Abstract
Background
The full leg x-ray is a widely used imaging modality for post-operative assessment of total knee replacement (TKR). However, these assessments require controlled conditions and precise measuring in order to be accurate. inter-observer reliability remains a matter of concern as well. This study examines whether intersurgeon differences are significant.
Method
Post-operative lateral and full-leg frontal x-rays of 26 patients were assessed by 6 surgeons according to a strict measuring protocol. Four measurements (Figure 1 and 2) were taken of which two were on the femur (Femoral Varus Angle FVA and Femoral Flexion Angle FFA) and two, on the tibia (Tibial Varus Angle TVA; Tibial Slope Angle TSA). A random effects, two-way ANOVA was performed on the data using Minitab (v 16.0, Minitab Inc., Pennsylvania, USA) to determine whether a surgeon has influence on the results (α = 0.05). Intra-class correlation coefficients (ICC) and standard error of measurements (SEM) resulting in smallest detectable changes (SDC) were also calculated [1].
Results
The resulting p-value for FVA is p = 0.246, p = 0.006 for FFA, p = 0.006 for TVA and p = 0.032 for TSA. The measured ICC of FVA is 0.88, 0.75 of FFA, 0.76 of TVA and 0.76 of TSA. Additionally, the SEM and SDC for FVA are respectively 0.26° and 0.72°, 0.84° and 2.36° for FFA, 0.45° and 1.26° for TVA and 0.66° and 0.86° for TSA.
Discussion
The ANOVA shows for FFA, TVA and TSA a significant influence of the surgeon on the measurement result. Difficulty in indicating the implant line and the sensitivity of short lines to angular measurement could be reasons for the significant influence on FFA and TSA, which are measured on short lateral films. TVA measurements on the full frontal x-ray are also affected by difficulty in indicating implant line where for a femoral implant a line tangent to the condyles is more consistent.
These results can be compared to the findings of Hirschmann et al [2] where, contrary to their data, our measurements on FVA and TVA correlate better, presumably because of the use of long leg film instead of short film. Our data is collected from 6 surgeons whereas the Hirschmann et al study represents 1 surgeon and 1 radiologist.
SDC calculations based on this variability study show the smallest detectable changes on x-ray measurements for FVA, TVA and TSA are approximately 1° and approximately 2° for FFA.
Conclusion
The results indicate that x-ray measurement variability should be taken into account when choosing a post-operative measurement technique. While x-ray measurement still has its place in daily practice, we advise more consistent methods of measuring for research.
