Variations among dentists in radiographic detection of occlusal caries

The quality of radiographic diagnosis of occlusal carries was examined with special attention to the diagnosis of lesions in dentin. Eighty-four extracted teeth, most of them with occlusal caries of different severity and some apparently sound, were radiographed. For validation purposes, the lesion depths were recorded during a drilling procedure. Ten dentists interpreted the radiographs. On average, 12% of the sound surfaces were wrongly diagnosed radiographically as having caries (false positives), 33% of the enamel lesions were identified on radiographs, and 62% of the small dentin lesions and 100% of the big dentin lesions were diagnosed radiographically (true positives). There was wide variation in the ability to identify dentin caries correctly between the observers; the sensitivity values ranged from 0.77 to 1.0 (mean 0.90) and the specificity values from 0.45 to 0.93 (mean 0.76). The dentist demonstrating the highest diagnostic quality (measured by the area beneath the receiver operating characteristics curve) had only about 9% false positives (1-specificity), while the dentist who had the lowest diagnostic quality scored caries in about 55% of the teeth in the control group. Both of them had 96% true positives. The tendency to make false-positive registrations in the dentin adjacent to enamel seems to be a major problem in radiographic diagnosis of occlusal caries. The radiograph has significant value in the diagnosis of occlusal dentin caries, but the great variations between observers indicate the need for training and calibration.