Crush Liberalism

5 Comments »

1. I believe the press is trying to make more out of this then is actually there. As a person who frequently uses these figures, I can confirm that there was no attempt to actually “cook the books”. In fact there are a few key reasons why the numbers were overestimated, and its important to recognize this.

   First, there was an assumption that the prevalence of HIV among men and women in Africa were relatively similar (women having only a slightly higher prevalence than men). The reason this is important was that up until a few years ago, pregnant women provided the only available data on HIV in the general population. When new surveys (DHS+)were conducted in the last few years, it was found that the pregnant women data was pretty comparable to the new data being collected among women. However, it was also found that the assumed prevalence among men was much lower than anticipated. In other words, women are much more likely to be infected than men in Africa. As a result, prevalence data based only on pregnant women will overestimate the numbers in the general population. The new surveys found in fact that new prevalence estimates were 30% lower than the original estimates based solely on pregnant women.

   Second, there is now better data for many countries in more rural areas. When surveys were originally conducted, the focus was largely on urban areas. This made sense, given the lack of resources to conduct large-scale surveys all over each country. In regions where there was no data available, it was necessary to make a best guess. In the case of Mozambique, for example, there was originally no data available in the north of the country. Therefore it was necessary to make an educated guess about the prevalence in that part of the country based on the data available in the south and central parts of Mozambique. When data did become available in the north, the prevalence estimates were revised downwards.

   Third, we now have better data for India. I worked in the area of HIV/AIDS in India a number of years ago, and everyone back then recognized that there was a complete lack of data on HIV prevalence in India (people would tell me there were 5 million people infected with HIV, but nobody could explain where this number came from). When you are dealing with a population of 1 billion, it obviously makes a big difference what prevalence estimate you use.

   Does this all mean that none of the numbers were inflated? I couldn’t say that they were all necessarily the best numbers possible, but I can say that the numbers come from the countries themselves. Countries are concerned about the financial implications of showing a drop in HIV prevalence. We see with countries such as Cambodia that a drop in HIV prevalence does result in less attention and resources by donors. However, I would say that most countries do actually make the best possible estimates based on the data which is available.

   My main complaint with UNAIDS numbers is not that they are inflated, but that they are poorly explained. Last year the numbers for the Caribbean (largely the island of Hispaniola) were revised downwards significantly, with little explanation about how or why this occurred. I hope that in countries like Honduras, which has recently cut their estimates by more than half, the explanation for this change will be more comprehensive.

   I also believe that we should be very careful about continuing to cite the same 3 critics of UNAIDS (Drs. Chin, Halperin and Green). I have read Dr. Chin’s book about UNAIDS estimates. While I agree that a revision downwards was necessary (although I’m skeptical of his estimate of 25 million), it is also obvious that Dr. Chin has his own ax to grind and has a vested interest in saying that the numbers were better when he was in charge of making those estimates at WHO. As for Dr. Halperin, I have worked with him and have noted some very significant problems in the way he analyzes data. As an anthropologist (not an epidemiologist), I believe that many of his conclusions have been deeply flawed (Dr. Halperin, for example, estimated large-scale occurrences of “dry sex” in the Dominican Republic, with no data to support these conclusions). As the press continues to cite only these 3 critics, I believe we are failing to pursue the truth.

   In conclusion, I think that we should be skeptical of those who claim that UNAIDS was being alarmist. The real answer, in my opinion, is that the estimates are getting better, and we are finding that some of the original estimates were influenced by assumptions which tended to overestimate the pandemic. Politics had very little to do with the overestimate, although we should all remain vigilant in assuring that the data which is used is applied properly.

   Comment by Steven | November 22, 2007

2. I still find the notion of women particularly black women having more aids than men to be preposterous!
   Women of a particular race having more aids than men of the exact same race is nonsense!
   Take into account that women usually prefer sexual encounters with men from their race rather than outsiders you’ll have to assume the whole race is more suspectible to the aids virus which is nonsense!

   I’m no scientist but I’m just being logical.

   Comment by Chilerkle | November 24, 2007

3. At first glance it does seem illogical that within a certain race, prevalence would be higher in one gender than another. However, it isn’t really so difficult to imagine. In Latin America, for example, the epidemic is clearly more common in men than women,due in large part to the effect of gay men and injecting drug users. The same is true for the US and most of Europe, as well as large parts of Asia.

   It does become more complicated when you have an epidemic which is largely heterosexual, such as exists in Sub-Saharan Africa. One could argue that women are more affected than men because of prostitution, but there is no evidence that prostitution is more common in Africa than it is in other regions. In fact I would suspect it is actually less common. It might be argued that women are more infected because they are more “promiscuous”, but again the evidence doesn’t support the argument. What does seem to explain this difference is concurrence. Both Epstein and Chin argue in their books that Africa’s epidemic (it isn’t really one epidemic; Africa has the most heterogeneous set of epidemics in the world) is driven by concurrence, which not only explains the high prevalence overall, but also explains why women are more infected than men.

   As for the actual data, I’ve included a quick review of 12 Demographic and Health Surveys (DHS) conducted in Sub-Saharan Africa. The data shows that prevalence is higher for women in 11 of the 12 countries. The only exception is Burkina Faso, which has a prevalence of 1.9% among men and 1.8% among women. Thus it appears pretty convincing that women in Africa are more affected than men (the average of the data below shows that women are 45% more likely to be infected than men, comparing 5.4% among men to 7.8% among women).

   Of course there is no guarantee that concurrence is the only explanation. What do you think explains the differences, Chilerkle?

   Male Female
   Population-based (Indicator Value )
   Sub-Saharan Africa
   Burkina Faso Demographic and Health Survey 2003 1.9 1.8
   Cameroon Demographic and Health Survey 2004 3.9 6.8
   Ethiopia Demographic and Health Survey 2005 0.9 1.9
   Ghana Demographic and Health Survey 2003 1.6 2.7
   Guinee Enquete Demographique et de Sante 2005 1.1 1.9
   Lesotho Demographic and Health Survey 2004 18.7 26.3
   Mali Enquete Demographique et de Sante 2001 1.3 2
   Rwanda Demographic and Health Survey 2005 2.2 3.6
   Senegal Demographic and Health Survey 2005 0.4 0.9
   Uganda HIV/AIDS Sero-Behavioural Survey 2004-05 5.2 7.3
   Zambia Demographic and Health Survey 2001-2002 12.6 17.8
   Zimbabwe Demographic and Health Survey 2005-06 14.8 21.1

   Comment by Steven | November 27, 2007

4. Well the question is are they bisexual or lesbian?

   If they are exclusively heterosexual than the data is suspect in my opinion but since I have no real way of knowing.
   There is also the possibility that the data maybe rigged to achieve that lop-sided AIDs statistics!

   Comment by Chilerkle | November 27, 2007

5. It is unlikely that a disproportionate share of women in Africa are lesbian or bisexual. Furthermore its not likely that this would explain the higher prevalence, since there is no increased risk associated with lesbianism (there is higher risk associated with male homosexuality).

   The DHS surveys are not “rigged” and they are not produced by UNAIDS (they are managed by a company in Maryland with funding from USAID).

   The explanation is that most women appear to be as “faithful” in their relationships as women in other countries. However, men in parts of Africa tend to have multiple concurrent partners. Since much of the transmission occurs when a person first becomes infected (generally about 6 weeks after infection), the most dangerous time for transmission is in the days after infection. If a woman becomes infected during these early stages, she may pass it on to her regular partner, but because most women are faithful, only one person at most becomes infected when a woman is positive. This is obviously not the case if the woman is a sex worker or if she has multiple partners. In fact in countries where everyone passes on the virus to fewer than 1 other person, the epidemic will eventually decline. The epidemic can only take off if people have multiple partners, which is not the case among most women in Africa. It is, however, the case among men in Africa (or so the argument goes). African men do not necessarily have more partners than men in other countries, but they do have more concurrent partners. Why does that matter? Well if most new infections occur during a 6 week period of time, then what really matters is how many partners you have at that time. If you engage in serial monogamy (have one partner, then another partner a year later, then another partner the next year), you are unlikely to create a lot of new HIV infections. If you have 3 girlfriends at the same time and you get infected, then you are likely to sustain the epidemic.

   If a man has multiple concurrent partners (his wife and his 2 girlfriends, for example), but the women are all “faithful” to this man, then many more women will become infected in the community than men.

   There are a number of ways to address AIDS, therefore. One is President Bush’s approach (based largely on data from Uganda, where the epidemic significantly declined starting in the late 1980s) in which you focus on abstinence, be faithful, or use a condom (ABC). As Epstein argues in her recent book, abstinence programs have not been particularly successful (although in the case of Uganda, it has been argued that raising the age at first sex among educated girls actually did have a very beneficial impact on the epidemic). Promoting faithfulness can also be difficult, but as both Epstein and Halperin argue, this method is likely to have the biggest impact. Finally condom programs have been successful, although more so in terms of the highest risk groups (its very hard to convince married couples to use condoms consistently).

   My question, however, concerns why you, Chilerkle, believe that the numbers are lop-sided? Is there one country you can point to where you think that the prevalence estimates have been inflated or at the very least, are overestimated?

   Comment by Steven | November 28, 2007

« Previous | Next »

Site info

Crush Liberalism
Theme: Andreas04 by Andreas Viklund. Clone this site at WordPress.com