Primary Knee Replacement: Management and Alternatives
The 8th Annual North American Hip and Knee Symposium
January 20-22, 2000
Course Directors: Robert B. Bourne, MD, FRCSC and Cecil H. Rorabeck, MD
Reviewer: Richard S. Laskin MD[Medscape Orthopaedics & Sports Medicine 4(2), 2000. © 2000 Medscape, Inc.]
Primary Knee Replacement: Management and Alternatives
The knee absorbs joint reactive load over a wide area of thick cartilage. In degenerative arthritis, which is the condition most commonly necessitating total knee replacement, the quality and quantity of the articular cartilage is diminished. As wear continues, there is more bone-on-bone contact, resulting in increasing pain. At this Symposium, innovative treatments and management strategies for total knee arthroplasty (TKA) were discussed.
The session began with a review of several papers describing the nonoperative treatment of patients with osteoarthritis. Dr. Johan Bellemans, of the University of Leuven in Belgium, presented an overview of the use of chondroitin sulfate and glucosamine for patients with osteoarthritis. A review of the published data indicates that these oral substances were superior to placebo in relieving pain, but were less effective than the nonsteroidal anti-inflammatory drugs. The best results were seen in patients with very early osteoarthritis; the results diminished as the severity of the osteoarthritis increased.
The use of injectable hyaluronic acid substitutes in the treatment of osteoarthritis was also discussed. A comparison of the various types indicates that those with the highest molecular weight and highest amount of cross-linking seemed more efficacious, but they also had a higher incidence of inflammatory reactions when compared with lower-molecular-weight, less cross-linked drugs. The propensity for an inflammatory reaction after injection did not appear to be related to patient age or diagnosis. Again, as was the case with chondroitin sulfate and glucosamine, the best results were observed in patients with early osteoarthritis; the results in patients with more advanced arthritis are much less predictable.
Two alternatives proposed for formal total knee replacement are high tibial osteotomy and unicompartmental replacement.
High tibial osteotomy. Dr. Robert Bourne of the University of Western Ontario in London, Ontario, reported a long-term survivorship follow-up of patients who had undergone high tibial osteotomy. A total of 106 tibial osteotomies were performed by 2 surgeons between 1976 and 1988 in patients who had primary degenerative osteoarthritic changes in the medial compartment of the knee and a varus deformity. Survivorship analysis revealed that the probability of survival was 73% at 5 years, 51% at 10 years, 39% at 15 years, and 30% at 20 years. Complications such as delayed union, nonunion, intra-articular fracture, and infection occurred in up to 10% of patients. Of the 106 high tibial osteotomies, 61 required conversion to a total knee and, of those, 11 (18%) had already required a second total knee replacement. They found that those patients who preoperatively had a lateral thrust or range of motion less than 120 degrees did not do well postoperatively; neither did those with a small body mass. Neither did the thinner patients. Survivorship of those with more than 120 degrees of flexion, no lateral thrust, and age younger than 50 was 95% at 5 years, 80% at 10 years, and 65% at 15 years. These investigators recommended the ideal candidate as being younger than 50 years of age and having a body mass index greater than 25 kg per square meter, isolated medial compartment osteoarthritis, no lateral thrust, a varus deformity less than 10 degrees, and a range of motion greater than 120 degrees.
Dr. Cecil Rorabeck, also from the University of Western Ontario, presented the results of patients with unicompartmental replacement who were followed for 10 years. They found that failures, when they did occur, occurred early (= 5 years), and that, in order for this procedure to function successfully, patients required an anterior cruciate ligament that was intact. Ten-year survival in this study was 85%. Total knee arthroplasty. Dr. Richard Laskin, of The Hospital for Special Surgery, discussed the key design criteria for a total knee replacement prosthesis for the year 2000. He cited the following as important factors in femoral component design: an anatomic right and left femoral component; a proper trochlear groove that is deep, long, and funneled; and prostheses that accommodate either retention or removal of the posterior cruciate ligament. The tibial component should be asymmetrical to maximize tibial coverage and to provide a proper capture mechanism for the polyethylene to prevent cold flow and lower surface wear. Because of the unknown and various scenarios that may occur at the time of surgery, it is important to have polyethylene inserts available to accommodate PCL retention, posterior stabilization, and the use of a conforming implant. Both Dr. Laskin and Dr. Leo Whiteside of St. Louis, Missouri, discussed the crucial importance of proper configuration of the trochlear surface of the femoral component to aid in patella tracking and to minimize contact stresses in that compartment of the knee. Dr. Whiteside suggested that with the proper femoral component and proper trochlear groove, one could consider not resurfacing the patella. He substantiated this suggestion with data demonstrating that 90% of his cases had excellent results. The concept of a medial pivot knee was also mentioned, and opinions were expressed as to whether it had any clinical relevance in the total knee patient in whom the ACL was removed. There was no consensus other than that further study is necessary to determine its clinical relevance. Polyethylene. Dr. Kevin Weaver of Memphis, Tennessee, presented the current status of cross-linked polyethylene and its use in total hip and total knee replacement (Figure 1). He presented data demonstrating that polyethylene that was irradiated in air and then allowed to age had inferior wear properties compared with polyethylene that was nonsterilized or treated by ethylene oxide sterilization. When polyethylene was gamma-radiated in air, numerous free radicals were formed which could cause breaks in the polyethylene chains. When polyethylene was irradiated in an inert atmosphere, oxidation did not occur; however, some cross-linking did occur. The optimal method to obtain cross-linking would be first to irradiate the polyethylene and then melt it to eliminate the free radicals. The material is then machined into a final component form.
Figure 1. This comparison chart shows the amount of bone resected for the femoral box in various total knee PS prostheses. The image is used with permission of Dr. Laskin.
A major concern regarding cross-linked polyethylene was discussed: the trade-off between wear properties and strength. The more polyethylene is cross-linked, the greater its wear properties — but its strength diminishes. Since the articulations of the hip and knee differ, one could not extrapolate results seen in the hip with cross-linked polyethylene to results in the knee. Theoretically, a weaker material might not be appropriate in the knee, because there is more localized loading and lesser congruity. Dr. Weaver concluded that all cross-linked polyethylenes are not the same and that trade-offs would have to be made between strengths and wear to find the optimal material.
All Posterior Stabilized Knees Are Not Alike
Dr. Laskin discussed the configuration of the femoral box and the amount of bone removed with the different posterior stabilized femoral components presently available. The volume of bone removal varied, ranging from a low of 5.9 cm3 to a high of 13.5 cm3. With increasing amounts of bone removal, there is increasing concern that fracture of the distal femur may occur. To decrease the probability of fracture, the posterior stabilized post must be correctly placed at a proper height. A post that is too anterior impinges on the femoral component when the knee is fully extended or slightly hyperextended. A post of insufficient height and improper placement could allow the femoral component to jump over the post and dislocate.
A different problem, known as patella clunk, occurred with the use of the initial Insall-Burstein II, a posterior stabilized knee design. It has not been seen with any other design of this genre. The initial Insall-Burstein II had a very short trochlear flange and a sharp change in radius of curvature between the trochlear and the distal femur. Any soft tissue that developed near the junction of the patella and the quadriceps tendon could get caught in this area. This design problem was resolved in later versions of the implant.
Dr. Laskin discussed the results of a study using an ultra-conforming implant, rather than a posterior stabilized polyethylene. He presented data that demonstrated that range of motion, ability to reciprocate stairs, and stability were comparable. He noted that any posterior stabilized design must be used in conjunction with proper ligament balancing in order to have a stable knee in all degrees of flexion and extension.
Biomechanics of the total knee. Dr. Scott Banks of West Palm Beach, Florida, discussed the motion of total knee replacements during gait and presented the results of his study performed in conjunction with Dr. Steven Haas, Dr. Richard Laskin, and Dr. James Otis at The Hospital for Special Surgery. Twenty patients who had undergone total knee replacement and had excellent Knee Society scores were studied. He found that treadmill gait was characterized by a slightly prolonged stance phase and a slightly reduced loading response in knee flexion. The cruciate retaining knees exhibited some anterior femoral sliding and external rotation during knee flexion, while the cruciate substituting knees exhibited femoral rollback and internal rotation during flexion. Neither group exhibited large condylar translations or rotations during the stance phase of gait. Most interesting was the fact that there was no evidence of condylar liftoff in the fluoroscopic analysis of patients with this particular prosthesis (Genesis II).
Mobile Bearing Knees
The final paper was presented by Dr. Robert Bourne and discussed whether mobile bearing knees had a place in today’s orthopaedic practice. Dr. Bourne drew upon his experience with the SAL prosthesis, which had a rotating and translating bearing, as well as results described with the LCS prosthesis, which had just a rotating bearing. In theory, the rotating movable bearings are highly congruent and therefore should have decreased wear and a decreased incidence of loosening. When he reviewed the results obtained with the SAL prosthesis, he found that one third of the patients had proper rotation and translation, one third tended to rotate only, and one third had a paradoxical translation.
He concluded that a fixed bearing was appropriate for most patients. A rotating platform could be considered in patients with a life expectancy of more than 15 years, a deformity less than 15 degrees, and in those who would not require a medial or lateral release. He also suggested that most patients would do best with a rotating or translating bearing, so he would save the rotating type of component only for patients with gross PCL deficiency, those with a prior patellectomy, or possibly those with severe inflammatory arthritis.
The final conclusion was that all rotating platform knees are not the same; they differ in terms of rotation and translation, axis of curvature, and contact stresses. Rotating platforms present a potentially new design for total knee replacements; they may increase longevity and function. However, before widespread use of these components occurs, more clinical studies are needed to determine who would benefit most from them and what those benefits are.
Surgical Technical Tips for Primary Total Knee Replacements
Dr. Michael Ries, of the University of California in San Francisco, discussed externally rotating the femoral component in total knee replacements. Although external rotation of the femoral component does balance the flexion space, there are potential problems that may occur, such as notching of the anterior lateral cortex and incongruity in rotation of the knee in extension. It was suggested that an implant with asymmetrical posterior condyles be used, so that it filled the trapezoidal flexion space without resulting in external rotation of the implant.
Both Dr. Ries and Dr. Laskin discussed long-term results using this type of implant. They experienced a lateral release rate of only 2%; all of these patients had pre-existing patellar subluxation. Dr. Laskin expanded on this by describing a study that evaluated the configuration of the flexion space postoperatively. The results demonstrated that in approximately 96% of cases, the space was adequately filled by this trapezoidal type of implant. It was determined that the presence of congruity of these knees both in flexion and in extension was an important factor and that this method of balancing the flexion space had many advantages over the traditional method of externally rotating the femoral component.
Dr. Laskin also discussed balancing the flexion and extension spaces and the surgical techniques needed to do this. He described the surgical technique of measured femoral resection and measured tibial resection combined with flexion and extension space balancing. At the completion of this surgical procedure, the flexion and extension spaces must be completely filled by their respective implants and should be equal in size or within 1 to 2 mm of each other in size. In situations where the flexion space is found to be smaller than the extension space with the trial implants in place, the flexion space should be enlarged by downsizing the AP diameter of the femoral component used. In situations where the flexion space is larger than the extension space, he recommended resecting a bit more femur and inserting a thicker implant to fill both the larger flexion and extension spaces.
Improving Range of Motion
Dr. Whiteside presented a paper on surgical tips to increase range of motion during knee replacement, such as (1) ensuring proper tracking of the patella by avoiding overstuffing of the patella femoral groove, and (2) removal of the posterior femoral osteophytes.
These osteophytes can limit both flexion and extension of the knee. They are easily removed by a curved osteotome passed around the posterior femoral condyles after the condyles themselves have been resected.
Several speakers described how to elevate some of the soft tissue from the posterior aspect of the femoral condyles, even in those knees that do not have osteophytes. This maneuver alone tends to increase the potential flexion space posteriorly and will allow further flexion.
Dr. Whiteside expressed his belief that posterior sloping of the tibial component increases flexion as well. Likewise, avoidance of oversizing of the femoral component was important to optimize flexion.
Dr. Cecil Rorabeck presented techniques to expose the tight knee, especially during revisions. His algorithm calls for performing a lateral release; if this is ineffective, a rectus snip should be attempted. If the patella still cannot be everted for adequate exposure, a formalized quadriceps turndown or a tibial tubercle osteotomy should be attempted.
Dr. Rorabeck said that he would perform the tibial tubercle osteotomy rather than the formalized quadriceps turndown because of the lower complication rate. Also, with a formalized quadriceps turndown, there is the potential for avascular changes in the patella. He described the method of subsequent fixation of the tubercle osteotomy using 3 double-stranded metal wires, one going through the tubercle fragment and 2 going around it.
Correction of soft tissue deformities. The final 3 papers related to the correction of soft tissue deformities during total knee replacement. Dr. Jan Victor of Brugge, Belgium, discussed the varus knee and the necessity for a sequential release of the structures that were tight. For those knees with a varus contracture in extension, the tight structures that would require release included the semimembranosis tendon, the posterior oblique ligament, and the posterior medial capsule.
Dr. Victor stated that, at times, there will be laxity on the lateral structures of the knee associated with the tight varus deformity. The choices in this situation are to accept the lateral laxity, continue to release the medial side, elongate it to the length of the lateral laxed structures, or advance the lateral structures to tighten them. He suggested that one should accept some mild laxity of the lateral structures in a severe varus knee after the main portion of the contracture is released.
Dr. Kelly Vince of the University of Southern California described his approach to the valgus knee and the selective releases that he performed for this. Although the valgus deformity was accentuated by the hypoplasia of the lateral femoral condyle both distally and posteriorly, he suggested that the disabling pathology was on the medial side, because that was the area that required advancement or tightening. Therefore, he suggests using the standard medium parapatellar approach for valgus knees rather than the lateral parapatellar approach. He was also concerned that use of the lateral parapatellar approach could jeopardize the extensive mechanism, because it required elevation of a portion of the patella tendon for a patella diversion. He expressed concern that closure was less secure with this approach when compared with the standard medium parapatellar approach.
After the correct static alignment is achieved, he suggested continuing a few degrees beyond neutral for the extremely severe valgus knee. Because the posterior condyles are often asymmetrically eroded, he believes that they are a poor landmark to use for femoral component rotation. He releases the ligaments by placing the knee in extension, placing 2 laminar spreaders to distract the knee, and then selectively releasing, with a pointed 11 blade, those tight structures that can be palpated. A lateral parapatellar release should be used if the patella is maltracking before surgery. However, the mere presence of a valgus deformity should not lead to an increased incidence of lateral release procedures.
The final paper of this session was presented by Dr. Johan Bellemans. He discussed the various methods used to release flexion contracture, including posterior stripping of the femur, removal of osteophytes, and balancing of the medial and lateral soft tissues. He stressed the importance of releasing tissues on the medial and lateral side that are tight in extension, namely the posterior oblique ligament and the semitendinosis on the medial side and the iliotibial band and posterior lateral capsule on the lateral side. He suggested that overresection of the distal femur should not be used unless soft tissue balancing is not successful.
Issues in Revision Total Knee Arthroplasty
Determining the Cause of the Failure
In this session, Dr. Charles Clark of the University of Iowa discussed the algorithm for evaluating a patient with a failed total knee replacement. Infection should be ruled out in all failed knees. The consensus is that aspiration is the most reliable method of determining the presence of infection and that bone scanning is the least reliable method.
Instability of the knee is a common cause of postoperative failure that needs to be assessed not only clinically but also by performing the appropriate stress x-rays. It is important for the surgeon to review the initial postoperative film after the index arthroplasty and compare it with the most recent film to determine whether there are any changes in component position, alignment, or bone quality. If the surgeon only reviews the most recent radiograph and does not compare it with the initial one, small changes will not be identified.
Dr. Cecil Rorabeck expanded on Dr. Clark’s presentation by discussing the common failure of mechanisms in TKA. Two of the more prevalent causes of failure are osteolysis and instability of the knee; aseptic loosening has becoming less common. He stressed that at the time of the revision, if he could restore soft tissue stability, he would use a posterior stabilized or, at most, a varus valgus constrained implant. He said he would reserve the use of an articulated implant for those occasional cases when he could not restore stability with soft tissue procedures. In his own practice, Dr. Rorabeck stated that he is able to perform a revision using a posterior stabilized implant approximately 55% of the time. The remaining knees are reconstructed using a varus valgus constrained implant.
Endosteal referencing. Dr. Michael Ries, of University of California San Francisco, discussed the use of endosteal referencing in performing revision total knee replacement. The intramedullary reams and intramedullary trials are used to position the cutting blocks in the proper position. He suggested that endosteal referencing of the femoral and tibial bone cuts and implantation of the components with press fit stems was a reliable and reproducible method to produce satisfactory alignment for the majority of revision total knee replacements. However, because this technique is based on the anatomy of the intramedullary canal of the femur and tibia, full-length radiographs should be obtained prior to surgery to determine if any bowing is present; this would affect alignment.
Bony defects. Dr. Richard Laskin discussed the management of bony defects at the time of revision total knee replacement. He stated that for small areas of bone defects, lowering the resection line on the tibia or raising the resection line on the femur might be sufficient. For larger contained defects, the use of bone or cement proved to be satisfactory.
Bone used for filling of such defects could either be autograph bone or allograft bone. Local autograph bone, although plentiful in primary knees, is somewhat deficient in revision knees. If a sufficient amount is available, it obviates the problem of a second donor site, which is potential disease transmission or patient reaction. Allograft bone, if necessary, could either be structural (femoral head, iliac crest, or entire distal femur and proximal tibia) or morselized. He described the technique of filling cavitary defects with morselized bone and seating the prosthesis on the rim of the tibia, supplemented by a long-press-fit intramedullary stem. This method is somewhat analogous to compaction bone grafting used in revision total hip replacement.
Dr. Laskin also discussed the Whiteside method of peripheral filling of defects with cancellous bone, which entails covering them with a soft tissue flap. This method works well only with a well-fitting press-fit femoral stem and support of the implant on the plateau or condyle on the opposite side of the defect. He discussed the use of bone-inductive agents such as bone morphogenic protein (BMP) or human growth hormone as well as scaffolding products such as Grafton to supplement the bone graft. The algorithm for filling defects is that if the defect is small and contained, use morselized bone graft. If the defect is large and peripheral, choose a prosthetic augment, and if it is cavitary, fill it with morselized, compacted bone graft supplemented by some type of scaffolding material.
Maximizing the revision procedure. Dr. Leo Whiteside reviewed the difficult problem of how to achieve correct implant size and ligament balance during revision total knee replacement. Due to severe bone destruction, the distal surface of the femur is no longer a reliable landmark to determine alignment and rotation. He uses the intramedullary canal and the anterior femoral cortex for these references. He believes that positioning of the joint is best accomplished by using the distal femoral resection line as a guide. He suggested conserving as much bone stock as possible in the distal femur proximal tibia, then inserting the femoral trial component with the thickest surface and a plastic tibial component thick enough to stabilize the knee to 90 degrees of flexion.
If the knee can extend fully and is stable, the surgeon can feel confident that the femoral component thickness is appropriate, Dr. Whiteside noted. If full extension cannot be achieved, less distal femoral thickness is required. This requires resecting slightly more femur or using a thinner distal femoral augment. On the tibial side, he suggested that the fibular head may occasionally be higher than the lowest surface of the tibia, and partial resection of the upper portion of the fibula would be necessary to achieve a flat upper surface. The fibula tibial joint should then be excised and packed with bone graft.
Septic arthritis. Dr. Kelly Vince reviewed the current treatment concepts for septic arthritis. One of these concepts is that no antibiotics should be given until the diagnosis is made; another is that aspiration should be performed to make the diagnosis. For an acute infection, treatment with just a debridement is possible; however, for an established infection, treatment with a 2-staged protocol is indicated.
The gold standard for infected total knee prostheses is a 2-stage revision. The results of Goldman, Scuderi and Insall and Hirakawa reveal a survivorship of approximately 75% to 80% at 10 years with a 2-stage revision. Hirakawa found that for high-variance organisms (methicillin-resistant Staphylococcus aureus [MRSA]), this survivorship diminished to approximately 65%.
In reviewing a series of articles, various risk factors were found to be associated with infection of a total joint. Bierbaum demonstrated an increased incidence of infection when banked blood was used. Winiarsky, Barth, and Lotke found a higher incidence in obese patients. Waldman, Mont, and Hungerford reviewed 3500 total knee replacements and found that there were 62 late infections, 7 of which were strongly associated with dental procedures. The dental procedures were considered extensive in nature, and 56% of these patients had associated systemic risk factors. None of these patients received antibiotic prophylaxis prior to their dental procedure. Present recommendations include the administration of first-generation cephalosporin, given 1 hour preoperatively and 8 hours postoperatively.
The American Dental Association and the American Academy of Orthopaedic Surgeons (AAOS) Advisory Statement for the Antibiotic Prophylaxis for Dental Patients with Total Joint Replacements (http://www.ada.org/adapco/jada/9707/pro-01.html) states that “bacteremias can cause hematogenous seeding of total joint implants both in the early postoperative period and for many years thereafter. It would appear that the most critical period was up to 2 years after joint replacement. In addition, bacteremias might occur in the course of normal daily life and concurrently with dental or medical procedures. It was likely that many more oral bacteremias were spontaneously induced by daily events than were by dental treatment. There is no scientific evidence to support the position that antibiotics prophylaxis to prevent hematogenous infection was required prior to dental treatment in patients with total joint prosthesis on a routine basis.”
The AAOS did state, however, that prophylaxis should be considered for patients with total joint replacement who are immunocompromised or immunosuppressed (inflammatory arthropathies or disease or drug-induced suppression). Other patients who may be at increased risk for hematogenous total joint infection are patients with insulin-dependent diabetes or previous prosthetic joint infections, those who are malnourished, those diagnosed with hemophilia, and patients who have undergone total joint arthroplasty within the past 2 years. It is speculated that the highest incidence of infection occurs with a procedure such as dental extraction, periodontal surgery, dental implantation, root canal, and prophylactic cleaning of teeth, or with implants where bleeding is anticipated.
Reviewer’s Comments: From a practical standpoint, antibiotic prophylaxis should be given to any patient within 2 years after having a total joint arthroplasty, if the patient is to undergo any of the high-risk procedures listed above. Subsequent to that point, the individual surgeon must balance the risk/benefit ratio with the costs of using antibiotics and their potential side effects.
Dr. Vince discussed Dr. Barrack’s work on the value of preoperative aspiration before total knee revision, published in Clinical Orthopaedics in 1997. In that article, the positive predictive value of aspiration was 85% and the negative predictive value was 84%. Preoperative aspiration should be performed before total knee revision. Only then should the patient receive antibiotics. Although there has been some interest in the use of polymerase chain reaction testing, this technology is not widely available for the diagnosis of infected joints. Intraoperative frozen biopsies have 97% specificity.
Dr. Vince also discussed 2 articles describing prolonged suppressive antibiotic therapy combined with multiple irrigation debridement and retention of components in infected total knees. In a series of 18 patients by Segreti, 15 of the 18 patients had good results. In a second paper published by Mont, 71% had no further sepsis after this treatment. However, it should be noted that in the Mont study, the patients developed their infections less than 30 days after total knee surgery.
Revision TKR. The final presentation, by Dr. Robert Bourne of University of Western Ontario, Dept of Orthopaedics, London Ontario, was a retrospective review of revision total knee replacement. Revision results published in the 1980s demonstrated satisfactory outcomes in approximately two thirds of cases. Dr. Bourne presented his results of 46 revisions performed with a Genesis I prosthesis during the early 1990s.
Exposure was possible with a standard approach in 38 of these cases. There was no need to use a rectus snip or tibial tubercle osteotomy. Results showed a 9% reoperation rate, including 1 case of sepsis, 1 case of marked poly wear, 1 periprosthetic fracture, and 1 patient who eventually required an arthrodesis. At 5 to 9 years, the mean knee society rating score was 147, and the mean flexion was approximately 105 degrees. All of these revisions were performed with implants using press-fit intramedullary stems and cemented components; metal augments rather than bone grafts were used to fill in bone deficits.
Goldman RT, Scuderi GR, Insall JN. 2-stage reimplantation for infected total knee replacement. Clin Orthop. 1996 Oct;(331):118-124.
Hirakawa K, Stulberg BN, Wilde AH, Bauer TW, Secic M. Results of 2-stage reimplantation for infected total knee arthroplasty. J Arthroplasty. 1998;13(1):22-28.
Bierbaum BE, Callaghan JJ, Galante JO, Rubash HE, Tooms RE, Welch RB. An analysis of blood management in patients having a total hip or knee arthroplasty. J Bone Joint Surg Am. 1999;81(1):2-10.
Winiarsky R, Barth P, Lotke P. Total knee arthroplasty in morbidly obese patients. J Bone Joint Surg Am. 1998;80(12):1770-1774.
Waldman BJ, Mont MA, Hungerford DS. Total knee arthroplasty infections associated with dental procedures. Clin Orthop. 1997 Oct;(343):164-172.
Barrack RL, Jennings RW, Wolfe MW, Bertot AJ. The Coventry Award. The value of preoperative aspiration before total knee revision. Clin Orthop. 1997 Dec;(345):8-16.
Segreti J, Nelson JA, Trenholme GM. Prolonged suppressive antibiotic therapy for infected orthopedic prostheses. Clin Infect Dis. 1998 Oct;27(4):711-713.
Mont MA, Waldman B, Banerjee C, Pacheco IH, Hungerford DS. Multiple irrigation, debridement, and retention of components in infected total knee arthroplasty. J Arthroplasty. 1997;12(4):426-433.