Methods: Twenty-four primary teeth were split up into two samples presenting, respectively, 70% and 40% of sound sites+enamel caries. 12 undergraduate and 12 graduate students assessed occlusal surfaces using ICDAS. Comparing to reference examiner, we checked the type of disagreements. To verify differences between samples with primary and permanent molars, we used other 75 primary and 75 permanent molars. Two experienced examiners evaluated the surfaces and types of divergences between them were also studied. Considering examiners’ experience and sample composition, the interexaminer reproducibilities and the consistency of exams were calculated by weigthed Kappa values and intraclass correlation coefficient. The number and types of divergences were compared by chi-square test. Bland-Altman analyses were performed to compare the scores assigned by examiners in all samples and check any trend regarding their divergences
Results: The interexaminer reproducibilities and consistencies were higher, respectively, than 0.70 and 0.90, with a slightly wider range of Kappa values for the undergraduate students. In primary teeth, Kappa values were a little higher than in permanent ones. The sample composition did not influence the overall number of errors independently of clinical experience (p<0.05). Divergences between scores 1 and 2 prevailed among undergraduate students in the sample comprising 70% of sound or enamel caries (p<0.001). No systematic trends concerning divergences were observed in Bland-Altmand analyses.
Conclusion: Samples comprising high proportions of initial caries can lead to high occurrence of divergences between ICDAS scores 1 and 2, specially among beginners. Examiners seem to disagree slightly less in primary molars than in permanent ones.