Validation of methods used in dental caries diagnosis.

Accurate diagnosis of caries is critical both in clinical practice and epidemiology. Current knowledge of the validity of conventional caries diagnostic methods is reviewed and some theoretical aspects of the design and conduct of validation studies discussed. Four studies of the validity of clinical diagnostic methods are described and their findings summarized and compared. The results indicate that a trained and experienced examiner using a visual diagnostic technique can detect dentine caries, when it is shown to be present experimentally in borderline lesions, with a sensitivity exceeding 0.6, and can return a negative finding where disease is absent with a specificity exceeding 0.8. It is suggested that a visual technique of diagnosis which emphasizes specificity at the expense of some loss of sensitivity is the clinical method of choice, given a climate of low prevalence and slow progression of disease, the perceptual inability of imperfectly standardized examiners, and the adverse consequences of false-positive diagnoses. Selecting teeth with borderline lesions and balanced numbers of diseased and non-diseased sites is recommended for validation studies to allow standardized comparisons to be made and benchmarks for diagnostic performance to be established. However, an uncritical extrapolation of experimental findings to the general population of teeth is likely to lead to spurious assumptions about the consequences--in particular, of false-positive treatment decisions.