Reliability of four different computerized cephalometric analysis programs: a methodological error

Sir, We were interested to read the paper by Erkan and Gurel published in the June 2012 issue of Eur J Orthod. The authors aimed to compare the traditional method of manual cephalometric tracing with four different computerized tracing programs (Dolphin Imaging, Vistadent, Nemoceph and Quick Ceph). They used multivariate analysis of variance and Box’s and Levene’s tests, showing no statistically significant difference between manual tracing and the computerized tracing programs (Erkan et al., 2012). The authors pointed out in their conclusions that the measurements obtained with the cephalometric analysis programs used in the study were reliable. However, they did not use any of the commonly used statistical tests (Intraclass correlation coefficient ICC or weighted kappa) to assess the reliability (Jeckel et al., 2007; Szklo and Nieto, 2007; Rothman et al., 2008). Therefore, we would like to point out that this conclusion may be misleading.