Doping tests in sport are commonplace today. Some recent 'nandrolone' cases in football involved high-profile international players who were tested 'positive' after UEFA Cup and Serie A matches. The outcomes of investigations into doping depend on the presence and (if a threshold is given) on the maximum concentration of a substance. Such thresholds are given if the substance may be present in the body for reasons other than the abuse of drugs.

Technical Considerations

Suppose that a test indicates a concentration of 5.2 ng per ml for the nandrolone metabolite 19-NA. The limit laid down is 2 ng per ml. It seems that the player has tested 'positive.' But how confident are we in the measurement made, bearing in mind that no test is perfect? The question is pertinent because, depending on the outcome, a player's career could be affected or even ruined. The other side of the coin is that a result could unfairly stand.

Therefore, it is necessary to make a statement about the quality of the measurement. One consideration is the difficulty of measuring small concentrations. The above limit is very small as regards accurately determining such concentrations, but an actual concentration that was appreciably greater would be likely to have a performance-enhancing effect. Is therefore 5.2 ng per ml "appreciably greater" than 2 ng per ml?

Internationally agreed minimum requirements for the competence of laboratories (ISO/IEC 17025, ILAC G8) recognise the difficulties of measurement, stating that a player is positive when the test result is 'significantly above the limit.' The laboratory concerned carries the burden of proof: only when a player is found positive should it report the case for sanctioning.

On the basis of repeating and analysing the measurements made and an understanding of the equipment used, the laboratory makes a statement about a spread of possible test results. The value of 5.2 ng per ml would indicate a 'likely value' within such a spread. Because of the mentioned difficulty of performing tests accurately at low concentrations, the spread could be broad. In fact, in this particular case it was concluded that the spread ranged from 0.4 ng per ml to 10.0 ng per ml. An accompanying statement was that the laboratory was 99% confident that the actual concentration lay in this range. Because the lower endpoint in this spread lay below the limit, the internationally agreed interpretation is that the result is not positive. Evidently, for a somewhat larger test concentration of, say, 7.2 ng per ml, and the same spread about this value, the result would be positive.

Conclusions

Statements such as the above concerning a range of values within which a measurement can be expected to lie form a major part of modern metrology (measurement science). The statement is known as an expression of uncertainty associated with a measurement, for which the accepted 'bible' is the Guide to the Expression of Uncertainty in Measurement published by the International Organisation for Standardisation. Clearly, the smaller the uncertainty, the greater the likelihood of obtaining a clear-cut statement concerning whether an actual concentration lies above or below a limit. The number of borderline cases in which there was genuine doubt would consequently be decreased. Metrologists continually strive to reduce measurement uncertainty and to determine it more reliably for the general benefit of the community.

Dr Adriaan van der Veen of the Dutch Metrology Institute (NMi), who has acted as expert witness in several cases involving the alleged abuse of 'nandrolone' in football, contributed to this article.