs 76%; χ21=9.5, P<.05); consistent condom use in anal sex with men in the previous 6 months increased (47% vs 61%; χ21=7.3, P<.05); the proportion of last-time condom use in anal sex with men increased (56% vs 68%; χ21=5.6, P<.05) | |
| Song et al [], 2017; Guangxi | 1-group pretest-posttest design; September 2014-December 2014; N=212 | SMS text messages | Intervention: SMS text messages with HIV knowledge, condom knowledge, and HIV testing information; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 6 months—2 messages every week | BL to 6-month FU—no change in the proportion of HIV testing in the previous 3 months; consistent condom use in anal sex with men in the previous 3 months increased (27% vs 57%; χ21=30.4, P<.05); the proportion of last-time condom use in anal sex with men increased (38% vs 73%; χ21=3.6, P<.05) | |
| Zhang et al [], 2014; Shandong | 1-group pretest-posttest design; August 2013-May 2014; N=468 | App—QQ | Intervention: HIV prevention intervention on HIV knowledge, risk of infection, condom use and safe sex practice, and HIV testing; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 6 months—did not report the intervention delivery and frequency | BL to 6-month FU—the proportion of lifetime HIV testing increased (30% vs 51%; df=1, P<.01); consistent condom use in anal sex with men in the previous 6 months increased (45% to 60%; df=1, P<.01); other outcomes: HIV knowledge increased (88% vs 94%; df=1, P<.01) | |
| Yan et al [], 2013; Heilongjiang | 1-group pretest-posttest design; October 2011-December 2012; N=400 | App—QQ | Intervention: CBO volunteer one-on-one HIV prevention intervention session via QQ; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 6 months—did not report the intervention delivery and frequency | BL to 6-month FU—the proportion of lifetime HIV testing increased (57% vs 68%; df=1, P<.01); consistent condom use in anal sex with men in the previous 3 months increased (45% vs 60%; df=1, P<.01); the proportion of last-time condom use in anal sex with men increased (66% vs 82%; df=1, P<.05) | |
| Tao et al [], 2020; Shaoxing | 1-group pretest-posttest design; April 2018-June 2018; N=209 | App—WeChat | Intervention: health messages delivered by a CBO and a free oral fluid–based HIVST kit; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 6 months—health messages were delivered biweekly | BL to 6-month FU—consistent condom use in anal sex with regular partners in the previous 6 months increased (52% to 64%; χ21=4.4, P<.05); consistent condom use in anal sex with casual partners met online in the previous 6 months increased (67% to 85%; χ2=8.5, P<.05); consistent condom use in anal sex with casual partners met in person in the previous 6 months increased (58% to 89%; χ21=5.6, P<.05); other outcomes: knowledge increased in several AIDS knowledge items | |
| Wang et al [], 2014; Chengdu | Pretest-posttest design using a repeated cross-sectional survey; October 2012-September 2013; n=370 at BL and n=236 at posttest | Web page and app | Intervention: multidimensional intervention model including online and venue-based interventions, as well as trained peer educators conducting online counseling and introducing men who have sex with men from online to in-person venues to receive condoms, lubricant, and HIV testing; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 12 months—did not report the intervention delivery frequency | BL to 12-month FU—consistent condom use in anal sex in the previous 6 months increased (46% to 64%; df=1, P<.01) | |
| Wang et al [], 2011; Yingtan | Pretest-posttest design using a repeated cross-sectional survey; August 2009-August 2010; n=135 at BL and n=134 at posttest | App—QQ | Intervention: peer education through QQ groups and on-site counseling; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 12 months—did not report the intervention delivery frequency | BL to 12-month FU—the proportion of lifetime HIV testing increased (10% vs 75%; χ21=48.4, P<.01); consistent condom use in anal sex with men in the previous 6 months increased (24% vs 78%; χ21=78.2, P<.01); the proportion of last-time condom use in anal sex with men increased (50% vs 90%; χ21=35.8, P<.01); other outcomes: knowledge increased in several AIDS knowledge items | |
| Wang et al [], 2009; Mianyang | Pretest-posttest design using a repeated cross-sectional survey; December 2006-January 2008; n=201 at BL and n=203 at posttest | Unclear | Intervention: POLw-delivered online and offline peer education; control: no control or comparison group | Did not report that any theory was used; did not report the process of intervention development | A total of 12 months—did not report the intervention delivery frequency | BL to 12-month FU—HIV testing in the previous 6 months increased (13% vs 54%; χ21=76.5, P<.01) | |
aResults reported by the study authors. Point estimates for nonsignificant outcomes are left out.
bRCT: randomized controlled trial.
cTPB: theory of planned behavior.
dBL: baseline.
eFU: follow-up.
fSBS: safe behavior and screening.
gIMB: information–motivation–behavioral skills.
hSTD: sexually transmitted disease.
iMI: motivational interview.
jHIVST: HIV self-testing.
kHIVST-OIC: HIVST with online real-time instructions and pretest-posttest counseling.
lHBM: health belief model.
mRR: relative risk.
nARR: absolute risk reduction.
oOR: odds ratio.
pSTI: sexually transmitted infection.
qUAI: unprotected anal intercourse.
rIRR: incident rate ratio.
sCDC: Center for Disease Control and Prevention.
tCBO: community-based organization.
uiPOL: internet popular opinion leader.
vDOI: diffusion of innovations.
wPOL: popular opinion leader.
Intervention Features
The development of approximately one-quarter of the interventions (7/24, 29%) was informed by a behavioral theory [,,-,]. All studies guided by behavioral theories were published in English. The theories used included conventional behavioral theories such as the health belief model [], the theory of planned behavior [], the diffusion of innovations theory, the information–motivation–behavioral skills model [,], and the parallel response model [], as well as an HIV-specific model—the AIDS Risk Reduction Model []. Regarding intervention content, over half (15/24, 62%) of the studies used a comprehensive intervention package rather than a single intervention [,,,-,,,-,-]. Most of these comprehensive interventions (14/15, 93%) included back-and-forth interaction with participants, such as screening for HIV-related risk behaviors, peer counseling, motivational interviewing, community-based organization–led activities, and online counseling with referrals for in-person HIV tests [,,,-,,,-,,]. Most of these interactions (12/15, 80%) were live conversations with a research team member or a trained volunteer [,,-,,,-,,]. A few studies (3/15, 20%) used automatic-reply text-based messages [,,].
Over half (14/24, 58%) of the studies delivered intervention content through social media platforms and direct message functionality, half (7/14, 50%) of which were WeChat based [,,,,,,]. A quarter of the studies (7/24, 29%) used SMS text messages, video messages, or email to deliver intervention content [,,,,,,]. The interventions in 21% (5/24) of the studies were delivered through web pages [,,,,]. In total, 25% (6/24) of the studies used 2 types of technologies in their interventions [,,,,,]. Three-quarters of studies (18/24, 75%) delivered their interventions multiple times; a third of those (6/18, 33%) did not report the frequency of intervention delivery. Among studies reporting their intervention delivery frequency, weekly (5/12, 42%) [,,,,] and biweekly (4/12, 33%) [,,,] were the most common frequencies. The length of the interventions varied from a single time to a year, and the most frequent intervention length was 6 months (9/24, 38%) [,,,,-]. In terms of community engagement, less than half (10/24, 42%) of the studies reported involving men who have sex with men in intervention development [,-,-,], and a third of the studies (9/24, 38%) reported engaging the men who have sex with men community in intervention delivery [,,,,,,-]. Only 12% (3/24) of the studies reported engaging the men who have sex with men community in both the intervention development and delivery process [,,].
Study Quality
On the basis of the quality assessment using the ICROMS global quality scores, over two-thirds of the studies (17/24, 71%) met the score requirement (). When considering the mandatory criteria, only 25% (6/24) of the studies met them, all of which were RCTs. On the basis of the quality review by study design, in the case of RCTs, 8% (1/13) of the studies did not meet the score requirement [], and 46% (6/13) did not meet the mandatory criteria of managing bias between groups with random allocation [,,,,,]. The controlled pretest-posttest design study did not meet either the score requirement or the mandatory criteria []. Regarding uncontrolled pretest-posttest design studies (10/24, 42%), half (5/10, 50%) did not meet the minimum score requirement [,,,,], and none of them met the mandatory criteria of mitigating the effect of no control group.
Publication Bias
The contour-enhanced funnel plots in 
Figures 2 and reveal an asymmetrical distribution of studies for both HIV testing uptake and consistent condom use, indicating possible publication bias. In particular, 12% (3/24) of the studies [,,] and 8% (2/24) of the studies [,] seemed to report much larger effect sizes compared to the rest of the studies for HIV testing uptake and consistent condom use, respectively. The asymmetry regarding nonsignificant results suggests that the asymmetrical distribution was probably caused by publication bias based on statistical significance. The robust Bayesian meta-analysis results document evidence of publication bias, with a Bayes factor of 21.04 and 4.37 indicating that the likelihood of publication bias was 21.04 and 4.37 times the likelihood of no publication bias for HIV testing uptake and consistent condom use, respectively. The Egger test indicated no small-study effects as the results were not statistically significant, with a z statistic of 0.24 and a P value of .81 for HIV testing uptake and a z statistic of 0.37 and a P value of .71 for consistent condom use.
Bayesian Meta-Analysis
For the meta-analysis, 62% (15/24) of the studies were included for the HIV testing uptake outcome, and 54% (13/24) of the studies were included for the consistent condom use outcome. Of the 15 studies included for the HIV testing uptake outcome, most (n=13, 87%) reported an improvement in HIV testing uptake, whereas 13% (2/15) of the studies reported no substantial change. Only 33% (5/15) of the studies estimated the effect size [,,,,]; other studies (10/15, 67%) conducted chi-square tests to assess significance. For condom use behavior, most studies (10/13, 77%) reported an improvement in consistent condom use, 23% (3/13) of the studies reported no significant change, and only 23% (3/13) of the studies estimated an effect size.
The overall effect sizes estimated from the Bayesian random-effects model were 0.20 (95% CrI 0.10-0.30) for HIV testing uptake and 0.15 (95% CrI 0.05-0.26) for consistent condom use. The probability that the mean difference exceeded 0 was >99% for both HIV testing uptake and consistent condom use despite the effect sizes for a few studies (2/15, 13% for HIV testing uptake; 3/13, 23% for consistent condom use) being estimated to be close to 0. In the HIV testing uptake model, the effect sizes of 20% (3/15) of the studies seemed much larger than those of the others [,,]. However, after using a robust model with t-distributed random effects, the pooled effect size was 0.19 (95% CrI 0.10-0.30), similar to the estimates from the original model. The probability that the mean difference exceeded 0 was still >99%. Similarly, adjusting the 15% (2/13) of the studies with large effect sizes [,] in the robust model for consistent condom use yielded close estimates (0.12, 95% CrI 0.05-0.23) compared to those of the original model.
The overall effect sizes were also estimated separately by study design (RCT and nonrandomized design) given their difference in study quality assessment. 
Figure 4 presents the estimated effect sizes for HIV testing uptake by RCT and nonrandomized designs. The pooled effect point estimate of RCT studies (0.16, 95% CrI −0.02 to 0.33) was smaller than that of nonrandomized studies (0.23, 95% CrI 0.07-0.38), but the CrIs largely overlapped. The 95% CrI for RCTs was slightly wider than that for nonrandomized studies, but they largely overlapped. Similarly, the pooled effect point estimate for consistent condom use for RCTs (0.10, 95% CrI −0.02 to 0.21) was also smaller than that for nonrandomized studies (0.19, 95% CrI −0.00 to 0.37), but these CrIs largely overlapped (
Figure 5). Detailed model results are presented in .
Covariate Analysis
We explored the association between behavior change outcomes (HIV testing uptake and consistent condom use) and study characteristics and intervention features (Table 2). Among the study characteristics and intervention features we assessed, the only characteristic that was associated with the behavior change outcome was the length of the intervention and study follow-up. Interventions lasting >6 months were associated with a 35% greater uptake of HIV testing (95% CrI 19%-51%) compared to interventions lasting 6 months. Studies followed up on for 12 months were associated with a 2% greater uptake of HIV testing (95% CrI 1%-39%) compared to those followed up on for 6 months.
| Study and intervention characteristics | HIV testing uptake (n=15 studies) | Consistent condom use (n=13 studies) | |||
| Values | Coefficientb (95% CrIc) | Values | Coefficient (95% CrI) | ||
| Publication year, median (IQR) | 2014 (2012-2018) | −0.01 (−0.03 to 0.01) | 2016 (2011-2020) | −0.01 (−0.04 to 0.02) | |
| Publication language, n (%) | |||||
| English | 7 (47) | Reference | 4 (31) | Reference | |
| Chinese | 8 (53) | 0.04 (−0.12 to 0.20) | 9 (69) | 0.10 (−0.07, 0.27) | |
| Outcome measure, n (%) | |||||
| Lifetime HIV testing | 6 (40) | Reference | —d | — | |
| HIV testing in the previous 3 months | 4 (27) | −0.04 (−0.24 to 0.15) | — | — | |
| HIV testing in the previous 6 months | 5 (33) | 0.09 (−0.11 to 0.27) | — | — | |
| Consistent condom use on the last anal sexual activity | — | — | 1 (8) | Reference | |
| Consistent condom use in the previous month | — | — | 1 (8) | −0.04 (−0.50 to 0.42) | |
| Consistent condom use in the previous 3 months | — | — | 5 (38) | 0.10 (−0.24 to 0.43) | |
| Consistent condom use in the previous 6 months | — | — | 6 (46) | 0.10 (−0.24 to 0.43) | |
| Intervention based on a behavior change theory, n (%) | |||||
| No | 10 (67) | Reference | 9 (69) | Reference | |
| Yes | 5 (33) | 0.05 (−0.12 to 0.21) | 4 (31) | −0.10 (−0.27 to 0.07) | |
| Intervention included back-and-forth interactions, n (%) | |||||
| No | 4 (27) | Reference | 5 (38) | Reference | |
| Yes | 11 (73) | 0.09 (−0.08 to 0.27) | 8 (62) | 0.07 (−0.09 to 0.24) | |
| >1 intervention session, n (%) | |||||
| No | 4 (27) | Reference | 8 (62) | Reference | |
| Yes | 11 (73) | −0.05 (−0.21 to 0.11) | 5 (38) | −0.02 (−0.20 to 0.15) | |
| Intervention development involved the target population, n (%) | |||||
| No | 9 (60) | Reference | 8 (62) | Reference | |
| Yes | 6 (40) | 0.04 (−0.13 to 0.19) | 5 (38) | −0.10 (−0.27 to 0.06) | |
| Intervention delivery engaged the target population, n (%) | |||||
| No | 8 (53) | Reference | 8 (62) | Reference | |
| Yes | 7 (47) | 0.04 (−0.12 to 0.21) | 5 (38) | 0.07 (−0.10 to 0.25) | |
| Length of the intervention, n (%) | |||||
| 6 months | 8 (53) | Reference | 5 (38) | Reference | |
| <6 months | 5 (33) | −0.06 (−0.17 to 0.06) | 5 (38) | −0.09 (−0.27 to 0.10) | |
| >6 months | 2 (13) | 0.35 (0.19 to 0.51) | 3 (23) | 0.08 (−0.15 to 0.31) | |
| Length of study follow-up, n (%) | |||||
| 6 months | 9 (60) | Reference | 7 (54) | Reference | |
| 3 months | 3 (20) | −0.14 (−0.30 to 0.02) | 3 (23) | −0.10 (−0.29 to 0.09) | |
| 12 months | 3 (20) | 0.02 (0.01 to 0.39) | 3 (23) | 0.10 (−0.12 to 0.33) | |
aUnivariate regression were conducted for each characteristic.
bCoefficients from univariate models exploring the relationship between variables in the first column and HIV testing uptake and condom use. For example, the first coefficient represents the coefficient of publication year in a model with only publication year as the covariate.
cCrI: credible interval.
dNot applicable.
Sensitivity Analysis
We conducted sensitivity analyses using different priors, such as a noninformative prior (μ~N[0, 10,000]) and an informative prior (μ~N[0.38, 1] for HIV testing uptake and μ~N[0.21, 1] for consistent condom use) on meta-analytic means, as well as an alternative heterogeneity prior (τ~Half-Cauchy [0,1]; Tables S1 and S2 in ). For both the HIV testing uptake and consistent condom use models, the effect estimates across the models were similar to those of the original models, suggesting that the estimated effects from the original models were stable.
Discussion
Principal Findings
Reviewing a total of 24 eligible studies published in the last 2 decades, we found promising effects for technology-based interventions designed to support HIV testing uptake and consistent condom use among men who have sex with men in China. The estimated pooled effect sizes from our primary Bayesian meta-analysis found a promising absolute effect on increasing both HIV testing and condom use—a 20% increase in HIV testing uptake and a 15% increase in consistent condom use. The probability that the effect size exceeded 0 was >99% for both HIV testing uptake and consistent condom use. To address potential outlier studies reporting larger effects compared to other reviewed studies, we used robust models for both outcomes that estimated pooled effect sizes similar to those of the original models. These promising behavior change effects of interventions in China align with findings of other meta-analyses worldwide. A previous global meta-analysis identified a significant effect in increasing HIV testing uptake (Cohen d=0.38) and reducing condomless anal intercourse (Cohen d=0.21) []. Another meta-analysis of computer-based interventions also estimated a significant effect in increasing condom use (Cohen d=0.26) []. To address concerns regarding prior selection, we conducted sensitivity analyses using different types of priors for both HIV testing uptake and consistent condom use. These models produced similar estimates to those of the original models. Therefore, we propose that technology-based interventions are likely to be effective in changing these 2 behaviors based on the reviewed studies.
We conducted subgroup analyses to explore the difference in estimated effect sizes across study designs to account for the varied quality of studies despite the fact that the benefit of Bayesian random-effects models is allowing for a combination of the effect sizes of studies with different designs. We found that the estimated pooled effect sizes for nonrandomized studies were slightly larger than those for RCT studies in both the HIV testing uptake and consistent condom use outcomes. The effect size dispersion of consistent condom use among RCTs seemed lower than that among nonrandomized studies. However, these 95% CrIs largely overlapped, indicating that the effect size differences between study designs were not significant or substantial. Previous meta-analyses examining the effect size difference among study designs have reported a significant estimated effect size difference in unprotected anal intercourse reduction across RCTs and nonrandomized designs, whereas the results were not significant for HIV testing uptake improvements [,]. These inconsistent findings suggest that study designs should be explored as an essential factor that could impact the intervention effect size in future meta-analyses. This also urges more rigorous study designs such as RCTs to fully demonstrate the effects of these technology-based interventions.
Previous reviews and meta-analyses have indicated that several intervention features might be associated with the intervention effect, including involvement of users in the design process, interactive interventions, multiple intervention sessions, longer treatment durations, and the use of combined technology modalities to deliver the intervention [,]. However, we did not find any significant associations between these intervention features and the effect size, which could be due to the known issue of small sample size for specific features (ie, the small number of studies included with these traits). Although our meta-regression analysis showed that interventions lasting >6 months seemed to have a significantly greater impact on HIV testing uptake compared to interventions lasting 6 months, it could be a coincidence as the 8% (2/24) of the studies that delivered interventions for >6 months reported much larger effect sizes than the rest of the studies [,]. Moreover, we did not find any other intervention features, such as theory-based intervention content, back-and-forth interactions, multiple intervention sessions, and community engagement, that could possibly explain the association between intervention duration and the effects on HIV testing uptake.
One of the unique features of these mobile health (mHealth) interventions in China is leveraging the same mainstream, all-in-one social networking app, WeChat. As mHealth becomes a more important and efficient tool to deliver health interventions, there are lingering questions regarding how to tailor interventions for target populations and how to scale up efficacious interventions to a broader range of populations [,]. Technology-based intervention modes have expanded from web-based formats to SMS text messages to social media [,,]. With 1.3 billion monthly active users (>80% of the population) in China [], WeChat is a convenient platform to distribute direct messages with text, pictures, and videos to specific users. It also offers a platform for people to develop WeChat-based apps that are less costly compared to developing a standard smartphone app []. In addition, using a mainstream social networking app instead of a men who have sex with men–focused app may reduce the concern of stigma by avoiding unanticipated outing or labeling. In our review, despite the fact that most studies that used WeChat only used its direct messaging function to deliver interventions, several studies (4/7, 57%) developed an official account or a mini app for risk assessment and HIV self-testing kit distribution [,,,]. Some researchers are also exploring approaches to leverage this social media platform to provide more comprehensive services together with men who have sex with men–focused community-based organizations and public opinion leaders [-]. Using this social networking platform, researchers may be able to enhance community engagement and provide more comprehensive health services to a larger target population.
Despite the meta-regression showing no significant association between effect size and intervention features, some patterns in intervention development and delivery can still be observed descriptively. For example, theory-based interventions have been reported to be more effective than non–theory-based ones []. In our review, less than a third of the studies (7/24, 29%) reported using a health behavior theory when developing the intervention. This low use of behavior change theories aligns with a previous review on mHealth studies in low- and middle-income countries []. This is not surprising because most of the health behavior theories were developed in the context of high-income countries, and some of the constructs, such as self-efficacy, could be difficult to apply without tailoring for the target population and local context. The insufficient use of health behavior change theories in the development of mHealth interventions highlights an opportunity to further improve the intervention effects with theory-guided intervention development and reinforces the need to develop or adapt these existing theories into culturally and contextually appropriate theories.
Limitations
Our analysis is limited by publication bias given that significant and positive intervention results are more likely to be published. Nonsignificant findings should be encouraged to be published. All the reviewed studies (24/24, 100%) were conducted in the context of China. Despite the fact that technology-based interventions in general have been found to be effective for HIV testing and condom use behavior, the application of technology and intervention contents should be tailored based on culture, context, and the needs of the targeted populations in different countries. In addition, the data available for analysis from original study reports were limited. Future intervention studies should be encouraged to report more details about the intervention features to allow for further analysis or scale up. Moreover, we did not identify any studies that targeted transgender women in China in this review. This indicates a lack of technology-based HIV prevention research focused on transgender women in China. The lack of research on and high HIV prevalence among transgender women worldwide indicate a need for HIV prevention efforts tailored to this key population []. Finally, this review searched for studies published before 2021. There could be a risk of excluding relevant studies published between 2022 and 2024. These studies could be included in future reviews.
Conclusions
Existing social networking platforms in China provide great opportunities for technology-based intervention development and distribution for HIV prevention. Technology-based HIV prevention interventions were found to have significant effects on health behavior change in both HIV testing uptake and consistent condom use among men who have sex with men in China across a broad array of studies and study designs. However, many study designs in this review were less rigorous, without a randomized design or a control group. More rigorous study designs, such as RCTs, and measurement of outcomes that address the limitations of self-report, such as picture verification of HIV self-tests, are needed to build up a more robust evidence base for the development and implementation of future technology-based intervention programs.
Acknowledgments
The authors would like to thank the support of the National Institute of Allergy and Infectious Diseases (R01AI143875). The authors also thank the Emory Center for AIDS Research (P30AI050409) for facilitating this work.
Data Availability
All data generated or analyzed during this study are included in this published article (and its supplementary information files).
Conflicts of Interest
None declared.
Multimedia Appendix 1
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist.
PDF File (Adobe PDF File), 69 KBReferences
- Wu Z, Xu J, Liu E, Mao Y, Xiao Y, Sun X, et al. National MSM Survey Group. HIV and syphilis prevalence among men who have sex with men: a cross-sectional survey of 61 cities in China. Clin Infect Dis. Jul 2013;57(2):298-309. [FREE Full text] [CrossRef] [Medline]
- Cui Y, Guo W, Li D, Wang L, Shi CX, Brookmeyer R, et al. Estimating HIV incidence among key affected populations in China from serial cross-sectional surveys in 2010-2014. J Int AIDS Soc. 2016;19(1):20609. [FREE Full text] [CrossRef] [Medline]
- Cai Y, Wang Z, Lau JT, Li J, Ma T, Liu Y. Prevalence and associated factors of condomless receptive anal intercourse with male clients among transgender women sex workers in Shenyang, China. J Int AIDS Soc. 2016;19(3 Suppl 2):20800. [FREE Full text] [CrossRef] [Medline]
- Shan D, Ning Z, Yu M, Zheng H, Yang J, Gong H, et al. HIV incidence and risk factors among transgender women and cisgender men who have sex with men in two cities of China: a prospective cohort study. Infect Dis Poverty. Mar 07, 2022;11(1):26. [FREE Full text] [CrossRef] [Medline]
- Yan H, Xiao W, Chen Y, Chen Y, Lin J, Yan Z, et al. High HIV prevalence and associated risk factors among transgender women in China: a cross-sectional survey. J Int AIDS Soc. Nov 2019;22(11):e25417. [FREE Full text] [CrossRef] [Medline]
- Baral SD, Poteat T, Strömdahl S, Wirtz AL, Guadamuz TE, Beyrer C. Worldwide burden of HIV in transgender women: a systematic review and meta-analysis. Lancet Infect Dis. Mar 2013;13(3):214-222. [CrossRef] [Medline]
- Knox J, Chen Y, He Q, Liu G, Jones J, Wang X, et al. Use of geosocial networking apps and HIV risk behavior among men who have sex with men: case-crossover study. JMIR Public Health Surveill. Jan 15, 2021;7(1):e17173. [FREE Full text] [CrossRef] [Medline]
- Cao B, Saha PT, Leuba SI, Lu H, Tang W, Wu D, et al. Recalling, sharing and participating in a social media intervention promoting HIV testing: a longitudinal analysis of HIV testing among MSM in China. AIDS Behav. May 10, 2019;23(5):1240-1249. [FREE Full text] [CrossRef] [Medline]
- Cheng W, Cai Y, Tang W, Zhong F, Meng G, Gu J, et al. Providing HIV-related services in China for men who have sex with men. Bull World Health Organ. Mar 01, 2016;94(3):222-227. [FREE Full text] [CrossRef] [Medline]
- Chen S, Zhang Q, Chan C, Yu FY, Chidgey A, Fang Y, et al. Evaluating an innovative HIV self-testing service with web-based, real-time counseling provided by an artificial intelligence chatbot (HIVST-Chatbot) in increasing HIV self-testing use among Chinese men who have sex with men: protocol for a noninferiority randomized controlled trial. JMIR Res Protoc. Jun 30, 2023;12:e48447. [FREE Full text] [CrossRef] [Medline]
- Wang L, Podson D, Chen Z, Lu H, Wang V, Shepard C, et al. Using social media to increase HIV testing among men who have sex with men - Beijing, China, 2013-2017. MMWR Morb Mortal Wkly Rep. May 31, 2019;68(21):478-482. [FREE Full text] [CrossRef] [Medline]
- Zhu X, Zhang W, Operario D, Zhao Y, Shi A, Zhang Z, et al. Effects of a mobile health intervention to promote HIV self-testing with MSM in China: a randomized controlled trial. AIDS Behav. Nov 2019;23(11):3129-3139. [FREE Full text] [CrossRef] [Medline]
- Yun K, Chu Z, Zhang J, Geng W, Jiang Y, Dong W, et al. Mobile phone intervention based on an HIV risk prediction tool for HIV prevention among men who have sex with men in China: randomized controlled trial. JMIR Mhealth Uhealth. Apr 13, 2021;9(4):e19511. [FREE Full text] [CrossRef] [Medline]
- Liu X, Jiang D, Chen X, Tan A, Hou Y, He M, et al. Mental health status and associated contributing factors among gay men in China. Int J Environ Res Public Health. May 24, 2018;15(6):1065. [FREE Full text] [CrossRef] [Medline]
- Ritchwood TD, He J, Smith MK, Tang W, Ong JJ, Oduro A, et al. "Getting to zero" among men who have sex with men in China: a review of the HIV care continuum. Curr HIV/AIDS Rep. Dec 02, 2019;16(6):431-438. [FREE Full text] [CrossRef] [Medline]
- Plantin JC, de Seta G. WeChat as infrastructure: the techno-nationalist shaping of Chinese digital platforms. Chin J Commun. Feb 21, 2019;12(3):257-273. [CrossRef]
- Shi Y, Qiu J, Yang Q, Chen T, Lu Y, Chen S, et al. Increasing the HIV testing among MSM through HIV test result exchange mechanism: study protocol for a cluster randomized controlled trial. BMC Infect Dis. Aug 06, 2021;21(1):764. [FREE Full text] [CrossRef] [Medline]
- Nguyen LH, Tran BX, Rocha LE, Nguyen HL, Yang C, Latkin CA, et al. A systematic review of eHealth interventions addressing HIV/STI prevention among men who have sex with men. AIDS Behav. Sep 2019;23(9):2253-2272. [FREE Full text] [CrossRef] [Medline]
- Horvath KJ, Walker T, Mireles L, Bauermeister JA, Hightow-Weidman L, Stephenson R. A systematic review of technology-assisted HIV testing interventions. Curr HIV/AIDS Rep. Aug 2020;17(4):269-280. [FREE Full text] [CrossRef] [Medline]
- Xin M, Viswanath K, Li AY, Cao W, Hu Y, Lau JT, et al. The effectiveness of electronic health interventions for promoting HIV-preventive behaviors among men who have sex with men: meta-analysis based on an integrative framework of design and implementation features. J Med Internet Res. May 25, 2020;22(5):e15977. [FREE Full text] [CrossRef] [Medline]
- Veronese V, Ryan KE, Hughes C, Lim MS, Pedrana A, Stoové M. Using digital communication technology to increase HIV testing among men who have sex with men and transgender women: systematic review and meta-analysis. J Med Internet Res. Jul 28, 2020;22(7):e14230. [FREE Full text] [CrossRef] [Medline]
- Hackenberger BK. Bayesian meta-analysis now - let's do it. Croat Med J. Dec 31, 2020;61(6):564-568. [FREE Full text] [CrossRef] [Medline]
- Higgins JPT, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-analysis. J R Stat Soc Ser A Stat Soc. Jan 2009;172(1):137-159. [FREE Full text] [CrossRef] [Medline]
- Kruschke JK, Liddell TM. The Bayesian new statistics: hypothesis testing, estimation, meta-analysis, and power analysis from a Bayesian perspective. Psychon Bull Rev. Feb 7, 2018;25(1):178-206. [CrossRef] [Medline]
- Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. Mar 29, 2021;372:n71. [FREE Full text] [CrossRef] [Medline]
- Schardt C, Adams MB, Owens T, Keitz S, Fontelo P. Utilization of the PICO framework to improve searching PubMed for clinical questions. BMC Med Inform Decis Mak. Jun 15, 2007;7:16. [FREE Full text] [CrossRef] [Medline]
- Zingg W, Castro-Sanchez E, Secci F, Edwards R, Drumright L, Sevdalis N, et al. Innovative tools for quality assessment: integrated quality criteria for review of multiple study designs (ICROMS). Public Health. Apr 2016;133:19-37. [CrossRef] [Medline]
- Sin J, Henderson C, Spain D, Cornelius V, Chen T, Gillard S. eHealth interventions for family carers of people with long term illness: a promising approach? Clin Psychol Rev. Mar 2018;60:109-125. [FREE Full text] [CrossRef] [Medline]
- Darzi AJ, Busse JW, Phillips M, Wykoff CC, Guymer RH, Thabane L, et al. Retina Evidence and Trials INternational Alliance (R.E.T.I.N.A.) Study Group. Interpreting results from randomized controlled trials: what measures to focus on in clinical practice. Eye (Lond). Oct 28, 2023;37(15):3055-3058. [CrossRef] [Medline]
- Hoffman MD, Gelman A. The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. J Mach Learn Res. 2014;15(1):1593-1623. [FREE Full text]
- Lau JT, Lau M, Cheung A, Tsui HY. A randomized controlled study to evaluate the efficacy of an internet-based intervention in reducing HIV risk behaviors among men who have sex with men in Hong Kong. AIDS Care. Aug 2008;20(7):820-828. [CrossRef] [Medline]
- Lau JT, Lee AL, Tse WS, Mo PK, Fong F, Wang Z, et al. A randomized control trial for evaluating efficacies of two online cognitive interventions with and without fear-appeal imagery approaches in preventing unprotected anal sex among Chinese men who have sex with men. AIDS Behav. Sep 2016;20(9):1851-1862. [CrossRef] [Medline]
- Wang Z, Lau JT, Ip M, Ho SP, Mo PK, Latkin C, et al. A randomized controlled trial evaluating efficacy of promoting a home-based HIV self-testing with online counseling on increasing HIV testing among men who have sex with men. AIDS Behav. Jan 2018;22(1):190-201. [CrossRef] [Medline]
- Chiou PY, Liao PH, Liu CY, Hsu YT. Effects of mobile health on HIV risk reduction for men who have sex with men. AIDS Care. Mar 2020;32(3):316-324. [CrossRef] [Medline]
- Ko NY, Hsieh CH, Wang MC, Lee C, Chen C, Chung A, et al. Effects of internet popular opinion leaders (iPOL) among internet-using men who have sex with men. J Med Internet Res. Feb 25, 2013;15(2):e40. [FREE Full text] [CrossRef] [Medline]
- Luo Q, Wu Z, Mi G, Xu J, Scott SR. Using HIV risk self-assessment tools to increase HIV testing in men who Have Sex with men in Beijing, China: app-based randomized controlled trial. J Med Internet Res. Sep 01, 2023;25:e45262. [FREE Full text] [CrossRef] [Medline]
- Tang W, Han L, Best J, Zhang Y, Mollan K, Kim J, et al. Crowdsourcing HIV test promotion videos: a noninferiority randomized controlled trial in China. Clin Infect Dis. Jun 01, 2016;62(11):1436-1442. [FREE Full text] [CrossRef] [Medline]
- Tang W, Mao J, Liu C, Mollan K, Zhang Y, Tang S, et al. SESH study group. Reimagining health communication: a noninferiority randomized controlled trial of crowdsourced intervention in China. Sex Transm Dis. Mar 2019;46(3):172-178. [FREE Full text] [CrossRef] [Medline]
- Tang W, Wei C, Cao B, Wu D, Li KT, Lu H, et al. Crowdsourcing to expand HIV testing among men who have sex with men in China: a closed cohort stepped wedge cluster randomized controlled trial. PLoS Med. Aug 2018;15(8):e1002645. [FREE Full text] [CrossRef] [Medline]
- Liu LQ, Li P, Li J. Analysis on effects of the AIDS infections intervention for MSM using I=internet. Chin J Dis Control Prev. Dec 2014;18(12):1232-1234. [FREE Full text]
- Song LP, Tang J, Zhang YL, Zhang ZK, Lan GH, Liu YL. Impact of text message intervention on HIV-related high risk sexual behavior of MSM. Chin J Dis Control Prev. 2017;23(10):932-934. [CrossRef]
- Tao JH, Fang YR, Lu QL, Yang ZK. Effect evaluation of follow-up intervention among internet-based self-testing for men having sex with men in Shaoxing. Chin J AIDS Std. 2020;26(9):958-961. [CrossRef]
- Yan HM, Gao C, Li Y, Tong X, Hui S, Yu L. Evaluation of QQ-based HIV high-risk behavior interventions for MSM. Chin J AIDS STD. 2013;(3):174-176. [FREE Full text] [CrossRef]
- Zhang TL, Liu Z, Sun SY. Evaluation of the effectiveness of network intervention on HIV/AIDS prevention and control among gay men. Prev Med Trib. 2014;20(8):609-610. [FREE Full text]
- Liu GW, Lu HY, Wang J, Cao XB. Evaluation of web-based HIV/AIDS interventions among men who have sex with men. Chin J AIDS STD. 2012;18(9):578-580. [CrossRef]
- Wang FX, Huang YL. Analysis on effect of AIDS related high risk behavior intervention among MSM in Yingtan city. Prev Med Trib. 2011;17(6):518-519.
- Wang XD, Yang D, Ma X, Fan G, Xu H, Lin H, et al. An initial explore of multidimensional and full coverage intervention mode for MSM. Pract J Clin Med. 2014;11(3):62-64. [FREE Full text]
- Wang Y, Zhang HB, Li ZJ, Xu J, Zhang GG, Dou Z, et al. Analysis of AIDS prevention services for MSM group and promotion effectiveness of standardize treatment for STD. Occup Health. 2009;25(15):1586-1588. [FREE Full text] [CrossRef]
- Xie YL, Chen BF, Zhang QL, Pan YJ, Luo HJ, Zhuo BG, et al. Evaluation of internet-based HIV /AIDS interventions among men who have sex with men in Dongguan. J Prev Med. 2018;34(3):273-277. [FREE Full text]
- Cheng W, Xu H, Tang W, Zhong F, Meng G, Han Z, et al. Online HIV prevention intervention on condomless sex among men who have sex with men: a web-based randomized controlled trial. BMC Infect Dis. Jul 19, 2019;19(1):644. [FREE Full text] [CrossRef] [Medline]
- Li XX, Zhao J, Liu Q, Xie XQ, Liu HW, Long QP, et al. A comparative study on the effects of "Internet+" AIDS network intervention and field intervention for men in contact groups. Systems Medicine. 2020;5(12):39-41. [FREE Full text] [CrossRef]
- Xiao ST, Chen P, Zhu LD, Jin YZ, Xin X. Effect evaluation of precise AIDS intervention on MSM by using WeChat official account. Chin J AIDS STD. 2020;26(6):611-614. [FREE Full text] [CrossRef]
- Noar SM, Black HG, Pierce LB. Efficacy of computer technology-based HIV prevention interventions: a meta-analysis. AIDS. Jan 02, 2009;23(1):107-115. [CrossRef] [Medline]
- LeGrand S, Muessig KE, Horvath KJ, Rosengren AL, Hightow-Weidman LB. Using technology to support HIV self-testing among MSM. Curr Opin HIV AIDS. Sep 2017;12(5):425-431. [FREE Full text] [CrossRef] [Medline]
- Muessig KE, Nekkanti M, Bauermeister J, Bull S, Hightow-Weidman LB. A systematic review of recent smartphone, internet and Web 2.0 interventions to address the HIV continuum of care. Curr HIV/AIDS Rep. Mar 2015;12(1):173-190. [FREE Full text] [CrossRef] [Medline]
- Cao B, Gupta S, Wang J, Hightow-Weidman LB, Muessig KE, Tang W, et al. Social media interventions to promote HIV testing, linkage, adherence, and retention: systematic review and meta-analysis. J Med Internet Res. Nov 24, 2017;19(11):e394. [FREE Full text] [CrossRef] [Medline]
- Thomala LL. WeChat - statistics and facts. Statista. URL: https://www.statista.com/topics/9085/wechat/#topicOverview [accessed 2024-04-29]
- Hu S, Lu Y, He X, Zhou Y, Wu D, Tucker JD, et al. Effectiveness of the secondary distribution of HIV self-testing with and without monetary incentives among men who have sex with men living with HIV in China: study protocol for a randomized controlled trial. BMC Infect Dis. Mar 14, 2023;23(1):160. [FREE Full text] [CrossRef] [Medline]
- Tan RK, Wu D, Day S, Zhao Y, Larson HJ, Sylvia S, et al. Digital approaches to enhancing community engagement in clinical trials. NPJ Digit Med. Mar 25, 2022;5(1):37. [FREE Full text] [CrossRef] [Medline]
- Zhang W, Hu Q, Tang W, Jin X, Mao X, Lu T, et al. HIV self-testing programs to men who have sex with men delivered by social media key opinion leaders and community-based organizations are both effective and complementary: a national pragmatic study in China. J Acquir Immune Defic Syndr. Aug 15, 2020;84(5):453-462. [CrossRef] [Medline]
- Michie S, Johnston M, Francis J, Hardeman W, Eccles M. From theory to intervention: mapping theoretically derived behavioural determinants to behaviour change techniques. Appl Psychol. Jul 08, 2008;57(4):660-680. [CrossRef]
- Cho YM, Lee S, Islam SM, Kim SY. Theories applied to m-Health interventions for behavior change in low- and middle-income countries: a systematic review. Telemed J E Health. Oct 2018;24(10):727-741. [FREE Full text] [CrossRef] [Medline]
Abbreviations
| CrI: credible interval |
| ICROMS: integrated quality criteria for review of multiple study designs |
| mHealth: mobile health |
| PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses |
| RCT: randomized controlled trial |
Edited by A Mavragani; submitted 11.06.24; peer-reviewed by S Chen, E Harriss; comments to author 21.12.24; revised version received 08.01.25; accepted 09.03.25; published 28.04.25.
Copyright©Wenting Huang, Daniel Stegmueller, Jason J Ong, Susan Schlueter Wirtz, Kunru Ning, Yuqing Wang, Guodong Mi, Fei Yu, Chenglin Hong, Jessica M Sales, Yufen Liu, Stefan D Baral, Patrick S Sullivan, Aaron J Siegler. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 28.04.2025.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.