The Voevodin letter.
David Canzi
about the Voevodin letter, to use it in an argument against Alex. I
scrapped the argument that I had started to write when I noticed it was
invalid. But what I found searching USENET and the web for information
about the Voevodin letter deserves some comment.
Voevodin's letter to the Lancet was cited by the Perth Group and others
to demonstrate how unreliable ELISA tests are. Actually looking up the
Voevodin letter and reading it reveals quite a different picture of the
ELISA test's reliability. Only 1 test result out of 1000 was a false
positive.
I wrote an article about it in 1994:
http://groups.google.com/groups?selm=14880%40sci.med.aids&oe=UTF-8&output=gplain
The last paragraph of that article deserves quoting here:
| All the relevant numbers needed to judge the accuracy of ELISA were
| there in Voevodin's letter, and the figures show that ELISA is very
| accurate. Somehow, with the real truth staring them in the face,
| Papadopulos-Eleopulos and her coauthors managed to omit from their
| paper just the figures that had to be omitted so that a statistical
| fallacy could drastically mislead their audience about the accuracy
| of ELISA.
The title of that article, "Damned Statistics", was inspired by this
quote: "There are three kids of lies: lies, damned lies, and statistics."
A search of the web and USENET for articles mentioning both "Voevodin"
and "ELISA" turns up many hits, almost unanimously denialist arguments
for ELISA's unreliability. (This reminds me of another quote: "A lie
can go around the world before the truth can get its boots on.")
This illustrates something about the nature of the denialist movement.
They have all copied this argument uncritically from each other. This
is no doubt how all their arguments spread. Denialists are a densely
interconnected and susceptible population. Bad arguments spread among
them like epidemics.
--
David Canzi If money is the root of all evil, marketing
must be the trunk and most of the branches.
Chris Noble
> While replying to Alex, I searched for an article I wrote years ago
> about the Voevodin letter, to use it in an argument against Alex. I
> scrapped the argument that I had started to write when I noticed it was
> invalid. But what I found searching USENET and the web for information
> about the Voevodin letter deserves some comment.
>
> Voevodin's letter to the Lancet was cited by the Perth Group and others
> to demonstrate how unreliable ELISA tests are. Actually looking up the
> Voevodin letter and reading it reveals quite a different picture of the
> ELISA test's reliability. Only 1 test result out of 1000 was a false
> positive.
>
> I wrote an article about it in 1994:
> http://groups.google.com/groups?selm=14880%40sci.med.aids&oe=UTF-8&output=gplain
Voevodin's letter really just states the obvious. Even tests with high
specificity will have a low positive predictive value in low
prevalence populations. Why do Denialists think that they were the
first to think of this?
If you test every male in the US with a pregnancy test then a number
will test positive (testicular cancer can cause false positives). I
would be willing to bet that exactly zero of these will be confirmed
as being pregnant.
Therefore according to Denialist logic pregnancy tests are completely
unreliable, there is no such thing as pregnancy and even if there was
it has nothing to do with sex.
Chris Noble
GMCarter
wrote:
LOL--nice to see you Chris!!
George
Alex
> wrote:
>
> >dmc...@watserv1.uwaterloo.ca (David Canzi) wrote in message
news:<aji5if$6fm$1...@tabloid.uwaterloo.ca>...
> >> While replying to Alex, I searched for an article I wrote years ago
> >> about the Voevodin letter, to use it in an argument against Alex. I
> >> scrapped the argument that I had started to write when I noticed it was
> >> invalid. But what I found searching USENET and the web for information
> >> about the Voevodin letter deserves some comment.
> >>
> >> Voevodin's letter to the Lancet was cited by the Perth Group and others
> >> to demonstrate how unreliable ELISA tests are. Actually looking up the
> >> Voevodin letter and reading it reveals quite a different picture of the
> >> ELISA test's reliability. Only 1 test result out of 1000 was a false
> >> positive.
> >>
> >> I wrote an article about it in 1994:
> >> http://groups.google.com/groups?selm=14880%40sci.med.aids&oe=UTF-8&output=gplain
> >
> >Voevodin's letter really just states the obvious. Even tests with high
> >specificity will have a low positive predictive value in low
> >prevalence populations. Why do Denialists think that they were the
> >first to think of this?
> >
> >If you test every male in the US with a pregnancy test then a number
> >will test positive (testicular cancer can cause false positives). I
> >would be willing to bet that exactly zero of these will be confirmed
> >as being pregnant.
Let's up the ante on that one. Let's devise a program that _only_ tests
men with testicular cancer.
Now, all of a sudden, you'll see a _huge_ pregnancy rate among testicular
cancer sufferers.
Let's only test pregnant women for HIV. Are they "low prevalence populations"
or not? They have a really high prevalence of pregnancy, which is one of the
reasons for returning a false positive on the ELISA test.
They will also have a high prevalence of exposure to: malaria, tuberculosis,
and heaven knows what other bacterial pathogen that will return a false positive.
The point is, that in the West, you're not allowed to conclude that someone
is infected with HIV, from just one positive return from an ELISA test.
You're supposed to do a follow up test, Wester Blot hopefully...
But no such luck for pregnant women showing up at antenatal clinics
in Zimbabwe, South Africa, etc.
Are they even reflective of the entire population?
How representative are men with testicular cancer for the people of the USA?
Should we deduce pregnancy rates in America by testing Americans with
testicular cancer?
> >Therefore according to Denialist logic pregnancy tests are completely
> >unreliable,
They are for men with testicular cancer...
> >there is no such thing as pregnancy and even if there was
> >it has nothing to do with sex.
In men with testicular cancer it hasn't...
Just as a single EIA test has very little to do with HIV infection in pregnant
women in Zimbabwe.
Darn, this is easy.
Alex
Dave
Doesn't hold water, I'm afraid.
Your assumption is that pregnancy causes false positive ELISAs for
HIV. I know you will try and quote some out of date references to
support this, but the reality is that it hardly ever happens.
Isn't it strange that although pregnancy is universal, high rates of
HIV seroprevalence are only found in antenatal screening programmes
where HIV is prevalent in the general population?
In my hospital, since the UK antenatal programme was brought in 2
years ago we have screened over 12000 pregnancies. We have had only
two ELISA positives that was shown to be false, and one was where a
lab error in sample labelling had occurred.
False positive rate = 0.008%
> They will also have a high prevalence of exposure to: malaria, tuberculosis,
> and heaven knows what other bacterial pathogen that will return a false positive.
>
And again your memory, and epidemiology in general, fail you - In
South Africa, apart from the north bit of Kwa-Zulu there is no
malaria. Countries with malnutrition, high rates of infection with TB
and malaria can have low rates of HIV (Afghanistan eg).
These inconsistencies have been pointed out to you in the past but you
ignore them.
Just like you will ignore the ridiculous claims you make over HIV
screening and predictive values (if you even have a grasp of what they
mean, which I doubt).
> The point is, that in the West, you're not allowed to conclude that someone
> is infected with HIV, from just one positive return from an ELISA test.
Oh yes you are for anonymised guthrie card testing of newborns which
is done throughout the whole of the UK, giving a complete picture of
maternal seroprevalence.
> You're supposed to do a follow up test, Wester Blot hopefully...
> But no such luck for pregnant women showing up at antenatal clinics
> in Zimbabwe, South Africa, etc.
I can just see it now......Imagine the medical conversation in the
antenatal clinic of Baragwanath Hospital..."Hmm, lets see, this month
we have found that 1385 women tested positive. 2 or 3 of these may be
false positives. We could do Western Blotting on them all (at a cost
of around $55 000) just to find out which 2 are really negative,....or
we could try and set up a service with all the money to give them all
some cheap (or free!)HIV prophylaxis with NVP. Gosh, this is a
difficult decision... Oh, to hell with it, lets go for the Western
Blots!!"
Even for someone with your limited understanding of the realities in
Southern Africa, it must be obvious what the correct choices are.
>
> Are they even reflective of the entire population?
In that pregnant women represent an accurate, near-perfect random
sample of sexually active female population, yes.
> How representative are men with testicular cancer for the people of the USA?
Not at all - thats another reason why your stupid comparison fails.
As you would say: darn, this is easy.
GMCarter
> We have had only
>two ELISA positives that was shown to be false, and one was where a
>lab error in sample labelling had occurred.
>False positive rate = 0.008%
Therefore the test is a worthless sham and lie! HIV does NOT exist! It
does exist and does NOT cause AIDS! It's all so CLEAR now!!!!!
I am CONVERTED!!!!
George "AC/DC" Carter
David Canzi
Alex <avdeele...@wanadoo.nl> wrote:
>On 16 Aug 2002 18:00:56 -0700, chris...@hotmail.com (Chris Noble)
>> wrote:
>> >Voevodin's letter really just states the obvious. Even tests with high
>> >specificity will have a low positive predictive value in low
>> >prevalence populations. Why do Denialists think that they were the
>> >first to think of this?
>> >
>> >If you test every male in the US with a pregnancy test then a number
>> >will test positive (testicular cancer can cause false positives). I
>> >would be willing to bet that exactly zero of these will be confirmed
>> >as being pregnant.
>
>Let's up the ante on that one. Let's devise a program that _only_ tests
>men with testicular cancer.
>Now, all of a sudden, you'll see a _huge_ pregnancy rate among testicular
>cancer sufferers.
>
>Let's only test pregnant women for HIV. Are they "low prevalence populations"
>or not? They have a really high prevalence of pregnancy, which is one of the
>reasons for returning a false positive on the ELISA test.
I seemed to recall that Voevodin's letter said something about
pregnant women, so I looked it up again.
"... in 1991 alone some 8000 false-positive results were reported in
pregnant women, with only 6 confirmations. A quarter of HIV tests
in Russia are done in this group..."
29.4 million tests were done in 1991. One quarter of that is about
7.4 million. 8000 false positives out of 7.4 million tests is 1.08
out of 1000. In the general population, it's 1 out of a thousand.
In pregnant women, 1.08 out of a thousand. The difference barely
exceeds rounding error.
HIV testing technology hasn't been standing still in the last 11
years. According to a web page at UNAIDS, current tests are more
specific than older ones but, irrirtatingly, it didn't give numbers.
Which brings us to...
>They will also have a high prevalence of exposure to: malaria, tuberculosis,
>and heaven knows what other bacterial pathogen that will return a false
>positive.
This is a standard denialist song and dance... some large number of
conditions are alleged to cause false positives. Well, we have seen
now what "pregnancy causes false positives" means -- damned little.
How much is this "high prevalence" of exposure to malaria, tuberculosis
and 57 varieties of unnamed microbes? You don't say, because you don't
know. How much does each of these causes affect the chance of a false
positive? You don't say, because you don't know. It might very well
not be enough to matter.
You have no numbers. You have no evidence. You have no journal
citations to actual observations. All you're doing is repeating the
rumours you've read at virusmyth and similar sites.
Did you hear the story about the kid who went to Holy Communion and,
instead of swallowing the host, took it home and stuck it with a pin,
and it bled all over the place? The story is bullshit, but when I was
a kid this story spread like wildfire among Catholic kids.
Why did it spread? Probably for the same reason that any argument that
seems to confirm denialist beliefs, no matter how bad it is, no matter
how dishonest it is, will spread rapidly among denialists.
Like the widespread misuse of Voevodin's letter that I commented on at
the start of this thread.
--
David Canzi If love of money is the root of all evil, marketing