Development and Validation of a Radiographic–Clinical Discordance Index for Risk Stratification of Incidental Jaw Lesions

Introduction: While existing methods mostly focus on the prevalence and radiographic features of these findings, they place only limited stress on the discrepancy between an apparently asymptomatic clinical situation and radiographic features that may be significant. Objectives: In this study, the Radiographic–Clinical Discordance Index (RCDI) was introduced as a systematic way of risk-stratifying incidental jaw lesions. Methods: A one-year retrospective observational study into diagnostic accuracy was carried out. Patients who had jaw lesions found accidentally during routine panoramic radiographic examinations were assessed by means of predetermined clinical and radiographic criteria. The features of the lesions—such as size, border, presence of cortical involvement, their relationship to neighbouring teeth, and any changes to the roots—were all included in the RCDI. The diagnostic performance of the index was evaluated using receiver operating characteristic (ROC) analysis, and sensitivity, specificity, predictive values, likelihood ratios, and interobserver agreement were also determined. Results: 180 patients were assessed, 62 of them (34.4%) having clinically significant lesions. The mean RCDI score was found to be significantly greater in cases of clinically significant lesions than in those with clinically insignificant lesions (8.1 ± 2.4 as compared with 3.6 ± 1.8; p < 0.001). The RCDI showed good discriminatory power (AUC = 0.891; 95% CI: 0.845–0.938). At a cut-off value of ≥6, the sensitivity and specificity were 87.1% and 81.4% respectively. Conclusions: The RCDI showed promising potential in the identification of jaw lesions which require further investigation. Because of its standardized and quantitative method, it could aid clinical decision-making and help in the prioritisation of incidental findings. However, multicentre prospective validation is necessary before it can be implemented routinely.