To the ongoing debates in total knee arthroplasty on whether to use cemented or cementless fixation, whether to preserve or replace the posterior cruciate ligament, whether or not to resurface the patella, and, more recently, what is the appropriate alignment of a total knee replacement (mechanical or “kinematic”), we now add whether the use of computer navigation in total knee arthroplasty will alter the outcome of the surgery. The article by Kim and colleagues in this issue of The Journal of Bone and Joint Surgery (American Volume) provides midterm clues, yet raises several important questions. The seminal issue raised by this paper is one that requires further examination: Are we asking the correct questions when we assess the effect of computer navigation on the outcome of a surgical procedure?
In their prospective randomized trial, Kim et al. reported the results of 520 patients who underwent bilateral total knee arthroplasty performed sequentially during the same anesthetic session by a single experienced surgeon. The patients were randomly assigned to undergo either total knee arthroplasty with the use of conventional guides or computer-navigated total knee arthroplasty on the first limb, followed by the opposite approach on the contralateral limb. The authors posed three study questions: (1) Was clinical function improved in the computer-navigated surgery group? (2) Did computer-navigated surgery improve alignment and, as a surrogate, total knee arthroplasty survivorship; and (3) Did malalignment of >3° deviation from the mechanical axis increase the rate of aseptic failure? At a mean follow-up of 10.8 years, there were no significant differences between the groups in the outcomes measured, including alignment, survivorship, and clinical function. The authors concluded that “the effect of computer-navigated total knee arthroplasty compared with conventional total knee arthroplasty on long-term implant survival remains unproven.”
In examining the “effect” of computer navigation on mechanical alignment and the reduction in “outliers” or the number of patients with >3° deviation from the mechanical axis, meta-analyses as well as clinical studies have shown that the use of computer navigation leads to fewer “outliers.”1-4 In the study by Kim et al., the pooled data for both groups demonstrated no significant differences, although there were more outliers in the conventional-guide group. Would a different outcome related to the variable studied—namely, computer navigation compared with conventional guides—have been obtained if each patient’s individual alignment outcomes were examined between the two knees? The authors could not comment on whether or not, when the computer was used for the first arthroplasty, information was gained that benefited the accuracy of the “conventional” surgery that was subsequently performed. One would reasonably assume that there was some information obtained; one would need to also assume some effect when conventional surgery was performed first. Of concern in this study is the prevalence of femoral notching (5% in the computer-navigated group and 1% in the mechanically guided group), which the authors believed could be attributed to registration errors. The computer is not a safety net against human error, as the computer receives its information from the surgeon.
Whether the effect of computer navigation in reducing mechanical alignment “outliers” is beneficial as a means of improving long-term survivorship remains an unanswered issue. In the fifteen-year total knee survivorship study by Parratte et al. from the Mayo Clinic, no difference in implant longevity was seen among patients who were “outliers” and those who were not5. The current study was unable to show a difference between the two surgical tools and concluded that the effect of the computer on implant survivorship was unproven. However, three separate conclusions and/or questions may appropriately be deduced from the study: (1) the conclusion is accurate as shown by the data; (2) is alignment a true surrogate for long-term survivorship?; and (3) can an opposite and accurate conclusion from the data be made that conventional total knee arthroplasty does not improve the outcome relative to that obtained with computer navigation? At this time, the issues of alignment and implant longevity and the inter-relationship of the two remain unknown.
Most importantly, the authors ask whether the clinical outcome is improved with use of the computer. If the assessment of clinical outcome is confined to the examination of knee motion, then no difference is shown, as prior studies have clearly demonstrated that it is preoperative motion that predicts postoperative motion. If alignment measured by radiographs and computed tomography scans is used to describe clinical outcome, then clearly a trend is shown toward improvement when computer navigation is used; this trend can be reasonably explained by the greater accuracy of a computer when employed correctly, with clear identification of accurate osseous landmarks. However, if the domains of pain, function, and activity level are being used to assess clinical outcome, one must ask why the use of computer guidance would lessen pain, improve function, or improve activity level. The key is that we do not yet have the tools to examine these parameters or the effect of a surgical tool on these domains.
Dr. Kim and his colleagues are to be commended for performing and reporting on this rigorous Level-I study, and for tempering their conclusion, as the data could have been reported as showing that computer navigation makes no difference. The reader must remember that the skill set of the surgeon is important to the outcomes described. Because individual results may vary among orthopaedic surgeons, the choice to use the computer as a surgical tool should never be based on any single study. This study highlights the fact that we still do not know if using the computer is of benefit, that we do not have outcome measures that can appropriately test beneficence, and that yet another debate in total knee arthroplasty rages on.
Disclosure: The author received no payments or services, either directly or indirectly (i.e., via his institution), from a third party in support of any aspect of this work. The author, or his institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, the author has had another relationship, or has engaged in another activity, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.