Preplanned Studies: HIV-Positive Men Are More Likely to Be Hyper Linked Within College Student Social Network — Northeast China, 2017–2018

The incidence of human immunodeficiency virus (HIV) cases in young male students who had sex with men (MSM) has reported an average annual 10% increase in HIV diagnoses from 5,841 in 2010 to 13,717 in 2019 in China (1). The rise and popularity of geo-social networking apps used by young MSM (YMSM) have a parallel increased risk for sexually transmitted infections (STIs), due to an increase in the number of partners and a reduction in sexual health behaviors (2). However, lowering community viral load through treatment as prevention (TasP) (3) has become a critical HIV response strategy, by reducing sexual transmission among persons living with HIV (PLHIV) (4). In this paper, we sought to describe and examine the association and magnitude of social contacts within the social network of an app that employs serosorting to identify the uneven social linkage among college MSM in Northeast China. In this study, we recruited 759 college MSM and 382 of their social contacts from the Yiyou app from April 2017 to June 2018. A total of 98 participants (8.6%) were confirmed HIV positive during baseline and follow-ups serology tests, of whom 34 were college MSM. For college MSM, HIV positive prevalence at baseline and incidence at follow-ups were 3.8% (29/759) and 2.9 per 100 person-years (PY) (5/170.50 PY). Network graphing visualization showed that HIV-positive MSM had comparatively more social contacts (social dynamic) than the seronegative MSM. Therefore, identifying certain individuals as hyper-linkages within college MSM social network can inform tailored HIV prevention strategies such as sending reminders for information about antiretroviral therapy (ART) and promoting safer sexual behaviors.

We conducted a prospective longitudinal study and recruited participants among Yiyou app (developed by Heilongjiang Kangtong Community [HKC]) users. HKC manages 21 offline voluntary counseling and testing (VCT) sites using the Yiyou app umbrella in 16 cities within Northeast China to facilitate serosorting and other offline services, by appointment. The Yiyou app users must register with a cell phone number that is later used to share testing information with a mutual agreement upon inquiry, i.e., partner informed serosorting purposes (Figure 1A). That is the main reason for the app’s popularity. We recruited participants via three methods, including online advertisements on the Yiyou app homepage, person-to-person sharing by HKC visitors, and public service announcements in offline VCT clinics of HKC. The participants’ recruitment and follow-up procedures were detailed in Figure 1B. A self-administered structured questionnaire was conducted via the Yiyou app on sociodemographic information, perceived risk of HIV infection, sexual behaviors, recreational drug use in the past 6 months, and HIV self-testing experience. Upon completion of the online survey, participants were asked to undertake dual rapid HIV and syphilis tests and a parallel rapid HIV test by trained workers in offline VCT sites and confirmed HIV testing at a provincial CDC. Descriptive analyses were performed to identify the distribution of sociodemographic and behavioral characteristics and a Mann–Whitney U test was used to identify the difference in number of social contacts between the two HIV statuses.

Figure 1. 

Illustration and flowchart of the study. (A) The illustration of Yiyou App installation and use process; (B) The flowchart of participants' recruitment and follow-up procedure of the study.

Abbreviation: HIV=human immunodeficiency virus; STI=sexually transmitted infection.

As per informed consent, we enhanced the Yiyou app information with a social parameter, using latitude and longitude from GPS data of all universities in Northeast China to verify the students’ identity by their geolocation sign-in, in addition to self-reported student identity. After a predesigned de-identification procedure, contact lists were linked to registered users of the Yiyou app and the social contacts were filtered by logical inference from the label information of phone contacts. Furthermore, we calculated the number of social contacts for each participant and determined the linkage for each participant as a snapshot of the whole social network at the end of the study, midnight of June 30, 2018. We used network analytic techniques to provide a “snapshot” to characterize what the network looks like as a static social structure. Nodes represent users and links represent interactions between users; the size of the node is proportional to the number of social contacts. The algorithm finds each node by minimizing the dispersion of the whole system. Microsoft Excel (Microsoft^®, Albuquerque, New Mexico, USA) and Statistical Analysis System (Version 9.4, SAS Institute Inc., Cary, North Carolina, USA) were used to manage the dataset and conduct analyzes and Gephi® software for Windows (V.0.9.2) was used to visualize the social network.

Of 3,698 persons approached, 1,141 (30.9%) became study participants aged 24 years old [interquartile range (IQR)=21–33]; 759 (66.5%) were college student and 382 (33.5%) non-student contacts. A total of 98 participants (8.6%) were confirmed HIV positive during baseline and follow-ups, of whom 34 were college MSM. For college MSM and their social contacts, HIV prevalence at baseline and incidence in follow-ups were 3.8% (29/759) and 2.9 per 100 PY (5/170.50 PY), 13.9% (53/382), and 10.5 per 100 PY (11/105.06 PY), respectively.

Among college MSM, 97% were Han ethnicity, 96% were not married to a woman, 44.3% were migrants without local residency permits (hukou), 61.7% lived in Northeast China for >2 years, and 75.6% self-identified as homosexual. Regarding risk perception and sexual behavior, 29.8% perceived their risk of HIV infection as moderate or high. In the past 6 months, 49.5% reported anal sex, 47.6% used condoms consistently, 25.8% had >1 regular partners, and 18.3% used recreational drugs (Table 1). The median number of social contacts of college students was zero compared to two of their social contacts. The number of social contacts showed a significant difference (Z=6.1, P<0.001) between 98 HIV positive MSM (median: 1, IQR: 1–2) and 1,043 HIV negative MSM (median: 0, IQR: 0–1).

Figure 2 allows us to identify highly connected individuals (i.e., hyper-linkages) with different HIV statuses among college students and their social contacts. It showed that HIV-positive MSM had comparatively more social contacts than their seronegative counterparts. However, some of the HIV positive students showed no social contact or declined to share social contact with us. Therefore, they are dispersed outside.

Figure 2. 

Network graph of the young men who have sex with men in Northeast China, 2017–2018, categorized by HIV sero-status and social identity (college students and social contacts, N=1,141).

Note: Nodes represent users and links represent interactions between users; the size of the node is proportional to the number of social contacts.

Abbreviation: HIV=human immunodeficiency virus.