Knee A 15-year follow-up study of 466 primary total knee replacements V. I. Roberts, C. N. A. Esler, W. M. Harper From Glenfield General Hospital, Leicester, England This is a 15-year follow-up observational study of 439 patients with 466 primary total knee replacements (TKRs) implanted in the Trent health region between 199 and 1992. The operations were performed in 21 hospitals, including both district general and teaching hospitals, with 77 different surgeons as named consultant. The main objective was to analyse the survival of the patients and of the prostheses, and to evaluate what impact different variables have on survival. In addition, the 148 patients (33.7%) (1556 TKRs) alive at 15 years following operation were sent a self-administered questionnaire which examined their level of satisfaction, of pain, and their quality of life at 15 years. Completed responses were received from 912 TKRs (58.6%). Three survival curves were constructed: a best-case scenario based on the patients entered into the life tables, another included failures not reported in the revision database, and a third worst-case scenario based on all patients lost to follow-up presumed to have had a failed primary TKR. In the best-case scenario survival at 15 years was 92.2%, and in the worst-case scenario was 81.1%. Survival was significantly increased in women and older patients (Mantel-Cox log-rank test, p <.5 and p <.1, respectively). Revision as a result of infection was required in 4 TKRs (18.8%) representing.87% of the original cohort. The limited information available from the questionnaire indicated that satisfaction was less frequent among men, patients with osteoarthritis and those who required revision (chi-squared test, p <.5, p <.5 and p <.1, respectively). With regard to pain, older patients, females and patients who still had their primary replacement in place at 15 years, reported the least pain (chi-squared test for trends, p <.5, p <.5 and p <.1, respectively). The reported quality of life was not affected by any variable. V. I. Roberts, MRCS, Research Fellow W. M. Harper, FRCS(Ortho), MD, Professor, Orthopaedic Surgeon Department of Orthopaedics Trent Arthroplasty Audit Group, Glenfield Hospital, Groby Road, Leicester LE3 9QP, UK. C. N. A. Esler, FRCS(Tr & Orth), Honorary Senior Lecturer, Consultant Orthopaedic Surgeon Leicester General Hospital, Gwendolene Road, Leicester LE5 4PW, UK. Correspondence should be sent to Miss V. I. Roberts; e-mail: vero_roberts1@hotmail.com 27 British Editorial Society of Bone and Joint Surgery doi:1.132/31-62x.89b11. 19783 $2. J Bone Joint Surg [Br] 27;89-B:1452-6. Received 4 June 27; Accepted 25 June 27 There have been several studies of European origin which have examined cumulative survival of total knee replacement (TKR) at ten years. 1-4 However, there is very little information on the survival of primary TKR in the longer term. We analysed the survival of primary TKRs performed in the United Kingdom at 21 hospitals and by 77 different consultant surgeons, using information stored on a regional arterioplasty database. The survival of the prostheses was analysed at five, ten and 15 years post-operatively. Previous studies have found that the age of the patient at the time of surgery has a significant effect on survival, with older patients less likely to require a revision procedure, 5-7 but that neither gender nor the primary disease had a significant effect on cumulative revision rates. 6,7 Accordingly, we examined the effect of these variables on our patient cohort. We also examined the rate of infection resulting in revision procedures. Patient satisfaction following TKR has been reported to range between 75% and 85%. 8-1 We evaluated the proportion of patients who were satisfied at 15 years following primary TKR, the frequency of pain, and the quality of life, and attempted to determine which variables contributed to these results. Patients and Methods The Trent (and Wales) Regional Arthroplasty Audit Group 11 was established to assess the outcome of joint replacement in one United Kingdom health region. Since the beginning of 199, and with the agreement of all consultant orthopaedic surgeons in the Trent region, all primary TKRs and total hip replacements (THRs) performed have been recorded prospectively and the details registered on the 1452 THE JOURNAL OF BONE AND JOINT SURGERY
A 15-YEAR FOLLOW-UP STUDY OF 466 PRIMARY TOTAL KNEE REPLACEMENTS 1453 8 6 4 2 5 15 2 Duration from total knee replacement (mths) Fig. 1 Survival curve for the entire cohort. Survival 25 database at the University of Leicester. The information recorded by the surgeon (not an author) at the time of the operation includes demographic, medical and operative details for each patient and implant. All information is validated by a peripatetic clerk visiting each unit. Since 1992, a similar database has been established for all revision procedures. Between 199 and 1992, a total of 466 primary TKRs in 439 patients were performed in the region. Patients were traced using the NHS Strategic Tracing Service. 12 A self-administered postal questionnaire was sent to all surviving patients at a mean of 15.2 years (13.8 to 16.7) postoperatively. This questionnaire sought the patients opinion of their TKR. It asked about their level of satisfaction, whether they experienced pain, if they have encountered any problems or required further operations, and used a visual analogue scale (VAS) to measure their quality of life. Any patients who had undergone a revision procedure were identified using the revision arthroplasty database. The cause of failure as stated by the surgeon, and the length of time until failure of the primary TKR, were established from this database. Failure was defined as any revision procedure. A total of 3989 TKRs (86.6%) were cemented but there was no information available concerning whether the patella was resurfaced. The tracing service was unable to trace 46, patients 2 of whom had bilateral TKRs (48 TKRs lost to follow-up). A total of 148 patients (33.7%) with 1556 TKRs were alive at a mean of 15.2 years (13.8 to 16.7) post-operatively, 76 of whom had undergone bilateral TKR. The mean age of the 439 patients in this study at the time of operation was 69.5 years (19 to 92), which is comparable to the results of both Swedish and Norwegian joint registers, 5,13 and 28 (63.8%) were female which is also comparable with the Swedish joint register. 6,14 The principal disorder leading to joint replacement was osteoarthritis (OA) with rheumatoid arthritis (RA) the next most common cause, involving 3398 (77.4%) and 814 (18.5%) patients, respectively. Statistical analysis was performed using SPSS version 14 (SPSS Inc., Chicago, Illinois) to assess the survival of the original population cohort, survival of the prostheses, infection rate, and the opinion of the patients at 15 years of follow-up. The Kaplan-Meier method was used for survival analysis. Results Survival analysis showed that at one month the mortality rate was.4%, and at one year was 1.8%, which is comparable with the results of the National Joint Register (1.3% and.4%, respectively). 11 Further analyses showed that five years after replacement 85.7% (3795) of patients were alive; after ten years, this figure was 65.6% (2873); and after 15 years 47.2% (268) (Fig. 1). Women were statistically more likely than men to be alive 15 years post-operatively (Mantel-Cox log-rank test, p <.1). The survival curves for the different age groups showed statistically significant differences between the groups (Mantel-Cox log-rank test, p <.1) (Fig. 2). Prosthesis survival. In total, 3 identifiable prostheses were used, but in 356 TKRs (7.7%), the manufacturer s information was not available. However, six prostheses alone were used for 3577 TKRs (77.7%). These were Freeman- Samuelson, 229 (5%) (Sulzer Orthopaedic Baar, Switzerland); Insall Burstein II, 46 (8.8%) (Zimmer, Swindon, United Kingdom); Kinematic, 677 (14.7%) (Howmedica, London, United Kingdom); Kinemax, 127 (27.6%) (Howmedica); Omnifit, 185 (4%) (Osteonics, Rutherford, New Jersey) and Press Fit Condylar, 81 (17.6%) (Johnson and Johnson Inc, Raynham, Massachusetts). A total of 239 TKRs (5.2%) had been revised at a mean of 76 months (1 to 185). From the questionnaires, a further 29 patients (.62%) reported undergoing a revision of their primary prosthesis, but no information was available to validate this on the revision database. Three survival curves have been constructed. One is based on those patients entered into the life tables and is a best-case scenario. Another included the additional 29 patients who stated that they have required a revision but had not been registered on the revision database. In this scenario, failure of the prosthesis was recorded as occurring before five years. The third curve is the worst-case scenario, in which all patients who could not be traced are assumed to have had a revision within five years of their primary TKR. VOL. 89-B, No. 11, NOVEMBER 27
1454 V. I. ROBERTS, C. N. A. ESLER, W. M. HARPER 8 6 4 2 Age groups SE <55 2.82 55 to 64 1.99 65 to 74 1.51 75 to 84 1.85 >85 6.37 98 94 92 9 5 15 2 25 5 15 2 25 Duration from total knee replacement (mths) Fig. 2 Survival curve for the entire cohort of 439 patients, according to age. Time to faliure (mths) Fig. 3 Survival curve of all prostheses used between 199 and 1992 (all prostheses, SE.635). 99 93 9 87 5 Fig. 4 15 Survival curve according to gender. Gender SE Female.725 Male 1.216 2 25 The best-case scenario of cumulative implant survival is shown in Figure 3. After five years, 97.5% (standard error (SE).3%) of prostheses were in situ. At ten years, the figure was 94.4% (SE.4%); and at 15 years, 92.2% (SE.5%). In the second scenario, the survival at five years was.6% (SE.3%), at ten years was 94.3% (SE.4), and at 15 years was 91.1% (SE.5%). However, in the final worst-case scenario, survival at ten years was 83.1% (SE.6) and at 15 years was 81.1% (SE.7). All subsequent analysis of survival of the implants, and the impact of different variables, is based upon life tables and the best-case scenario. No statistical significance was observed in the relationship between survival and diagnostic indication for TKR, either OA or RA (Mantel-Cox log-rank test, p =.27). However, on plotting separate survival curves for gender, there was a significant difference in the survival of the prostheses (Mantel-Cox log-rank test, p <.1), with female patients undergoing fewer revisions (Fig. 4). Age at the time of the primary procedure also had a significant inverse relationship on survival of the prostheses (Mantel-Cox log-rank test, p <.1), with younger patients requiring more revision procedures (Fig. 5). Separate survival curves were plotted for each of the six most commonly-implanted prostheses, revealing that there was a significant difference in survival depending on the prosthesis used Mantel-Cox log-rank test (p <.5) (Fig. 6). However, separate analysis of other identified variables associated with a reduced survival was not undertaken for each implant design. Therefore, these results may have been confounded. Cause of failure. Revision was a result of aseptic loosening in 81 TKRs (34%), to infection in 4 TKRs (16.7%) and to wear of polyethylene in 33 TKRs (13.8%). The remaining 85 failures were caused by instability, patellofemoral pain, technical problems, peri-prosthetic fractures, or the cause was unknown. Failure attributed to aseptic loosening occurred at a relatively constant rate over the 15-year follow-up (Fig. 7). THE JOURNAL OF BONE AND JOINT SURGERY
A 15-YEAR FOLLOW-UP STUDY OF 466 PRIMARY TOTAL KNEE REPLACEMENTS 1455 99 97.5 93 9 87 Age groups SE <55 2.628 55 to 64 1.515 65 to 74.976 75 to 84.955 >85 2.11 95. 92.5 9. 87.5 Prosthesis SE Omnifit 1.71 IB2 1.92 PFC 1.35 Kinemax 1.32 Freeman 3.15 Kinematic 1.91 5 15 2 25 5 15 2 25 Fig. 5 Survival curves according to age. Fig. 6 Survival in the six most commonly-used prostheses. The vertical lines and crosses on the line graphs refer to censored total knee replacements (artefact of SPSS). Numbers at risk at final follow-up included 74 (Freeman), 15 (IB2), 19 (Kinematic), 44 (Kinemax), 48 (Omnifit). 8 6 4 2 5 Fig. 7 15 Time distribution of failures according to cause. Implant/ poly failure Aseptic loosening Infection 2 Infection as a cause of failure, however, occurred most commonly within the first 12 months, whereas implant failure or failure of the polyethylene occurred most commonly at around eight years post-operatively. It is possible to consider the infection rate as a percentage of surviving or presumed surviving patients, or as a percentage of the original cohort. This gives two different results, a worst-case and a best-case scenario, respectively. By three years post-operatively, 12 patients had required revision because of infection, representing.28% of surviving TKRs, or.26% of the original cohort. By five years, 19 TKRs had been revised for infection representing.49% of surviving TKRs, or.41% of the original cohort of 466, By 15 years, 4 patients had undergone a revision for infection, representing 2.57% of surviving TKRs or.87% of the original cohort of TKRs. Population outcome measures. The 15-year follow-up questionnaire sought the opinion of 148 surviving patients (1556 TKRs) regarding their operation; the responses concerning 912 TKRs were returned (58.6%). The three main outcome measures used were satisfaction, frequency of pain, and quality of life. Of the respondents 778 patients (85.3%) were satisfied, which is comparable to other studies. 8,9 However, contrary to other studies, 8 the results indicated that the patients satisfaction was not affected by their age at primary surgery but was affected by gender, the diagnosis leading to TKR, and the need for revision. Men, patients who had OA, and those who had undergone a revision procedure were statistically more likely to be dissatisfied (chisquared test, p <.5, p <.5 and p <.1, respectively). With respect to the frequency of pain, age, gender and failure of the primary TKR had a significant effect on the results. The results indicate that older patients, females, and patients whose primary prosthesis was still in situ at 15 years experienced pain less often (chi-squared test for trends, p <.5, <.5, and <.1, respectively). Patients quality of life, as measured on the VAS (mean VAS 58.9 ( to )), was not affected by any variable. Therefore, gender, age at primary replacement, failure of VOL. 89-B, No. 11, NOVEMBER 27
1456 V. I. ROBERTS, C. N. A. ESLER, W. M. HARPER Table I. The relationship between different variables and the outcome measurements of satisfaction, pain and quality of life Satisfaction Pain (frequency) Quality of life (VAS * ) Gender p <.5 p <.5 p =.515 Age at TKR p =.155 p <.5 p =.94 Diagnosis p <.5 p =.1 p =.327 Failure of TKR p <.1 p <.1 p =.352 Cemented TKR p =.321 p =.49 p =.938 Intra-operative complication p =.415 p =.527 p =.429 Post-operative complication p <.1 p <.5 p =.72 Further operation within one year p <.1 p =.134 p =.957 * VAS, visual analogue scale TKR, total knee replacement TKR and diagnosis leading to TKR did not have any bearing on quality of life at 15 years following TKR. The relationships between different variables and the outcome measurements of satisfaction, pain, and quality of life are presented in Table I. Discussion The survival analysis of cohort showed that women and younger patients were statistically more likely to survive to 15 years following TKR. The mortality rates at both three months and one year post-operatively, are comparable to contemporary results for England and Wales. 15 The best-case survival curve at 15 years following TKR showed that 92.7% of prostheses were still in situ. In the worst-case survival, 81.1% were in situ. However, our bestcase scenario findings differed from those of the Swedish Knee Arthroplasty register 5,6 with respect to the influence of gender, with 89.5% of prostheses surviving to 15 years in males and 93.5% in females. Survival of the prosthesis was significantly affected by the age of the patient, as concluded in previous studies. 4-6 In patients aged less than 55 years at the time of surgery, only 87.% of prostheses survived to 15 years; but as age at the time of TKR increased survival of the TKR also improved (Fig. 5). The infection rates at three, five and 15 years based on the whole cohort were.26%,.41% and.74%, respectively. These findings were comparable with those of other long-term studies. 11 Although limited by the large number of non-responders, a high level of satisfaction with the TKR was still felt 15 years following the replacement. Males and patients who had undergone a revision were statistically more likely to be dissatisfied. Similarly, men reported pain from their TKR more often than did women. It was disappointing that 29 of the revisions (.62%) had not been documented on the database. This may have occurred when patients in the original cohort moved out of the region. This is one of the first long-term studies from the United Kingdom to look at the survival of primary TKR in one simultaneous cohort without selecting specific prostheses. 7 It shows that survival at five and ten years is comparable to the results of similar studies from other countries. 1-3 However, there is very little information on the survival of knee prostheses beyond ten years but the results from the Trent Regional Arthroplasty Study are better than have been previously published, with more than 9% of prostheses surviving to 15 years. 2,6 The authors would like to acknowledge the surgeons of the Trent region for participating in Trent (and Wales) Regional Arthroplasty Study (the then-titled Trent Regional Arthroplasty Study). The authors would also like to acknowledge Dr Nick Taub (PhD), research fellow in medical statistics, Trent RDSU. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. References 1. Furnes O, Espehaug B, Lie SA, et al. Early failures among 7174 primary total knee replacements: a follow-up study from the Norwegian Arthroplasty Register 1994-2. Acta Orthop Scand 22;73:117-29. 2. Tayot O, Ait Siselmi T, Neyret P. Results at 11.5 years of a series of 376 posterior stabilised HLS 1 total knee replacements: survivorship analysis and risk factors for failure. Knee 21;8:195-25. 3. Himanen AK, Belt E, Nevalainen J, Hamalainen M, Lehto MU. Survival of AGC total knee arthroplasty is similar for arthrosis and rheumatoid arthritis: Finnish Arthroplasty Register report on 8,467 operations carried out between 1985 and 1999. Acta Orthop 25;76:85-8. 4. Oliver MC, Keast-Butler OD, Hinves BL, Shepperd JAN. A hydroxyapatite-coated Insall-Burstein II total knee replacement: 11-year results. J Bone Joint Surg [Br] 25;87-B:478-82. 5. No authors listed. Annual Report 25. The Swedish Knee Register, 26. www.knee.nko.se/english/online/uploadedfiles/14_skar%225_engl.pdf (date last accessed 29 August 27). 6. 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 21;72:53-13. 7. Pradhan NR, Gambhir A, Porter ML. Survivorship analysis of 3234 primary knee arthroplasties implanted over a 26 year period: a study of 8 different implant designs. Knee 26;13:7-11. 8. Noble P, Conditt MA, Cook KF, Mathis MB. Patient expectations affect satisfaction with total knee arthroplasty. Clin Orthop 26;452:35-43. 9. Barrack RI, Engh G, Rorabeck C, Sawhney J, Woolfrey M. Patient satisfaction and satisfaction after septic versus aseptic revision total knee arthroplasty. J Arthroplasty 2;15:99-3. 1. Baker PN, van der Meulen JH, Lewsey J, Gregg PJ. The role of pain and function in determining patient satisfaction after total knee replacement: data from the national Joint Registry for England and Wales. J Bone Joint Surg [Br] 27;89-B:893-9. 11. No authors listed. Trent (and Wales) Regional Arthroplasty Audit Group. http:// www.2.le.ac.uk/departments/healthsciences/extrone+/ecc/orthopaedicsurgery/ TWAAG (date last accessed 22 September 27). 12. No authors listed. NHS strategic Tracing Service. www.connectingforhealth.nhs.uk/ systemsandservices/nsts/index.html (date last accessed 22 September 27). 13. No authors listed. Surgical and orthopaedic implant information for total hip and total knee replacement procedures performed in Canada May 21-March 22. Canadian Joint Replacement Register. www.cihi.ca/cihiweb/en/downloads/ services_cjrr_bl_may22.pdf (date last accessed 29 August 27). 14. No authors listed. Third Annual Report of National Joint Register of England and Wales. 26. www.njrcentre.org.uk/documents/reports/annual/3rd/njr_ar2.pdf (date last accessed 23 September 27). 15. Phillips JE, Crane TP, Noy M, Elliott TS, Grimer RJ. The incidence of deep prosthetic infections in a specialist orthopaedic hospital: a 15 year prospective study. J Bone Joint Surg [Br] 26;88-B:943-8. THE JOURNAL OF BONE AND JOINT SURGERY