Total knee arthroplasty (TKA) and unicompartmental knee arthroplasty have been used successfully to treat patients with osteoarthritis, rheumatoid arthritis, and similar conditions for several decades. The numbers of primary and revision surgeries attributable to osteoarthritis are increasing exponentially worldwide causing strain on hospitals and economic resources, while the number of primary surgeries attributable to rheumatoid arthritis is decreasing slightly.7 Reasons for the increase in surgery include many factors, but increasing lifespan, obesity, a more active lifestyle of the patients, and safer anesthesiologic procedures are some of the primary contributors. The success rate of the longevity of the implants has increased with time and surgical techniques and implant components have been improved dramatically.
However, the success rate for younger patients treated with TKA or unicompartmental knee arthroplasty has been a topic for discussion for a long time. In many countries implant surgery will not be considered unless the patient is 60 to 65 years old.10 Several studies have been done to clarify how the success rate is related to the patient’s age at initial surgery and have shown different results. Diduch et al3 observed 88 patients, younger than 55 years, treated with TKA and reported a survival rate of 87% at 18 years. Duffy et al5 studied 74 TKAs in 54 patients younger than 55 years and reported an estimated survival rate of 99% at 10 years. Lonner et al10 examined 32 patients with osteoarthritis younger than 40 years who were treated with TKAs and reported an aseptic failure rate of 12.5% at 8 years. Ranawat et al13 observed 62 patients younger than 55 years who were treated with TKAs and reported a survivorship rate of 96% at 10 years. 13 These authors all stated that their results were comparable to the survival rate of older patients.
These results did not agree with experiences in Sweden. As part of a research project in which the goal was to custom design knee implants for younger patients, a new study was done to clarify the mixed results from previous studies. It was hypothesized that younger patients would have a lower survival rate than older patients and that a larger number of patients included in the study would show a different and more accurate result. To test the hypothesis a group of younger patients were compared with a group of older patients who during the same time had been treated in Sweden for similar conditions. Some of the previous studies included a mix of patients with rheumatoid arthritis, osteoarthritis, and similar conditions, which could have affected the results.3,5,13 Therefore only patients with osteoarthritis and similar conditions were included in our study. In addition to comparing the risk for revision between the two age groups, we compared the risk for revision between genders and looked at how the risk for revision is affected by the year of surgery.
MATERIALS AND METHODS
The analyses were done using data available in the Swedish Knee Arthroplasty Registry, which contains information about most knee arthroplasties done in Sweden since 1975.7 According to Swedish law no written consent from the patients was necessary and the study was done following the rules stated by the Swedish Data Inspection. The Swedish Knee Arthroplasty Registry is a unique registry with more than 90,000 arthroplasties registered. The primary interest group for this study was younger patients who had TKAs or unicompartmental knee arthroplasties. According to the results from the previous-mentioned study,7 the number of arthroplasties done for patients with osteoarthritis is increasing rapidly whereas the number of arthroplasties done for patients with rheumatoid arthritis has decreased slightly with time. Although patients with osteoarthritis and rheumatoid arthritis are receiving the same treatment of their joint disease, the causes of their arthritis are different. When designing this analysis it was decided to include only patients treated with TKA or unicompartmental knee arthroplasty for osteoarthritis (2349 in the group of younger patients and 32,378 in the group of older patients) and similar conditions such as osteochondritis (21 in the group of younger patients and 306 in the group of older patients), osteochondritis (four in the group of younger patients and three in the group of older patients), and posttraumatic arthritis (232 in the group of younger patients and 564 in the group of older patients). When patient information is submitted to the registry, the treating physician specifies the cause for surgery. It is suspected that in some cases the cause is specified as osteoarthritis when in fact it is attributable to a different cause such as osteonecrosis, osteochondritis, or posttraumatic arthritis. Because some cases of these conditions can be hidden in the osteoarthritis group, it was decided to include all patients in the study. The age groups were defined as patients 60 years or older and patients younger than 60 years. The hypothesis for the analysis was that the group of younger patients would have a higher cumulative rate of revision surgery than the older patients. Early experiences with unicompartmental knee arthroplasties discouraged their use in the United States because of the higher revision rate. However, the development and improvements of unicompartmental knee arthroplasty components have led to their increasing use in Europe.9 Robertsson et al14 compared 15,437 primary TKAs and 10,624 unicompartmental knee arthroplasties done between 1985 and 1995. The results showed that despite the higher cumulative revision rate the unicompartmental knee arthroplasty could be recommended for patients with unicompartmental arthritis because it has a lower cost, is a less complicated procedure, and has a lower number of serious complications. Because unicompartmental knee arthroplasties are not widely used in the United States, the two were analyzed separately. Table 1 shows the number of patients in the different groups included in the analysis. The average age for the group of younger patients was 55.18 years with a range from 25.62 to 59.99 years. The average age for the group of older patients was 73.21 years with a range from 60.0 to 95.83 years.
;)
Table 1: Number of Patients Included in the Analysis From Each Age Group
Sixty-eight percent of patients with TKAs and 62% of patients with unicompartmental knee arthroplasties were female. To exclude the effect of obsolete and inferior models on results, only data from 1988 to 1997 were included in this study. The current study was designed to represent the average implant surgery but the majority of TKAs and unicompartmental knee arthroplasties were done using cemented components and the majority of TKAs were done with retained posterior cruciate ligaments.
The cumulative revision rate was calculated using the life-table method with 1-month intervals.1,8 The 95% confidence intervals were achieved using the Wilson quadratic equation with Greenwood and Peto effective sample-size estimates.4 The curves were cut off when fewer than 40 patients remained at risk. Cox regression also was used to calculate the risk-ratio between the two age groups. Year of surgery and gender also were included as factors. Cox regression often is used when survival data are analyzed in medical or healthcare research projects.2,6,11 For categorical variables in the model (age, group, and gender), one category is defined as a reference having the relative risk of 1.0, with which the other categories are then compared. For numerical variables (year of operation), the risk ratio relates to the change in risk if the variable increases by one unit. Three analyses were done separately for TKAs and unicompartmental knee arthroplasties. The first analysis considered all causes for revision as the end point. The second analysis only considered aseptic loosening as the end point because earlier studies showed loosening as the most common cause for failure. The third analysis only considered revision because of infection as the end point, which used to be a serious problem for patients having arthroplasties.
RESULTS
The results from the first analysis (Fig 1) show a higher cumulative revision rate for the group of younger patients (mean, 13% at 8.5 years) compared with the group of older patients (mean, 6% at 8.5 years) treated with TKAs when any cause for revision is included. The Y axis in the life-table graph shows the cumulative revision rate in percent and the X axis shows the number of years after the primary arthroplasty. Because the curves were cut off when fewer than 40 patients remained at risk for revision, the younger group ends after 8.5 years but the older group ends after 10 years.
Fig 1.:
(A) The cumulative revision rate is shown for patients 60 years or older treated with TKAs. All reasons for revision are included in this analysis. (B) The cumulative revision rate is shown for patients younger than 60 years treated with TKAs. All reasons for revision are included in this analysis.
The regression analysis (Table 2) shows that the risk for revision in the group of older patients is lower (risk ratio, = 0.49; confidence interval, 0.38–0.62; p = 0.0000) compared with the risk for the group of younger patients. Regression analysis also shows the risk for revision decreased, attributable to the year of surgery (risk ratio, 0.92; confidence interval, 0.89–0.96; p = 0.0000). In the first case the risk for revision was set to 1.0 for the group of younger patients with which the risk for revision for the group of older patients was compared. In the second case the risk for revision attributable to the year of surgery was set to 1.0 and the risk for the following year was compared with this. The risk for revision attributable to the year of surgery decreased each year for both groups, probably because of better implant systems and improved surgical techniques. The risk ratio attributable to gender, considering all reasons for revision, is not statistically significant (p = 0.6874) and no conclusions can be made. The results from the second analysis (Fig 2) show a higher cumulative revision rate for the group of younger patients (mean, 6% at 8.5 years) compared with the rate for the group of older patients (mean, 2.5% at 8.5 years) treated with TKAs when only revisions attributable to loosening of components are included. The Y axis in the life-table graph shows the cumulative revision rate in percent and the X axis shows the number of years after the primary arthroplasty.
Table 2: Cox Regression Table for Revision Attributable to Any reason
Fig 2.:
(A) The cumulative revision rate for patients 60 years or older treated with TKAs includes only revisions attributable to loosening of components in this analysis. (B) The cumulative revision rate for patients younger than 60 years treated with TKAs includes only revisions attributable to loosening of components in this analysis.
The regression analysis (Table 3) shows that the risk for revision attributable to loosening of components in the group of older patients is lower (risk ratio, 0.41; confidence interval, 0.27–0.62; p = 0.0000) compared with the risk for the group of younger patients, and the risk for revision decreased attributable to the year of surgery (risk ratio, 0.87; confidence interval, 0.82–0.94; p = 0.0001). In the first case the risk for revision was set to 1.0 for the group of younger patients, and the risk for revision for the group of older patients was compared with that. In the second case the risk for revision attributable to the year of surgery was set to 1.0, which the risk for the following year was compared with. The risk for revision attributable to the year of surgery decreased each year for both groups. The risk ratio attributable to gender considering only loosening of components as the reason for revision is not statistically significant (p = 0.1395) and no conclusions can be made.
Table 3: Cox Regression Table for Revision Attributable to Loosening of Components
The third analysis shows a low cumulative revision rate for both groups of patients treated with TKAs when only revision surgery attributable to infection is considered. The low number of revisions done because of infection agrees with results of current studies showing a decreasing problem with patients having infection.15 The regression analysis shows that there is no difference between the two age groups regarding risk of revision surgery because of infection. The risk for revision attributable to infection decreased each year for both age groups (risk ratio, 0.91; confidence interval, 0.85–0.96; p = 0.0015). Men are more likely to have revision surgery (risk ratio, 1.64; confidence interval, 1.23–2.18; p = 0.0007) attributable to infection compared with women.
The same analyses were done on the data for the patients with unicompartmental knee arthroplasties and similar results were achieved but with a slightly higher cumulative revision rate. The results from the first analysis (Fig 3) show a higher cumulative revision rate for the group of younger patients (mean, 22% at 9.2 years) compared with the rate for the group of older patients (mean, 14% at 9.2 years) treated with unicompartmental knee arthroplasty when any cause for revision is included. The Y axis in the life-table graph shows the cumulative revision rate in percent and the X axis shows the number of years after the primary arthroplasty. Because the curves were cut off when fewer than 40 patients remained at risk for revision, the younger group ends after 9.2 years but the older group ends after 10 years.
Fig 3.:
(A) For patients 60 years or older (B) and younger than 60 years treated with unicompartmental knee arthroplasties, reasons for revision were included in this analysis.
The regression analysis (Table 4) shows that the risk for revision in the group of older patients is lower (risk ratio, 0.55; confidence interval, 0.45–0.65; p = 0.0000) compared with the risk for the group of younger patients, and the risk for revision decreased attributable to the year of surgery (risk ratio, 0.94; confidence interval, 0.91–0.97; p = 0.0001). In the first case the risk for revision was set to 1.0 for the group of younger patients, which the risk for revision for the group of older patients was compared with. In the second case the risk for revision attributable to the year of surgery was set to 1.0, which the risk for the following year was compared with. The risk for revision attributable to the year of surgery decreased each year for both groups because of better implant systems and improved surgical techniques. The risk ratio attributable to gender considering all reasons for revision is not statistically significant (p = 0.7122) and no conclusions can be made.
Table 4: Cox Regression Table for Revision Attributable to Any reason
The results from the second analysis (Fig 4) shows a higher cumulative revision rate for the group of younger patients (mean, 8% at 9.5 years) compared with the rate for the group of older patients (mean, 6.5% at 9.5 years) treated with unicompartmental knee arthroplasty when only revisions attributable to loosening of components are included. The Y axis in the life-table graph shows the cumulative revision rate in percent and the X axis shows the number of years after the primary arthroplasty.
Fig 4.:
(A) For patients 60 years or older and (B) for patients younger than 60 years treated with unicompartmental knee arthroplasties, only revisions attributable to loosening of components were included in this analysis.
The regression analysis (Table 5) shows that the risk for revision attributable to loosening of components in the group of older patients is lower (risk ratio, 0.63; confidence interval, 0.48–0.83; p = 0.0012) compared with the risk for the group of younger patients. Regression analysis also shows the risk for revision decreased attributable to the year of surgery (risk ratio, 0.91; confidence interval, 0.87–0.96; p = 0.0002). In the first case the risk for revision was set to 1.0 for the group of younger patients, which the risk for revision for the group of older patients was compared with. In the second case the risk for revision attributable to the year of surgery was set to 1.0, which the risk for the following year was compared with. The risk for revision attributable to the year of surgery decreased each year for both groups attributable to better implant systems and improved surgical techniques. The risk ratio attributable to gender considering only loosening of components as the reason for revision is not statistically significant (p = 0.2314) and no conclusions can be made.
Table 5: Cox Regression Table for Revision Attributable to Loosening of Components
The third analysis shows a low cumulative revision rate for both age groups when considering only infection as the cause for revision; however, the group of younger patients did have a slightly higher cumulative revision rate. The regression analysis does not show a statistically significant result regarding revision attributable to infection when comparing the two age groups and the year of surgery. The risk for men to have revision surgery because of infection is higher (risk ratio, 1.88; confidence interval, 1.13–3.14; p = 0.0156) when compared with women.
DISCUSSION
This study was done to clarify if younger patients with osteoarthritis treated with TKA or unicompartmetnal knee arthroplasty have a higher cumulative revision rate compared with older patients. Several studies done on a small number of patients have shown little or no difference in results between younger and older patients.3,5,13 To ensure that the results from the current study are statistically significant, more than 35,000 patients were included. The patient data that were used for this study were taken from the Swedish Knee Arthroplasty Registry. When the patient information is reported to the registry, the treating physician decides the cause for revision, which might not be correctly stated in all cases. However, all cases of osteoarthritis and similar conditions were included in the study and the large number of data assure that the results are reliable.
The results from the current study agree with the hypothesis and they show that patients younger than 60 years have a higher risk of early revision compared with patients older than 60 years. The reasons for the varying results between the current study and earlier studies can be discussed but obviously they have different designs. Ranawat et al 13 did a study on 62 patients (93 knees) younger than 55 years. The reasons for TKAs were rheumatoid arthritis (76 knees) or osteoarthritis (17 knees). The results showed a 96% cumulative survivorship at 10 years, which is similar to results for older patients. Despite the excellent results from this study, Ranawat et al expressed concern about treating younger patients with TKAs. In their study the majority of the patients were treated for rheumatoid arthritis, which might explain the high survivorship because these patients tend to be less active. Diduch et al3 did a similar study on 88 patients (114 TKAs) with an average age of 51 years. In their study, all patients were treated for severe osteoarthritis, the same components were used on all patients, and two physicians did all surgeries. The overall survival rate varied between 94% and 87% at 18 years dependent on which failure causes were included. The small number of patients included in their study could have affected the results and other factors. Because the same two surgeons did all arthroplasties the patients most likely came from the same geographic area. The current study included patients from all of Sweden, with different living conditions. The surgeries were done by a large number of surgeons using a wide variety of implant components. Duffy et al5 studied 74 TKAs in 54 patients who were 55 years or younger with an average age of 43 years. Forty-seven knees were treated for rheumatoid arthritis and the remainder of the knees were treated for gonarthrosis (12 knees), posttraumatic arthritis (six knees), osteonecrosis (10 knees), and other related diseases (six knees). The average followup was 13 years (range, 10–17 years), and the estimated survival was 99% at 10 years and 95% at 15 years after primary surgery. Lonner et al10 reported on patients 40 years and younger with osteoarthritis treated with TKAs. Thirty-two patients were included in the analysis with a mean followup of 7.9 years. The results showed an aseptic failure rate of 12.5% after 8 years, which agrees with the results from the current study. Several of the referenced articles describe the higher activity level of younger patients as the main reason for early failures. The referenced studies all were done using a small number of patients and it was not stated whether the results were statistically significant.3,5,10,13 None compared the group of younger patients with a group of older patients and only compared their results with other studies done with older patients. There also might be differences attributable to the geographic location and social culture. The American society depends on using a car, leading to a less active lifestyle. The society in Sweden and most parts of Europe is different and much more walking is required for daily living. Also, according to common knowledge, problems with obesity are not as frequent in Europe as in the United States, which also leads to a more active lifestyle. We think that the current study is more complete than other studies and the results are more reliable because of the large number of patients included. The reason for excluding patients treated for rheumatoid arthritis was that these patients are not as active as patients treated for osteoarthritis because of the severity of their disease. Patients with rheumatoid arthritis usually experience additional pain caused by symptoms other than deteriorating cartilage in the joints. The results from the analyses confirmed previous findings that aseptic loosening is a major reason for revision surgery regardless of age. Infection is a decreasing problem for both age groups and the significant results of gender do not prove that men are more likely to get infections, just that men more often need revision surgery because of infection than women. It is possible that women get infections as often as men but they are treated successfully through nonsurgical procedures. We think that the significant difference between the two age groups should be recognized and treatments using current designs of knee implants on younger patients should be used carefully. The fact that, regardless of age, unicompartmental knee arthroplasties have a higher cumulative revision rate than TKAs suggests that these procedures should be considered. However, according to a study done by Newman et al,12 patients treated with unicompartmental knee arthroplasty have a larger range of motion than patients treated with TKA. The increased mobility allows the patients to live a more active life. Robertsson et al14 concluded that revising a unicompartmental knee arthroplasty with a TKA typically provides a very good result. Based on these results and the fact that unicompartmental knee arthroplasties are less invasive, less costly, have a lower risk for complications, and result in a shorter recovery period, we recommend this treatment as a good solution for primary knee arthroplasties only.
Acknowledgment
We thank the Swedish Knee Arthroplasty Registry for cooperation and support.
References
1. Armitage P, Berry G: Statistical Methods in Medical Research. Survival Analysis, 2nd Edition. Oxford, Blackwell Scientific Publications 469–492, 1994.
2. Collett D: Modeling Survival Data In Medical Research. In Chatfield C, Zidek JV (eds). Time-Dependent Variables. London, Chapman & Hall 223–235, 1994.
3. Diduch D, Insall J, Scott W, Scuderi GR, Font-Rodrigues D: Total knee replacement in young, active patients: Long-term follow-up and functional outcome. J Bone Joint Surg 79A:575–582, 1997.
4. Dorey F, Nasser S, Amstutz H: The need for confidence intervals in the presentation of orthopaedic data. J Bone Joint Surg 75A:1844–1852, 1993.
5. Duffy G, Trousdale R, Stuart M: Total knee arhtroplasty in patients 55 years old or younger: 10- to 17-year results. Clin Orthop 356:22–27, 1998.
6. Kleinbaum DG: Survival Analysis. A Self-Learning Text. In Dietz K, Gail M, Krickeberg K, Singer B (eds). The Cox Proportional Hazards Model and Its Characteristics. New York, Springer 84-120, 1996.
7. Knutson K, Lewold S, Robertsson O, Lidgren L: The Swedish Knee Arthroplasty Register: A nation-wide study of 30,003 knees 1976-1992. Acta Orthop Scand 65:375–386, 1994.
8. Lee ET: Statistical Methods For Survival Data Analysis. Nonparametric Methods of Estimating Survival Functions, 2nd Edition. Belmont, CA Lifetime Learning Publications 66–99, 1980.
9. Lewold S, Robertsson O, Knutson K, Lidgren L: Revision of unicompartmental knee arthroplasty: Outcome in 1,135 cases from the Swedish Knee Arthroplasty Study. Acta Orthop Scand 69:469–474, 1998.
10. Lonner JH, Hershman S, Mont M, Lotke PA: Total knee arthroplasty in patients 40 years of age and younger with osteoarthritis. Clin Orthop 380:85–90, 2000.
11. Marubini E, Valsecchi MG: Analysing Survival Data from Clinical Trials and Observational Studies. The Cox Regression Model. Chichester, John Wiley & Sons Ltd 171-200, 1995.
12. Newman JH, Ackroyd CE, Shah NA: Unicompartmental or total knee replacement? Five-year results of a prospective randomized trial of 102 osteoarthritic knees with unicompartmental arthritis. J Bone Joint Surg 80B:862–865, 1998.
13. Ranawat C, Padgett D, Ohashi Y: Total knee arthroplasty for patients younger than 55 years. Clin Orthop 248:27–33, 1989.
14. Robertsson O, Borgquist L, Knutson K, Lewold S, Lidgren L: Use of unicompartmental instead of tricompartmental prostheses for unicompartmental arthrosis in the knee is a cost-effective alternative: 15,437 primary tricompartmental prostheses were compared with 10,624 primary medial or lateral unicompartmental prostheses. Acta Orthop Scand 70:170–175, 1999.
15. 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 72:503–513, 2001.
