This study measured the three bony axes usually used for femoral component rotation in total knee arthroplasty and compared the accuracy and repeatability of different measurement techniques.

Fresh cadaveric limbs (n=6) were used. Three observers (student, trainee and consultant) identified the posterior condylar (PCA), anteroposterior (AP) and the transepicondylar (TEA) axes, using a computer navigation system to record measurements. The AP axis was measured before and after being identified with an ink line. The TEA was measured by palpation of the epicondyles both before and after an incision was made in the medial and lateral gutters at the level of the epicondyles, allowing the index finger to be passed behind the gutters. In addition the true TEA was identified after dissection of all the soft tissues. Each measurement was repeated three times. For all axes and each observer the repeatability coefficient was calculated.

The identification of the PCA was the most reliable (repeatability coefficient: 1.1°) followed by the AP after drawing the ink line (4.5°) then the AP before (5.7°) and lastly the TEA (12.3°) which showed no improvement with the incisions (13.0°). In general the inter-observer variability for each axis was small (average 3.3°, range 0.4° to 6°), being best for the consultant and worst for the student. In comparison to the true TEA, the recorded TEA and AP axis averaged within 1.5° whilst the PCA was consistently 2.8° or more internally rotated.

This study echoed previous studies in demonstrating that palpating the PCA intra-operatively is highly precise but was prone to errors in representing the true TEA if there was asymmetrical condylar erosion. The TEA was highly variable irrespective of observer ability and experience. The line perpendicular line to the AP axis most closely paralleled the true TEA when measured after being identified with an ink line.