Thursday, September 29, 2011

HIV Concurrency Theory is Dead; Can't You Smell it Yet?

In their desperation to explain why HIV prevalence is extremely high in African countries, hundreds of times higher than it is in many other countries, UNAIDS have flailed around and grabbed at any new theory that came their way. One of those theories was that overlapping sexual partnerships are responsible for levels of transmission that cannot be explained by any scientific data that is currently available. This is called the 'concurrency' hypothesis.

There remained the problem of showing that concurrency was exceptionally high in areas where HIV prevalence was high. This is still a problem and strong correlations between concurrency and HIV prevalence, even weak ones, have yet to be identified. But a more acute problem was the question of whether high rates of concurrency really would result in high rates of HIV transmission; is concurrency "especially effective in spreading HIV". The answer is 'no'.

Larry Sawers, Alan Isaac and Eileen Stillwaggon used a modified version of the mathematical model and data used by earlier researchers and added in an element that was missing from earlier work; 'coital dilution'. If someone has many partners, they have less sex per partner, on average. They found that, even with very high levels of concurrency, higher than those found in any population, once you add in slightly lower levels of coital frequency, the result is 'epidemic extinction'.

The orthodox view that Africans have inordinate amounts of sex, and inordinate amounts are required to explain massive HIV epidemics found in some African countries, is not supported by logical or empirical evidence. Only simulation models that use non-empirical evidence, or a very selective use of evidence, can 'explain' very high levels of heterosexual HIV transmission. And models that do use empirical evidence show that the orthodox view is wrong.

The authors conclude that "concurrency cannot be an important driver of HIV epidemics in subSaharan Africa. Alternative explanations for HIV epidemics in sub-Saharan Africa are needed." The concept of coital dilution even explains why "polygyny appears to protect populations from HIV", though UNAIDS often cite polygyny as a risk factor in the spread of HIV (note, this is not an argument for the promotion of polygyny, just one showing that it does not contribute significantly to HIV epidemics in Africa).

Indeed, other phenomena cited as factors that increase the spread of HIV are also put into perspective in this paper. Mobile populations, migrants, especially internal and temporary migrants have often been targeted by HIV campaigns. High rates of HIV in, for example, mining populations, were said to be a result of the sexual behavior of those infected. In the light of this research, miners are more likely to have been infected through shoddy health care practices provided by mine owners, who tend to score very badly in the field of corporate social responsibility.

The graphic depictions of epidemics driven by heterosexual sex alone becoming extinct are fascinating and I recommend the article. But there are other questions that were in need of answering before these researchers did this piece of work: why does the orthodox view of HIV, promulgated by UNAIDS and most of the other wealthy HIV institutions, fly in the face of evidence? And how can a view that is so inherently racist be accepted by so many?

But there's no point in wringing our hands in despair now that we know we have spent such a long time and so much money barking up the wrong tree. What are the 'alternative explanations' that these researchers suggest are needed? Eileen Stillwaggon herself has a whole set of suggestions, after all, why should there only be one way to reduce HIV transmission?

Plenty of other suggestions can be found on the Don't Get Stuck With HIV website, which looks at medical and cosmetic risks for HIV transmission. But at least we don't have to subscribe to the orthodox view of Africans as mere victims of their own stupidity, unsafe sexual behavior and lack of consideration for those around them, including their closest friends and relations.



Anonymous said...

Skepticism is a great thing, but it's worth being a bit skeptical of the critiques leveled at the concurrency hypothesis as well. The key error in your summary above is the line that "There remained the problem of showing that concurrency was exceptionally high in areas where HIV prevalence was high." Concurrency does not need to be exceptionally high. A little goes a long way, because it creates "connectivity" at any point in time (unlike serial monogamy, which completely breaks up a network at any point in time). Take a look at this network movie to see the effect: In this example, well under 10% of both groups have concurrent partners at any point in time, and only 5% of all partnerships are concurrent over a 10 year period, but these concurrencies account for over 50% of the possible infections. Intuitively, I think we all know that when our partner has other (unprotected) partners, we are put at risk.

Ironically, the concurrency hypothesis was first proposed as the way to explain how hyperepidemics can be established in populations that do *not* have lots of sexual partners. It was *a critique of* the argument that the epidemics in Southern and Eastern Africa were proof those folks had lots of sexual partners.

Simon said...

Well Larry Sawers and Eileen Stillwaggon have been putting the 'theory' in perspective over a number of years and several substantial articles, the latest of which appeared in the IAS:

We can differ as to what is most deserving of skepticism. The suggestion you make is simply not borne out by any evidence; it remains a hypothesis for which there is precious little evidence supporting and a hell of a lot of evidence opposing. Is that not how one exercises skepticism?

Intuitively, we know when our partner takes risks, we are at risk, but if our partner is a woman and it is she who is taking the risk, she is much less likely to infect her male partner than the other way around. At least, that seems to be the case in Western countries.

There was pressure to come up with something like a concurrency hypothesis because it was so obvious that there are promiscuous people in many places, not just in high HIV prevalence countries, but HIV prevalence often seemed relatively unrelated to sexual behavior, safe or otherwise.

There were all sorts of other attempts; dry sex, 'traditional' practices, widow inheritance, etc, but they collapsed the same way concurrency did as hypotheses. Circumcision is another commonly cited reason for high HIV transmission and if you use the evidence that supports the hypothesis and discard or ignore evidence that doesn't, you've got a watertight theory, albeit a completely unscientific one. But hey, if there's money in it, why not?