The Company You Keep: Do Social Networks Influence HIV Status?
A shoulder to cry on, someone to hang out with, a friend to talk to about money and health or with whom to grab a bite: All of these things make up a life and a community of support. But could they also play a role in HIV status?
That was the question Wenjing Zheng, a biostatistics Ph.D. student from UC Berkeley, sought to answer when she started looking at relationships between people in three Kenyan villages. The data was presented at the Conference on Retroviruses and Opportunistic Infections (CROI 2017) in February. And the verdict? There is some relationship.
"We have found that having HIV-positive contacts in your social network predicts your baseline HIV status, even after adjusting for individual risk factors," she said.
However, the association wasn't the same between genders or for people of different ages.
Still, the results, which put a social spin on the phylogenetic mapping the CAPRISA team presented at IAC 2016 on HIV networks in South Africa, point to a way to target future test-and-treat interventions.
"Going forward, having longitudinal social network and seroconversion data will allow us to investigate the potential of social networks in predicting seroconversions in order to inform targeted prevention and self-testing," Zheng said. "[It also holds] potential to enhance the care cascade through peer support."
Networks of Care, Networks of Risk
The analysis used data from the SEARCH trial in Kenya and Uganda, meaning that almost all the participants were treatment naïve and therefore didn't have the protective benefit of suppressed viral loads. More than 15,000 people registered for the trial and, in so doing, volunteered names of people they didn't live with but with whom they ate, shared emotional support, discussed health or money issues and passed free time.
Zheng and her colleagues were able to connected about 40% of those names to other people in the study and their demographic and census data. Voila: A network was born. At the conference, Zheng displayed a graph that looked more like a universe or an atom than a topographic map: purple dots forming a tight nucleus with blue and green dots cascading out from it, and lines connecting all of them. This is what these networks of care looked like when the researchers charted them. And within each of them -- closely connected to each of them -- were the about 16% of people who were living with HIV who also gave and got emotional, health and other forms of support.
"These networks are very dense, and they are highly interconnected large clusters," she said.
So, Zheng and her team started analyzing these connections for confirmed HIV status and whether these networks were associated with an increased likelihood of a participant being HIV positive herself or himself.
When they looked initially, without controlling for demographic or risk data, there was a clear association. Any contact with a friend or partner living with HIV could increase a man's chance of also being HIV positive more than twofold.
But, when they adjusted for a man's own behavior? The association largely disappeared. Only interactions with any woman living with HIV were associated with a 1.5-times increased chance of him also being HIV positive.
For women, the associations were a little different. Before adjusting for demographics or personal risk factors, a woman was more likely to be HIV positive herself if she spent time with a man who was at least 10 years her senior (a 1.9 unadjusted risk ratio [uRR]), if she spent time with any man outside her household who was living with HIV (uRR: 1.4) or even if she hung out with any woman living with HIV (uRR: 1.4).
After adjustment, only interactions with any man living with HIV (aRR:1.4) or a man living with HIV who was also a decade or more older (aRR: 1.6) were associated with significant, increased likelihood that a woman herself would be HIV positive.
Specific Care and Support
But, the analysis didn't just find that generalized interaction was associated with likelihood of testing HIV positive at baseline; sharing specific forms of support was, too -- but only for women.
"We did this analysis in men, and we didn't have any significant findings," Zheng said. "But for women, we found some interesting associations."
For instance, a woman had an 80% increased chance of being HIV positive herself if she talked about emotional issues with any man living with HIV. And if she discussed health matters specifically with a man a decade her senior, she was also 80% more likely to herself be HIV positive.
"And interestingly, talking about peer support, we also found that if she has a positive female contact outside her household with whom she discusses health matters, then she is also more likely to be positive herself," said Zheng.
Other, similar associations emerged when she looked at young women specifically. They were more likely to be HIV positive if they shared food with a man 10 years their senior, if they spent free time with a man 10 years their senior or if they discussed health matters with that same man.
Still, the data was cross-sectional -- that is, a snapshot of time -- and doesn't provide more details about the specific contexts that increase people's risk for HIV. Continuing to follow these participants may help with that.
But the data raised questions about their significance among researchers. Some asked whether these networks were just markers for unsuppressed HIV in the community. Others asked whether breaking bread with someone was just a surrogate for sex partners.
"One would imagine," said Steven Goodreau, Ph.D., of the University of Washington, "discussing food or health with one's sex partners."
"I'm really glad you brought that up," Zheng said. "It's a great way to provide a perspective for these analyses."
In short, she said, it's too soon to know exactly how networks of support influence HIV status. They held true even after the researchers adjusted for overall HIV community prevalence. So, it's not just a marker for unsuppressed HIV in the community. There may be something specifically important about these forms of support. But, whether someone is simply getting a bite to eat while on a date and then later have sex, she didn't know.
"Some of it could be a surrogate for sexual networks," she said. "But we also found some significance between, for example, females who are positive. This could be a pathway where we are capturing discussions in health matters that were possibly sparked by a shared risk factor or behavior that we didn't measure in the variables."