Inter-observer reliability of measurements performed on digital long-leg standing radiographs and assessment of validity compared to 3D CT-scan

Background: Long-leg radiographs (LLR) are often used in orthopaedics to assess limb alignment in patients undergoing total knee arthroplasty (11(A). However, there are still concerns about the adequacy of measurements performed on LLR. We assessed the reliability and validity of measurements on LLR using three-dimensional computed tomography (3D CT)-scan as a gold standard. Methods: Six different surgeons measured the mechanical axis and position of the femoral and tibial components individually on 24 LLR. Intradass correlation coefficients (ICC) were calculated to obtain reliability and Bland-Altman plots were constructed to assess agreement between measurements on LLR and measurements on 3D CT scan. Results: ICC agreement for the six observer measurements on LLR was 0.70 for the femoral component and 0.80 for the tibial component. The mean difference between measurements performed on LLR and 3D CT-scan was 0.3 degrees for the femoral component and - 1.1 degrees for the tibial component. Variation of the difference between LLR and 3D CT-scan for the femoral component was 1.1 degrees and 0.9 degrees for the tibial component. 95% of the differences between measurements performed on LLR and 3D CT-scan were between - 1.9 degrees and 2.4 degrees (femoral component) and between - 2.9 and 0.7 (tibial component). Conclusion: Measurements on LLR show moderate to good reliability and, when compared to 3D CT-scan, show good validity.