Abstract
The accuracy of measurement in computer-assisted total knee arthroplasty is dependent on the quality of data acquisition at the start of the procedure; errors in landmark identification could lead to misalignment and therefore poorer longterm outcomes.
Some navigation systems require the surgeon to explicitly identify the femoral epicondyles in order to calculate the trans-epicondylar axis, whereas other systems are able to interpolate the epicondylar location based on a number of points acquired from the distal femoral surface. Significant inter-observer variability in landmark identification has been previously reported in dry bone studies. The purpose of this study was to test the accuracy of identification of the epicondyles during a simulated total knee replacement on a fresh cadaveric specimen.
An unfixed fresh cadaveric left lower limb was used to perform a navigated total knee replacement using the Orthopilot® (B|Braun-Aesculap, Tuttlingen, Germany) image-free navigation system.
Sixteen surgeons attending an advanced navigation training course were invited to take part. A single consultant surgeon performed initial dissection and pin placement, up to the point of landmark acquisition. Each subject was then asked to use a pointer tool to identify the medial and lateral epicondyles, as they would in an operative situation. Data were recorded by the Orthopilot® system, and exported as a 3D array for further analysis.
Initial visualisation with a 3D scatter plot showed that points were evenly distributed within a circular pattern around each epicondyle. The length of a vector between each point on each epicondyle was calculated in turn. The maximum distances between points were 15.6mm for the medial epicondyle, and 19.9mm for the lateral epicondyle.
We then calculated the length and angulation of the trans-epicondylar axis (TEA) for each observer, equivalent to the vector between each pair of points (medial and lateral epicondyle). An average TEA was calculated, and the range and standard deviation of angulation were determined. In the x axis the range was 16.3° (–8.3° to 7.9°, SD 5.1°), in the y axis the range was 18.7° (–8.7° to 10°, SD 5.2°) and in the z axis the range was 20.5° (–10.1° to 10.4°, SD 6.5°). Range of recorded TEA length was 64.5 to 74.9mm (mean 70.6mm, SD 3.3mm).
We conclude that in this simulated operative scenario, surgeons exhibited considerable variability when locating the epicondyles. Range of angulation of the TEA exceeded 16° (SD > 5.1°) in all 3 planes. We cannot recommend the use of a trans-epicondylar axis determined from 2 single points, as a reliable landmark in navigated total knee replacement.
Correspondence should be addressed to Mr K. Deep, General Secretary CAOS UK, 82 Windmill Road, Gillingham, Kent ME7 5NX UK. E Mail: caosuk@gmail.com
