There has been increased interest in the design and use of elbow replacement in the last several years with a number of new devices appearing on the market. This chapter focuses on what is known and documented. It thus serves to complement the description of other implants, especially the newer designs, by distinguishing what is established from what is hypothesized. Today, rather than characterizing devices as resurfacing or hinge, constrained or unconstrained, the most appropriate terminology is that of a linked (coupled) or unlinked (uncoupled) device. The reason for this is that some devices can be very highly constrained because of the configuration of the articular surface, yet they are not mechanically linked. Other devices can be mechanically linked but provide significant play, laxity, or less constraint of the linked articulation. Further, the amount of bone removed for unlinked implants may be considerable and thus not accurately described as resurfacing or conservative.

The potential advantages of an unlinked implant are stated in chapter 22. In general, the perceived value includes an opportunity for a greater amount of polyethylene at the articulation, allows for the option or requirement to replace the radial head, and is more anatomic in concept and thus might theoretically provide a more functional joint. The unproven value of the radial head is a potential disadvantage in some cases. The implications of increased technical difficulty and potential complications associated with the greater demands placed on the accuracy of insertion are other concerns regarding the new devices. At this time, the absence of outcome data requires the use of these implants to be based on theoretic advantages. Finally, prior designs for which data do exist have all suffered from an instability rate of around 5%, a concern that must be addressed by technique or design of the uncoupled device.

The advantage of a linked device is that the joint is rendered immediately stable. This means a broad spectrum of pathology may be addressed. It also implies the potential for a relatively simple procedure of insertion. Hence, in our view there are a significant number of advantages to such a design. These features imply that a rather broad spectrum of conditions can be treated by a relatively large and expanding number of surgeons.

Depending on the design, the amount of polyethylene at the articulation may be limited with the linked implant, giving rise to the potential for wear and osteolysis. Some linked implants may require more bone to be removed than the unlinked designs, causing difficulty at revision. If the condyles are sacrificed, resection arthroplasty, as a salvage procedure, is unpredictable. Because the forces across the joint are transmitted to the implant/bone interface, linked or unlinked designs that are more constrained have the potential for increased humeral component loosening.

The indications for the linked implant are identical to those of unlinked devices and are consistent with the indications for other joint replacements. Specifically, elbow dysfunction that is considered a major limiting factor in the patient’s day-to-day existence justifies consideration of linked implant replacement. This dysfunction typically is the result of pain, but at the elbow the dysfunction can also be attributed to instability or even stiffness. The Mayo experience with the use of a single design since 1982 in more than 1,000 patients reveals the most common indication for total elbow arthroplasty is rheumatoid arthritis (Table 21-1). As noted, the patient with rheumatoid arthritis represents less than 40% of the 1,021 procedures that have been performed at the Mayo Clinic. This is in contrast to most reports regarding elbow replacement in which the patient with rheumatoid arthritis accounts for 80% or more of these procedures. The reason for this is that the linked implants do not require osseous or ligamentous integrity, whereas the unlinked devices do require these anatomic features to be present for implant stability. The indications for a linked implant are thus rather broad and include those with ligamentous deficiency and those with traumatic conditions, which include acute fractures, established posttraumatic arthrosis, distal humeral nonunion, posttraumatic ankylosis, and posttraumatic instability. This spectrum of potential candidates reduces the relative frequency of replacement for rheumatoid arthritis. Finally, revision of a failed total elbow arthroplasty is most reliably managed by a linked device because bone and ligaments are almost always compromised in this setting. The optimum design goal is that of a reliable “off the shelf” implant capable of addressing all these pathologic presentations.

The contraindications for a linked arthroplasty are those for any total elbow—that is, the presence of infection or the lack of adequate motors to gain the anticipated efficacy of the implantation. There are no unique contraindications to the linked implant compared to the unlinked one.

The literature reflects experience with limited semiconstrained/linked devices to be increasing in the last few years. One in-depth experience is that of the GSBIII design (Fig. 21-1). Gschwend and colleagues (1) report 59 procedures, of which 51 were for rheumatoid arthritis. Of the 36 elbows available for long-term assessment at a mean of 13.5 years, 88% of these implants were still in situ. The
authors report a 9% incidence of infection and a similar 9% incidence of loosening of one or more components.

A French experience with 20 GSBIII implants for rheumatoid arthritis (2) reports subjective satisfaction in 19 of 20 elbows (95%) a mean of 3 years after surgery. This experience is in contrast to that of Schneeberger and colleagues, which documented that 9 of 14 patients with a mean 5.8 years after surgery had rheumatoid arthritis, but only 7 of 14 procedures were considered satisfactory (3). Loosening was the major problem occurring in 29% of these patients.

In general, and especially in Mayo’s experience, the results of the linked implant have been satisfactory not just for the patient with rheumatoid arthritis but also for the posttraumatic conditions and even in revision settings (4, 5, 6, 7, 8, 9, 10, 11, 12). The overall success is approximately 90% at 5 years.