From Bench to Bedside: Robotics and Navigation in Orthopaedics—Rise of the Machines or Just Rising Costs?

The machines of the digital age—the electronic medical record, digital monitors, digital imaging, online patient portals, and “phones” that are used for pretty much everything but making calls—have invaded our operating rooms. But we’ve had time to prepare for the takeover. The first robot-assisted surgery (a brain biopsy) was performed at least as early as 1985 [8].

A discussion of the varied flavors of robotics (direct or indirect, autonomous, haptic or boundary control [4]) and navigation assistance (passive, imageless, etc…) is beyond what I can cover here. And I won’t try to split hairs between navigation and robotic surgery—the latter generally implies, requires, and includes some use of the former. Rather, I will group both robotics and navigation assistance into “machine-assisted” procedures. In general, the goal of machine-assisted surgery is to prevent errors (improve accuracy and precision), improve outcomes (through more-reliable, less-invasive procedures), and save time and money. So how are we doing?

In my opinion, just so-so. Some studies [3, 4] suggest that machine-assisted hip and knee arthroplasty improves precision and reduces “outliers”—components positioned outside some predefined target zone. But the best studies [7, 13] and systematic reviews [3] that we have suggest that this does not matter at all to patients in terms of pain, outcomes scores, or a reduced risk of revision. While less-robust work [2, 11] has suggested that machine assistance can nudge functional outcomes in the direction we’d like it to go, we have to weigh the results of long-term followup of randomized trials [7, 13] and systematic reviews [3] more heavily, and based on those, conclude that this technology has not delivered on its promise. Orthopaedic applications of machine-assistance outside of arthroplasty deserve mention, as well, but those studies paint no clearer a picture. Research on machine-assistance for orthopaedic tumor resection and reconstruction [1], orthopaedic trauma [5], and spine instrumentation [12] has demonstrated early promise but without adequate comparison cohorts, mixed findings, and insufficient evidence to support the machine-assisted techniques, respectively.

The main disadvantage of machine-assisted surgery is cost—for the machine, system or robot itself, any additional preoperative or intraoperative imaging required, and operating room time. Machine-assisted surgery requires more, not less, operative time [3, 4], and a 2005 study [14] estimated that operating room time costs about USD 66 a minute on average (and rising) in the United States. Further, most machine-assisted systems have a substantial learning curve [15]. The exact cost burden or profitability, of course, is highly variable based on the hospital or surgeon’s payor mix, his or her operative speed and proficiency with the system, and what kind of deal has been negotiated with the vendor(s). Even as a marketing tool—“come get your machine-assisted surgery here (or by Surgeon X)— in order to be profitable for a hospital, surgery center, or surgeon, machine-assistance will need to bring in either more patients, which a facility or practice must be able to accommodate, or patients with a better payor mix.

Cost and profitability are both different than cost-effectiveness. Whether or not machine-assisted techniques are cost effective depends on improved reconstruction longevity and/or clinically meaningful improvements in functional outcomes. But if a technology has not been proven effective [3, 7, 13], by definition it cannot be cost effective.

A final area to explore is considering machine-assistance as a training tool. Even if these robots and navigation systems do not deliver more-accurate surgery, more-durable reconstructions, or happier patients, might they still make us better surgeons? In terms of reducing operating room errors and teaching trainees, and even improving experienced surgeons, the answer appears to be “yes” [6, 10]. But if we as surgeons are being honest with ourselves, we don’t know the answer to this question, either. Reverse the question—can these technologies make us worse surgeons? The answer to this question conversely, and somewhat perversely, may have an affirmative answer as well [9]. We must consider the potentially higher rate of unintended complications [3], the known issues such as femoral notching [7], and the possible deterioration of surgical skills due to overreliance on the machine(s). At a minimum, it makes sense that having a foundation in nonmachine-assisted procedures will remain important for residents and staff surgeons alike. A further, largely unanswered question at this point, is whether (having gained this foundation), machine assistance can improve our surgical technique (similar to surgical simulation) such that the machine itself becomes unnecessary?

While not an active user myself, I have directly used or assisted on procedures involving a variety of machine-assisted techniques. And, to be clear, I do not think that machine-assisted surgeries are going away; however, given that our best current evidence does not support the use of these technologies, their persistence in the near term likely has more to do with market pressures, advertising and misperceptions of evidence by surgeons and patients alike as any actual benefit. That is, given current technologies and evidence, they probably should go away until there is compelling evidence that they are better and can make us better. Newer, “better” systems would ideally be restricted to cites studying these systems to prove that they are better, worthwhile, and cost effective. If we are, collectively, to continue down the path of machine assistance, we need to keep working to improve them, monitor the outcomes, be wary of the next big thing (particularly given the long tail of arthroplasty longevity studies), and be cognizant of developing, growing, and maintaining our own basic skill sets in the process.