Orthopaedic surgeons report variable success with unicondylar knee arthroplasties performed during the 1980s and 1990s.1,2,5-7,9,12,14-17,19,20,22,24,25,28,29,32,34-38 As an example, reports from one national registry estimate the cumulative rate of revision at 10 years after unicompartmental arthroplasties performed within this timeframe to be as high as 20% in some regions and as low as 10% in others.38 Factors that could explain the inconsistent longevity of unicondylar arthroplasties performed during the previous two decades include the age of the patient at surgery, the weight of the patient, initial thickness of the tibial component, features intrinsic to the implant design used (articular surface curvature and constraint, type of polyethylene used, available tibial component thickness variations, characteristics of implant fixation surface), polyethylene shelf age (most unicondylar bearings placed during this era were sterilized with gamma radiation in air, and the material properties of polyethylene tibial bearings sterilized with this method deteriorate with time),10,11 and surgeon factors (operative experience, philosophical perspectives on implant and postoperative knee alignment).
Based on considerable evidence, we presumed younger patient age,9,17,18,29,30 increased weight,3,15,17,18,22 use of a thinner tibial component,15,22,23,27,29 and a longer polyethylene shelf age9,24 are key factors that adversely affected survival of medial unicondylar arthroplasties performed during the 1980s and 1990s. Additionally, we speculate some unicondylar systems have had better survival than others. We do not believe gender has influenced arthroplasty survival.30 In confirming the evidence, we intend to answer the question puzzling many knee reconstruction surgeons today-why have the results of this operation been so variable with unicondylar systems used during the preceding two decades?
MATERIALS AND METHODS
We retrospectively reviewed a consecutive series of unicondylar arthroplasties performed by one surgeon over a 15-year period. The lone outcome measure was arthroplasty survival free of component revision. Multivariate statistical analysis (proportional hazards regression) was used to isolate factors making revision more likely from those not influencing the risk of revision. Risk factors under evaluation included three related to the patient (age, weight, gender) and three related to the implants (design name, tibial component composite thickness, polyethylene shelf age).
One surgeon (GAE) performed 411 unicondylar arthroplasties in 323 patients as an index reconstruction in knees with medial compartment osteoarthritis (OA) from 1984 to 1998. Morbid obesity (subjectively defined) was considered a contraindication to unicondylar arthroplasty within this timeframe. A flexion contracture in excess of 15° was also a contraindication. Age and weight at arthroplasty averaged 67 years (range, 45-89) and 83 kg (range, 49-123), respectively (Table 1). Two hundred twenty-nine of the 411 arthroplasties were performed in men. All 411 arthroplasties were entered into the study; no knees were excluded on the basis of followup time. We searched the United States Social Security Death Index to determine if patients not known to have undergone revision died. When a patient died 3 or more years after their last followup, a relative was telephoned to verify if the patient had undergone revision arthroplasty. Living patients who were not examined in 3 years were interviewed by telephone to confirm survival of their arthroplasty and encouraged to continue with followup examination. Of 16 patients (16 knees) for which neither patient nor relative could not be located, nine patients (nine knees) had at least 5 years of followup.
On the preoperative erect anteroposterior (AP) radiograph, no knee had more than 10° of tibiofemoral varus, more than 10 mm of lateral subluxation of the tibia on the femur, or less than 4 mm of remaining lateral compartment joint space width. The surgeon performed a medial parapatellar subvastus or midvastus exposure. The anterior cruciate ligament (ACL) and posterior cruciate ligament (PCL) were intact by visual inspection in all cases. Grade III chondromalacia26 (area of cartilage fragmentation and fissuring > 1.25 cm in diameter) was accepted on the lateral tibiofemoral compartment articular surfaces. Grade IV changes26 (erosion of cartilage to subchondral bone) were accepted in the patellofemoral compartment if eburnation was isolated to just one side of opposing patellofemoral articular surfaces (complaints of anterior knee pain or pain on stair climbing were not considered in the decision to perform unicondylar arthroplasty). On average, the unicondylar knees were aligned in slight mechanical varus (postoperative tibiofemoral valgus averaged 4.5° ± 2.8°, mean ± standard deviation [SD]).
The surgeon used 12 fixed-bearing tibial component unicondylar designs from six manufacturers (prior literature reports detail features of six designs: Robert Brigham® [Johnson and Johnson Orthopaedics, Raynham, MA],12,17,23,27 Single Compartment Replacement® [Osteonics, Allendale, NJ],9,13 Miller-Galante® [Zimmer, Warsaw, IN],2,6,25,28,29 PFC® [Johnson and Johnson Orthopaedics],33 Duracon® [Howmedica, Rutherford, NJ],21,24 and Microloc® [Johnson and Johnson Orthopaedics]4,37). Four hundred two of 411 (98%) of the tibial components and 375 of 411 (91%) of the femoral components were fixed with polymethylmethacrylate (PMMA) cement. The initial thickness of the tibial components averaged 8.5 mm (range, 6-12.5 mm), and 87% (358 of 411) had a metal backing (polyethylene initial thickness of at the center of the metal-backed tibia averaged 6.1 mm with a SD of 1.4 mm and range of 4-10.5 mm). Three hundred sixty-nine of 411 (90%) of the tibial bearings were sterilized with gamma radiation in air. The mean polyethylene shelf age was 1.3 years (maximum, 5.3 years).
The survival time of each arthroplasty was defined as the number of elapsed years from surgery to the date on which it was last known in place. The proportion of surviving arthroplasties was evaluated up to the mean survival time of 9 years (SD, 4.5 years; range, 0.1-19.3 years) using the Kaplan-Meier method with component revision as the endpoint (SPSS 8.0 for Windows, SPSS, Inc, Chicago, IL).
Cox proportional hazards regression was used to examine whether revision was statistically associated with six factors: (1) age at surgery; (2) weight; (3) gender; (4) unicondylar system name; (5) tibial component initial composite thickness; and (6) polyethylene shelf age. The multivariate approach tabulated patient factors, implant factors, survival time (confirmed years in vivo), and survival status (revised or in vivo) for each case to determine the significance of each patient-related and implant-related factor while controlling for the other factors. We then simplified the regression by only including the factors associated with revision (per the SPSS Wald backward method) to determine an odds (risk) ratio increase associated with arbitrarily changing each variable. This was performed to estimate how the specified change would sway the odds of a given knee being revised when other risk factors were held constant. Significance was set at the p < 0.05 level.
Because the Single Compartment Replacement® design had been used in 148 of the 411 knees (and we had already reported outcomes of a series of Single Compartment Replacement® knees including 89 arthroplasties in the current study9), the survival and proportional hazards regression analyses were also performed for the remaining 263 cases to verify our statistical findings had not been skewed by the prevalence of this one implant system.
Once we had identified which variables were statistically associated with the event of revision, we examined the relative improvement in survival when these variables had been controlled (especially those that could be prospectively controlled or altogether avoided today such as increased weight, thinner tibial component initial thickness, longer polyethylene shelf age).
Eighty-five patients (96 knees) underwent revision, including 18 (18 knees) who chose to have their knee revised elsewhere. Failure mechanisms were aseptic and usually isolated to the tibial component (ie, polyethylene fatigue wear or loosening, or a combination). Two knees were revised specifically because of narrowing of the lateral tibiofemoral compartment joint space. No revisions were performed because of patellofemoral symptoms.
Survival free of revision decreased from 96% at 3 years postoperatively to 93% at 5 years postoperatively and to 80% at 9 years postoperatively (95% confidence intervals [CI] 94-98%, 90-95%, and 76-85%, respectively) (Table 2).
Four of the five factors presumed to influence arthroplasty survival were associated (p < 0.01) with revision: younger patient age, thinner initial tibial component thickness, longer polyethylene insert shelf age, and unicondylar system name (Table 3). Implanting a 1 year older polyethylene bearing had a similar effect on survival to implanting a 2 mm thinner bearing or performing the operation in a 10 years younger patient; each change made the arthroplasty 2.1 to 2.6 times more likely to be revised. Of the two factors not associated with revision (weight and gender), increased weight had been presumed to make revision more likely.
We identified differences in survival between some of the more commonly used designs. Single Compartment Replacement® knees (Osteonics, Allendale, NJ) were more likely to be revised than Omnifit® knees (Osteonics) (p < 0.01), Miller-Galante® knees (Zimmer Inc, Warsaw, IN) (p = 0.02), or PFC® knees (Johnson and Johnson, Raynham, MA) (p = 0.05). Robert Brigham® knees (Johnson and Johnson) were more likely (p = 0.04) to be revised than Omnifit® knees (Osteonics).
Exclusion of the most prevalent design (Single Compartment Replacement® [Osteonics]) from the proportional hazards regression did not alter which factors were associated (p < 0.02) with revision; patient age, tibial component initial thickness, polyethylene insert shelf age, and unicondylar system name remained associated (p < 0.02) with revision. Exclusion of this design did improve 9-year survival to 85% (95% CI, 80-90%) though, as those 148 knees tended to have longer and more variable polyethylene shelf ages (mean ± SD, 1.7 ± 1.3 years).
When tibial component initial thickness was 7.5 mm or greater and polyethylene shelf age was less than 1 year (154 of 411 knees satisfied these two criteria), 5-year survival improved to 99% and 9-year survival improved to 94% (95% CI: 97-100% and 89-98%, respectively). In this subcohort, the mean patient age was 67 years (SD, 8 years), the mean tibial component thickness was 8.8 mm (SD, 0.9 mm), the mean polyethylene shelf age was 0.5 years (SD, 0.3 years), and 10 knee systems were represented (Robert Brigham® [Johnson and Johnson]: 62 knees; Single Compartment Replacement® [Osteonics]: 59 knees; Miller-Galante® [Zimmer]: 11 knees; PFC® [Johnson and Johnson]: 5 knees; Omnifit® [Osteonics]: 4 knees; PFC Sigma® [Johnson and Johnson]: 4 knees; Synatomic® [DePuy, Warsaw, Indiana]: 3 knees; AMK® [DePuy]: 3 knees; Mod III® [Richards, Orthez, France]: 2 knees; Keane® [DePuy]: 1 knee).
Due to concerns about loosening of all-polyethylene unicondylar (and total) knee tibial components used in the 1970s,7,15,16,18,19,22,34,36 a philosophical design shift occurred in the 1980s leading to widespread production of metal-backed tibial components. The large number of (fixed-bearing) metal-backed tibial components and gamma-irradiated-in-air polyethylene bearings in our series reflects prevailing North American tibial component design philosophies and implant sterilization practices of the mid- to late-1980s and early- to mid-1990s. As we had hypothesized, operating on a younger patient, placement of a thinner tibial component, implantation of a polyethylene bearing with a longer shelf age, and the choice of knee system considerably swayed the odds of revision of a medial unicondylar arthroplasty performed at our institution during this timeframe. We speculate these four factors may explain much of the variable success reported to date with unicondylar knee arthroplasty.1,2,5-7,9,12,14-17,19,20,22,24,25,28,29,32,34-38
We note several limitations. We could not, or did not, evaluate more subjective measures of success, such as outcome or satisfaction scores. Though our survival analysis employed an objective failure endpoint (component revision), the decision to revise an arthroplasty is subjective and influenced by the opinion and expertise of the surgeon, patient preference, and other factors (especially patient age and health status). Some knees may have undergone revision in the interval that elapsed after we last examined or surveyed the patient. Survival of many designs could not be compared due to variable number of cases performed with each. The influences of a metal backing on the tibial component and polyethylene sterilization method could not be studied owing to scarcity of all-polyethylene components and bearings sterilized with methods other than gamma-radiation-in-air. All tibial components were of a fixed-bearing design, and our findings may not be relevant to mobile-bearing devices. Though the importance of knee and component alignment factors were not examined (because a full series of preoperative and early postoperative radiographs was unavailable for one in four cases), the rate of failure (especially those related to polyethylene wear and tibial component loosening) may have been lessened if more complete correction of the preoperative varus deformity had been achieved. Data from the Swedish Knee Arthroplasty Register38 begs the question of whether other surgeons could reproduce the results of one performing a large volume of cases.31
We include a large sample size, a broad spectrum of component designs from multiple manufacturers, knowledge of polyethylene sterilization data (including shelf age), one surgeon (consistent case selection criteria and surgical technique), access to up-to-date confirmed implant survival information (no arthroplasty was assumed to be in situ), and use of a multivariate statistical method (proportional hazards regression) capable of sorting out the relative importance of many, often-interconnected risk variables.
In 2001, Robertsson et al30 reported on 12,282 unicondylar arthroplasties performed for osteoarthritis (OA) from 1988 to 1997, entered into the Swedish registry, and followed by mail survey for up to 10 years. Cox proportional hazards regression showed younger patient age was associated with revision while gender was not.30 Using similar methods, we found age influenced the risk of revision whereas gender did not. These findings echo those of the Swedish registry study and confirm two of our hypotheses. Implanting an arthroplasty into a ten years younger patient made it 2.1 times more likely to be revised; thus, a 55-year old surgical candidate would be at 4.4 (or 2.12) times greater risk of revision compared to a 75-year old candidate (presuming other risk factors constant). The inverse statistical association between patient age and revision may perhaps be inescapable, as the young will tend to live and remain active longer, placing greater physical demands on their tibial component than the elderly. At the least, one would hope the relative increase in revision risk associated with operating on a younger patient would decrease with continued evolution of unicondylar arthroplasty design and polyethylene processing technologies.
Contrary to what we presumed, patient weight (a variable not assessed in the Swedish registry) was not associated with revision. In 1993, Heck et al15 reviewed early Marmor-style unicompartmental knees featuring allpolyethylene unicondylar tibial components. They found higher revision rates in patients who weighed more than 81 kg (mean, 103 kg; range, 50-136 kg).15 Possible explanations include that, in our study, 87% of tibial components had a metal backing and the average patient weighed 83 kg (range, 49-123 kg).
Heck et al15 also found knees with a 6 mm tibial implant were more likely to be revised than patients who received thicker components. In our study, implanting a 2 mm thinner tibial component into a knee made it 2.6 times more likely to be revised. It seems logical a thinner gamma-irradiated-in-air bearing would usually be revised sooner than a thicker one when other factors influencing polyethylene wear were controlled for (such as patient age and polyethylene shelf age). Current recommendations of the United States Food and Drug Administration tend to restrict the minimum thickness of an all-polyethylene component to the range of 7 to 8 mm and a fixed bearing metal-backed component to 9 to 10 mm.8
We did find some designs had better survival than other designs (as we had hypothesized), but it is difficult to pinpoint the reasons with certainty. Disparate survival rates of unicondylar (and total) knee systems in registry reports13,30,38 (including institutional registry reports like ours) should be interpreted with caution, as better survival may not necessarily reflect better design principles. The knee series being compared may have different quality polyethylene (a combination of forming resin, fabrication method, sterilization method, and shelf age), tibial component thickness options, patient factors (age), and knee or implant alignment practices. Robertsson et al attributed better survival to surgeon familiarity with the operative technique.31 Our experience tends to prioritize polyethylene factors. Had the unicondylar systems in our single-surgeon series been used with the same frequency, same varieties of polyethylene, and similar polyethylene shelf age and thickness distributions, we wonder whether the remaining interdesign differences would be consequential enough to substantially influence survival rates.
Shelf age was the only polyethylene quality factor evaluated, as some manufacturers were unable to provide fabrication data. Holding other factors constant (for a given knee), implanting a 1 year older polyethylene bearing increased the likelihood of revision by a factor of 2.2 (implanting a two years older bearing would place the knee at 4.8, or 2.22, times greater risk of revision). Polyethylene fatigue failure has been isolated to gamma-irradiated-in-air bearings to date, but not to any specific system or those with metal backings (Fig 1). Because every bearing implanted before 1996 was sterilized with gamma-radiation-in-air, longer followup is needed to assess the benefits of using components sterilized with other methods (none of 42 knees with bearings sterilized with a gamma radiation in inert environment method or nonradiation method have undergone revision).
Some surgeons may use survival estimates from unicondylar series to decide or justify whether to perform the operation. When evaluating the contemporary role of unicondylar arthroplasty, we encourage surgeons to contemplate factors historically compromising implant survival and anticipate how current trends (eg, polyethylene sterilization with gamma radiation in an inert environment or nonradiation methods, use of less invasive surgical approaches, increased popularity of all-polyethylene tibial components inspired by less invasive surgery) may alter their importance. Implanting aged or thinner gamma-irradiated-in-air polyethylene components are historically important, yet underappreciated, factors limiting survival of unicondylar arthroplasties placed during the 1980s and 1990s. We suspect many medial unicondylar arthroplasty failures attributed to lateral tibiofemoral compartment pain or tibial component loosening may be rooted in the polyethylene failure tending to develop rather rapidly in knees with gamma-irradiated-in-air unicondylar bearings (especially since many revising surgeons have limited or no experience with unicondylar arthroplasty, have not had access to polyethylene sterilization data or understood its importance until recent years, have not had access to serial radiographs, and had no knowledge of the condition of the lateral compartment articular surfaces at the time of the initial arthroplasty).
With use of less aged polyethylene (typically sterilized with gamma-radiation-in-air) and tibial components thicker than 7 mm, we found the chances of a study arthroplasty remaining in place into the 10th postoperative year improved from 80% to 94%. This 94% value may make for a more reasonable preoperative expectation today for an experienced surgeon than would the 80% figure, given the scarcity of gamma-irradiated-in-air polyethylene tibial bearings and current standards regarding tibial polyethylene thickness minimums. We believe longer term success with the contemporary medial unicondylar arthroplasty will depend on the durability of the polyethylene, the technical quality of the knee reconstruction, and maintenance of tibial component fixation. Polyethylene issues to be resolved over the next thirty years of unicondylar arthroplasty include: (1) the relative advantages of using nonradiation versus radiation sterilization methods; (2) the optimal radiation dose to limit burnishing losses of irradiated bearings without compromising mechanical properties (strength, elongation) of the polymer; (3) whether or how best to eliminate free radicals generated when bearings are subjected to radiation; and (4) analysis of survival and bearing wear data to confirm the importance of shelf aging has been limited by these refinements in implant packaging and sterilization.
The authors wish to acknowledge the cooperation of Stryker (Mahwah, New Jersey), DePuy, a Johnson and Johnson Company (Warsaw, Indiana), Zimmer (Warsaw), and Smith and Nephew (Memphis, Tennessee) in furnishing polyethylene component sterilization information.
1. Ackroyd CE, Whitehouse SL, Newman JH, Joslin CC. A comparative study of the medial St Georg sled and kinematic total knee arthroplasties: ten-year survivorship. J Bone Joint Surg Br
. 2002; 84:667-672.
2. Argenson JN, Chevrol-Benkeddache Y, Aubaniac JM. Modern unicompartmental knee arthroplasty with cement: a three to ten-year follow-up study. J Bone Joint Surg Am
3. Barrett WP, Scott RD. Revision of failed unicondylar unicompartmental knee arthroplasty. J Bone Joint Surg Am
. 1987;69: 1328-1335.
4. Bartley RE, Stulberg SD, Robb WJ 3rd, Sweeney HJ. Polyethylene wear in unicompartmental knee arthroplasty. Clin Orthop Relat Res
5. Bergenudd H. Porous-coated anatomic unicompartmental knee arthroplasty in osteoarthritis: a 3- to 9-year follow-up study. J Arthroplasty
6. Berger RA, Meneghini RM, Jacobs JJ, Sheinkop MB, Della Valle CJ, Rosenberg AG, Galante JO. Results of unicompartmental knee arthroplasty at a minimum of ten years of follow-up. J Bone Joint Surg Am
7. Capra SW Jr, Fehring TK. Unicondylar arthroplasty: a survivorship analysis. J Arthroplasty
8. Center for Devices and Radiological Health, Food and Drug Administration. Class II special controls guidance document: knee joint patellofemorotibial and pemorotibial metal/polymer porous-coated uncemented prostheses; guidance for industry and FDA. Available at: http://www.fda.gov/cdrh/ode/guidance/1418.pdf
. Accessed April 21, 2005.
9. Collier MB, Engh CA Jr, Engh GA. Shelf age of the polyethylene tibial component and outcome of unicondylar knee arthroplasty. J Bone Joint Surg Am
10. Currier BH, Currier JH, Collier JP, Mayor MB. Effect of fabrication method and resin type on performance of tibial bearings. J Biomed Mater Res
11. Currier BH, Currier JH, Collier JP, Mayor MB, Scott RD. Shelf life and in vivo duration. Impacts on performance of tibial bearings. Clin Orthop Relat Res
12. Emerson RH Jr, Hansborough T, Reitman RD, Rosenfeldt W, Higgins LL. Comparison of a mobile with a fixed-bearing unicompartmental knee implant. Clin Orthop Relat Res
13. Gioe TJ, Killeen KK, Hoeffel DP, Bert JM, Comfort TK, Scheltema K, Mehle S, Grimm K. Analysis of unicompartmental knee arthroplasty in a community-based implant registry. Clin Orthop Relat Res
14. Hasegawa Y, Ooishi Y, Shimizu T, Sugiura H, Takahashi S, Ito H, Iwata H. Unicompartmental knee arthroplasty for medial gonarthrosis: 5 to 9 years follow-up evaluation of 77 knees. Arch Orthop Trauma Surg
15. Heck DA, Marmor L, Gibson A, Rougraff BT. Unicompartmental knee arthroplasty: a multicenter investigation with long-term follow-up evaluation. Clin Orthop Relat Res
16. Insall J, Walker P. Unicondylar knee replacement. Clin Orthop Relat Res
17. Kozinn SC, Marx C, Scott RD. Unicompartmental knee arthroplasty. A 4.5-6-year follow-up study with a metal-backed tibial component. J Arthroplasty
18. Kozinn SC, Scott R. Unicondylar knee arthroplasty. J Bone Joint Surg Am
19. Laskin RS. Unicompartmental tibiofemoral resurfacing arthroplasty. J Bone Joint Surg Am
20. Lindstrand A, Stenstrom A. Polyethylene wear of the PCA unicompartmental knee: prospective 5 (4-8) year study of 120 arthrosis knees. Acta Orthop Scand
21. Lindstrand A, Stenstrom A, Ryd L, Toksvig-Larsen S. The introduction period of unicompartmental knee arthroplasty is critical: a clinical, clinical multicentered, and radiostereometric study of 251 Duracon unicompartmental knee arthroplasties. J Arthroplasty
22. Marmor L. Unicompartmental arthroplasty of the knee with a minimum ten-year follow-up period. Clin Orthop Relat Res
. 1988;228: 171-177.
23. McCallum JD 3rd, Scott RD. Duplication of medial erosion in unicompartmental knee arthroplasties. J Bone Joint Surg Br
. 1995; 77:726-728.
24. McGovern TF, Ammeen DJ, Collier JP, Currier BH, Engh GA. Rapid polyethylene failure of unicondylar tibial components sterilized with gamma irradiation in air and implanted after a long shelf life. J Bone Joint Surg Am
25. Naudie D, Guerin J, Parker DA, Bourne RB, Rorabeck CH. Medial unicompartmental knee arthroplasty with the Miller-Galante prosthesis. J Bone Joint Surg Am
26. Outerbridge RE. The aetiology of chondromalacia patellae. J Bone Joint Surg Br
27. Palmer SH, Morrison PJ, Ross AC. Early catastrophic tibial component wear after unicompartmental knee arthroplasty. Clin Orthop Relat Res
28. Pennington DW, Swienckowski JJ, Lutes WB, Drake GN. Unicompartmental knee arthroplasty in patients sixty years of age or younger. J Bone Joint Surg Am
29. Perkins TR, Gunckle W. Unicompartmental knee arthroplasty: 3- to 10-year results in a community hospital setting. J Arthroplasty
30. Robertsson O, Knutson K, Lewold S, Lidgren L. The Swedish Knee Arthroplasty Register 1975-1997: an update with special emphasis on 41,223 knees operated on in 1988-1997. Acta Orthop Scand
31. Robertsson O, Knutson K, Lewold S, Lidgren L. The routine of surgical management reduces failure after unicompartmental knee arthroplasty. J Bone Joint Surg Br
32. Rougraff BT, Heck DA, Gibson AE. A comparison of tricompartmental and unicompartmental arthroplasty for the treatment of gonarthrosis. Clin Orthop Relat Res
33. Schai PA, Suh JT, Thornhill TS, Scott RD. Unicompartmental knee arthroplasty in middle-aged patients: a 2- to 6-year follow-up evaluation. J Arthroplasty
34. Scott RD, Cobb AG, McQueary FG, Thornhill TS. Unicompartmental knee arthroplasty: eight- to 12-year follow-up evaluation with survivorship analysis. Clin Orthop Relat Res
35. Skyrme AD, Mencia MM, Skinner PW. Early failure of the porous-coated anatomic cemented unicompartmental knee arthroplasty: a 5- to 9-year follow-up study. J Arthroplasty
. 2002;17: 201-205.
36. Squire MW, Callaghan JJ, Goetz DD, Sullivan PM, Johnston RC. Unicompartmental knee replacement: a minimum 15 year followup study. Clin Orthop Relat Res
37. Swank M, Stulberg SD, Jiganti J, Machairas S. The natural history of unicompartmental arthroplasty. An eight-year follow-up study with survivorship analysis. Clin Orthop Relat Res
. 1993;286: 130-142.
38. Swedish Knee Arthroplasty Register. Available at: http://www.ort.lu.se/knee/indexeng.html
. Accessed January 10, 2005.