Social Network Theory in HIV Prevention, the Role of Metaphor and Scope

With the realization that social context plays a large role in HIV transmission in addition to personal behaviors, some people have turned to social network theory to help explain why HIV infection happens the ways it does, and to whom it does.
In turn, researchers, health promoters and activists have been struggling with (and arguing about) the implications for social network theory-based analyses on how to inform public health practice (for an example, check this out).
This post comes out mainly of a series of conversations with Michael Scarce & Peter Keogh, with my own special twist.

Wrong toolbox?
I'm going to argue that there is a profound disconnect between thinking about what social context means and how to design public health strategies that address social contextual factors, and social network theory as an approach for understanding the interpersonal dynamics of disease transmission.

Social network theory
From what I can gather, social network theory is about describing the linkages between people, and in the disease transmission context, using those linkages to help explain why a disease moves in the way that it does.

Social networks are often invoked to explain why rates of HIV infection are different in different demographic identified groups, despite similar occurrences of personal behaviors across these groups. But usually the evidence for that interpretation is the observation that the groups have different rates of HIV infection, a circular logic. I'm not aware of anyone having done the kind of individual mapping necessary to show how (over the span of decades, with constantly changing linkages) these communities came to have different HIV infection rates.

The metaphor of a social network as a map of nodes and lines, in this case describing who is friends with whom in a 4th grade class in the 1930's, has a remarkable similarity to the 'web of causation' models elaborated in epidemiology, especially since WWII.

Epidemiology's 'web of causation'
The metaphoric imagery of the web of causation is so vivid that a web to show linkages between causes and effects for any specific study is rarely actually drawn out. Usually, someone trained in public health sees one in their textbook at some point, then that's it. For illustrative purposes, here's an example from a 1966 US Government report. If you can reduce hyperlipidemia by reducing fat in a population's diet, or increasing physical exercise, this will reduce atherosclerosis, which in turn will reduce the clinical manifestations of heart attacks and strokes. The power of the 'web' among public health folks is that interrupting any of the pathways between causes, intermediaries, and finally disease outcomes can have a large impact on public health.

The similarity of these metaphorical representations I think is partly at fault for the way interventions based on the invocation of social network theory sound so strange.

When someone from Public Health sees a network of nodes connected by lines, our gut reaction is to start thinking about how to break them, so in this case, that translates into breaking social networks apart in order to limit disease transmission.

In the specific case of HIV prevention, this could mean, for instance, trying to interrupt HIV transmission by limiting the sexual vectors between younger and older men, since older men are more likely to have HIV.

But in a social network, the lines represent the substance, the sociality, of social networks. They are the sources of information, of resources, of resilience, not just vector lines across which bits of infectious protoplasm are transmitted.

So, any program designed to tear these networks apart, or prevent them from forming, seems counter-productive. Ripping the heart out of the communication of ideas and resources, the means men have to get support from one another, just to interrupt sexual connections between younger and older men is an exchange that even Faust would have been hesitant to make.

A kinder, gentler form of social engineering designed to limit social network growth has been proposed, the idea being to present young men with neutral and unbiased information about higher HIV infection rates among older men than younger men, the implication being that they should choose to associate only with other younger men.
The impact is the same, even if the intent is somewhat more benevolent. No matter how you dress it up in neutral language, the message that "older guys represent a sexual danger to you, young man, so avoid them" comes through loud and clear.

So, the invocation of social network theory, and it's strong metaphorical resemblance to the 'web of causation' probably has played a role in the development of such bizarre and counter-productive ideas for addressing the social contextual factors that influence HIV transmission.

Lace, Macrame, or Felt?
Perhaps we need a new metaphor to describe social networks. At the Chicago LGBTI Health Summit, Marshall (Feldman?) described the network of social relations in San Francisco as 'lace', its fragility due to the high turnover of people in this transitory town. This metaphor may be more useful for us public health types - why would you cut lace? If anything, wouldn't it be better to bolster the lace with new connections, tougher and stronger connections.

Perhaps the goal of health promotion would be to turn lace into macrame, the knitting craze that filled American houses with hanging plants in the 1970's. But there's something too vertical and hierarchical about macrame for my taste as a metaphorical substitution.
Someone suggested that felt, made by mashing fibers together, rather than weaving or knitting them, might be a good metaphor, it invokes the idea of the strands having multiple points of contact with one another, not fixed into a pre-arranged shape, and the tighter the felt, the more points of connection between the fibers, the more difficult it is to rip apart.

At any rate, using fabric as a metaphor pushes one to think about strengthening ties, not rending them apart.

Social networks as units of analysis
Another way to think about using social network theory is to design studies and interventions around the idea that social networks are discrete entities. This may be plausible when describing the strictly sexual links that have been used to trace STD transmission among heterosexuals, whose sexual networks tend to be dyadic, or only slightly more extensive than that. But the idea of discrete, separable social networks quickly falls apart when thinking about social ties other than strictly sexual ones, or even "purely" sexual networks in the vibrant sexual culture that gay men have fought so hard to create.
Despite this practical limitation, one could theoretically think about how to design a study using an individual's perceptions of their social network as the unit of analysis, and ask which types of social networks are more successful than others at help that individual avoid HIV, or avoid exposing others in the context of secondary prevention.
Doing this sort of research would imply that interventions should be designed to intervene in social networks that are identified as being dysfunctional. How to do that without simultaneously stigamtizing the very groups one is trying to assist seems difficult to me. In simpler terms, "Hi, we're here becuase we think your friends might let you down, so we want to help you make new friends" could be taken as very offensive to people who have a lot invested in their social networks, no matter how dysfunsctional social science research deems that style of network to be.

Broadening scope
So, perhaps the main problem isn't one of metaphor, but of scope. Rather than focus on the ties that bind individuals together within discrete social networks, a more fruitful way to think about the role of intentionally altering social contexts might be to go a bit broader, and engage the structural forces of gender hierarchy, material deprivation, and racial ideologies that work a broad social level. These forces certainly get reflected within social networks of friends and family, but their source, and the means for addressing them require a broader focus; on laws, policy, and public opinion. It requires engaging the total population, not just the relatively small population that is 'at risk' for whatever reasons.
With this broader focus, lobbying, grassroots organizing, school-based anti-bullying campaigns, and social marketing campaigns designed to address stigma become practical, efficient ways to think about intervening in the realm of social context to reduce HIV infection rates.

Throw out the tools if they aren't working
Don't get fooled by the work 'social' in social network theory. Social network theory may be a good way for infectious disease epidemiologists to understand disease spread through a time- and space-limited population (such as measles through the Faroe islands), but it's probably not a good guide when thinking about how to address the social contextual factors that influence HIV infection spread.

Get some new tools instead.