Reliability of the Metaphyseal-Diaphyseal Angle in Tibia Vara as Measured on Digital Images by Pediatric Orthopaedic Surgeons

Abstract

Levine and Drennan described the tibial metaphyseal-diaphyseal angle (MDA) in an attempt to identify patients with infantile Blount's disease. Pediatric orthopaedic surgeons have debated not only the use, but also the reliability of this measure. Two techniques have been described to measure the MDA. These techniques involved using both the lateral border of the tibial cortex and the center of the tibial shaft as the longitudinal axis for radiographic measurements. The use of digital images poses another variable in the reliability of the MDA as digital images are used more commonly. The radiographs of 21 children (42 limbs) were retrospectively reviewed by 27 staff pediatric orthopaedic surgeons. Interobserver reliability was determined using the intraclass correlation coefficients (ICCs). Nine duplicate radiographs (18 duplicate limbs) that appeared in the data set were used to calculate ICCs representing the intraobserver reliability. A scatter plot was created comparing the mean MDA determined by the 2 methods. The strength of a linear relationship between the 2 methods was measured with the Pearson correlation coefficient. Finally, we tested for a difference in variability between the 2 measures at angles of 11 degrees or less and greater than 11 degrees by comparing the variance ratios using the F test. The interobserver reliability was calculated using the ICC as 0.821 for the single-measure method and 0.992 for the average-measure method. The intraobserver reliability was similarly calculated using the ICC as 0.886 for the single-measure method and 0.940 for the average-measure method. Pearson correlation coefficient (0.9848) revealed a highly linear relationship between the 2 methods (P = 0.00001). We also found that there was no statistically significant variability between the 2 methods of calculating the MDA at angles of 11 degrees or less compared with angles greater than 11 degrees (P = 0.596688). There was excellent interobserver reliability and intraobserver reliability among reviewers. Using either the lateral diaphyseal line or center diaphyseal line produces reasonable reliability with no significant variability at angles of 11 degrees or less or greater than 11 degrees. Level IV.