Reproducibility in the Assessment of Caries Lesion Behaviour: A Comparison between Conventional Film and Subtraction Radiography

Abstract

The purpose of this study was to compare reproducibility in the assessment of caries lesion behaviour in digital subtraction images and conventional radiographs. Ninety–seven pairs of conventional film bite–wings were included. The radiographs had been taken with a Kwik–Bite filmholder without further standardisation. The selection criterion was that at least one surface imaged in a bite–wing pair should show progresssion of a caries lesion from the first exposure to the second. The time interval between the bite–wings was 1–2 years. The radiographs were scanned into a personal computer and the two images from the same patient subtracted. Seven observers scored the subtraction images and the conventional films. The scale was based on no change/ change in surface appearance scored in the following categories: 0 = no change; 1 = development of a new caries lesion in enamel; 2 = development of caries lesion in enamel and dentine/progression of an existing enamel lesion into dentine/progression of a dentinal lesion further into dentine; 3 = development of secondary caries; 4 = new filling/change of filling/extension of filling. Cohen’s κ coefficients were calculated for intra– and inter–observer agreement for the two radiographic methods separately. The κ value served as the statistical sampling unit for testing differences between the two methods (Wilcoxon’s test for ranked pairs). The average intra–observer κ value was 0.875 (range 0.775–0.958) for the subtraction images and 0.758 (range 0.560–0.890) for the conventional radiographs. The difference was statistically significant (p<0.05). For inter–observer agreement, the κ values were lower than for intra–observer agreement. The average value was 0.678 (range 0.485–0.754) for the subtraction images and 0.701 (range 0.497–0.817) for the conventional radiographs (p>0.1). This study suggests a new way of comparing reproducibility in terms of κ values between diagnostic methods in clinical studies where accuracy cannot be evaluated. The subtraction method may be useful in the assessment of caries lesion behaviour in the clinic.

A novel classification system for assessment of approximal caries lesion progression in bitewing radiographs

“…9 Raters included in this study represented different professional experiences working in different institutions as well as different time lengths of expertise (long, intermediate and short time of experience). Yet, interrater agreement of 10 raters for caries lesion progression indicated substantial agreement (kappa 0.61), which is comparable with the results for 7 raters (kappa 0.70) in the study of caries lesion development by Wenzel et al 7 Pairwise interrater agreement varied, but substantial agreement was reached for many pairs of raters. The average intrarater agreement was 0.76, identical to that presented by Wenzel et al 7 for the assessment of caries lesion behaviour.…”

“…Yet, interrater agreement of 10 raters for caries lesion progression indicated substantial agreement (kappa 0.61), which is comparable with the results for 7 raters (kappa 0.70) in the study of caries lesion development by Wenzel et al 7 Pairwise interrater agreement varied, but substantial agreement was reached for many pairs of raters. The average intrarater agreement was 0.76, identical to that presented by Wenzel et al 7 for the assessment of caries lesion behaviour. Although difficult to compare, these results indicate higher interrater agreement than for caries detection by bitewing radiography reported in studies of high and medium quality.…”

“…35 It may, however, be dubious to base a decision of operative intervention on caries risk assessment, as evidence concerning the validity of existing methods for identifying individuals at high risk is limited. 36,6 Table 3 Intrarater agreement expressed as percentage of agreement (%) and kappa with confidence interval (CI) for the assessment of caries lesion progression of approximal surfaces (n 5 65) in bitewing radiography of adolescents Raters Agreement All surfaces assessed to present progression (5,6,7,9,10,12) a Surfaces assessed to present progression within enamel (5,6,9) a Surfaces assessed to present progression in outer half of dentine (7,10,12) Numbers refer to categories presented in Figure 1.…”

“…To our knowledge, there is only one previous publication on the reproducibility of radiographic assessment of caries lesion progression. 7 In view of the importance of the radiographic assessment of caries lesion progression, an established classification system for this assessment could prove valuable. It would have the potential to facilitate clear communication among clinicians and researchers.…”

Re-evaluation of methods to identify individuals with increased risk of caries

“…The level of agreement among raters for radiographic assessment of caries lesion progression is however, largely unknown. Only one previous study was found regarding the agreement of caries lesions progression (Wenzel et al, 2000). Based on the reported variation among dentists in the detection and depth estimation of caries lesions, it can be expected that this variation may also exist for the assessment of caries lesion progression.…”