Radiographic caries diagnosis. A study of caries progression and observer performance.

A detailed index and score system was employed in radiographic studies of the progression of proximal caries during a three and a six year interval after the termination of the regular school dental care. It was demonstrated that the score system offers advantages compared with systems for estimating caries progression only taking into account the number of new lesions, since the former also reveals progression of already existing lesions. Most of the new lesions being developed during a six year interval was found in the enamel. The progression of already existing carious lesions was slow. Fewer new lesions and a slower progression of already existing lesions were found with the increasing age of the patients. Wide variations between observers were found at radiographic examinations of extracted teeth. The influence of such variations on the results of epidemiological caries investigations was elucidated. The importance of minimizing the rate of false positive diagnoses in investigations comparing the caries prevalence between different groups of patients was demonstrated. Data from the dental literature were used to estimate the probabilities of clinical cavities at different extents of the radiographically registered carious lesions. These probabilities were lower, the lower the prevalence of cavities and the smaller the extent of the radiographic lesion. Receiver operating characteristic curves based on the extents of the radiographic carious lesions were employed in assessing optimal criteria for restorative treatment, taking into account the prevalence of carious cavities and the costs of errors in the decision-making. The lower the cavity prevalence and the higher the cost of overtreatment, the greater the extent of the radiographic lesion that should be used as criterion for restorative treatment. The results of radiographic examinations of carious lesions were found to be greatly influenced by information given to the examiners and by the opinions of other observers. Such an influence, if occurring in epidemiological investigations, may give misleading results. Different densities in radiographs of different groups of patients to be compared may also give rise to inaccurate results. Ways of minimizing the influence of observer variations are discussed.