[afro-nets] Not A&B but C&D reduce HIV prevalence (2)

[Edward C. Green <EGreendc@aol.com>]

Not A&B but C&D reduce HIV prevalence (2)
-----------------------------------------

That would be nice if true, because it would justify that many
hundreds of millions of dollars that major donors have put into
condom social marketing as their primary intervention for AIDS
prevention. It would justify the Western medical risk-reduction-
only paradigm, developed by Americans for an American epidemic.
Some time we should discuss how well this model worked in Amer-
ica. But the previous posting was about Uganda.

Alas, the "CD but not AB" is not true for Uganda, or Africa.
Which African country has the highest levels of C&D? Botswana.
Need I say more?

Perhaps a bit more. Comments by people familiar with the data
can be found at:

British Medical Journal
http://bmj.bmjjournals.com/cgi/eletters/330/7490/496-a

New York Times
http://www.nytimes.com/2005/03/14/opinion/l14uganda.html?
http://www.nytimes.com/2005/02/26/opinion/l26aids.html?

Here is the letter by Jim Shelton of USAID, for example: Partner
Reduction Remains the Predominant Explanation

8 March 2005 - James D. Shelton, Senior Medical Scientist Agency
for International Development, Washington, DC 20523
Send response to journal:
Re: Partner Reduction Remains the Predominant Explanation

Partner reduction and fidelity, the "B" in ABC was clearly the
predominant factor in the decline in HIV incidence that occurred
in Uganda from the late 1980's to about 1994. Dr. Wawer's pres-
entation does not seriously challenge that. Unfortunately the
BMJ article and others confuse incidence with prevalence. More-
over, by 1994, the starting point for the reported data from Ra-
kai, the decline in incidence nationally, was essentially over.
So the reported Rakai data come too late to help us understand
that decline. I'm afraid the current batch of news articles
based on Dr. Wawer's presentation represent more of the point-
less A versus C debate, when in fact B is much more important in
generalized epidemics. Bear in mind, these are very highly re-
spected researchers and the Rakai study is the premier of such
studies on HIV in Africa. While Rakai has been a site of inten-
sive effort, it is probably somewhat reflective of Uganda over-
all.

However: 1. To assess prevention efforts, we are really inter-
ested in incidence, not prevalence. Prevention efforts can only
work on new infections (i.e. incidence.) In a generalized epi-
demic, the large reservoir of people who are already infected
dominate prevalence. 2. Death is essentially the only way preva-
lence can decrease when you are dealing with a lifelong infec-
tion. (Setting aside subtle effects of demographic change.) So
when death is described as the major contributor to decline in
prevalence it is no surprise. (However for this to happen, inci-
dence has to be lower than mortality.) 3. Most importantly,
1994-2003 is too late to explain the earlier large decline in
incidence nationally in Uganda. While incidence estimates have
to be inferred from other data, in all likelihood incidence
peaked in Uganda about 1989, declined dramatically and stabi-
lized around 1994 (though there may have been some subtle
changes thereafter.) But prevalence continued to decline through
2000 as people died who were infected years earlier. (See Stone-
burner RL and Low-Beer D. Science 2004:714-718.)

So data from Rakai beginning in 1994 miss the major decline in
incidence. What the Rakai data can describe quite well is the
dynamic of B and C from 1994 onward. But it is actually worri-
some re prevention and the condom. The Rakai study is powerful
in that it includes ongoing HIV incidence and also sexual behav-
ior. Over the 1994 -2003 time period, HIV incidence varied, but
was fairly stable. So the previous downward momentum on HIV in-
cidence was largely lost. But during that period, there appears
to have been some substitution of condoms for B, or in other
words "compensating" behavior of more C and less B. This is wor-
risome. It may reflect so-called "disinhibition" of risky sex
practice as has occurred among gay men in the U.S. This argues
for promoting B,C and A in a mutually supportive way to maximize
overall risk reduction. But B is the most powerful in a general-
ized epidemic, especially reduction of concurrent partners.


Edward C. Green, Senior Research Scientist Harvard Center for
Population and Development Studies, Cambridge, MA 02138 USA
Send response to journal:
Re: Rakai study does not invalidate the ABC model

This summary of the Rakai findings calls into question the Bush
administration's ABC policy (Abstain, Be faithful, use a Con-
dom). But the summary is misleading. There were fundamental
changes in sexual behaviour in the Rakai district of Uganda, and
in Uganda as a whole, between 1987 and the mid-1990s, i.e., less
casual sex, more monogamy, and fewer youths ages 15 to 19 sexu-
ally active (1-3). In an earlier study in Rakai by at least one
of the same authors as the new study, "Only 4.4% reported con-
sistent condom use and 16.5% reported inconsistent use during
the prior year" (4). Thus during the period when HIV incidence
fell in Rakai (5), there were higher levels of "A and B"
Competing interests: None declared

--
Edward C. Green
mailto:EGreendc@aol.com