Is A a false positive or is B a false negative?


This Forum was created to divert traffic from Current Events at the height of the BALCO scandal. It comes and goes as "needed"; it's back to being locked.

Is A a false positive or is B a false negative?

Postby drbop » Thu Sep 07, 2006 2:56 am

Don Catlin estimates that the EPO test produces at least 10 false negatives
for every true positive. One of the tests is wrong. Statistically, which one
do you think it is? Testing has been set up on the philosophy of tolerating
many, many false negatives but no false positives.
drbop
 
Posts: 46
Joined: Sat Oct 08, 2005 4:32 am

Re: Is A a false positive or is B a false negative?

Postby RJMB_1 » Thu Sep 07, 2006 3:02 am

drbop wrote:Don Catlin estimates that the EPO test produces at least 10 false negatives
for every true positive. One of the tests is wrong. Statistically, which one
do you think it is? Testing has been set up on the philosophy of tolerating
many, many false negatives but no false positives.


I guess we'll never know as A is unlikely to be retested, as it could be said to be contaminated - although it would be interesting if it could be done using 'stricter' B test methods.
RJMB_1
 
Posts: 261
Joined: Sat Oct 08, 2005 4:32 am
Location: POKE 9450,173, Unlimited Kilobytes

Postby El Toro » Thu Sep 07, 2006 5:45 am

Arrrghhh, there are no 'stricter B test methods". It's the same fallible test interpreted by (usually) different but still fallible people and subject to variation of interpretation. That's why there is provision for a closely monitored second test to remove(as much as possible) accusations of incompetence and/or bias.

Although, it should be noted that it is possible for two perfectly conducted tests to still be wrong due to poorly maintained samples or outright documentary errors.
El Toro
 
Posts: 903
Joined: Sat Oct 08, 2005 4:32 am

Postby 26mi235 » Thu Sep 07, 2006 5:59 am

El Toro wrote:Arrrghhh, there are no 'stricter B test methods". It's the same fallible test interpreted by (usually) different but still fallible people and subject to variation of interpretation. That's why there is provision for a closely monitored second test to remove(as much as possible) accusations of incompetence and/or bias.

Although, it should be noted that it is possible for two perfectly conducted tests to still be wrong due to poorly maintained samples or outright documentary errors.


See gh's comments about the testing procedure. He comments that the first test is done in a more assembly-line manner and not necessarily by the labs best technicians and does not have "unlookers". The implication is that the "B" test process probably leads to a more reliable result.

That said, one potential approach (I do not know if it is used) is to use a "broader net" (lower threshold/standard) for the "A" sample and thus get a second measurement to help clarify on the with the "B" sample.

I am not a Jones apologist, but I have definitely move from my position that it is very likely that she had been using (in 2006, at least) back into the "I do not know' camp; tafnut, is there room?
26mi235
 
Posts: 16333
Joined: Sat Oct 08, 2005 4:32 am
Location: Madison, WI

Postby El Toro » Thu Sep 07, 2006 6:09 am

NO, the method is not stricter. It's always the same method. There may be different levels of focus or competence between those conducting the tests but nothing changes in the methodology. Anyway, there is nothing to guarantee that the person conducting the B test is necessarily better than the person conducting the A test. Anybody ever worked with an incompetent boss who thought, "If I do it, it must be right" ? Yeah, we've all seen it.....
El Toro
 
Posts: 903
Joined: Sat Oct 08, 2005 4:32 am

Postby RJMB_1 » Thu Sep 07, 2006 6:58 am

El Toro wrote:NO, the method is not stricter. It's always the same method. There may be different levels of focus or competence between those conducting the tests but nothing changes in the methodology. Anyway, there is nothing to guarantee that the person conducting the B test is necessarily better than the person conducting the A test. Anybody ever worked with an incompetent boss who thought, "If I do it, it must be right" ? Yeah, we've all seen it.....


Of course the test is the same and the way it is applied is the same and testing criteria will be the same. But the conditions under which a B test is done will be different from the A test, as you indicate a greater level of focus or scrutiny is part of the overall procedure (see others comments). Hence stricter was written as 'stricter'.
RJMB_1
 
Posts: 261
Joined: Sat Oct 08, 2005 4:32 am
Location: POKE 9450,173, Unlimited Kilobytes

Postby DrJay » Thu Sep 07, 2006 1:09 pm

Since I've not heard otherwise, I assume the A and B tests are done by the same testing method. Does anyone here know that with certainty? There are situations in medicine where a two-step process like this uses completely different tests for each step. When you donate blood, they test your blood for, among other things, hepatitis C, a viral liver infection. The method used for that test (ELISA or Western Blot, I believe) is one that is very sensitive, meaning out of 100 people with hepatitis C, it will be positive close to, if not actually, 100% of the time. That's to try and prevent any infected blood from getting through and being used for transfusion. However, that test is not 100% specific, meaning it will have a moderate percent of false positives. If a test has a specificity of, say, 97%, then out of 100 tests that are abnormal, 97 will be true postives while 3 will be false positives. In the blood donating scenario, if you test positive for hepatitis C, they will not accept your blood. They will tell you that you tested positive for hepatitis C, then you will see your physician. He/she should then order the next test, done by a different method (RIBA) that has a lower sensitivity (could miss some cases of hepatitis C if used as the initial screen for the blood supply) but is highly specific, meaning very few false positives. I've had a number of patients who tested positive on the first test and negative on the RIBA test. They did not have hepatitis C.

So I wonder if any of the PEDS testing involves a different, more specific but less sensitive method for the B sample compared to the A sample.
DrJay
 
Posts: 5485
Joined: Sat Oct 08, 2005 4:32 am
Location: Woodland Park, CO

Postby Snation » Thu Sep 07, 2006 2:34 pm

To answer your question, you need to know the data from the lab. You need to have sensitivity, specificity, positive predictive value, and negative predicitve value. (and a bunch more)

The lab adjusts the cutoff values to maximize the above.
a= true +
b= false +
c= true -
d= false -
Sensitivity = a / (a + d)
Specificity= d / (b + d)
Positive Predictive value = a / (a + b)
Negative Predictive value = d / (c + d)
false-positive rate = (b / (b + d)

And there is a ton more. As I wrote this I realized how complicated it is.
http://www.medal.org/visitor/www/inactive/ch39.aspx

So sensitivity = 1/11 = not very sensitive
------------
false-negative rate = (c / (a + c))
where:
• c = false negatives
• (a + c ) = sum of (true positives, false negatives) = all people with disease

If Don Catlin said that a=1 and c=10 then the false negative rate is 10/11 or over 90%.

The test would be horribly unsensitive; but likely very specific.

The error would be with the 'B' tests, I am guessing.
Last edited by Snation on Thu Sep 07, 2006 2:43 pm, edited 1 time in total.
Snation
 
Posts: 207
Joined: Mon Aug 28, 2006 11:03 am
Location: sunny iowa

Postby eldrick » Thu Sep 07, 2006 2:35 pm

El Toro wrote:NO, the method is not stricter. It's always the same method. There may be different levels of focus or competence between those conducting the tests but nothing changes in the methodology. Anyway, there is nothing to guarantee that the person conducting the B test is necessarily better than the person conducting the A test. Anybody ever worked with an incompetent boss who thought, "If I do it, it must be right" ? Yeah, we've all seen it.....


i think we can assume a lowly lab technician batch testing a dozen-or-so tests is not going to give as reliable a result as the boss ( likely with his "spock" next to him ), who's had a coupla weeks anticipation/mental preparation of the date he'll have to do the b test, knowing a professor in this field, from harvard will be one of the athlete's team observing team
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby eldrick » Thu Sep 07, 2006 2:45 pm

Snation wrote:The test would be horribly unsensitive; but likely very specific.

The error would be with the 'B' tests, I am guessing.


catlin's claim has always been that his "b" is 0

it's not a sensitivity/specificity argument, just the outrageous claim of a zero "b"
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby Snation » Thu Sep 07, 2006 2:54 pm

How hard is it to detect EPO? It's a beeeetch! I have highlighted several different kinds of EPO.

Recombinant Erythropoietin and Analogues: A Challenge for Doping Control.

Basel Proceedings
Therapeutic Drug Monitoring. 26(2):175-179, April 2004.
Pascual, J. A. PhD *+; Belalcazar, V. BSc *+; de Bolos, C. PhD ++; Gutierrez, R. PhD *+; Llop, E. BSc *+; Segura, J. PhD *+

Abstract:
Erythropoietin (EPO) increases the number of circulating erythrocytes and thus muscle oxygenation. The availability of the recombinant protein (rEPO) has increased the risk of its illegal use in sports, its detection being a difficult challenge. Five different hematopoietic parameters were initially chosen as indirect markers of rEPO abuse: concentration of serum EPO, concentration of serum-soluble transferrin receptors (sTFr), hematocrit, percentage of reticulocytes, and percentage of macrocytes. New models considering only hemoglobin, serum EPO concentration, and percentage of reticulocytes are simpler and seem to be more sensitive when low doses of rEPO are used. A more direct method of urine analysis (isoelectrofocusing, double blotting, and chemiluminescent detection) based on the charge differences between rEPO and endogenous EPO, related to their carbohydrate composition, provides proof of rEPO use. Furthermore, this approach permits the detection of darbepoetin, a direct analogue of EPO also known as NESP ("new erythropoiesis stimulating protein"). Recently a protein conjugate, "synthetic erythropoiesis protein" (SEP), containing precision-length, monodisperse, negatively charged polymers instead of oligosaccharides has been synthesized. Finally, EPO-mimetics are molecules capable of acting as EPO in dimerizing the EPO receptor. Two kinds of EPO-mimetics have been described: peptides and nonpeptides. The enhancement of oxygen availability to muscles by rEPO, analogues, and mimetics constitutes one of the main challenges to doping control. Major steps have already been developed for detection of rEPO and some analogues. In the near future, the transfection to an athlete's body of genes that code for erythropoietin might be an emerging doping issue, and sports authorities have incorporated "gene doping" among the prohibited practices.
Snation
 
Posts: 207
Joined: Mon Aug 28, 2006 11:03 am
Location: sunny iowa

Postby Snation » Thu Sep 07, 2006 2:56 pm

I don't understand the '0' of the 'B' test. Do you mean 0 errors in the 'B' test? Is that for EPO?
Snation
 
Posts: 207
Joined: Mon Aug 28, 2006 11:03 am
Location: sunny iowa

Postby eldrick » Thu Sep 07, 2006 3:06 pm

zero false-positives for epo
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby eldrick » Thu Sep 07, 2006 3:14 pm

from horse's mouth :

http://www.wada-ama.org/en/dynamic.ch2? ... ory.id=527

relevant quote is :

Questions were raised by certain individuals who are not well versed in the science of EPO detection in relation to a phenomenon that is rare but well understood by anti-doping experts.


In certain rare circumstances, normal endogenous EPO shifts into the recombinant EPO area. This phenomenon is clearly identified by accredited laboratories so that these rare profiles are labelled properly and are not reported as adverse results due to EPO doping.


WADA was informed of this phenomenon by accredited laboratories in the spring of 2005.


implication being that not all labs were aware of this errator until early-'05

therefore the test before this date was potentially erroneous & therefore grounds to challenge all epo convictions prior to this date
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby Snation » Thu Sep 07, 2006 3:51 pm

I see your point. WADA (not Catlin, for he did not originate the rhEPO test) says no false+.

Now, check my thinking on this:
1. If there are zero false+ then the specificity is 1 or 100%
2. If it is true that there is 1 true+ to 11 false-, the sensitivity is low (picks up about 10% of true+)

Thus, the problem is with the B test. It is a false-, which appears to happen alot.

Catlin says he has tested 2600 doping samples. A and B samples have come back 'positive' in 9.

.According to this, if the athlete is taking rhEPO he/she has a 1/10 chance of getting caught. Is that right????
Snation
 
Posts: 207
Joined: Mon Aug 28, 2006 11:03 am
Location: sunny iowa

Postby eldrick » Thu Sep 07, 2006 4:32 pm

Snation wrote:I see your point. WADA (not Catlin, for he did not originate the rhEPO test) says no false+.


no

catlin has made that claim for his lab - for ALL tests, not just epo

Now, check my thinking on this:
1. If there are zero false+ then the specificity is 1 or 100%
2. If it is true that there is 1 true+ to 11 false-, the sensitivity is low (picks up about 10% of true+)

Thus, the problem is with the B test. It is a false-, which appears to happen alot.

Catlin says he has tested 2600 doping samples. A and B samples have come back 'positive' in 9.

According to this, if the athlete is taking rhEPO he/she has a 1/10 chance of getting caught. Is that right????


i'm gearing up for blake v fed, so i can't guarantee sens v spec, but, an epo test used to cost ~ $500 ( probably still close to that seeing as it takes 3/7 & half-dozen+ steps ), so we are probably never going to see the contra of every -ve epo test re-tested for the b result to ascertain the ?sens/spec figure
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby JRM » Thu Sep 07, 2006 5:46 pm

Eldrick, I'm curious: why this offensive witch hunt against Catlin? What did he do to deserve such scorn?
JRM
 
Posts: 2625
Joined: Sat Oct 08, 2005 4:31 am
Location: Woodland Hills, CA

Postby jammin » Thu Sep 07, 2006 5:57 pm

Damm she is UGLY
Image
jammin
 
Posts: 1000
Joined: Sat Oct 08, 2005 4:32 am

Postby eldrick » Thu Sep 07, 2006 6:15 pm

jammin wrote:Damm she is UGLY
Image


will the idiot ever be sent on a long vacation ?
eldrick
 
Posts: 14147
Joined: Sat Oct 08, 2005 4:31 am
Location: 19th hole st andrews

Postby EPelle » Thu Sep 07, 2006 9:33 pm

eldrick wrote:zero false-positives for epo

Yes, eldrick, noll false-positives for EPO.

As a matter of fact, the only time any false-positive for EPO has occured in athletics - ever - is with Bernard Lagat, and that was conducted and analysed here in Europe.

Telegraph.uk.co wrote:It is highly unusual for a B test to contradict the initial findings since the analysis is conducted on two halves of the same urine sample.

The only other time it has happened with EPO-testing in athletics involved the former Kenyan runner, Bernard Lagat, who now represents the United States.


We:ve gone through his case test and its procedures. I doubt this mistake would ever be replicated in those steps again.
EPelle
 
Posts: 21442
Joined: Sat Oct 08, 2005 4:32 am

Postby RJMB_1 » Fri Sep 08, 2006 12:21 am

eldrick wrote:
jammin wrote:Damm she is UGLY
Image


will the idiot ever be sent on a long vacation ?


If you were to remove the ponytail she does look like a bloke. But when she's not running facially she looks quite feminine. The face does move around a bit when sprinting - giving the impression of a picaso painting!
RJMB_1
 
Posts: 261
Joined: Sat Oct 08, 2005 4:32 am
Location: POKE 9450,173, Unlimited Kilobytes


Who is online

Users browsing this forum: No registered users and 1 guest