Reliability of leg alignment using the OrthoPilot system depends on knee position: a cadaveric study.

Despite the increase in clinical use of navigation systems in total knee arthroplasty, few studies have focused on the reproducibility of these systems. The aim of the present study was to assess the influence of knee position and observer experience on intra- and inter-observer agreement in limb alignment assessment with the OrthoPilot system. Limb alignment in the coronal plane and extension range of the knee were assessed in four embalmed cadaveric specimens by five independent observers and measurements were repeated four times to determine intra- and inter-observer agreement, expressed as intraclass correlation coefficients (ICCs). Additionally, navigation results were compared against figures from conventional measurement of leg alignment (ground truth). Intra- and inter-observer agreements were excellent for assessing the extension range (ICC, 0.97 and 0.95) and the coronal femuro-tibial axis in knee extension (ICC, 0.92 and 0.88) but were generally worse in knee flexion (ICC, 0.62 and 0.55). There was an increased tendency of intraobserver errors in observers with less clinical experience. Mean correlation with conventional measurements was fair (Spearman's rho 0.61). The OrthoPilot system showed excellent reproducibility for assessment of extension range and coronal limb alignment. However, assessments of coronal limb alignment in flexion were prone to error and caution should be taken when relying on these measurements.