Badgett's TCO2 Excess

[TGJB]

Bob-- Actually, since you are here, maybe you (or NC Tony) can explain something to the people on this board. The TCO2 threshhold level is set at 37 because that is 2 (?) standard deviations above what would normally be a real positive. You can explain to us what that means, and the chance of getting a false positive.

And that\'s before we get to the 39 they use in NY.

Sight-- I went through the math on this a few days ago, explaining how unlikely it is that one trainer could get all or almost all of his horses to jump up that much, that abruptly. If Dutrow is doing this by worming his horses, he has a hell of an eye for horses that have worms.

One of the many problems with that type of logic is this-- when we look at what Frankel did in 2001, or Scott Lake does every day, we are dealing with older horses with previously established form and limits. To believe that you can get an individual 5yo to all of a sudden run much faster than he ever has before (as opposed to returning to his best, which is what good horsemen often do), you would have to believe that he has had that exact set of problems for every start of his whole career, and now suddenly-- and I mean instantly-- this guy has solved them, and that it\'s true of 15 horses in a row. Not likely. Not even remotely.

[sighthound]

>> Bob-- Actually, since you are here, maybe you (or NC Tony) can explain >>something to the people on this board. The TCO2 threshhold level is set at 37 >>because that is 2 (?) standard deviations above what would normally be a real >>positive. You can explain to us what that means, and the chance of getting a >>false positive.

The TCO2 is set at 37 because that level cannot be attained naturally in nearly all horses.

A level set at two standard deviations from the mean should include 95% of horses\' natural levels.  

The mean is NOT \"what would normally be a real positive\".  It is NORMAL horses.

The mean is an arthmatic calculation (add them all together and divide by the number of horses you are looking at)  but it is not necessarily a good representation of \"average\" (as horse sale results prove)

The standard deviation simply tells you how tightly or loosely your data points are grouped.

Unless statistics have changed since I\'ve been a scientist.

[TGJB]

Sight-- you could pick a lot of numbers that horses could not naturally attain, the question is why this one. I remember it had something to do with standard deviations, I think two, but I could be wrong about that.

[miff]

\"Dr. Scott Stanley, associate professor at the University of California-Davis Animal Health and Food Safety Laboratory, discussed TCO2 testing July 14 as part of the National Horsemen\'s Benevolent and Protective Association Medication Committee meeting in Minneapolis. He said officials in California already have noticed trends.

From more than 60,000 samples, TCO2 readings have been as low as 18.3 millimoles and as high as 44.8 millimoles, according to Stanley\'s report. The mean reading is 31.5, plus or minus 2.06\"

[TGJB]

Miff-- that\'s with lasix, which makes readings higher, and is for all horses, independent of whether they have been given alkalizers, I assume.

[sighthound]

I think that the 37 mmol/L is more logically three standard deviations from the mean.  

That level of 37 would then account for 99% of horses normally falling under that established level.

That leaves the rare horse that could still fall above 37 mmol/L it normally, with no cheating.  

However most jurisdictions have provision for later proving that is the horses normal level.

[sighthound]

You have to also know where and how those horses were sampled:  in their stalls, pre-race holding barn, very hot day, etc. as so much can affect the results.  

I assume it\'s pre-race holding barn, as that is the only time the track could reasonably legally get a needle into a horse.  

Wonder if it includes post-race samples, however?

[TGJB]

From what I understand, some tracks have been testing post-race, which will get lower readings, and keep some drugged horses under the limit. The Jockey Club covered this in their recommendations.

As far as I know, all horses tested in cal are pre-race.

[bobphilo]

TGJB Wrote:
-------------------------------------------------------
> Bob-- Actually, since you are here, maybe you (or
> NC Tony) can explain something to the people on
> this board. The TCO2 threshhold level is set at 37
> because that is 2 (?) standard deviations above
> what would normally be a real positive. You can
> explain to us what that means, and the chance of
> getting a false positive.
>
> And that\'s before we get to the 39 they use in
> NY.
>

OK Jerry. Wish I knew how to post a picture of a Bell Curve for a visual presentation but hopefully my words will give the picture.

The mean is just the average value for some quality - in this case TCO2levels. In a normal arrangement or distribution all the other values are clustered either above or below this value symmetrically. The Standard Deviation is a measure of variability in a population and is roughly the average amount that each value (or measurement) varies from the mean in both directions - higher or lower. The 1st SD encompasses 68% of the population and therefore 32% will be higher or lower than the 1st SD. Since the distribution is normal, these 68% will be distributed with 16% above the upper limit of the SD and 16% below the lower limit. These areas are called tails because that’s what they look like at each end of the Bell Curve
Now if we widen this to 2 SDs, this will encompass about 95% of the population with 5% left. 2.5% will be above the 37 upper limit (right tail) and 2.5 below the lower limit (left tail). Now I don’t have the mean or value of the SD but I do have enough information to calculate the possibility of a horse not given alkalizing agents ending up in the right tail and thus being a false positive, is 2.5% or p=. 025 Of course if the upper limit is raised to 39 that would put less horses in the right tale and further reduce the possibility of false positives. If the upper limit were set at 3SDs the probability of false positives would be 0.3% or p=.003  If you give me more information, such as what the mean and SD were I could also calculate the probability (p) of false positives at the 39 upper limit.

Sighthound raised the important issue that this test is only appropriate if the population is normally distributed. For example, people’s incomes and horse sale prices are not normally distributed but skewed towards the big prices. However biological and physiological values like height, weight, and glucose levels tend to be normally distributed so I would expect TCO2 levels to be normally distributed as well.
There are also test for normalcy that can be applied to answer this question.

Bob

[bobphilo]

sighthound Wrote:
-------------------------------------------------------
> I think that the 37 mmol/L is more logically three
> standard deviations from the mean.  
>
> That level of 37 would then account for 99% of
> horses normally falling under that established
> level.
 
I got the same impressions when doing the calculations. Jerry if you can get me the mean and standard deviation I can do a more complete test at 3 SDs. The numbers of false positives would be dramatically lower.

Bob