| |
Several new methods for the detection of occlusal caries have emerged in recent years. The purpose of the present study was to compare the reproducibility and accuracy of some of those methods. Ninety- six extracted permanent molar teeth were selected. Three examiners were trained, and subsequently carried out the examinations, using visual examination (score, depth), radiographic examination (presence, depth), ECM, QLF and DIAGNOdent. Twenty-five percent of the teeth were re-examined for repeatability. Stereomicroscopic examination served as gold standard. Intra- and interexaminer agreements were assessed using weighted . statistics, except for ECM and DIAGNOdent (ICC). Methods were compared for differences in accuracy, sensitivity, and specificity using generalized estimating equation methods applied to logistic regression models. Intra-examiner agreement ranged from substantial to almost perfect. Interexaminer agreement was fair to moderate. Areas under the ROC curves were 0.82, 0.61, 0.79, 0.84, and 0.88, respectively, for visual, radiographic, ECM, QLF and DIAGNOdent examinations. Examiners also rated their recommended treatment after only visual examination and after obtaining data from the other methods. When all data were at hand, the examiners recommended more invasive treatment, and the accuracy of the recommendations made after reviewing all data relative to lesion presence was significantly lower (p = 0.01). In conclusion, the accuracies of all methods, except radiographs, were similar. Having data from all methods increased recommendations for invasive treatment and lowered the accuracy of the treatment recommendation relative to the gold standard. |