Reliability of Cephalometric Landmark Identification and Sagittal Discrepancy Measurements Across Skeletal Classes I, II, and III: A Comparative Study

Background/Objectives: Cephalometric analysis remains the principal tool for diagnosing sagittal jaw discrepancies. Its clinical reliability depends on the accuracy of landmark identification and the resulting horizontal and vertical dispersion in the Cartesian system. The aim of this study was to assess the inter-rater reliability of 12 cephalometric landmarks and of six measurements used for sagittal discrepancy assessment (ANB, Wits, Tau, Yen, Sar and W) across three skeletal classes. Methods: Twenty-four lateral cephalograms (eight per skeletal class) were assessed twice, seven days apart, by 15 orthodontists trained in a 5-hour calibration course. Landmark coordinates were normalized against a reference value derived from two expert raters. Reliability of horizontal (X) and vertical (Y) coordinates and of the resulting sagittal measurements was quantified using the intraclass correlation coefficient ICC(2,1). Between-class differences were tested with Fisher Z-transformed Z-tests; the Benjamini–Hochberg false-discovery-rate (FDR) procedure was applied to control for multiple comparisons. Results: Horizontal coordinates of all landmarks showed excellent inter-rater agreement (ICC 0.91-0.96) regardless of skeletal class. Vertical coordinates showed considerably greater variability (ICC 0.52-0.94). Among sagittal measurements, Wits demonstrated the highest reliability across all classes (ICC 0.87-0.90), followed by Yen (ICC 0.76-0.86). Tau, Sar and W reached near-perfect agreement only in Class III patients (ICC 0.93-0.95). The ANB angle showed the lowest reliability, particularly in Class I (ICC =0.55, 95% CI 0.32-0.84). Conclusions: Vertical and horizontal dispersion of cephalometric landmarks materially influences the diagnostic accuracy of sagittal measurements. Wits and Yen are the most reliable parameters across all skeletal classes, whereas Tau, Sar and W are particularly trustworthy in Class III. The traditional reliance on the ANB angle as a gold standard should be reconsidered, as it was the least reproducible measurement in our cohort.