Accuracy of two visual criteria for the assessment of caries around restorations: a delayed-type cross-sectional study.

The aim of this study was to compare the performance of two visual criteria used for the detection of caries around restorations in permanent teeth. In this delayed-type cross-sectional study, the study sample was randomly allocated to one of two visual criteria for the assessment of restorations, as follows: a) FDI criteria (International Dental Federation; considers marginal staining, marginal adaptation, and caries recurrence), and b) CARS (Caries Associated with Restorations or Sealants) criteria, defined by the International Caries Classification and Management System. A calibrated examiner assessed the restorations using two reference standards, as follows: i) for restorations requiring operative interventions (repair/replacement), the restoration was partially or totally removed and the presence or absence of carious tissue was assessed; and ii) for restorations requiring non-operative intervention, follow-up for a period of one year was recommended to allow identification of the presence of new lesions not detected at baseline. The sensitivity, specificity, area under Receiver Operating Characteristic (ROC) curve (Az), and accuracy of the criteria were assessed. The study included 305 restorations. The FDI marginal staining parameter had the lowest Az value (Az = 0.501), while similar sensitivity was observed between the CARS (62.0%), FDI presence of caries (65.0%), and FDI marginal adaptation (74.0%) parameters. CARS exhibited the highest specificity (88.3%) and accuracy (85.6%). The CARS criteria exhibited better specificity and accuracy in detecting caries around restorations, followed by the FDI criteria for caries recurrence and marginal adaptation. Considering marginal staining or combining multiple marginal features to assess secondary caries resulted in an increased risk of false-positive outcomes and over-treatment.