A GUM-compliant uncertainty budget for the analysis of total carbon dioxide (TCO2) in equine plasma

Samples of plasma from chosen horses are collected prior to racing for the analysis of total carbon dioxide (TCO2). Five or six (dependent on jurisdiction) tubes of blood are taken as required for a screen, followed (if positive) by confirmation (two tubes in quadruplicate) and, further confirmation (two tubes in quadruplicate) at a second referee laboratory. The threshold limit prescribed by the rules of racing is an amount concentration of 36.0 mM, with further action ensuing if the reported concentration is greater than 37.2 mM. Using QC data and an analysis of the measurement procedure, the combined standard measurement uncertainty of a measurement result was estimated to be 0.22 mM. This leads to a conclusion that the odds of incorrectly deciding that a sample at the threshold limit exceeded the action concentration, are 13 000 000:1 against. The combined standard measurement uncertainty could be reduced to 0.20 mM by combining results from the first and referee laboratories. An allowance for uncertainty of measurement of 1.0 mM is proposed, which leads to odds of 185 000:1 against concluding a sample is over the threshold limit when, in fact, it is not.