This prospective study describes the outcome of the first 1000 phase 3 Oxford medial unicompartmental knee replacements (UKRs) implanted using a minimally invasive surgical approach for the recommended indications by two surgeons and followed up independently. The mean follow-up was 5.6 years (1 to 11) with 547 knees having a minimum follow-up of five years. At five years their mean Oxford knee score was 41.3 (sd 7.2), the mean American Knee Society Objective Score 86.4 (sd 13.4), mean American Knee Society Functional Score 86.1 (sd 16.6), mean Tegner activity score 2.8 (sd 1.1). For the entire cohort, the mean maximum flexion was 130° at the time of final review.
The incidence of implant-related re-operations was 2.9%; of these 29 re-operations two were revisions requiring revision knee replacement components with stems and wedges, 17 were conversions to a primary total knee replacement, six were open reductions for dislocation of the bearing, three were secondary lateral UKRs and one was revision of a tibial component. The most common reason for further surgical intervention was progression of arthritis in the lateral compartment (0.9%), followed by dislocation of the bearing (0.6%) and revision for unexplained pain (0.6%). If all implant-related re-operations are considered failures, the ten-year survival rate was 96% (95% confidence interval, 92.5 to 99.5). If only revisions requiring revision components are considered failures the ten-year survival rate is 99.8% (confidence interval 99 to 100).
This is the largest published series of UKRs implanted through a minimally invasive surgical approach and with ten-year survival data. The survival rates are similar to those obtained with a standard open approach whereas the function is better. This demonstrates the effectiveness and safety of a minimally invasive surgical approach for implanting the Oxford UKR.
The Oxford Knee (Biomet, Swindon, United Kingdom) is a unicompartmental knee replacement (UKR) with a fully congruent mobile bearing, designed to minimise wear.1,2 Phase 1 and 2 Oxford Knees were implanted through a standard approach, with patellar dislocation, as for a total knee replacement (TKR). Good results have been achieved when used in the medial compartment through an open approach: the survival rate at ten years has been between 90% and 100% in most published series1,3–7 (Table I⇓). In addition, Price and Svard8 have reported a survival rate of 91% at 20 years.
The phase 3 Oxford Knee was introduced in 1998. This was similar to the phase 2 knee, but it introduced a larger range of sizes and the instrumentation was designed so that the procedure could be performed through a short skin incision without eversion or dislocation of the patella. We have previously shown that the minimally invasive surgical approach results in quicker recovery and improved function.9 However, there was concern that the limited access might result in more complications and a higher rate of revision. The aim of this study was to report the clinical outcome and ten-year survival of the phase 3 Oxford medial UKR.
Patients and Methods
All patients who have undergone a phase 3 Oxford UKR under the care of two surgeons (DWM, CAFD) were followed prospectively in a dedicated research clinic run by independent physiotherapists. The principal indications for the use of the Oxford UKR are anteromedial osteoarthritis (OA) and medial spontaneous osteonecrosis. The full details of the recommended indications and methods of assessment have been given by Goodfellow et al.10 In anteromedial OA there should be full-thickness cartilage loss on both sides of the medial compartment, with bone-on-bone contact. There should also be preservation of full-thickness cartilage in the lateral compartment.11 Both of these are best visualised on varus and valgus stress radiographs. The medial collateral ligament (MCL) should be functionally normal, as demonstrated by a correctable intra-articular varus deformity at 20° of flexion. This is also best shown on a valgus stress radiograph. The anteror cruciate ligament (ACL) should be functionally intact. The presence of a chondral ulcer on the medial side of the lateral femoral condyle can be ignored,12 as can the patient’s age, weight, level of activity and the presence of chondrocalcinosis, which are not considered to be contraindications.1,13 Similarly, the state of the patellofemoral joint and anterior knee pain can be ignored, provided that there is not severe lateral patellofemoral OA with bone loss and subluxation.1,14,15
The outcomes of the first 1000 consecutive phase 3 cemented medial Oxford UKRs with the recommended indications were analysed. All knees except those that were lost, revised or where the patient had died, had a minimum follow-up of one year. The knees were implanted between June 1998 and March 2009. During this period, an additional 97 knees received a medial Oxford UKR for indications that are not recommended; 37 were implanted in knees that did not have a functionally intact ACL, 32 with a friable and fragmented ACL and five with an absent ACL, and/or had other problems, such as central ulceration in the lateral compartment; 20 were implanted in patients with partial-thickness loss of cartilage in the medial compartment, seven in patients with medial osteochondritis dessicans; five with a previous high tibial osteotomy (HTO); three with laxity of the posterior cruciate ligament; two in patients with von Willebrand’s disease and one with a previous medial collateral ligament repair. In addition, during the recruitment phase, 22 young patients were treated with a combined ACL reconstruction and UKR for end-stage OA with ACL deficiency, the outcome of which has been previously reported.16 These 97 knees were not included in the study; however, their results are summarised in Table II⇓.
Patients were assessed using a standard protocol before surgery and at one, five, seven and ten years following surgery. All were sent questionnaires, which included the Oxford Knee Score (OKS),17 the Functional American Knee Society Score (AKSS-F)18 and the Tegner Activity Score.19 In addition, when patients were seen for a clinical follow-up appointment, they were assessed using the Objective American Knee Society Score (AKSS-O).18 Assessment of range of movement and alignment was made using a long arm goniometer in a standardised fashion.20 In order to assess the final outcome of the patients, they were all contacted within the 18-month period between September 2008 and March 2010 to confirm whether the knee had been revised or not, its current status (using the OKS), and their perception of the outcome. Patients were asked: ‘Overall, how pleased have you been with the outcome of your unicompartmental knee replacement?’, with the possible options of ‘very pleased’, ‘fairly pleased’, ‘not very pleased’, or ‘very disappointed’.
If, for any reason (social, geographical or medical), patients were unable to attend for follow-up and did not return the questionnaire, they were contacted by telephone and the relevant clinical information was obtained (revision status, OKS, Tegner activity score and AKSS-F). For patients who had died, information was gathered from hospital notes, general practitioner’s records and relatives to establish whether the patient had undergone any further surgery on the knee. Operative data, including surgical findings and components used, were recorded using a standard form. Any complications were noted. A survival analysis was undertaken using the life-table method for various definitions of failure. These were (a) implant-related re-operations, which included any re-operations in which components were changed, in which the meniscal bearings were replaced for dislocation, and any re-operations in which new components were inserted; (b) conversion to total knee replacement (TKR) and (c) revision surgery using revision TKR implants. We also calculated 95% confidence intervals (CI) using the method described by Peto et al.21
The first 1000 phase 3 medial UKRs with recommended indications were implanted in 818 patients. There were 636 unilateral and 182 bilateral procedures, of which 22 were simultaneous and 160 were staged. The mean age of the patients at the time of operation was 66 years (32 to 88); 425 (52%) were women and 393 (48%) were men. The procedure was undertaken in 977 knees for primary anteromedial OA and for medial spontaneous osteonecrosis in 23 (20 femoral, three tibial). Four patients (four knees) were lost to follow-up, all lived abroad and were lost in the first year. The outcome of the remaining 996 (99.6%) knees was known, and questionnaire data was collected at all post-operative time points (one, five, seven and ten years), as well as at their final review. A pre-operative OKS was available in 810 (81%) knees. At each time-point an average of 61% of patients were examined and had an AKSS-O. The mean follow-up was 5.6 years (1 to 11) with 547 knees having a minimum follow-up of five years.
All 1000 knees were followed for at least one year, except for those that were revised (seven) or where the patient had died during the first year (five) or were lost to follow-up (four). Every patient except those that had died (72), been revised (29) or lost (four) was contacted in the last 18 months to ascertain their final status. The OKS of these knees (895) and the last OKS of those who had died are summarised in Figure 2⇓. The mean OKS was 24.7 (sd 8.7) pre-operatively and 41.0 (sd 7.5) at final review. Of these, 728 patients (895 knees), 560 (688 knees, 77%) were very pleased, 123 (152 knees, 17%) were fairly pleased, 29 (36 knees, 4%) were not very pleased and 14 (19 knees, 2%) were very disappointed. At the final review, the mean flexion was 130° (85° to 152°) compared with a mean pre-operative flexion of 117° (25° to 145°). The clinical data on all living patients (excluding those who had been revised or were lost to follow-up) at intervals of one, five, seven and ten years, are summarised in Table III⇓.
There were 29 implant-related re-operations, the details of which are given in Table IV⇓. Two patients required revision using revision TKR components with stems and/or wedges. The ten-year cumulative survival rate with revision surgery using revision TKR components as the definition of failure was 99.8% (95% CI 99 to 100). There were 17 patients who needed conversion to primary TKR components. The ten-year cumulative survival rate using conversion to primary TKR or revision using revision TKR components as the definition of failure was 97.1% (95% CI 94.1 to 100). There were ten other re-operations related to the implant, of which six were dislocations. The bearing required replacement in five of these dislocations, and exchange of the bearing and a new femoral component in the sixth. In three cases of progression of arthritis in the lateral compartment a lateral UKR was implanted, as the medial UKR was secure and the ACL was intact. In one patient who presented with unexplained pain and possible tibial loosening, the tibial component, which was found to be secure, was revised. The ten-year cumulative survival rate using implant-related re-operation as the definition of failure was 96% (95% CI 92.5 to 99.5) (Table V⇓, Fig. 1⇑). During the tenth year 121 implants were available for analysis. In the worst-case scenario, in which all patients lost to follow-up were considered to be failures, the ten-year cumulative survival rate using implant-related operation as the definition of failure was 95.2% (95% CI 91.5 to 98.9).
The most common cause for re-operation was arthritis of the lateral compartment. This occurred in nine knees at a mean of five years (two to ten) after the index procedure. All patients presented with recurrence of pain and swelling in the knee, and impaired walking ability. In six cases, the ACL was found to be ruptured at the time of revision surgery, with progression of OA in the lateral compartment. In these cases the medial UKR was revised to a primary TKR. In the remaining three cases the ACL was found to be intact and the UKR was functioning well. A lateral UKR was therefore implanted using a lateral parapatellar approach through the original minimally invasive surgery medial skin incision, which was extended.
Six UKRs were complicated by dislocation of the bearing. The mean time to dislocation after the index procedure was 24 months (3 to 67). The dislocations were posterior in two cases and anterior in four. Three were associated with trauma, and one with impingement. In the other two cases no obvious cause was found. The trauma was usually a twisting injury to the flexed knee. In each case, the dislocated bearing was replaced after confirming that there was no associated problem which had contributed to the dislocation. In one case impingement was noted, and this was addressed at the time the bearing was exchanged. In one case of posterior dislocation of the bearing, the femoral component was dislodged while retrieving the dislocated bearing, and a new femoral component was therefore inserted.
In all, six UKRs in six patients were revised to a primary cruciate-retaining TKR for unexplained pain. Three were revised elsewhere and three were revised in our hospital. The mean time to revision was 37 months (9 to 108). The patient’s symptoms improved in three cases, two substantially and one only moderately. In the other three cases the knee was significantly worse than before the revision procedure. One patient developed a deep infection and required a two-stage revision knee replacement with a gastrocnemius flap. Another is wheelchair-bound owing to persistent pain in the revised knee.
Infection requiring revision occurred in five knees (four patients) at a mean of 16 months (5 to 34) after the index procedure. In two (in the same patient) of these five knees, no organisms could be cultured. However, as there was a strong clinical suspicion of infection these were categorised as infection. In all cases the infection was treated with two-stage revision. In four cases, a primary cruciate-retaining TKR was implanted, and in one case a posterior-stabilised TKR with a stemmed tibial component was used.
One patient had an ACL reconstruction at another hospital for a traumatic rupture of the ACL, 25 months after the UKR procedure. The knee became infected and a two-stage revision to a cruciate-sacrificing TKR with a stemmed tibial component was undertaken. One patient developed avascular necrosis (AVN) of the lateral femoral condyle nine months after the index procedure, and the knee was revised to a cruciate-retaining primary TKR. One patient developed loosening of the tibial component and OA of the lateral compartment. The knee was revised to a cruciate-retaining primary TKR 105 months after the index procedure.
A total of 14 knees in 12 patients required manipulation under anaesthesia to improve flexion, and two needed a second manipulation under anaesthesia. The range of movement improved from a mean of 70° (40° to 80°) to 129° (125° to 135°). Arthroscopy was undertaken in seven knees for persistent pain and/or swelling. The findings at arthroscopy included a lateral meniscal tear in four knees, AVN of the lateral femoral condyle in one knee, which subsequently needed revision, and recurrent synovitis in two knees. In four of these seven cases the symptoms settled completely, whereas in the other two the pain was well-managed with oral medication. One patient developed a superficial methicillin-resistant Staphylococcus aureus (MRSA) infection requiring debridement of the wound and a prolonged course of antibiotics, but had a good outcome.
Complications not requiring further surgery.
No patient died as a result of their surgery, but 67 patients (72 knees) died from unrelated causes. In each case the status of the knee at the time of death was established by contacting the family and/or the patient’s general practitioner. Three patients (four knees) had a pulmonary embolus confirmed by a ventilation/perfusing scan. Two patients had severe angina. There were no other major medical complications.
There were an additional 97 patients who had medial Oxford UKRs implanted for indications that are not recommended (Table II⇑). At ten years the number remaining (three) was too small for a reliable survival analysis. At eight years the survival rate was 88% (95% CI 73 to 100) when there were 15 remaining at risk. Had these patients not been excluded, the ten-year survival for the combined series based on all implant-related re-operations would have been 95% (95% CI 91.2 to 98.8) with 124 remaining.
This is a large series of UKRs that gives a relatively accurate assessment of outcomes. The survival rate of 96% at ten years, achieved using a minimally invasive approach, is similar to that reported by Svärd and Price4 and most other series in which the Oxford UKR was implanted through a standard open TKR approach (Table I⇑). The 20-year survival with the minimally invasive approach is therefore likely to be similar to that reported by Svärd and Price,8 which was 92%. As a result, the Oxford UKR, implanted with either a minimally invasive or an open approach using the appropriate indications and surgical technique, should be considered to be a definitive knee replacement and not a pre-TKR. The function and range of movement is however better after minimally invasive rather than an open approach.9
This study reports the outcome of patients who satisfy the recommended indications and observed the contraindications for the Oxford UKR.10 As the outcome is good, it suggests that these indications and contraindications are appropriate. The indications are well defined and are based on pathology. In the two conditions in which the Oxford UKR is recommended, anteromedial OA and spontaneous osteonecrosis, the ACL and MCL should be functionally normal and the lateral side should have full thickness cartilage. In anteromedial OA there should be bone-on-bone contact. There are virtually no contraindications and we ignore those recommended by other groups.13 We do, however, consider severe lateral patellofemoral joint arthritis with bone loss, grooving and subluxation to be a contraindication. In the past we have estimated that about one-third of patients who need a knee replacement are suitable for the Oxford UKR; however, in our practice, for the last five years, over 60% have had a UKR. Although our practice is skewed because of our particular interest, this does suggest that the proportion of knee replacement patients who could benefit from UKR is higher than one-third.
For various reasons in 97 knees we implanted a medial UKR for indications other than those recommended. These patients were excluded from the study and represent a small number (< 10%) compared to the main series. In this group the eight-year survival rate was 88% (95% CI 73 to 100). In view of the relatively small number, had they been included, they would not have affected the overall ten-year survival (95%). The relatively low survival in the group confirms that these patients were not ideally suited to UKR, but nevertheless most of these patients have done well. Further work is therefore needed to identify whether there are any subgroups for whom UKR could be recommended. For example, young patients with medial OA secondary to deficiency of the ACL could probably be successfully treated with combined ACL reconstruction and UKR,16,22 whereas a previous HTO and partial-thickness disease should probably still continue to be considered as contraindications.23,24
The primary outcome measure was the OKS, which is patient-based and has a maximum possible score of 48.25 A mean OKS of 41.0 and 94% of patients pleased with the outcome of their operation is good, and probably better than that which would be achieved following a TKR.26,27 However, it is impossible to be certain of this without a randomised controlled trial. A survival rate of 96% at ten years is similar to that expected with a TKR. With a TKR, revisions are usually complex and almost always require revision components with stems and wedges. In contrast, in this series the majority of the implant-related re-operations were straightforward, with 20% being replacement of a dislocated bearing, 60% being conversion to a primary TKR and only 10% requiring revision TKR components. In order to compare the rates of revision for UKRs and TKRs it is therefore perhaps best to compare the revisions that require revision TKR components. In this series of UKR the ten-year survival rate with failure defined as the use of revision TKR components was 99.8%, which compares favourably with TKR. There are also a number of other advantages of UKR over TKR, including lower rates of infection and lower peri-operative morbidity and mortality.28
The outcome of the revisions was good in all but three cases. The three patients with a poor outcome were converted to a TKR for unexplained pain. This supports the view that revision should not be undertaken unless a mechanical problem is identified. The bearing dislocations tended to occur early and simply required replacement of the bearing without conversion to TKR, and all had a good outcome. The five infections were all treated with two-stage revision and had a good outcome, which may be a reflection of the small amount of bone damage associated with an infected UKR.
The main strengths of this study are that it is large and well-documented; the appropriate indications were used and the surgeons were experienced in the procedure. It therefore represents the results that other experienced surgeons could expect to achieve using the same indications and techniques. The main limitation of the study is that it is not representative of the overall results that would be achieved nationwide. This information is best obtained from national registers,28 which quote a higher revision rate than in our series. As the same implant is used, the higher rate of revision is likely to be a manifestation of poor technique, or inappropriate indications for the primary UKR or for its revision. Unfortunately, registers collect incomplete information about these factors, so it is difficult to be certain which is most important.
This study demonstrates that, when used for appropriate patients, minimally invasive mobile-bearing UKR is safe and effective.
The authors wish to thank B. E. Marks and J. Copp for their assistance with this study. The study has been supported by the NIHR Biomedical Research Unit into Musculoskeletal Disease, Nuffield Orthopaedic Centre and the University of Oxford. Financial support has been received from Biomet.
The author or one of more of the authors have received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article. In addition, benefits have been or will be directed to a research fund, foundation, educational institution, or other nonprofit organisation with which one or more of the authors are associated.
- Received August 25, 2010.
- Accepted November 1, 2010.
- © 2011 British Editorial Society of Bone and Joint Surgery