Design
As the intervention in our study will involve two levels—the community level and the individual level, and MSM in the same city often communicate with each other, we adopt a cluster-randomized controlled trial with one intervention arm and one control arm (Fig. 1). The cluster unit will be the city. Surveys will be conducted at baseline and every three months thereafter (Fig. 2). A total of ten clusters, including city A, city B, city C, city D, city E, city F, city G, city H, city I, city J, city K from Shandong Province were chosen based on the following criteria: (1) CDC MSM sentinel surveillance sites; (2) capacity for recruiting sufficient MSM. City J and a nearby city K are in the same cluster because the number of estimated MSM in both cities is small.
Randomization
We randomly assigned cities in each block (1:1) to either the intervention group or the control group with a block design, stratified by the city’s number of people living with HIV in 2018, average GDP in 2017, cumulative number of HIV-positive from 2012-2018 and population at the end of 2017. We have a total of five blocks involving ten clusters: Block 1 consists of city A and city F; Block 2 consists of city B and city G; Block 3 consists of city C, city J and city K; Block 4 consists of city D and city H; Block 5 consists of city E and city I. Then the ten clusters were randomized with five (city A, city B, city C, city D, city E) in the intervention arm and five (city F, city G, city H, city I, city J, city K) in the control arm. Random numbers were generated using SAS 9.4 software. Cities in the intervention arm will implement the digital crowdsourced intervention while cities in the control arm will implement conventional interventions organized by local CDCs. Participants and data analysts will be blinded in this trial.
Participants and eligibility criteria
Eligibility criteria for participants were: (1) aged at least 18 years; (2) had anal sex with men in the past one year; (3) currently living and planning to live in the city for the next 12 months; (4) born biologically male; (5) agree to provide phone number (just for follow up); (6) willing to participate in the follow-up survey every 3 months; (7) fully understand the facts about the trial with the informed consent. The informed consent form will be shown to the potential participants at the beginning of the online questionnaire. Only those who clicked the “agree” option may be included in the study. Participants living with HIV will be invited to complete the baseline survey but will not be followed.
Study measures and Recruitment
We have built an online questionnaire (see Additional file 1) using Sojump Survey Software. The questionnaire is divided into two parts. Part 1 includes the informed consent form and six items related to study eligibility. Part 2 includes the main items related to this trial. MSM who are determined eligible in the first part proceed to the second section. They will be required to add us as contacts on WeChat (the largest social mobile phone app in China) so that we can establish an online cohort. Participants who do not meet all requirements can scan the QR codes to get 0.14USD (1RMB) on WeChat as an inducement. MSM will be recruited through banner advertisements on Blued, the largest gay dating app in China. CBOs in each study city will also assist with recruitment. In addition to direct recruitment through digital approaches, eligible participants will be invited to refer no more than five friends from their social networks of MSM and can receive a 1.44USD (10RMB) per person as incentive if their friends meet the requirements of this trial. All participants who enroll in the trial will receive a 7.22USD (50RMB) incentive for baseline survey and 7.22USD (50RMB) for each follow-up. Surveys will be conducted at baseline and every 3 months thereafter. Those who complete all five surveys will have an opportunity to win an iPad mini.
All the information in the questionnaire is self-reported. The questionnaire used in MSM surveillance sites run by local CDCs in each city will include additional questions about sexual partners, social media engagement, HIV/syphilis testing and HIV/syphilis test results (see Additional file 2). Cell phone numbers will be used to link CDC and online survey data sets so the self-report data can be triangulated with CDC surveillance data during the same period.
Intervention
The intervention includes 24 images and 4 short videos about HIV testing and safe sexual behaviors. We will also send self-test kits to the participants in intervention group with their consent to evaluate the effectiveness of self-test kit in improving HIV testing rate at the 2nd and 3rd follow-up. Images and videos were developed through a series of participatory crowdsourcing contests before this study. These included a sprint-like designathon that identified digital HIV self-testing as a priority [35] and an open call for images and videos [31]. The 24 images aim to address the importance of HIV testing to improve HIV testing uptake. The purpose of the 4 short videos of 50s-1min in duration is to enhance the awareness of safe sexual behaviors like condom use among MSM and encourage them to take regular HIV testing. The images and videos can be found in Additional file 3. The intervention will be implemented at the individual level through “WeChat”, and at community level through WeChat-group and WeChat-moments (a function on WeChat that people can share their life with friends).
The intervention will be implemented in five clusters (city F, city G, city H, city I, city J, city K) in the intervention arm. For individual-level intervention, the images will first be shown at the end of the baseline questionnaire, and then we will send one image every 2 weeks and one video every 3 months to the participants via WeChat. We will also send self-test kits to the participants in the intervention group at the 2nd and 3rd follow-up. As part of the community-level intervention, we will set up several WeChat groups in each city and invite 30-40 participants in each group. Messages about HIV testing and safe sexual behavior from authoritative facilities (hospital, CDCs, CBOs, etc.) will be distributed to these groups and WeChat-moments every 2 weeks. We will encourage members in these groups to talk about HIV through these private groups. We will invite volunteers from local CBOs to join the groups. Team members of this study will participate in discussion if the participants have any question. Participants in other five cities (city A, city B, city C, city D, city E) will receive routine intervention by local CDCs. All concomitant care is acceptable with no restrictions on what participants may seek.
Follow up
All participants will receive an online survey at baseline and follow-up surveys will be conducted every three months. Each man will participate in a total of five surveys. At the 2nd and 3rd follow-up, we will send self-test kits to the participants in the intervention group to evaluate the effectiveness of self-test kit in improving HIV testing rate.
Outcomes
The primary outcome of this study will be the uptake of self-reported HIV testing uptake after 12 months. We have set ten monitoring sentinels in the eleven cities so that the self-reported HIV testing rate can be triangulated with the rate from surveillance data during the same period. In addition, participants who receive the self-test kit are required to send their results to us on WeChat so that we can verify that the self-reported results are consistent with the self-test results. An increase of 20% in testing rate (assuming a proportion of HIV testing of 70% in intervention arm and 50% in control arm) was chosen as the superiority margin. This choice was based on the data of MSM sentinel surveillance collected by CDCs of Shandong Province in last one year. Secondary outcomes include condom use, syphilis testing, social network characteristics and others (see Additional file 4).
Sample Size
A two-arm, cluster-randomized controlled trial will be used. The sample size was calculated based on the primary outcome. We assumed that a digital crowdsourced intervention will be superior to CDC routine intervention in promoting HIV testing rate among MSM who have not been tested in the past three months. Assuming the HIV testing rate is 70% in intervention cities and 50% in the control cities, a total of ten clusters (five in intervention group and five in control group), an intraclass correlation coefficient of 0.02 (usually between 0.01 to 0.03), two-sided α=0.05, power=0.85, and the loss of follow-up is 30%, the total sample size is 500 men (50 for each cluster). The calculation was made using the software PASS 15.
Timeline
This study will last 17 months. The first five months will be the baseline survey. Given the total population of MSM are small in some cities, we wanted to ensure sufficient time for participant recruitment. CDC surveillance surveys will also be conducted in participating cities. The following 12 months will be the intervention phase. The intervention will be implemented in City F, City G, City H, City I, City J and City K following the cluster RCT design outlined in Fig.2. All the participants in the eleven cities will be surveyed at baseline and every 3 months thereafter. In the 17th month, the last follow-up survey will be implemented.
Data Collection and Management
The Sojump Survey Software will be used for data collection and storage. In this study, we will collect information about socio-demographics, sexual behaviors, HIV and syphilis testing behaviors, social network characteristics and psychological characteristics through an online questionnaire. The online tool and questionnaire were used in our original stepped wedge cluster randomized controlled trial [31]. Socio-demographic characteristics include age, occupation, education level, annual income, marital status, sexual orientation, and sexual orientation disclosure. Behavioral variables consist of sex partners, frequency of sex, condom use, self-reported HIV testing (including both facility-based and self-testing), syphilis testing and social networking. Psychological characteristics include HIV testing self-efficacy and HIV stigma.
To improve data quality, we have set up a verification mechanism. When a questionnaire has been completed, we will check it first. If it has only one logical error, we will verify with the participant. When there is more than one logical error in the questionnaire, the participant will no longer be included in the study. All questions in our online survey have been set as compulsory questions, which can improve data completeness.
Final survey data completed by all participants will be saved in a secure computer only used for saving data, discarding other privacy issues such as open network environments. Passwords and firewalls will also be used to protect the data.
An internal advisory committee composed of STI experts has been established. The committee will guide us in study design and data analysis, and meet regularly to review and assess progress in data collection and research.
Analysis
The sociodemographic characteristics will be analyzed using descriptive statistics. Comparison in baseline characteristics between the intervention and control arm will be conducted using a chi-squared test adjusted for clustering. The primary outcome of this trial will be the uptake of self-reported HIV testing rate after 12 months. Generalized linear mixed models (GLMM) will be used for the analysis. Intervention status and time will be considered fixed effects, while sites and individual participants with multiple measurements across the four follow-ups will be considered random effects.
The secondary outcomes will include a series of binary variables (continuous variables will be categorized into binary variables), such as number of sexual partners, frequency of anal sex, sex without condoms, frequency of HIV testing (including both facility-based and self-testing), awareness of syphilis status, social network engagement and others. These will be analyzed similar with the primary outcomes. Furthermore, the interaction effect of intervention between individual level and community level will also be analyzed.
After the 2nd follow-up, we will conduct an interim analysis to evaluate the effectiveness of digital images and videos. After the 4th follow-up, we will also evaluate the effectiveness of digital media and self-test kit using data collected from the 3rd and 4th follow-up.
The data cleaning, sorting and analysis will be carried out by team members who do not participate in the randomization and city administration and do not know the treatment allocation. Data analysts were not allowed to modify or delete data. If an outcome is missing for < 15% of participants, analyses will use a complete-case approach. Otherwise, multiple imputation will be used. Participants who miss the follow-up will be included and assumed not achieve the primary and secondary outcomes during the missed follow-up period.
We will also carry out a subgroup analysis based on age (>30 years old versus ≤30 years old) and way of HIV testing (facility-based versus self-testing).
In order to measure contamination, we will conduct per protocol sub analyses to measure whether our results have been affected by contamination.
Results of this study will be distributed to target cities and national stakeholders.
Adverse events and others
We will capture potential unexpected adverse events via spontaneous self-report. We build one WeChat account for each city. Participants can contact us on WeChat so we can capture questions and potential unexpected adverse events. Additionally, some participants will be recruited through local CBOs. The respective CBO leaders can also help us to collect adverse events.
Adverse events will be fully and truly reported on a regular basis to the study PI.
If any modifications to the protocol are required at any point over the course of the trial, these changes will be communicated to relevant parties by email.