Horse racing authorities impose a limit on the concentration of plasma 'total carbon dioxide' (TCO2), typically 36 mM with action taken above 37 mM, as measured by an electrochemical gas analyzer. It is of interest to understand the distribution of TCO2 in a 'normal' population of racehorses and determine probabilities of members of this population exceeding these current regulatory and action limits. TCO2 levels in equine plasma samples have been modelled for 12 months (2011-2012) of thoroughbred (3076 measurements) and standardbred (3788 measurements) data in Australia. The two populations have a common seasonal pattern, while the non-seasonal distributions differ. A single Gaussian distribution about the seasonal pattern explains the thoroughbred data, but there is evidence for a second Gaussian component for the standardbred horses. A Gaussian mixture model for standardbred horses gave a main component that matched the thoroughbred distribution, which was centred about 30.2 mM, and a smaller (about 20 % of the total density) Gaussian centred at 32.3 mM. The existence of a second, higher-meaned population of standardbred horses points to increased use of alkalinizing salts among a minority of trainers, whom still, however, maintain mostly legal levels of TCO2. Identification of this group can be used to direct intelligence-based testing with a view to limiting use of these products. Probabilities of exceeding limits are affected by seasonality, but the current rules remain conservative.
Keywords: equine plasma; expectation maximization (EM) algorithm; horse racing; sports drugs; total carbon dioxide (TCO2).
Copyright © 2014 John Wiley & Sons, Ltd.