Repeated measures analysis of geometrically constructed and directly determined cephalometric points.

Cephalometric landmarks were directly determined visually from bony anatomy on lateral cephalometric radiographs and compared with geometrically constructed cephalometric landmarks for repeated measures reliability of identification. Three groups of four orthodontists, grouped by level of experience, identified 18 landmarks on each of four cephalometric radiographs. This measurement procedure was repeated four times in 1-week intervals. The variability, main associations and interactions among the variables of tracer experience, quality of radiograph traced, and the anatomic landmarks traced were assessed by means of the repeated measures analysis of variance and follow-up Tukey HSD test. The coefficient of variation was used to compare the relative variability of replicate identification of the 18 cephalometric landmarks. Rank order and statistically significant groupings by variability are listed. The level of observer experience and quality of radiograph were statistically unrelated to landmark variability on replicate examination. The geometrically constructed landmarks tested were not statistically different from the reliability of the directly determined bony-based landmarks. This finding would lend support to the techniques of geometric construction used in the Ricketts cephalometric analysis.