To better target HIV prevention, identify people whose social contacts have high viral loads

Rather than relying on assessment of an individual’s sexual behaviour or of ‘community viral load’, targeting of pre-exposure prophylaxis (PrEP) and other prevention interventions could in part be based on the proportion of a person’s social contacts who have unsuppressed HIV. There is a correlation between young gay men having HIV and their ‘network viral load’, according to a study published online ahead of print in the Journal of Acquired Immune Deficiency Syndromes.

Britt Skaathun and colleagues at the University of Chicago believe that the concept could eventually be used by public health departments in the US, by combining data from partner services (contact tracing) with viral load data from individuals in care. “NVL [network viral load] could have substantial public health implications for persons most at risk for HIV infection given that this novel metric avoids overreliance on individual level behavior or broad community indices,” they write.

Nonetheless, the current study only provides proof of principle and the authors acknowledge that their data would need to be replicated in larger, longitudinal studies.

Community or network?

The data for the study come from a cohort of young black American men who have sex with men (MSM) in Chicago.


odds ratio

Comparing one group with another, expresses differences in the odds of something happening. An odds ratio above 1 means something is more likely to happen in the group of interest; an odds ratio below 1 means it is less likely to happen. Similar to ‘relative risk’. 


The transition period from infection with HIV to the detectable presence of HIV antibodies in the blood. When seroconversion occurs (usually within a few weeks of infection), the result of an HIV antibody test changes from HIV negative to HIV positive. Seroconversion may be accompanied with flu-like symptoms.


longitudinal study

A study in which information is collected on people over several weeks, months or years. People may be followed forward in time (a prospective study), or information may be collected on past events (a retrospective study).

It is already well established that, in relation to the black HIV epidemic in the United States, the sexual and drug-using behaviour of an individual gives only a partial indication of their risk of acquiring HIV. There appear to be factors related to a person’s social network and the wider community in which they live which may also increase their vulnerability to HIV. For example, MSM who have dense social networks mostly made up of other black MSM are at greater risk of acquiring HIV.

The Centers for Disease Control and Prevention (CDC) has endorsed the concept of ‘community viral load’ as a measure which reflects some of the HIV risks in the wider community. Community viral load is the average viral load of people with HIV in a given geographic area – a higher average reflects a greater number of people whose treatment is not fully effective and so are infectious. Researchers have shown that in several settings, community viral load has fallen at the same time as HIV diagnoses.

However, community viral load has been criticised on several counts – it is generally based on viral load measurements of people who are diagnosed and attending medical services, whereas the people who are most likely to be infectious are either undiagnosed or have dropped out of care. It does not take HIV prevalence into account – as well as considering the proportion of people with HIV who are infectious, it’s also important to consider how many people in the population have HIV at all.

Furthermore, it assumes a geographically bounded community, taking no account of the make-up of a person’s sexual network.

Earlier this year, researchers from the same group at the University of Chicago published a study which used phylogenetic analysis (examination of the genetic make-up of the virus) to identify clusters of infection among newly diagnosed people in Chicago. It found that young black MSM were especially likely to form part of a cluster of newly diagnosed people, most often primarily made up of other young black MSM. But the study also found that clusters typically included people from all parts of Chicago, rather than being restricted to one geographical area.

The researchers suggested that dating apps and themed nights at bars and clubs may facilitate sexual encounters throughout the city. Whatever the explanation, the study showed that an individual’s sexual network is unlikely to be restricted to his own neighbourhood. 

Network viral load in Chicago

Britt Skaathun’s study is based on data from uConnect, a cohort study which recruited black MSM aged 16 to 29 who were living in south Chicago.

Each participant was asked to recruit a number of friends or peers who were also eligible for the study, according to a methodology known as respondent-driven sampling. It is often used to help researchers engage populations that might be otherwise hard to reach. In this case, it has also helped them map part of each man’s social and sexual network. The researchers say that young black MSM’s networks are dynamic, with social connections frequently becoming sexual connections and vice versa.

Data from 457 participants were analysed. They had an average age of 23 and 39% were already living with HIV. While 66% identified as gay, 27% identified as bisexual and 4% as straight (all had recently had sex with a man). Half had health care coverage, a third were unemployed, a quarter had been homeless in the past year, and just under half had ever been involved with the criminal justice system.

Looking at each man’s social networks (of up to 7 other study participants), the researchers found that:

  • 46% of men had a social network in which all members were HIV negative
  • 10% had a social network in which less than half of members were HIV positive
  • 44% had a social network in which more than half of members were HIV positive.

To assess ‘network viral load’, the researchers calculated the proportion of each man’s social network that was viremic, in other words with a viral load above 20,000 copies/ml. Take, for example, a man who had social connections with five other study participants. If two of these peers were HIV negative, one was HIV positive with an undetectable viral load, and two were HIV positive with high viral loads, then 40% of his social network would be viremic.

Looking at the data:

  • 46% of men had a social network in which all members were HIV negative
  • 37% had a social network in which fewer than 10% were viremic
  • 17% had a social network in which more than 10% were viremic.

Moreover, network viral load was associated with having HIV. After adjusting for other factors which could influence the results (such as sexual behaviour, drug use, age and education), men with a social network in which more than 10% of members were viremic were three times more likely to be HIV positive (adjusted odds ratio 2.75) than men with an HIV-negative network. Similarly, men with a social network in which fewer than 10% were viremic were twice as likely to be HIV positive (adjusted odds ratio 1.85). The ‘community viral load’ of the area in which a man lived was not associated with his HIV status.

However, those results are cross-sectional, from one point in time only. More persuasive would be longitudinal results, if they demonstrated that men with a high network viral load at the beginning of the study were more likely to acquire HIV later on.

A separate analysis of the same uConnect dataset does show links between men’s social networks and their own risk of seroconversion. From Ethan Morgan and colleagues at the University of Chicago, it is also published online ahead of print in the Journal of Acquired Immune Deficiency Syndromes.

Morgan had follow-up data from 343 young black MSM who were HIV negative when they enrolled in the study; 33 of them (10%) seroconverted during the 18 months of follow-up. This analysis considered the number of men in their social network who had either acquired HIV in the previous nine months or had long-term HIV infection. During recent infection, people have exceptionally high viral loads which make HIV transmission more likely.

Looking at the social networks the participants described when they entered the study, having more recently infected contacts was associated with a greater risk of seroconversion. After adjusting for other factors that could influence the result, each additional recently infected social contact multiplied the risk of seroconversion by 13 (adjusted odds ratio 12.96).

On the other hand, the number of network members with long-term HIV infection (who are more likely to be taking treatment) made no difference to men’s risk of acquiring HIV.

For each additional social contact who was HIV negative, the risk of seroconversion was lowered (adjusted odds ratio 0.14). And having more PrEP users in one’s social network also lowered the risk (adjusted odds ratio 0.44). 


Ethan Morgan’s paper shows that having more people with recent HIV infection in one’s social network is associated with a greater risk of acquiring HIV. He says that an understanding of social networks could help better target health interventions. As has recently been shown in London, early HIV diagnosis and immediate treatment can disrupt HIV transmission through high-risk networks.

Britt Skaathun believes her approach could be used by public health departments in the United States. Data from surveillance of viral loads of people with HIV who are engaged in care could be combined with data from partner services (which identify large numbers of people who are HIV negative). People being tested or treated for HIV would also need to be asked about their social and sexual connections.

A calculation of network viral load could help public health practitioners to target PrEP and other interventions to those at greatest risk of acquiring HIV. Current assessments of PrEP eligibility are based on an individual’s personal behaviours and don’t take networks into account. “A paradigm shift in how we use PrEP that includes characteristics of the network, which could include NVL [network viral load], is needed,” she says. 


Skaathun B et al. Network Viral Load: A Critical Metric For HIV Elimination. Journal of Acquired Immune Deficiency Syndromes, online ahead of print, 2017.

Morgan E et al. Are HIV Seroconversions Among Young Men Who Have Sex With Men Associated With Social Network Proximity To Recently Or Long-Term HIV-Infected Individuals? Journal of Acquired Immune Deficiency Syndromes, online ahead of print, 2017.

Morgan E et al. HIV-1 Infection and Transmission Networks of Younger People in Chicago, Illinois, 2005-2011. Public Health Reports 132: 48-55, 2017.