Cone-beam computed tomography transverse analyses. Part 2: Measures of performance

摘要

Introduction: The aim of this study was to compare the predictability of the cone-beam transverse (CBT), jugale (J-point), and transpalatal width measurement (TWM) analyses in identifying clinical crossbite. Methods: From a pool of patients with cone-beam computed tomography scans who came for orthodontic treatment, a sample of 133 patients was identified, with 54 in posterior crossbite (28 boys, 26 girls) and 79 not in crossbite (77 boys, 110 girls). No patient had dental compensation in this sample. After correcting for lateral mandibular shift, 33 of the 54 posterior crossbite patients had a bilateral crossbite, and 21 had a unilateral crossbite with no shift. The CBT, J-point, and TWM analyses were done for each patient from a coronal cross-section through the middle of both the maxillary and mandibular first molar crowns. The landmarks and measurements used were described in detail in a previous study. Posteroanterior cephalograms were constructed to simulate the geometry of the conventional cephalometric radiographs. All 3 analyses were performed on the same data set to predict whether crossbite was present. We used 2 assessments of diagnostic predictability: sensitivity and specificity, and positive and negative predictive values. While the 2 methods answer different questions, the prevalence of crossbite in a population will affect the positive and negative predictive values, but the sensitivity and specificity will not change. Results: Of the 133 patients studied, 54 had a clinical crossbite, and 79 had no crossbite. The J-point analysis accurately predicted that 38 patients would have a crossbite, and 45 would not. This resulted in a positive predictive value of 52.78%, a negative predictive value of 73.77%, sensitivity of 70.4%, and specificity of 57%. The TWM analysis accurately predicted that 53 patients would have a crossbite, but it falsely predicted that an additional 68 patients would have crossbite. This resulted in a positive predictive value of 43.8%, a negative predictive value of 91.67%, sensitivity of 98.1%, and specificity of 13.9%. The CBT analysis correctly predicted a crossbite in 47 patients and accurately predicted no crossbite in 73 patients. This resulted in a positive predictive value of 88.68%, a negative predictive value of 91.25%, sensitivity of 87.0%, and specificity of 92.4%. Conclusions: This study showed that although the TWM analysis had slightly better negative predictive and sensitivity values, the CBT analysis was overall better at both predictive value and sensitivity/specificity because of the limitations in J-point landmarks and the extent of the TWM analysis. Furthermore, the CBT analysis can distinguish between skeletal and dental discrepancies. Further work will test the analysis on additional samples with differing prevalences of crossbite.