Saturday, September 25, 2010

The ‘Unknown Knowns’ of HIV Epidemiology

In his book ‘The Black Swan’, Nassim Nicholas Taleb distinguishes between two realms, Mediocristan and Extremistan. The law of Mediocristan is: when your sample is large, no single instance will significantly change the aggregate or the total (p32). The law of Extremistan is: inequalities are such that one single observation can disproportionately impact the aggregate, or the total (p33).

An example from Mediocristan that Taleb gives is of sex workers, who can’t scale up their workload beyond a certain number of clients per hour or day (p27). Their earnings, no matter how high, are limited. In contrast, a successful financial trader can earn (or lose) more in one day than in their entire career. The financial trader’s job is scalable and is from the realm of Extremistan.

Wars used to belong to Mediocristan but modern technology makes it (theoretically) possible to wipe out whole populations, cities or even humanity in a single act (33-4). I would argue that it is also possible to infect huge populations with a virus through modern technology. There have been many documented instances of this.

Therefore, HIV, being difficult to transmit heterosexually, is stuck in Mediocristan. Despite the best attempts of UNAIDS and others to claim that sex workers can become infected and infect many others in a short space of time, a HIV epidemic driven by sexual transmission alone would never reach very high levels. The HIV epidemics of most countries, UK, Germany, USA, Australia, etc, are safely inside Mediocristan.

But there are countries where levels of HIV transmission can not be explained by heterosexual transmission alone. Prevalence figures in these countries, especially in certain groups within these countries, are extreme. In fact, they appear to be from the realm of Extremistan. There is a limit to how high they can go and, thankfully, none have reached 100% yet. But the highest prevalence figures can be hundreds of times higher than the lowest.

An example that stands out comes from Nairobi in the mid 1980s, only a few years after HIV was identified and not long after the virus probably reached Kenya. Prevalence among sex workers was found to be 4% in 1981 and it had increased to 61% by 1985. By what means were these women infected? In order for four sex workers to (indirectly) infect 61 sex workers (to keep the numbers simple) in the space of four years, how many men would they have to infect? Because the 61 would have been, according to the sexual transmission theory, infected by HIV positive men.

In 1980 in Nairobi, a particular sample of men being treated for a sexually transmitted infection was found to contain no HIV infections. In 1985, a similar sample had a prevalence of 15%. Unless the sex workers were all having enormous amounts of sex with a handful of men, these findings are completely inexplicable. Even over a longer period of time, such as fifteen or twenty years, such high sexual transmission rates are not credible. And over that length of time, most of those originally infected would be long dead.

Even less explicable are the HIV prevalence figures for people who are neither sex workers nor sex worker clients. Prevalence for Swaziland as a whole is in the mid twenties. For young, sexually active women, figures are in the thirties and forties, though they remain consistently lower for men at any age. What kind of amounts of sex could these women be having? And with whom could they be having so much sex? In order to become infected sexually, they don’t just need to have lots of sex, they also have to have lots of sex with HIV positive men.

Rather than imputing inhuman feats to Africans, as UNAIDS and their friends have been doing for many years, it is worth looking at where very high rates of transmission could come from. And the obvious source for transmission rates that could only come from the realm of Extremistan is unsafe health care (and possibly unsafe cosmetic practices).

A clinic could give hundreds of injections a day. Vaccination programs can cover tens of thousands, even hundreds of thousands of people, in a very short period of time. It could take years, even decades, for one HIV positive person to infect a few more, and for them to go on and infect others. But unsafe injections could, potentially, infect hundreds or thousands in far less time.

I’m not claiming that current HIV epidemics in African countries are Black Swans, because the risk is not incomputable. It’s just very high. The knowledge required to prevent these transmission events from happening does not consist of ‘unknown unknowns’. Far from it, the knowledge required consists of ‘unknown knowns’, information about the efficiency of transmission through unsafe health care, coupled with information about the quality of health care available in high HIV prevalence countries.

UNAIDS and their collaborators have gone through a process of ‘unknowing’ things that have long been well-established, sometimes by their own employees. But rather than investigating the many instances of health care transmission of HIV, they have built up a literature that simply ignores the very possibility of such transmission. Occasionally, nosocomial and iatrogenic transmission are mentioned, only to be dismissed as very slight and unimportant.

Documented outbreaks of HIV from medical procedures have identified hundreds, even thousands of infected people. Disturbingly, most of these outbreaks have never been investigated to find out how many people have been infected. One of the most infamous outbreaks occurred in Romania in the 1980s, thought to have infected about 10,000 children. No search was made to identify all those infected. An estimated 100,000 were infected in China but again, no effort has been made to identify all those infected.

There are plenty of other outbreaks like these. A number of African countries have estimated the proportion of incident HIV infections caused by unsafe injections. The median is 19% but Malawi estimated in the early 2000s that 54% of HIV infections in women were attributable to injections. That’s a lot higher than the UNAIDS estimates of 2.5% or less for Kenya and other countries.

Where HIV prevalence figures are extreme, as they are in many African countries and regions, transmission rates must also be extreme. It is not possible for such extreme rates to occur from even the highest levels of sexual transmission imaginable. But such high rates could come from unsafe medical practices. And this is what needs to be investigated.

It may be a mammoth task to investigate and eliminate such practices. But it is a mammoth and pointless task to shake a warning finger at half a billion Africans and tick them off about their sex lives when sexual behaviour is clearly not the main problem.

Such investigation is vital because the one or two people that a sexually active HIV positive person might infect in their lifetime is dwarfed by the hundreds or thousands that health facilities may be causing or may cause some time in the future. Everyone infected, whether sexually or non-sexually, can go on to infect others sexually or non-sexually. If they infect others sexually, the resulting number may be low. But if their infection becomes part of a health facility based outbreak, as any single infection could, the numbers infected as a result could be very high.


No comments: