Surgeons
The trial was approved by relevant national and local research ethics committees and was registered in a public trials registry (registry number, 1SRCTN45837371). Orthopaedic surgeons were eligible to take part if they performed knee replacements routinely. Before participating in the trial, the surgeons chose the comparisons to which they would contribute. We recognized that surgeons would differ in terms of the comparisons to which they would allow their patients to be randomly allocated. One hundred and sixteen surgeons in thirty-four centers in the United Kingdom participated.
Patients
All patients under the care of a collaborating surgeon were potentially eligible for inclusion if a decision had been made for them to have primary total knee replacement. A patient was not eligible for a trial comparison if the surgeon considered a particular type of operation to be clearly indicated (for example, if a patient required a highly constrained knee replacement to replace function of the collateral ligaments). A patient remained eligible only if the surgeon remained convinced that there was no indication for one particular choice within the trial; for example, a patient with a thin or osteopenic patella would not be eligible for the patellar resurfacing comparison because the surgeon would not have chosen patellar resurfacing for such a patient.
When possible, patients scheduled for a total knee replacement were sent information about relevant aspects of the study in advance of their hospital admission.
While it was anticipated that most participants would be enrolled in a single comparison, individuals could be recruited for more than one comparison if that was clinically appropriate. Only a minority of participants were included in more than one comparison, and they were randomized within each relevant comparison with use of a partial factorial design to ensure balance of allocation within and across comparisons.
Surgical Procedures
Within the randomized comparisons, all prostheses had suitable alternative designs. Surgeons followed their standard practice—i.e., the technique that they utilized did not require any modification for the purposes of the trial, and the outcomes were thus not influenced by a so-called learning-curve effect. We did not influence surgeons regarding whether they should utilize cruciate-retaining or substituting implants. All other aspects of care, such as prophylaxis against deep vein thrombosis, were left to the discretion of the responsible surgeon.
Principal Outcome Measures
The principal outcome measures were functional status as measured with the Oxford Knee Score14 (OKS), quality of life as measured with the Short Form-1215 (SF-12) and the EuroQol-5D16,17 (EQ-5D), and intraoperative and postoperative complications including the need for additional surgery. Secondary outcomes, including costs and cost-effectiveness, are also being assessed but are not reported here. The OKS was selected as a primary outcome measure because it had been developed specifically to measure outcomes of knee replacement and had been shown by a range of independent studies to perform very well compared with alternative instruments18-20.
Sample Size Estimation
The size of the effect on the OKS sought in each comparison (and hence the sample size chosen) was based on the size of the difference in the OKS that seemed likely, as judged on the basis of current experience, and the size of the effect that was likely to offset any adverse effects and cost differences of the prosthetic design variable. The difference in OKS sought was 3 points for the comparisons involving the tibial metal backing and the mobile bearing, with 350 participants providing 80% statistical power and 470 participants providing 90% power to identify this difference (p < 0.05). The difference sought was 1.5 points for the patellar resurfacing comparison, with 1400 participants providing 80% power to detect this difference (p < 0.05).
Randomization
If the surgeon thought that a patient was eligible to participate in a comparison for which the surgeon had registered, fuller details of the trial were provided and the patient was asked to sign an informed consent form to participate. For the patients who joined the trial, the relevant aspect(s) of the replacement were chosen prior to the surgery by means of random allocation. An automated centralized telephone randomization service was called and, after basic identification had been given over the telephone, the patient was allocated to the relevant comparison or combination of comparisons. The randomization was stratified by surgeon, with minimization (randomization balanced with respect to specified variables) according to the patient's age (less than sixty years, sixty to seventy-nine years, or eighty years or more), sex (male or female), and site of disease (one knee, both knees, or general arthritis).
Data Collection
Data were collected prospectively on standard forms to record preoperative, operative, and postoperative information. Data describing functional status and quality of life were collected directly from questionnaires sent by mail to the participants. Follow-up questionnaires were completed at approximately three months, one year, and two years after the operation. One reminder was sent if necessary, and, if the questionnaire was still not returned, this was followed by a telephone-call reminder, with the patient then offered the option of completing the questionnaire over the telephone. The questionnaire included the OKS14, the SF-1215, the EQ-5D16,17, and questions about any additional hospital admissions and surgery.
Statistical Analysis
The three comparisons were analyzed as separate trials. Data were analyzed on the basis of the procedure allocation irrespective of the method of replacement that was actually used (intention-to-treat principle). The functional status and quality-of-life outcomes within each trial comparison were compared by using analysis of covariance that adjusted for baseline scores and the minimization factors. Readmission rates within each trial were analyzed with use of logistic regression analysis. The Mann-Whitney U test was used to compare operative times between the trials, and the chi-square test was used to compare binary outcomes (grade of surgeon) between the trials. Descriptive statistics are presented when appropriate, and effect sizes are presented with the associated 95% confidence intervals estimated with robust standard errors to account for potential surgeon effects.
Source of Funding
The Knee Arthroplasty Trial is funded by the NIHR Health Technology Assessment Programme (project number 95/10/01). Additional funding for research support in clinical centers was provided by Howmedica Osteonics; Zimmer; DePuy, a Johnson and Johnson company; Corin Medical; Smith and Nephew Healthcare; Biomet Merck; and Wright Cremascoli. The Health Services Research Unit is core-funded by the Chief Scientist Office of the Scottish Government Health Directorates.
Participant Flow and Recruitment
From July 1999 to January 2003, 4070 potentially eligible patients were identified and 2374 (58%) gave their consent and were randomized. The main reasons for nonrandomization were the patient's refusal to take part in the study (546; 32%), the surgeon not wanting the patient to be randomized (462; 27%), a missed opportunity to recruit a scheduled patient (351; 21%), cancellation or deferral of the surgery or nonattendance on the part of the patient (146; 9%), ineligibility of the patient (eighty-four; 5%), the surgeon undertaking the procedure not being registered to participate in the trial (thirty-eight; 2%), unavailability of necessary equipment (twenty-four; 1%), and unknown reasons (forty-five; 3%). Twenty-two patients were subsequently found to have been randomized in error: fourteen were randomized twice, three were not eligible, three were treated by surgeons who were not registered to participate in the comparison, and two were excluded for other reasons. This left 2352 patients formally in the trial: 409 were included in the comparison assessing the metal backing; 1715, in the comparison assessing the patellar resurfacing; and 539, in the comparison assessing the mobile bearing. Background information for the CONSORT statement can be found in Figure 1.
Baseline Data
A table in the Appendix provides a description of the groups at the time of entry into the trial. The overall mean patient age (and standard deviation) at the time of the index arthroplasty was 70 ± 8 years (range, twenty-two to ninety-three years), 43.7% (1014 of 2318) were men, and the mean body mass index was 29.7 kg/m2. Within the randomized comparisons, demographic and clinical data were well balanced at baseline although there were small differences across the three comparisons; participants recruited into the patellar resurfacing group tended to be healthier as judged by their American Society of Anesthesiologists (ASA) grade21 (p = 0.004).
Management and Operation Details (see Appendix)
The majority of the subjects (83.1%; 1927 of 2318) underwent the procedure as planned. Intraoperative complications were observed in only a small percentage of the patients (2.7%; fifty-nine of 2201), and the operative procedure caused problems in few patients (1.2%; twenty-seven of 2201). Overall, there were no differences between the randomized groups in these respects. The median operative time in the group allocated to the comparison of the metal and non-metal-backed components was significantly lower than that in the other two comparison groups (p < 0.001), but there was no evidence of within-group differences. There was evidence that the group in which the metal and non-metal-backed components were compared was more likely than the other two comparison groups to have been operated on by a fully trained specialist orthopaedic surgeon (holding a certificate of completion of specialist surgical training in orthopaedics or the equivalent in the United Kingdom) (p < 0.001). Whether the surgeon was fully trained or still in supervised training did not differ within each trial comparison. Lateral patellar retinacular release was performed most commonly in the patellar resurfacing group (17.1%; 136) and least commonly in the group treated with a non-metal-backed component (10.2%; twenty).
In-Hospital Care and Short-Term Complications (see Appendix)
Postoperative complications were reported in 14.9% (328) of 2207 patients; however, specific problems such as wound infection, septicemia, deep vein thrombosis or pulmonary embolism, cerebrovascular accident, and myocardial infarction were all rare. Overall, 1.8% (forty) of 2206 participants had additional knee surgery. Four had knee dislocation. One participant allocated to be treated with both patellar resurfacing and a fixed-bearing prosthesis, but who actually received a mobile-bearing prosthesis, required closed reduction of the joint because of dislocation of the rotating insert four days after the initial operation. The participant had another dislocation two weeks later and was readmitted for revision of the spacer and femoral component. One participant allocated to the patellar resurfacing group had a subluxation of the bearing and required a reoperation for replacement of the platform insert. The remaining two participants who had dislocation (one allocated to both the no-patellar-resurfacing group and the mobile-bearing group and the other allocated to the no-patellar-resurfacing group) required manipulation under anesthesia. Six participants died in the immediate postoperative period: two died from a pulmonary embolism; one, from a myocardial infarction; one, from ischemic heart disease; one, from pneumonia; and one, from a cerebrovascular accident. Overall, 95.2% (2101) of 2207 participants were discharged directly to their home. The median length of the hospital stay was nine days. There were no differences between the randomized groups with regard to any of the above factors.
Patient-Assessed Outcomes (see Appendix)
Functional status and quality-of-life scores were low at baseline and improved markedly across all trial groups following the knee replacement. The mean overall OKS was 18.0 points at baseline, 30.5 points at three months, 34.2 points at one year, and 34.8 points at two years. Thus, most of the change was observed by three months after the surgery, although additional small improvements were observed at one and two years (Figs. 2, 3, and 4). Within the individual trial comparisons, there was no evidence of differences in functional status or quality-of-life measures between the randomized groups at two years.
Complications After Surgery (see Appendix)
Overall, 9.9% (230) of 2318 patients were readmitted to the hospital for reasons related to the surgery on the knee and 5.8% (135) of 2318 had additional knee surgery. Three participants had an above-the-knee amputation: two within two months after the initial operation and one, three months after the surgery. One of the amputations (in a participant allocated to receive a non-metal-backed component) was due to a knee infection that developed after an open repair of the quadriceps tendon following a fall, another (in a participant allocated to no patellar resurfacing) was due to vascular insufficiency one month after the initial operation, and the third (in a participant allocated to receive a metal-backed component) was due to diabetic ischemia. Fifteen participants underwent staged revisions due to infection. Fifteen participants had single-stage revision. Seven participants who had been allocated to have no patellar resurfacing subsequently had the patella resurfaced. The principal reasons for readmission related to the surgery were suspected deep vein thrombosis or pulmonary embolism and infection such as pneumonia or urinary tract infection. Within each trial comparison, there were no significant differences in the number of patients requiring readmission: the odds ratio was 1.50 (95% confidence interval, 0.84 to 2.70) for the comparison of metal and non-metal-backed components, 1.08 (95% confidence interval, 0.87 to 1.35) for the comparison of patellar resurfacing and no patellar resurfacing, and 0.83 (95% confidence interval, 0.52 to 1.33) for the comparison of mobile and fixed bearings.
This multicenter trial of patients treated in the United Kingdom showed that, as a group, patients who had undergone primary total knee replacement had substantial pain relief and improvement in function, as assessed with condition-specific measures designed to evaluate the results of total knee replacement and with more generic measures of health-related quality of life. Benefits were observed by three months after the surgery, but there were additional, small improvements in pain, physical function, and health-related quality-of-life scores up to two years after the surgery. No significant differences in outcome were observed within any of the three randomized comparisons. This is what might be expected, at this stage, in the comparison assessing metal backing and that assessing the rotating platform, as those design features were developed in an effort to obtain a longer-term benefit. In contrast, patellar resurfacing might have been expected to reduce pain and improve early function, but that was not observed. The lack of a difference in improvement in scores irrespective of whether the patella was resurfaced or not is consistent with the findings of Pakos et al.6. The rate of knee-related readmissions in their study was perhaps higher than might have been expected, although the most common reason for readmission was manipulation with the patient under anesthesia.
The current trial was pragmatic in design; such trials are unusual in orthopaedic surgery but widely used to evaluate other health-care issues. The aim of pragmatic trials is to evaluate interventions in a usual-care context in terms of outcomes that are most important to patients. Entry criteria are relatively flexible so that a range of patients are recruited, enhancing generalizability. The trial involved a large number of orthopaedic surgeons in the United Kingdom and hence reflects a wide range of practice in the United Kingdom, geographically and in terms of types of centers and details of technique and strategy. Surgeons chose the comparisons into which their patients could be recruited. For each comparison, randomization was stratified by surgeon to ensure a balance between the trial groups in this respect. We focused on outcomes that are of concern to patients and used validated measures, independent of surgeon opinion. Each comparison included several hundred participants, so the estimates of differences were relatively precise. With respect to the OKS, in comparisons (B) and (C), the upper bound of the estimated confidence intervals provided evidence against the prespecified minimum clinical differences. In the smallest comparison (A), the confidence interval is slightly wider but also indicated the prespecified minimum difference unlikely. From this, we can conclude that there was no evidence of superiority of metal-backed tibial components over non-metal-backed tibial components, patellar resurfacing over not resurfacing, or mobile-bearing over fixed-bearing designs at two years. These are short-term results, however, and superiority over the longer term cannot be ruled out. It should also be recognized that not every participant received the prosthesis that had been allocated to him or her and the prosthetic types used in each of the groups may have subtle design variations; secondary subgroup analyses are planned to explore this possibility.
The KAT study demonstrated that large simple-in-design trials across a range of practices in orthopaedics are feasible. However, the study also demonstrated the need for individual surgeons to be clinically uncertain about the appropriate treatment choice.
The short-term results in the three trials are broadly consistent with evidence from other studies3,10. At this stage, it is not possible to suggest to health-care providers that one particular design or variation in design has clinical advantages that warrant limiting surgeon choice. It is generally accepted that variations in the outcomes of different knee prostheses tend to emerge with longer-term follow-up, and hence recommendations based on generic aspects of design may well be possible in the future.
The patients in this study will continue to be followed annually for at least ten years. While the main comparisons at the time of follow-up will be between the generic variants of total knee replacement, possible effects of subtle difference in design between manufacturers and between surgeons will also be investigated. Data from the KAT study will also be used to estimate the costs of each intervention and hence the relative cost-effectiveness within each randomized comparison.
Note: Data Monitoring Committee: Gordon Murray (Chair), Rajan Madhok, and Hamish Simpson. Central Trial Office (Aberdeen): Marion Campbell, Susan Campbell, Adrian Grant, Mark Kelaher, Anne Langston, Graeme MacLennan, Kirsty McCormack, Craig Ramsay, Sue Ross, Luke Vale, and Allan Walker. Regional Coordinators (Dundee and Oxford): Kim Clipsham (Oxford), Linda Johnston (Dundee), Doug McGurty (Dundee), Lesley Morgan (Oxford), and Sarah Poulter (Oxford). Other collaborators: Alan Price, Julie Rowsell (Alexandra Hospital, Redditch), Rose Finley, Sue Gardner, Richard W. Parkinson (Arrowe Park Hospital, Wirral), Liz Jackson, Iain Lennox, Timothy Peckham, John Targett, Rob Wakeman (Basildon and Thurrock University Hospital, Basildon), Stephen Hughes, Karen Humby, Carol Quick (Birmingham Heartlands Hospital), Jennifer Burbidge, Tony Chapman, Nicola Sheehan (Calderdale Royal Hospital, Halifax), Polly Emmitt, Marek Karpinski, Margaret Newman, Andre T. Plotka, Javed Salim, Kevin P. Sherman (Castle Hill Hospital, Hull), Ian Braithwaite, David Campbell, Janet Durrans, Karen Edwards, Sandra D. Flynn, Andrew Phillipson (Countess of Chester Hospital), Debbie Carpenter, Charles Grant, Linda Smith (Diana, Princess of Wales Hospital, Grimsby), Anthony Brewood, Carmel Cliffe, Ronan McGiveney, Diane Ross (Fairfield General Hospital, Bury), Lesley Plummer, Lavinia Psarras, Timothy Tasker, Norma White, Andrew Williams (Gloucestershire Royal Hospital), Julie Cunningham, Jane Hopkins (Goole and District Hospital), Kathleen Duncan, Robert Allan Dunsmuir, Alberto Gregori (Hairmyres Hospital, East Kilbride), Samir N. Amarah, Carol Donald, Peter Sewell, Timothy Vaughan-Lane, Alison Rosen (Hinchingbrooke Hospital, Huntingdon), Ian Archer, Stuart Calder, Mark Emerton, Gillian Johnston, David MacDonald, Martin Stone (Leeds General Infirmary and St. James University Hospital), Susan Finch, Graham Keys, Susan Smith (Macclesfield District General Hospital), Paul Gregg, Anthony Chi Wing Hui, Ian Wallace, Lisa Wood (Middlesbrough General Hospital), Anna David, Malcolm Downes, Ceri Hodinott, Mark Holt, Tim James, Kath Law, Robert Leyshon, Sharon Maggs, David Newington, Neil Price (Morriston Hospital, Swansea), Graham Foubister, Amir Jain, Linda Johnston, Doug McGurty, Manhal Nassif, David Rowley (Ninewells Hospital, Dundee), Nagi Darwish, William Farrington, Nigel Giles, Sunil Jain, Debbie Ludwell, Christopher Mills, Michael Podmore, Nicholas Treble, Peet Van Der Walt (North Devon District Hospital, Barnstaple), Lesley Boulton, David Miller (North Tees General Hospital, Stockton-on-Tees), Gavin Bowden, Kim Clipsham, Chris Dodd, Max Gibbons, Damion Griffin, Roger Gundle, Peter McLardy Smith, Lesley Morgan, David Murray, Sarah Poulter, Rob Sterling (Nuffield Orthopaedic Centre, Oxford), Arthur Espley, Jamie McLean, Lorna O'Donnell, Audrey Reilly (Perth Royal Infirmary), Katrina Boeree, Peter Cox, Keith Eyres, Graham A. Gie, Nigel Giles, Matthew Hubble, Peter Schranz, John Timperley (Princess Elizabeth Orthopaedic Centre, Exeter), Anthony Fogg, Michael Foy, John Ivory, Ian M.R. Lowden, Eve Middleton, David M. Williamson, David Woods (Princess Margaret Hospital, Swindon), Tim Cane, Hugh Clark (Queen Alexandra Hospital, Portsmouth), Ganapathyraman Mani, Anthony Percy, Sudhir Rao, Colin Smart, Mark Rowntree, Helen Stanger (Queen Mary's Hospital, Sidcup), Nick Fiddian, Gwen Newton (Royal Bournemouth Hospital), John Davidson, Simon Journeaux, Jill Pope (Royal Liverpool University Hospital), Janet Jessop, Una Jude, Louise Mitchell, Peter Molitor, Karen Watts (Scunthorpe General Hospital, Scunthorpe), Benjamin Bolton Maggs, Grahame Robertson (St. Helens and Knowsley Hospitals NHS Trust), Richard Buckley, Sarah Jane Keogh, Pete Rickhuss, Val Sutherland, Neil Valentine (Stracathro Hospital, Brechin), Colin M. Mainds, Clark Dreghorn, Eric G. Gardner, Peter D. Scott, Rhona Shields (Victoria Infirmary, Glasgow), G. Paddy Ashcroft, Ann Galt, Peter H. Gibson, Jimmy D. Hutchison, Alan Johnstone, David Knight, William Ledingham, Anne Potter (Woodend Hospital, Aberdeen), Noor Ahmed, Tracey Dennehy, Alison Lawrence, E. Rouholamin (Worcester Royal Infirmary Trust, Worcester), Laura Hobbs, Geoffrey Taylor, and Kenneth Wise (Wycombe General Hospital, High Wycombe).