The influence of observer performance in radiographic caries diagnosis.

Experienced oral radiologists were asked to report whether the proximal surfaces of extracted permanent premolars were radiographically positive or not with regard to carious lesions. The percentages of true positive and false positive diagnoses for each observer were plotted in a graph, displaying a reciprocal relationship. On the basis of differences between observers theoretical calculations were made of the relative number of carious lesions that would have been reported by different observers in populations differing with respect to caries prevalence. It was shown that differences in reported prevalences do not only depend on variations between observers but also upon underlying true prevalences. The implications with regard to epidemiological studies are discussed.