Friday, February 3, 2012

PEPFAR Committee Pulls Rug Out From Under Its Own Feet

About a year ago PEPFAR carried out a technical consultation on 'mixed' HIV epidemics. These describe country epidemics where HIV prevalence is relatively high in the general population (compared to non-African countries), and high or very high among members of groups known globally to be most at risk, such as men who have sex with men and intravenous drug users. The consultation involved representatives from 12 countries but I'll just look at one right now, Ethiopia.

To make things clearer, in every country with a HIV epidemic, in every continent, HIV prevalence is found in high risk groups, such as the ones mentioned above. But it is only in some countries in sub-Saharan Africa that HIV prevalence is high or very high outside of these groups. In fact, in most countries in the world, HIV is rare outside of high-risk groups. And it is only in some countries that HIV prevalence is inordinately high among commercial sex workers, most of them in Africa.

So all countries with a serious HIV epidemic could be called 'mixed', whether HIV prevalence is high or low in the general population, that is, outside of high-risk groups. The problem of explaining why a virus that is difficult to transmit through heterosexual sex appears to be high among people who only engage in heterosexual sex is as perplexing for the highest prevalence countries, such as Swaziland, Botswana, Zimbabwe and South Africa, as it is for Ethiopia, Kenya, Uganda or Tanzania, where prevalence is a lot lower.

HIV prevalence in Ethiopia, at 2.4%, is far lower than in countries such as Kenya, Uganda or Tanzania, where it currently hovers at around 6%. But given the huge population of Ethiopia there could be almost as many people living with HIV there as in each of the other three countries. An estimated 137,000 people become newly infected with HIV every year. According to PEPFAR, 87% of these new infections are transmitted through heterosexual sex, with another 10% transmitted from mother to child.

This leaves a mere 3% for other modes of transmission, whether they be through men having sex with men, intravenous drug use or unsafe healthcare (something PEPFAR people tend not to mention). There are three women infected for every two men and, while prevalence is less than 1% in rural areas it is nearly 8% in urban areas. The vast majority of people, over 80%, live in rural areas and there a lot more males than females living in urban areas.

The consultation identified several 'risks' for HIV, including multiple, concurrent partnerships, inconsistent use of condoms, transactional or commercial sex, intergenerational sex (between older men and younger women), early initiation of sex for females, high prevalence of sexually transmitted infections, etc. At least, these are thought to be risks for sexual transmission. Injecting drug use and men having sex with men are said to be 'emerging' in Ethiopia.

But it's interesting to compare those 'risks' with the 2005 Demographic and Health Survey for Ethiopia. This survey suggests that there are some non-sexual risks that should be examined, especially among women who received "Birth and delivery care by [a] professional", where HIV prevalence is 9.9%, compared to only 1.2% for those who did not receive care by a professional and 2% for those who had not given birth in the last three years. HIV prevalence was 3.5% among those receiving ante-natal care, compared to 1% among those not receiving care and 2% among those who hadn't given birth in the last three years.

With some of the figures, it might be wondered which men are infecting women. Among women who have had 'higher risk' sex in the past year, HIV prevalence is 12.3%, but among men it is only 1.8%. And one of those figures that should be disturbing to condom manufacturers is that prevalence is 20% among women who have used condoms compared to 2% among women who have never used them. Prevalence among uncircumcised men is only 1.1%, compared to .9% among circumcised men, hardly a massive difference.

As is commonly found in such surveys, women in the wealthiest quintile and those with the highest levels of education are significantly more likely to be infected. In fact, the bulk of infections among women, which consititute the bulk of heterosexual infections, are among wealthier, better educated women. Wealthier and better educated men are also far more likely to be infected than poor and less well educated men.

As UNAIDS say, 'know your epidemic, know your response'. So should Ethiopeans reduce education and poverty reduction programs? Should fewer women attend ante-natal or post natal care? Perhaps they should avoid cities, where most health facilities are? Should health facilities be extended out to the rural areas, or would that increase the risk that women in rural areas face? One certainly wouldn't expect condoms to increase HIV risk among those engaging in heterosexual intercourse.

The data clearly show that HIV risk is not closely correlated with sexual practices. On the contrary, it is correlated with non-sexual risks, such as post and ante-natal care. Many of the vulnerabilities PEPFAR identify are not vulnerabilities to HIV; FGM (which is actually generally correlated with low HIV prevalence) is far more common in rural areas, where HIV prevalence is very low; poverty is clearly correlated with low HIV prevalence; early and intergenerational marriage is also more common in rural areas. There may be some true sexual risks, but the 87% for heterosexual transmission mentioned above can be no more than a delusion.

Before escalating their usual response, which is to assume that almost all transmission is heterosexual despite evidence to the contrary, PEPFAR should remember how closely HIV tends to cluster around main roads, close to rural centers and, crucially, close to health facilities. HIV prevalence is not evenly distributed throughout countries like Ethiopia, whereas most of the 'risk' factors listed are very general and probably quite evenly distributed; many are likely to be just as common in non-African countries.

The most at risk population in Ethiopia, as in many other high prevalence countries, is female, urban dwelling, wealthier and better educated. That does not suggest a mainly heterosexually driven epidemic and shows that many of the 'risk factors' identified by PEPFAR are red herrings. Things that are true of Ethiopia may well be true of the other 11 countries. But PEPFAR will continue to concentrate on sexual risk and ignore non-sexual risk because that is what they appear to have set out to find.

It's odd that PEPFAR should have called Ethiopia's epidemic 'mixed', only to then claim that 87% of the virus is heterosexually transmitted and another 10% is, presumably, indirectly heterosexually transmitted. But they go on to say that neither men who have sex with men nor intravenous drug use contributes much to the epidemic. While it's hard enough to explain very high levels of heterosexual transmission anywhere, the idea that the virus is almost never transmitted through unsafe healthcare or cosmetic practices in a country with deplorable living conditions and 1.2 million HIV positive people is simply untenable.

[For more about non-sexual HIV risks, such as through unsafe healthcare practices, see the Don't Get Stuck With HIV site.]



Petit Poulet said...

So how many unexplained anomalies are needed before we change the paradigm? I think we have exceeded the normal limits.

There is an explanation that incorporates what has been observed and doesn't rely on racial stereotypes. Too bad no one want to talk about except for you.

Simon said...

Well, there are others talking about it, especially on the Don't Get Stuck With HIV site, to which I also contribute. Yes, anomalies occur, the problem is that so many of them remain uninvestigated.

Joseph said...

"FGM (which is actually generally correlated with low HIV prevalence) is far more common in rural areas, where HIV prevalence is very low;"

Just out of curiosity, what sources are you referring to for this information?

Simon said...

Hi Joseph, that sort of data is collected by MeasureDHS:

It's published in their Demographic and Health Surveys or their Aids Indicator Surveys. Try the data for Kenya and Tanzania, if you have trouble finding it let me know.

Anonymous said...

When we go for a HIV test or any other blood test like diabetes or cholesterol, what steps should we take to ensure that we don't get infected by contaminated equipment? How do we know that the equipment hasn't been fraudulently resealed / repackaged by corrupt officials and suppliers?

Simon said...

Hi Anonymous, chech the Healthcare Risks section of Some procedures are very low risk; equipment reuse is probably more likely in developing countries.

Anonymous said...

HIV prevalence among women who used hospital and home delivery was 9.9% and 1.2% respectively. I don't believe that the women who used hospital delivery are 10 times promiscuous than those who used home delivery.

Simon said...

Anonymous: "I don't believe that the women who used hospital delivery are 10 times promiscuous than those who used home delivery."

Quite; non-sexual risks need to be investigated in such instances.

Anonymous said...

Using World Bank data for the countries with the top 12 prevalence rates in the world, there seems to be an inverse relationship between hospital births and HIV prevalence. Check out these charts.

By the way, my stats training isn't beyond KCSE.

Simon said...

Thank you Anonymous, the relationship is certainly in need of analysis as the high prevalence countries do tend to have higher access to skilled medical staff.