Computer-assisted orthopaedic surgery (CAOS) improves mechanical alignment and the accuracy of surgical cuts in the context of total knee arthroplasty. A simplified, CAOS enhanced instrumentation system was assessed to determine if the same effects could be achieved through the use of a less intrusive system. Two cohorts of surgeons (experienced and trainees) performed a series of total knee arthroplasty resections in knee models with and without navigation-enhanced instrumentation. The percentage of resections that deviated from the planned cut by more than 2°or 2mm (outliers) was determined by post-resection advanced imaging for six unique outcome metrics. Within each experience level, the use of the CAOS enhanced system significantly reduced the total percentage of outliers as compared to conventional instrumentation (Figure 1). The experienced users improved from 35% to 4% outliers overall (p < .001) and the trainees from 34% to 10% outliers (p < .001). Comparing across experience levels, the experienced surgeons performed significantly better in only a single resection metric with conventional instrumentation (Figure 2A), varus/valgus tibial alignment, with 8.3% outliers compared to the trainee's 63% outliers (p = .004). The use of CAOS enhanced instrumentation eliminated any differences between the two user groups for all measured resections (Figure 2B). Comparing CAOS enhanced to conventional instrumentation specifically between anatomical deformity types revealed that there is significant improvement (p < .05) with the use of enhanced instrumentation for all three deformity types (Figure 3). These results suggest that non-intrusive CAOS enhanced instrumentation is a viable alternative to conventional instrumentation with possible benefits. This trial also demonstrates that additional experience may not correlate to improved surgical accuracy, and outliers may be less a result of individual surgeon ability or specific anatomic deformities, and more so related to limitations of the instrumentation used or other yet unidentified factors.

Incorrect registration during computer assisted total knee arthroplasty (CA-TKA) leads to malposition of implants. Our aim was to evaluate the tolerable error in anatomic landmark registration. We incorrectly registered the femoral epicondyles, femoral and tibial centers, as well as the malleoli and documented the change in angulation or rotation. We found that the distal femoral epicondyles were the most difficult anatomic landmarks to register. The other bony landmarks were more forgiving. Identification of the distal femoral epicondyles has a high inter- and intra-observer variability. Our observation that there is less than 2 mm of safe zone in the anterior or posterior direction during registration of the medial and lateral epicondyles may explain the inability of CA-TKA to improve upon the outcomes of conventional TKA.

Restoration of the joint line of the knee during primary and revision total knee arthroplasty is one of many critical steps that directly influence patient outcomes. Fifty MRI scans of normal atraumatic knees were analyzed to determine a quantitative relationship between the joint line of the knee and the bony landmarks of the knee joint: femoral epicondyles, metaphyseal flare of the femur, tibial tubercle, and proximal tibio-femoral joint. We describe the relationship of these six anatomic landmarks about the knee in a gender and size independent manner. This description supports a simple three-step algorithm allowing orthopaedic surgeons to calculate, instead of estimate, the location of the joint line of the knee.