A cross-sectional study was conducted between May 2018 and April 2019 in Zhejiang Province, China. The survey used baseline data collected for a cohort study from 2018 to 2020, including a 6-monthly follow-up to monitor new infections among MSM. In total, 731 MSM were recruited at baseline, with the inclusion criteria of assigned male sex at birth; having had sex with another male; age of ≥16 years and being HIV negative. Participants aged 18 years or older, who had casual sex with partners found through the internet in the last 6 months were selected. Individuals who had sought casual sexual partners through the internet in the past 6 months were defined as internet-based MSM. The analysis included a total of 281 participants.

Participants were enrolled using convenience sampling from MSM venues and the internet by volunteers at nongovernmental organization or community-based organizations as well as VCT staff. Participants enrolled from MSM venues were recruited using published posters in VCT clinics and outreach services (eg, bathrooms, bars, and gardens). Participants enrolled from the internet were recruited through advertisements regarding the study in chat groups on Blued (Beijing Blue City Brothers Culture Media), WeChat (Shenzhen Tencent Computer Systems), and Tencent (Shenzhen Tencent Computer Systems). All participants completed an electronic questionnaire by scanning a 2D code.

The questionnaire covered demographic characteristics, experience with HIV testing services, sexual behaviors with male partners, internet-based behaviors, HIV serostatus communication, perceptions of the HIV risk posed by internet-based partners and HIV-related knowledge. Casual sexual behavior was evaluated by asking “Did you have casual sex in the last 6 months with someone you met through the Internet?” Participants who replied “yes” were recruited, even if they had other types of partners. The main outcome measure (number of times one was HIV tested and self-tested) was assessed through the question “How many times have you undergone a HIV test and how many times have you performed an HIV self-test during your lifetime?” For the analysis, the data were transformed into a binary variable (“Tested once or less” or “Tested twice or more”).

Age was categorized as 18-24, 25-34, and ≥35 years. The residence was dichotomized (Zhejiang vs other provinces). Educational attainment was dichotomized (below university vs university and above). Annual income was self-reported and was classified as ≤30,000, 30,000-80,000 and >80,000 Chinese Yuan (approximately ≤US $429, US $429- $1143 and >US $1143).

HIV-related knowledge was assessed using 3 questions: “Are men who have sex with men the group most seriously affected by AIDS in China at present?”; “Does infection with other sexually transmitted diseases (STDs) increase the risk of HIV infection?” and “Does the use of drugs such as rush, methamphetamine, ecstasy, and k powder increase the risk of HIV infection?” Replies of “Yes” were defined as correct and “No” as incorrect. The answers of each respondent were categorized as mostly wrong (2 or more wrong or unknown answers), somewhat wrong (1 wrong or unknown answer) and correct. Sexual behaviors with male partners in the last 6 months were assessed based on the sexual role (only receptive vs ever insertive sex), sex with a regular partner (no vs yes), and sex with a venue-based casual partner (no vs yes).

Internet-based behaviors were evaluated in terms of the login frequency on MSM-specific social networking applications and the viewing of erotic videos. Participants were asked, “How frequently do you log in to MSM-specific social networking applications?” The possible response included: 3-7 times a week, 1-2 times a week, once a week, and less frequency. For analysis, the option “3-7 times a week” was defined as > 2 times a week, while the remaining responses were classified as ≤ 2 times a week. The participants were also asked, “How frequently do you view erotic videos?” with the same response options.

HIV serostatus communication was evaluated through the question, “In the past 6 months, did you ask your Internet-based partners about their HIV serostatus before sex?” The possible responses were “always or usually” and “little or none.”

The perceived HIV risk of internet-based partners was determined by asking “How do you perceive the HIV risk of Internet-based partners?” with the response options of “very high,” “high,” “average,” and “low.”

Before data analysis, variables were inspected for missing values. The proportions of missing data ranged from 0% to 3.91% among independent variables. The proportions of missing data were 0% and 16.73% for outcome measures of “frequent HIV testing” and “frequent HIV self-testing.” The demographics of the participants undergoing HIV testing and self-testing are described as frequencies. Uni- and multivariate logistic regression were used to explore the associations of HIV testing and self-testing by complete case analysis. We calculated the crude and adjusted odds ratios (ORs) and their 95% CIs. All factors in the univariate analysis were included in a multivariable regression model, together with age, education, and residence. The final multivariate logistic model was conducted using forward elimination. In addition, we used multiple imputation (5 imputations) to conduct multivariable regression models to assess the sensitivity of results to missing data. We performed these analyses using SPSS software (version 18.0; IBM Corp). We used a mixed effects logistic regression model with a random effect for the respondents ID to assess the sensitivity. We processed the analysis using R software (version 4.4.1; R Foundation for Statistical Computing).

This study was reviewed and approved by the ethics committee of Zhejiang Provincial Center for Diseases Control and Prevention (2018-033), and all methods were performed in accordance with the relevant guidelines and regulations. Informed consent was obtained electronically from all participants. The raw data did not contain any personal identifying information that can be linked to particular individuals and was anonymized before its use. Participants were given a gift (body shampoo or facial cleanser) worth 30 Chinese Yuan (approximately US $5).

In total, 281 internet-based MSM were enrolled, among whom 93.6% (262/280) reported having sought sexual partners through MSM-specific social networking applications. The demographic characteristics are summarized in Table 1. Most (127/281, 45.2%) were aged 25~34 years, 63.0% (177/281) had a university or higher education and 59.8% (168/281) were native to Zhejiang Province. Nearly half (126/281, 44.8%) of the respondents had annual incomes between 30,000 and 80,000 RMB and most (67.3%, 189/281) reported their sexual orientation as gay.

Among participants little or none communicating about the HIV serostatus of the internet-based partner before sex, 52.1% (73/140) of the participants were 25-34 years old which is higher than that among participants always or usually communicating about the HIV serostatus (Table S1 in Multimedia Appendix 1). In total, 22.2% (30/140) of the participants little or none communicating about the HIV serostatus of the internet-based partner before sex acquired the average or low perspective of HIV-infected risk toward internet-based partners, higher than that among participants always or usually communicating about the HIV serostatus.

Of the 281 participants, 174 (61.9%) reported frequent HIV testing. Of the 234 participants who reported the information of self-testing, 50.9% (n=119) underwent frequent HIV self-testing.

We conducted a cross-sectional survey of internet-based MSM in Zhejiang Province to describe the rates and associations of frequent HIV testing and self-testing. Overall, 61.9% of internet-based MSM underwent frequent HIV testing and 50.9% performed frequent self-testing. Communication about HIV serostatus of the internet-based partner before sex was significantly associated with frequent HIV testing and self-testing among internet-based MSM. Our results add to the literature on the associations among internet-based behaviors, HIV serostatus communication, and perceived HIV infection risk of internet-based partners with the frequency of HIV testing and self-testing.

HIV testing and self-testing were not performed by approximately 40% and 50% respectively of the internet-based MSM. In a large STI clinic in the Netherlands, the proportion was 55.2% [23], while it was 57.8% among black MSM in the United States [24], suggesting that there is room for improvement in testing frequency. The World Health Organization recommends HIV self-testing as an additional approach because of its convenience and confidentiality. The US Centers for Disease Control and Prevention recommends at least annual HIV tests for high-risk populations (eg, MSM), with some experts suggesting that more frequent (eg, every 3-6 months) testing benefits individuals at elevated HIV risk [25]. Increasing the acceptability and feasibility of HIV self-testing for this population is considered highly desirable [26]. However, HIV self-testing often lacks face-to-face pre- or posttest consulting, confirmatory testing, and subsequent referral to specialist care. Therefore, it is important to emphasize the benefits and limitations (eg, window period) of self-testing [27,28] when applying this approach.

Of our participants, more than half reported communicating about HIV serostatus of an internet-based partner before sex. That behavior was associated with increased odds of frequent HIV testing and self-testing. Determining HIV serostatus allows individuals to make informed decisions about how to engage in sexual behaviors. However, studies have found that most disclosures occur between steady partners [29]; only a small proportion of men were informed of the HIV serostatus of their most recent casual male partner [30]. Therefore, it is important for MSM to actively communicate their HIV serostatus with internet-based partners. However, merely knowing the HIV serostatus of sexual partners is insufficient to reduce the risk of HIV transmission; knowledge of one’s own HIV status is also important. Unfortunately, MSM who test for HIV infrequently might misperceive their positive HIV status as HIV-negative. Therefore, HIV serostatus communication and frequent testing are both urgent issues for MSM.

An increasing number of MSM use the internet to find partners, especially MSM-specific social networking applications [31], making them an effective target for MSM interventions [11]. Blued, the largest gay male–oriented social media platform and geosocial networking mobile app in China had approximately 27 million registered users and 12 million monthly users in China as of 2016 [32]. Other MSM-specific social networking apps, such as Jack’d and Grindr, have also been downloaded by millions of people, offering new ways for MSM to engage in peer-to-peer communication [33]. Many of the apps require information regarding HIV disclosure, basic demographics, and preferred sexual behaviors in the personal profile. Studies have shown that only 55% of the users report using the HIV disclosure option in the US [34], and in India, few individuals disclose their HIV status on their profiles because of the stigma in the culture [35]. Status disclosure is often easier through the internet than face-to-face. However, doubts persist regarding the authenticity of the results uploaded on apps by the users themselves. It is recommended that apps implement rules to ensure the security and the veracity of disclosures. Apps can also encourage users to regularly update the HIV test results (eg, every 3 months). In general, dating apps are valuable targets for intervention promoting HIV serostatus communication.

This study had several important limitations. First, because of its cross-sectional design, we cannot infer causality in the relationships of HIV serostatus communication with frequent HIV testing and self-testing. Second, participants were recruited through convenience sampling from MSM venues and the internet, and might not be representative of the entire MSM population. Third, information bias, especially recall bias, was possible due to the reliance on self-report methods. Finally, some important factors that could affect the testing behavior of the targeted population were not included in the questionnaire, such as social stigma, cultural diversity, and psychological factors. Future studies should address these limitations, and aim to replicate our findings in both similar populations and different sociocultural groups.

There is a need to improve the frequency of HIV testing and self-testing among internet-based MSM. Communication about the HIV serostatus of the internet-based partner before sex was significantly associated with frequent HIV testing and self-testing among internet-based MSM. HIV serostatus communication should be improved within the context of social networking apps to promote HIV frequent HIV testing among internet-based MSM.