IADR Abstract Archives
Reproducibility and Accuracy of ICDAS Radiographic Coding System in Detecting Proximal Caries
Objectives: To assess inter- and intra-examiner reproducibility and accuracy of ICDAS radiographic coding system to detect proximal caries by dental faculty
Methods: Sixty-four unrestored surfaces of permanent teeth in 25 human subjects with a range of caries status according to ICDAS radiographic coding system (0-RC6) were selected. Twenty-two dental faculty members who had previous training in the International Caries Detection and Assessment System (ICDAS) coding using digital images, extracted teeth and patients examined the identified bitewing radiographic images of the 25 subjects and coded each surface according to the ICDAS radiographic coding system of the selected tooth surfaces. Inter-rater reliability was assessed by calculating the agreement between faculty ratings and the gold standard (ICDAS expert). After a 1 hour break, 22 faculty members re-examined these sixty-four surfaces on bitewing radiographs of 25 patients. Intra-rater reliability was assessed by calculating the agreement between the first and second ratings. The results were subjected to weighted kappa analysis using a quadratic approach.
Results: The inter-examiner reproducibility that benchmarked the faculty scores with those of an ICDAS expert ranges 0.44 - 0.82 with 6 (60%) faculty having weighted kappa values above 0.6 for restorative dentists (n=10) and 0.42 - 0.70 with 5 (42%) faculty having weighted kappa values above 0.6 for non-restorative dentists (n=12). The weighted kappa values for intra-examiner reproducibility ranges 0.42 - 0.85 with 17 (77%) faculty having reproducibility ranges from moderate to excellent.
Conclusions: There is moderate to excellent intra-examiner reproducibility and accuracy using the ICDAS radiographic coding system in the clinical setting (in vivo). There is good agreement of restorative dentists with the ICDAS expert whilst for non-restorative faculty, a wider range of agreement from moderate to good was demonstrated. Some non-restorative faculty require more training and clinical exercises to detect proximal carious lesions using the ICDAS radiographic coding system.
Methods: Sixty-four unrestored surfaces of permanent teeth in 25 human subjects with a range of caries status according to ICDAS radiographic coding system (0-RC6) were selected. Twenty-two dental faculty members who had previous training in the International Caries Detection and Assessment System (ICDAS) coding using digital images, extracted teeth and patients examined the identified bitewing radiographic images of the 25 subjects and coded each surface according to the ICDAS radiographic coding system of the selected tooth surfaces. Inter-rater reliability was assessed by calculating the agreement between faculty ratings and the gold standard (ICDAS expert). After a 1 hour break, 22 faculty members re-examined these sixty-four surfaces on bitewing radiographs of 25 patients. Intra-rater reliability was assessed by calculating the agreement between the first and second ratings. The results were subjected to weighted kappa analysis using a quadratic approach.
Results: The inter-examiner reproducibility that benchmarked the faculty scores with those of an ICDAS expert ranges 0.44 - 0.82 with 6 (60%) faculty having weighted kappa values above 0.6 for restorative dentists (n=10) and 0.42 - 0.70 with 5 (42%) faculty having weighted kappa values above 0.6 for non-restorative dentists (n=12). The weighted kappa values for intra-examiner reproducibility ranges 0.42 - 0.85 with 17 (77%) faculty having reproducibility ranges from moderate to excellent.
Conclusions: There is moderate to excellent intra-examiner reproducibility and accuracy using the ICDAS radiographic coding system in the clinical setting (in vivo). There is good agreement of restorative dentists with the ICDAS expert whilst for non-restorative faculty, a wider range of agreement from moderate to good was demonstrated. Some non-restorative faculty require more training and clinical exercises to detect proximal carious lesions using the ICDAS radiographic coding system.