The reliability of radiographic measures of total ankle replacement position: an analysis from the OARS cohort.

Abstract

There is no validated radiographic measurement to diagnose prosthetic complication(s) following total ankle replacements (TARs) although a number of angular and linear measurements, used to define the TAR position on postoperative radiographs, have been recommended to detect prosthetic loosening. The aim of this study was to test the intra- and interobserver reliability of these measurements. This is a prospective study embedded within a multicentre cohort study. Following sample size calculation, 62 patients were analysed. Six measurements were performed on the first postoperative anteroposterior and lateral ankle radiographs: angles α and β, and length "a" defined the craniocaudal position of the tibial component, while angle γ, and lengths "b" and "c" defined the angular position of the talar component. Measurements were recorded by three independent observers. Inter- and intraobserver reliability was assessed with intraclass correlation coefficient (ICC), Bland-Altman plots, and within-subject coefficients of variation (CV). The intrarater ICC was "almost perfect" (ICC 0.83-0.97) for all six measurements. The interrater ICC was "substantial" to "almost perfect" (ICC 0.69-0.93). The mean difference in intrarater angular measurements was ≤ 0.6° and ≤ 0.8mm for linear measurements, and ≤ 2.2° and ≤ 2.1mm for interrater measurements. Maximum CV for the interrater linear measurements (≤ 17.7%) more than doubled that of the angular measurements (≤ 8.0%). The maximum width of the 95% limits of agreement was 6.5° and 8.4mm for intrarater measures, and 8.9° and 10.6mm for interrater measurements. Angular measures are more reliable than linear measures and have potential in routine clinical practice for TAR position assessment.