[Skip to Content]
[Skip to Content Landing]
November 13, 1996

Drug Testing in Sport: Bayes Theorem Meets Advanced Technology

Author Affiliations

Sayville, NY

JAMA. 1996;276(18):1471. doi:10.1001/jama.1996.03540180027021

To the Editor.  —In their article on testing for performanceenhancing drugs, Drs Catlin and Murray1 tell us that most athletes who test positive for drugs deny drug use. Instead, the athletes attribute their positive result to collection mistakes, breaks in the chain of custody, laboratory error, or sabotage. None of them, it seems, has used Bayes theorem to attack their result.Except for tests that have specificity of 100%, false-positive results will increasingly outnumber true-positive results as the prevalence of the condition decreases in the population being tested. Furthermore, unless the prevalence is known in some other way, the test results cannot be interpreted. Here, then, are 2 grounds on which to discredit a positive result.The key point is that such false-positive results do not come from the kind of errors the athletes cite and, thus, are not correctable by attention to technique. Rather, they are the normal