Where’d You Get That Number? The Process Behind Therapeutic Drug Guidelines In Racing: Paulick Report 4/27/16

By: Natalie Voss

As the Racing Medication and Testing Consortium continues its fight for uniform medication guidelines across states, it is perhaps no surprise that the group has encountered opposition. Some horsemen and veterinarians have protested the practicality of the guidelines, complaining that they do not allow for adequate treatment of actual health problems with therapeutic medications. Others question the amount of time spent on regulating therapeutic drugs while synthetic opioids and other performance enhancers continue to be available to interested cheaters online. And some question how the RMTC arrives at its recommendations for testing thresholds and withdrawal guidelines.

Concerns about the process behind model rule-making is often born out of worries that a withdrawal guideline is inadequate to help well-meaning trainers avoid rule violations, or that a threshold is too strict, not allowing enough wiggle room for cross-contamination or for an individual horse’s slow metabolism.

The RMTC has made an effort in recent months to clarify the statistical analysis that translates raw data into the numbers in the recommendations on its website. At its outset in the early 2000s, the organization was tasked with establishing thresholds for the most commonly-used therapeutic medications. At the time, much of the existing research on therapeutics had been conducted in the 1980s on small groups of retired Standardbred mares. Dr. Rick Sams, laboratory director at HFL Sport Science Laboratory and member of the RMTC’s Scientific Advisory Committee, said he and other members were concerned that inactive, older horses of a different breed may not be a good guide for testing how drugs moved through a working, young Thoroughbred’s body. Some of the old research was also conducted on as few as one or two horses, and results were arbitrarily rounded up.

The logical move seemed to be to commission new research, but Sams and his colleagues had a problem. Research on exercised Thoroughbreds in a university setting is very expensive, and sample sizes are usually small. The thresholds put into place for anabolic steroids were drawn from the sampling of thousands of horses internationally, and as a result, researchers felt very confident about drawing conclusions from the data. That kind of broad-scale information wouldn’t be available for other substances, however. Instead, the researchers collaborated with the Food and Drug Administration, which has been using Tolerance Interval Tests for its studies on foreign substances in food animals for years.

“I thought if it was good enough for food suppliers and the FDA, that was better than something we just ‘thought’ was the right approach,” said Sams.

It’s a tall order to use observations from a handful of research horses to make predictions about the population of Thoroughbreds that need to comply with drug regulations on the racetrack. Fortunately, there are a number of tests that can help researchers take a small number of results and responsibly expand them to get a guideline that can service the whole population of horses on the racetrack.

First, researchers choose a time after administration of the drug they’re testing and check the concentrations of the drug in their test subjects’ blood. For example, nine horses had blood taken at 48 hours after being injected with Drug XYZ, and were found to have between 3 mg/ml and 9 mg/ml in their blood. The 9 mg/ml is the number researchers are concerned with — a recommended threshold at 48 hours for Drug XYZ will need to be higher than 9 mg/ml.

Nine horses isn’t a very large group, however, so that 9 mg/ml is multiplied by a pre-determined factor to compensate for the small sample size. This widens the range of acceptable drug concentrations, since it’s likely that a larger group of horses would provide a greater range of results. This pre-determined multiplier shrinks when the number of study horses is larger. By this time, the 9 mg/ml has probably increased, let’s say to 12 mg/ml.

At this point, researchers have a range of concentrations that apply to a large number of horses, but they need to add more wiggle room into the threshold because it will be used for future racehorses in addition to the current population and there could be even more variability in horses’ metabolisms. An additional statistical test allows them to transform that 12 mg/ml to a number that will be higher than a given percent of the population. It also allows them to set a number representing their confidence in the application of that number. The standard the FDA researchers had been using was 95 percent, meaning the tests would provide a maximum that they were 95 percent confident could apply to 95 percent of the current and future population of racehorses. Running these tests will make that 12 mg/ml increase again. Boosting either of those percentages to say, 98 percent or 99.5 percent causes the maximum allowed concentration to increase even further.

At the end of the interval testing, the researchers have a suggested maximum threshold for the drug they’re testing. The RMTC Scientific Advisory Committee will often use that suggestion and round up, to allow for even more wiggle room in the unlikely event that a horse has a really slow metabolism or some other issue affecting the results.

That’s where regulators must tread a line — if you raise the limit too high in an effort to avoid unintentional positives, you face the risk of being too lenient and allowing an amount in the horse’s system which could manipulate the outcome of the race. If you try to limit the use of the drug too much, you may set a standard that could be easily violated by the rare horse with slow metabolism or other factors. That’s where the judgment of the committee comes in.

“Every one of these thresholds has been discussed by the Scientific Advisory Committee, a value obtained by the 95/95 tolerance interval has been reported to the committee and is a starting point for discussion,” said Sams. “In every case, the value has been raised

[from the calculated results].”

If the process seems convoluted, Sams points out that it’s fairly different from the way drug positives are detected (and hence arbitrated) in other countries. Many international labs have screening limits for the drugs they’re testing for, including therapeutic medications. A fair number of those screening limits are based on mathematical calculations from drug administration instructions, not studies with real horses. Also, screening limits result in a positive or a negative drug test, so laboratories do not have to disclose to a horse’s trainer what drug concentration it found. That means for horsemen in those countries, there is very little latitude to appeal drug overages, and it can generate the impression that the “line” horsemen must avoid crossing is a moving target.

The Tolerance Interval Test process may take a little extra concentration to follow, but statistical tests like it have been used to make predictions about large populations based on small ones in a variety of settings for years. Until large-scale research on exercised Thoroughbreds in an academic setting becomes more practical and affordable, it would seem that statistical tests are the most objective way to bridge the gap.