Enhancing Partner Services Among Men Who Have Sex With Men Living With HIV: A Pilot Randomized Controlled Trial
Enhancing Partner Services Among Men Who Have Sex With Men Living With HIV: A Pilot Randomized Controlled Trial
Individuals will participate in a pilot randomized controlled trial (RCT) that implements and evaluates the feasibility, acceptability, and effectiveness of crowdsourced partner services (PS) or conventional PS among Chinese men who have sex with men (MSM) living with HIV.
The pilot RCT will include 120 newly identified MSM HIV cases who were born biologically male, aged 18 years old or older, newly identified as HIV positive, had oral or anal sex with a man, had at least one sexual partner in the previous 6 months and live in Guangzhou. Participants will undergo a series of computer-based interviews (baseline and 2 months after enrollment) and will be randomly assigned into two groups in 2:1 ratio (intervention: control) and receive crowdsourced PS and conventional PS, respectively.
This is a pilot RCT, and the primary outcomes of this study are: 1) to test the feasibility and acceptability of the crowdsourced intervention, and 2) to pilot the preliminary impact of crowdsourced PS intervention in promoting HIV testing among partners of the newly identified HIV cases. In addition, this pilot RCT aimed to collect preliminary data for the designing of a full RCT, which aimed to measure the efficacy of the crowdsourced PS intervention in promoting PS. Thus, the sample size of this study is not designed and will be not powered enough to test the difference between the two arms. However, with a sample size of 120 (80 for intervention arm and 40 for the control arm, respectively), this pilot study would be able to observe the differences between the two arms and collect preliminary data for adequate sample size calculation of a full RCT. With a sample size of 120 (ratio of 2:1), alpha=0.05, power=0.80, and the partner return rate in the conventional PS arm is 30% (around 31% in 2016 in Guangzhou), the minimal detectable effect size would be 26.7% for this pilot study.
Data analyses: For the primary outcomes of feasibility and acceptability of the intervention: this will be analyzed through descriptive analysis to assess the feedback on the feasibility and acceptability of the crowdsourced PS intervention at baseline and the follow-up survey. For the primary outcome of the proportion of partners returning for HIV testing within 2 months of the index participant's enrolment in the study, log-binomial regression (generalized linear model with binomial distribution and log link) will be used to calculate 95% confidence intervals for the proportion of notifiable partners visiting by arm. Risk differences and risk ratios will also be calculated in these models. As this study is a pilot RCT, adjustment for covariates is not necessary and will not be conducted in the primary analysis. In secondary analyses, pre-specified potential confounding variables will be adjusted that might bias the estimate if not accounted for (e.g. index age, index marital status). If a partner reports more than one index referred him, he will be counted as a returning partner for each identified index.
Subgroup analyses will be performed by partner sex, partner type (e.g. main male partner vs. casual male partner) and sexual orientation. Time to presentation will also be assessed among all notifiable partners using Cox proportional hazards with a robust variance estimator. In the time-to-event analyses, censoring will occur at the end of 2-months. Note that for the primary outcome, the assumption is that any partner without an identifiable record at the community center or in the centralized electronic medical record (EMR) did not present for HIV testing. Therefore, this study assumes no missing outcome data. Sensitivity analyses will be conducted using alternative definitions of presentation for testing, including using the self-reported data from index cases and partners. For count variables expressed as the mean number of partners, the investigators will use either Poisson regression or negative binomial regression if overdispersion is identified.
Inclusion Criteria:
Exclusion Criteria: