On understanding the validity of diagnostic tests.

For clinical practice to be professionally responsible, any diagnostic tests used need to be valid because, if a test lacks validity the information that it provides is wrong. Of the several subtypes of validity, the crucial one is construct validity, which determines how well a diagnostic test discriminates simultaneously between the presence and the absence of the condition being diagnosed. Its key parameters are the sensitivity and specificity of the test, and its (positive) likelihood ratio. The likelihood ratio serves mathematically as a coefficient in an equation that measures the confidence one can have that a positive result of the test is true-positive, given the prevalence of the condition being diagnosed. There is no ideal value for likelihood ratios that make a diagnostic test worthwhile. The value required depends on how much confidence a physician needs to have in a diagnosis before undertaking treatment, which must be calculated using the likelihood ratio and prevalence of the condition being diagnosed.