snip>
>You mean the Boston Globe is not a credible source?
>What evidence do you have for that?
>
>Or are you saying that "of course", HIV tests in Africa
>are not 99% accurate and how silly of me to think so?
Alex, it is not about the test per se. The sensitivity and specificity
of HIV testing depends on how it is done in part, but is VERY high.
The issue is that a lot of the prevalence of HIV is undertaken by
sentinel surveillance sites, antenatal clinics and the like. As such,
a lot of the WHO and other statistics on the numbers of HIV-infected
individuals in a community are estimates.
As ever, I indeed hope and pray that this is the case that the
estimates are somewhat on the high side!! That would be great news.
By contrast, it does not matter if there is but ONE infection--and
there are way, way, way more than one...we need to have access to
care. We need to improve healthcare access for the poorest in the
world. We need to give a damn and make some changes to make this a
better world.
Or do you think it's OK to say--gee, it ain't that many so we don't
have to worry about 4 million people dying this year of AIDS? It's
fewer than we though, so let those kids waste away and die? Let the
mothers die. Let the men die? The miners, teachers, doctors, soldiers,
government officials, farmers, wives, truckers, husbands,
grandmothers...it's OK. Ain't THAT many. Let 'em die. Is that what you
mean?
George M. Carter
Alex - 22 Jun 2004 07:17 GMT
> snip>
> >You mean the Boston Globe is not a credible source?
[quoted text clipped - 5 lines]
> Alex, it is not about the test per se. The sensitivity and specificity
> of HIV testing depends on how it is done in part, but is VERY high.
I would say a single ELISA is much more specific in an environment
(Europe, North America) that is pretty much free of the pathogens
that are _known_ to increase the number of false positives, than they
are in an environment where such pathogens (malaria, tb, leprosy,
and who knows even malnutrition) are rife (Africa).
And I'm not even talking about the effect of only using pregnant
women (ANC studies).
> The issue is that a lot of the prevalence of HIV is undertaken by
> sentinel surveillance sites, antenatal clinics and the like. As such,
> a lot of the WHO and other statistics on the numbers of HIV-infected
> individuals in a community are estimates.
They are all estimates. The difference is that some estimates
are based on surveys that more closely represent the statistical
ideal.
There are lots of reasons why ANC studies of women at antenatal
clinics say far less about the general population, than DHS population
surveys which take representative samples of the entire population.
However, one big issue remains, and that is that both ANC and
DHS surveys do not use Western Blot as a confirmation test.
At best, they use a second ELISA if the first one was positive.
In the US, it is mandatory for testing facilities to confirm a positive
ELISA with a positive Western Blot, before telling the patient that
he or she is HIV positive.
There is no excuse at all, for not using the same rigor when doing
surveillance testing.
If a DHS tests 20,000 people, and 10% test positive (as they
did roughly in South Africa), that would only mean 2000 people
whose samples require further testing.
There is no excuse not to use Western Blot on those samples.
It will be more expensive than all ELISA testing, but the data
will be much more valuable.
That's my beef as far as surveillance data is concerned.
Alex