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

Background Men who have sex with men (MSM) in China hold a low-risk perception of acquiring HIV. This has resulted in an inadequate HIV testing rate. Objective This study aims to investigate whether administering HIV risk self-assessments with tailored feedback on a gay geosocial networking (GSN) app could improve HIV testing rates and reduce sexual risk behaviors in Chinese MSM. Methods We recruited MSM from Beijing, China, who used the GSN platform Blued in October 2017 in this 12-month double-blinded randomized controlled trial. From October 2017 to September 2018, eligible participants were randomly assigned to use a self-reported HIV risk assessment tool that provided tailored feedback according to transmission risk (group 1), access to the same HIV risk assessment without feedback (group 2), or government-recommended HIV education materials (control). All interventions were remotely delivered through the mobile phone–based app Blued, and participants were followed up at 1, 3, 6, and 12 months from baseline. The number of HIV tests over the 12-month study was the primary outcome and was assessed using an intention-to-treat analysis with an incident rate ratio (IRR). Unprotected anal intercourse (UAI) over 6 months was assessed by a modified intention-to-treat analysis and was the secondary outcome. All statistical analyses were conducted in SAS 9.3 (SAS Institute, Inc.), and a P value <.05 was considered statistically significant. Results In total, 9280 MSM were recruited from baseline and were randomly assigned to group 1 (n=3028), group 2 (n=3065), or controls (n=3187). After follow-up, 1034 (34.1%), 993 (32.4%), and 1103 (34.6%) remained in each group, respectively. Over 12 months, group 1 took 391 tests (mean of 2.51 tests per person), group 2 took 352 tests (mean of 2.01 tests per person), and controls took 295 tests (mean of 1.72 tests per person). Group 1 had significantly more HIV testing than the control group (IRR 1.32, 95% CI 1.09-4.58; P=.01), while group 2 did not differ significantly from the controls (IRR 1.06, 95% CI 0.86-1.30; P=.60). The proportion of UAI was not statistically different among different groups, but all 3 groups had UAI, which declined from baseline. Conclusions Repeated HIV risk assessments coupled with tailored feedback through GSN apps improved HIV testing. Such interventions should be considered a simple way of improving HIV testing among MSM in China and increasing awareness of HIV status. Trial Registration ClinicalTrials.gov NCT03320239; https://clinicaltrials.gov/study/NCT03320239


INTRODUCTION 2a-i) Problem and the type of system/solution
Yes.Despite the high burden of HIV infections in MSM communities, the HIV testing rate among MSM is low in China, where only half of MSM were reported to ever have tested for HIV.2a-ii) Scientific background, rationale: What is known about the (type of) system Yes.This is thought to be from social and psychological barriers, including a lack of trust in facility-based services, hesitancy to use facilities, and anxiety about knowing one's HIV status [11].Low perceived risk of HIV transmission is also problematic for HIV testing uptake [12-14], with almost 70% of MSM underestimating their risk of HIV [15].Underestimation of HIV risk can significantly impact HIV testing, leading to under-utilization of preventative services and riskier sexual behaviors [16].Therefore, effective interventions to expand HIV testing in MSM populations and prevent these deterrents are needed.

Does your paper address CONSORT subitem 2b?
Yes.Therefore, a randomized controlled trial was conducted where HIV risk assessment and feedback were delivered by a popular GSN, Blued in Beijing, China.The goal of this study was to determine the effect of this intervention on HIV testing numbers and unprotected anal intercourse (UAI) among Chinese MSM.

3a) CONSORT: Description of trial design (such as parallel, factorial) including allocation ratio
Yes.This double-blinded, triple-arm randomized controlled trial was conducted among MSM in Beijing, China between October 2017 and September 2018.3b) CONSORT: Important changes to methods after trial commencement (such as eligibility criteria), with reasons Not applicable.No change has been made after trial comencement.3b-i) Bug fixes, Downtimes, Content Changes 4a) CONSORT: Eligibility criteria for participants Yes.The inclusion criteria were as follows: MSM who were18 years or older; had anal intercourse with a man in the past six months; self-reported negative or unknown HIV status; resided in Beijing; logged onto the app at least once in the past week before enrollment; and agreed not to share the study materials with others in the study.Participants were excluded if they self-reported injection drug use within six months before enrollment, participated in another online intervention during the study period, or planned to move away from Beijing within the following year.4a-i) Computer / Internet literacy 4a-ii) Open vs. closed, web-based vs. face-to-face assessments: Yes.Recruitment messages were privately sent via the app to Blued users from a Blued administrative account.4a-iii) Information giving during recruitment 4b) CONSORT: Settings and locations where the data were collected Yes.All questionnaires were conducted online via a unique website developed for the study.If participants did not respond to the initial follow-up surveys, a second link was sent as a reminder one week later.Participants were considered lost to follow-up if they had not responded to the second reminder within one week.4b-i) Report if outcomes were (self-)assessed through online questionnaires Yes.The number of rapid HIV tests taken by participants was collected by working staff at drop-in testing sites run by Blued one year after randomization.

5-vii) Access
Yes.The message briefly introduced the study and provided a link to participate.MSM who clicked the link were directed to eligibility screening, the consent form, and baseline survey.Group 1 and Group 2 took a HIV RISK Assessment Tool at baseline and six months after randomization.

5-viii) Mode of delivery, features/functionalities/components of the intervention and comparator, and the theoretical framework
Yes. Group 1 also received tailored feedback based on their responses to the HIV RISK Assessment tool, which included their HIV risk score, probability of acquiring HIV, high-risk sexual behaviors contributing to their risk of HIV infection, and personalized measures to reduce their risk of HIV infection.An example of this bespoke feedback is captured in Figure 2. Participants in the Control group did not receive HIV risk assessment or feedback via the validated tool, and instead answered eight questions on their HIV knowledge based on a questionnaire established by the China CDC at baseline and six months after randomization [26].

5-xi) Report any prompts/reminders used
In our intervention, we did not provide any reminders for the use of the application.

5-xii) Describe any co-interventions (incl. training/support)
Yes.All participants regardless of randomization received a link at baseline and six months post-randomization to sign up for free HIV testing at convenient, drop-in testing sites managed by Blued.6a) CONSORT: Completely defined pre-specified primary and secondary outcome measures, including how and when they were assessed Yes.The primary outcome of interest was the cumulative number of rapid HIV tests taken over twelve months, described as the mean number of HIV tests per person who attended Blued testing sites for testing, per year.The secondary outcome was self-reported UAI at each follow-up, which was calculated as the number of participants who reported UAI in the past three months divided by the number of participants who completed at least one follow-up.6a-i) Online questionnaires: describe if they were validated for online use and apply CHERRIES items to describe how the questionnaires were designed/deployed 6a-ii) Describe whether and how "use" (including intensity of use/dosage) was defined/measured/monitored 6a-iii) Describe whether, how, and when qualitative feedback from participants was obtained 6b) CONSORT: Any changes to trial outcomes after the trial commenced, with reasons Yes.All questionnaires were conducted online via a unique website developed for the study.If participants did not respond to the initial follow-up surveys, a second link was sent as a reminder one week later.Participants were considered lost to follow-up if they had not responded to the second reminder within one week.7a) CONSORT: How sample size was determined 7a-i) Describe whether and how expected attrition was taken into account when calculating the sample size 7b) CONSORT: When applicable, explanation of any interim analyses and stopping guidelines Yes.The primary outcome of interest was the cumulative number of rapid HIV tests taken over twelve months, described as the mean number of HIV tests per person who attended Blued testing sites for testing, per year.The secondary outcome was self-reported UAI at each follow-up, which was calculated as the number of participants who reported UAI in the past three months divided by the number of participants who completed at least one follow-up.8a) CONSORT: Method used to generate the random allocation sequence Yes.Participants were simply randomized in a 1:1:1 ratio by a computerized randomization algorithm with SAS 9.3 (SAS Institute, Inc. USA) into three groups (Gropu1, Group2, and the Control group).Assignment of group allocations was masked to both the study staff and participants to ensure doubleblinding.8b) CONSORT: Type of randomisation; details of any restriction (such as blocking and block size) Yes.Participants were simply randomized in a 1:1:1 ratio by a computerized randomization algorithm with SAS 9.3 (SAS Institute, Inc. USA) into three groups (Gropu1, Group2, and the Control group).Assignment of group allocations was masked to both the study staff and participants to ensure doubleblinding.9) CONSORT: Mechanism used to implement the random allocation sequence (such as sequentially numbered containers), describing any steps taken to conceal the sequence until interventions were assigned Yes.Participants were simply randomized in a 1:1:1 ratio by a computerized randomization algorithm with SAS 9.3 (SAS Institute, Inc. USA) into three groups (Gropu1, Group2, and the Control group).Assignment of group allocations was masked to both the study staff and participants to ensure doubleblinding.10) CONSORT: Who generated the random allocation sequence, who enrolled participants, and who assigned participants to interventions Yes.Participants were simply randomized in a 1:1:1 ratio by a computerized randomization algorithm with SAS 9.3 (SAS Institute, Inc. USA) into three groups (Gropu1, Group2, and the Control group).Assignment of group allocations was masked to both the study staff and participants to ensure doubleblinding.11a) CONSORT: Blinding -If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how 11a-i) Specify who was blinded, and who wasn't Yes.Participants were simply randomized in a 1:1:1 ratio by a computerized randomization algorithm with SAS 9.3 (SAS Institute, Inc. USA) into three groups (Gropu1, Group2, and the Control group).Assignment of group allocations was masked to both the study staff and participants to ensure doubleblinding.11a-ii) Discuss e.g., whether participants knew which intervention was the "intervention of interest" and which one was the "comparator" 11b) CONSORT: If relevant, description of the similarity of interventions Not applicable.This is an ehealth trial.12a) CONSORT: Statistical methods used to compare groups for primary and secondary outcomes Yes.The mean number of rapid HIV tests per person, per year in each group was calculated and analyzed using zero-inflated Poisson regression analysis, estimating the ratios of HIV testing rates in the two intervention groups compared to the control group, and reported as incident rate ratios (IRR) and 95% confidence intervals (CIs).The primary analysis was conducted with an intention to treat analysis.12a-i) Imputation techniques to deal with attrition / missing values Yes.Adjusted multilevel logistic regression modeling evaluated for differences in UAI among study groups.UAI analyses were done with the modified intention to treat analyses, excluding those who were never followed.12b) CONSORT: Methods for additional analyses, such as subgroup analyses and adjusted analyses Yes.Adjusted multilevel logistic regression modeling evaluated for differences in UAI among study groups.UAI analyses were done with the modified intention to treat analyses, excluding those who were never followed.RESULTS 13a) CONSORT: For each group, the numbers of participants who were randomly assigned, received intended treatment, and were analysed for the primary outcome Yes.70,000 Blued-app users received a recruitment link containing information about the study.Of these 70,000 users, 29,316 were screened for eligibility, and 9,280 MSM met the inclusion criteria.Following randomization, Group 1, Group 2, and Control groups consisted of 3028, 3065, and 3187 participants, respectively (Figure 3).Of the 3,028 participants in Group 1, 1,034 (34.1%) finished at least one follow-up, of the 3,065 participants in Group 2, 993 (32.4%) finished at least one follow-up, and of the 3,187 participants in the Control group, 1,103 (34.6%) finished at least one follow-up.13b) CONSORT: For each group, losses and exclusions after randomisation, together with reasons Yes. Figure 3. Study flow chart.

13b-i) Attrition diagram 14a) CONSORT: Dates defining the periods of recruitment and follow-up
Yes.This double-blinded, triple-arm randomized controlled trial was conducted among MSM in Beijing, China between October 2017 and September 2018.Participants were recruited in October 2017 via the popular Chinese gay dating app, Blued.14a-i) Indicate if critical "secular events" fell into the study period 14b) CONSORT: Why the trial ended or was stopped (early) Yes.Follow-up was conducted at one month, three months, six months, and twelve months following randomization, and the same baseline questionnaire was to be completed, excluding demographic information.The baseline survey was not administered at twelve-month post-randomization due to low followup rates.

15) CONSORT: A table showing baseline demographic and clinical characteristics for each group
Yes. Table 1 summarizes the baseline characteristics of the three study groups.15-i) Report demographics associated with digital divide issues Not applicable.In our study, we recruited participants who were all Blued users.16a) CONSORT: For each group, number of participants (denominator) included in each analysis and whether the analysis was by original assigned groups 16-i) Report multiple "denominators" and provide definitions Yes.The proportion of self-reported UAI per study arm is shown in Figure 4.At baseline, the proportion of UAI in the past three months was 43.53% in Group 1, 43.86% in Group 2, and 44.14% in the Control arm.At one-month follow-up, UAI increased to 46.37% in Group 1 and 44.99% in Group 2 but declined to 42.42% in the Control arm.At three months, 33.11% of Group 1, 34.74% of Group 2, and 35.08% of the Control reported UAI.At six months, the proportion of UAI in Group 1 decreased to 26.67%.Reported UAI also decreased to 30.22% in Group 2 and 30.95% in the Control arm.

16-ii) Primary analysis should be intent-to-treat
4b-ii) Report how institutional affiliations are displayed 5) CONSORT: Describe the interventions for each group with sufficient details to allow replication, including how and when they were actually administered 5-i) Mention names, credential, affiliations of the developers, sponsors, and owners 5-ii) Describe the history/development process 5-iii) Revisions and updating 5-iv) Quality assurance methods 5-v) Ensure replicability by publishing the source code, and/or providing screenshots/screen-capture video, and/or providing flowcharts of the algorithms used