Whether transverse lines (TLs) could be reliably estimated on x-ray by two trained observers was examined. Male and female left tibiae were randomly selected from the Terry Collection. Their ages at death are between 20-83, with a median age of 31 years. Tibiae were x-rayed in both their anteroposterior (AP) and mediolateral (ML) orientations. One hundred tibiae were used for practice and another thirty were used for the test. Ratings were scored once independently by each rater, and then collectively rated by both observers twice. TLs were scored for their absence, presence, grouped, and continuous frequencies in the proximal and distal shaft for both the AP and ML orientations. Descriptive statistics showed that the TLs were non-normally distributed. Results for their reliability, according to Cohen’s kappa (k) and Spearman’s rank-order (Rho) correlation coefficients, showed that TLs could be reliably used with excellent results for presence and absence (k > .75). Reliabilities for grouped and continuous frequencies were good to excellent (k = .40-0.75; Rho > .80, p = 0.0001). Anteroposterior and ML orientations appeared equally reliable (Rho > 0.90, p < 0.0001). However, there were significant differences between proximal and distal shafts (Rho = 0.33-63; p = 0.08-0.0001), with the distal portion having a greater number of TLs. Log transformed TL data and paired t-tests also showed that there were no significant differences between AP and ML orientations. Thus, TLs could be reliably rated as they appeared on x-ray under constant conditions by two trained observers.
Clark, George A. and Mack, Mark
"Reliability Assessment of Transverse Lines,"
2, Article 10.
Available at: https://digitalcommons.wayne.edu/humbiol/vol60/iss2/10