Inter-examiner Reliability of Two Methods for Scoring Post-Orthodontic White Spot Lesions from Digital Photographs.

Abstract

To compare the reliability of two scoring systems for detecting white spot lesions (WSLs) from clinical photographs captured during debonding of fixed orthodontic appliances. Digital images of 58 healthy adolescents (34 females and 24 males) were examined, depicting 384 buccal surfaces of maxillary incisors, canines, and first premolars. Three trained examiners (E1, E2 and E3) independently evaluated the fully anonymized photos in a randomized order using the Gorelick index (GI) and the modified International Caries Detection and Assessment System (ICDAS II). A 1-2-week interval separated the scorings. Spearman's rank correlation coefficient, Fisher's z-test, and the interclass correlation coefficient (ICC) were applied to compare the scoring methods and express examiner agreement. The two scoring systems showed a moderate to strong positive relationship, but inter-examiner variations were significant (p<0.05). We found moderate to good reliability (ICC 0.60 to 0.84) with the ICDAS II system and good to excellent values with the GI (ICC 0.72 to 0.94), depending on the examiner. The agreement concerning the sound surfaces and the most severe WSLs was perfect, whereas the scoring of the milder lesion stages appeared more uncertain. A moderate to strong positive relationship was demonstrated between the two methods when scoring the presence and severity of WSLs from digital images. Significant inter-examiner variations affected reliability.