Despite the availability of many hip and knee replacement prostheses with excellent clinical track records, new prostheses are continually being introduced into the market1,2. Most new prostheses have limited or no clinical evidence to justify their use or to enable surgeons to determine their outcome compared with those of established prostheses. The expectation of new technology is that it will provide a better outcome than what is already available. Unfortunately, there is evidence to suggest that this may not be the situation with hip and knee replacements1,3-5.
This study was undertaken to determine if the introduction of new hip and knee prostheses is associated with an improved outcome at a national level over a five-year period.
This study was undertaken by the Australian Orthopaedic Association National Joint Replacement Registry (AOA NJRR). The Registry was established in 1999 and was completely implemented nationally by mid-2002. It receives information from all hospitals (public and private) performing joint replacement. It collects a defined minimum data set that enables the outcomes to be determined on the basis of a defined number of variables such as patient characteristics, prosthesis type and features, methods of prosthesis fixation, and surgical techniques.
This study involved the analysis of all new hip and knee arthroplasty prostheses introduced to the Australian market between January 1, 2003, and December 31, 2007, that had been used in at least 100 procedures and had a minimum duration of follow-up of one year. A new prosthesis was defined as a prosthesis whose first use was recorded by the AOA NJRR during the study period.
The cumulative percent revision of the prosthesis was calculated with use of Kaplan-Meier estimates of survivorship6. The 95% confidence interval was calculated with use of point-wise Greenwood estimates. The performance of each new prosthesis was compared with the combined result of the three best performing prostheses in its class. Determination of the three best performing prostheses was based on the cumulative percent revision of each prosthesis used in more than 1000 Registry-recorded procedures and followed for a minimum of five years.
Hazard ratios from Cox proportional hazard models, adjusted for age and sex, were used to compare the rate of revision of each new prosthesis and the combined rate of revision of the best three established prostheses in the corresponding group. Hazard ratios were determined over the entire period of use for each new prosthesis. All tests were two-tailed at the 5% level of significance.
Analysis was performed with use of SAS version 9.2 (SAS Institute, Cary, North Carolina).
Source of Funding
The AOA NJRR is funded by the Federal Australian Government through a contract with the Australian Orthopaedic Association.
Hip prostheses included conventional and resurfacing total hip implants. Conventional hip implants were subcategorized into four groups: cementless femoral, cementless acetabular, cemented femoral, and cemented acetabular. Resurfacing hip implants were subcategorized as femoral or acetabular resurfacing.
Knee prostheses included unicompartmental and primary total knee implants. Unicompartmental implants were subcategorized as femoral or tibial. Primary total knee implants were subcategorized as cementless femoral, cementless tibial, cemented femoral, or cemented tibial.
There were 266 new hip and knee (167 hip and ninety-nine knee) prostheses introduced to the Australian market in the five-year period between January 1, 2003, and December 31, 2007. Most of these new prostheses were not used in large numbers of procedures; only 22.9% were used in more than 100. This low use rate was evident for both new hip and new knee prostheses, with only 19.8% of the hip implants and 28.3% of the knee implants used in more than 100 procedures.
With regard to conventional hip replacement, the three best performing cementless femoral stems were the Citation, Secur-Fit Plus, and Summit. Fifty-eight new cementless femoral stems were introduced during the study period; eleven were used in over 100 procedures. Two, the Anca Fit and Profemur Z, had a significantly higher rate of revision than the comparators. The three best cementless acetabular components were the Mallory Head, Option, and Vitalock. There was a total of forty-six new cementless acetabular prostheses; twelve were used in over 100 procedures. Three, the ASR, Bionik, and Plus 2000, had a significantly higher rate of revision than the comparators. There were twenty-one new cemented femoral stems, but only one—the Adapter—was used in over 100 procedures. This had a significantly higher rate of revision than the comparators (the CPCS, Exeter, and MS-30). There were twenty-seven new cemented acetabular prostheses, but none were used in more than 100 procedures. The three best performing cemented acetabular prostheses were the Charnley Ogee, Exeter, and Reflection.
The three best resurfacing prostheses for both the acetabulum and the femur were the BHR, Conserve Plus, and Cormet 2000. Eight new resurfacing acetabular prostheses were introduced during the study period, and five were used in more than 100 procedures. Of these, the ASR and Durom had a significantly higher rate of revision than the comparators. There were seven new resurfacing femoral prostheses, and four were used in over 100 procedures. Of these, the ASR and Durom had a significantly higher rate of revision.
The comparative analysis of all new hip prostheses used in more than 100 procedures demonstrated that ten of the thirty-three prostheses had a significantly higher rate of revision compared with the best performing prostheses in their class. Importantly, none of the new hip prostheses performed better than the established prostheses (Table I).
Seven new unicompartmental femoral components were introduced during the study period. Four of these were used in over 100 procedures. The three comparator prostheses were the M/G, Repicci, and Unix. Two unicompartmental femoral prostheses, the AMC and balanSys, had a significantly higher rate of revision than the comparators. Eleven new unicompartmental tibial components were introduced, and four of them were used in over 100 procedures. The comparators and the tibial components with a higher rate of revision were the same as those found in the analysis of the unicompartmental femoral prostheses.
Eighteen new cementless femoral total knee prostheses were introduced; five were used in more than 100 procedures and one of them, the Columbus, had a significantly higher rate of revision than the comparators (the AGC, Natural Knee, and NexGen). Thirteen new cementless tibial prostheses were introduced, and four of them were used in over 100 procedures. One, the Columbus, had a significantly higher rate of revision than the comparators (the Coordinate, Natural Knee, and NexGen).
Twenty-five new cemented femoral total knee and twenty-five new cemented tibial total knee prostheses were introduced into the market. Six of the cemented femoral and five of the cemented tibial prostheses were used in over 100 procedures. The comparators for the cemented femoral components were the Kinemax Plus, Natural Knee, and NexGen, and the comparators for the cemented tibial components were the AGC, Natural Knee, and NexGen. None of the cemented femoral prostheses and two of the cemented tibial prostheses (Rocc and Optetrak RBK) had a significantly higher rate of revision than the comparators.
The comparative analysis of all new knee prostheses used in more than 100 procedures demonstrated that eight of the twenty-eight prostheses had a significantly higher rate of revision compared with the best performing prostheses in their class. Similar to the new hip prostheses, none of the new knee prostheses performed better than the comparators (Table II).
Over 75% of new hip and knee prostheses introduced to the Australian market were not used in large numbers. The reason for this almost certainly relates to a lack of a clinical or theoretical advantage attributable to these individual prostheses. It is likely that clear evidence that a prosthesis provides a substantial advantage in a particular clinical situation may be necessary to ensure substantial use. However, most new prostheses do not have any supportive clinical data specific to them. Despite this lack of clinical evidence, almost 25% of new prostheses were used in over 100 procedures. None of these prostheses had a better outcome compared with the best performing prostheses within the same class. Most (71%) performed equally well; however, the remainder (29%) had a significantly higher rate of revision. The percentage that did not perform as well was similar for both new hip (30%) and new knee (29%) prostheses.
It could be argued that if no new hip or knee prostheses had been approved for use during the study period, then the overall outcome for both hip and knee replacement surgery may have been better. Not only has the introduction of this new technology been potentially detrimental to patient care, but the current approach may also be an important driver of increased health-care costs—both because of the potential for increased revision rates and because new technology usually costs more than existing technology2. The current approach to the introduction of new hip and knee replacement prostheses does not appear to be either clinically efficacious or cost-effective.
This is not the first study to link the introduction of new joint replacement prostheses to less satisfactory outcomes1,3-5. The importance of this study is that it is the most extensive and comprehensive one that has been undertaken to date. The authors examined the introduction of new prostheses for a number of different joint replacement procedures nationally over a five-year period.
The findings have potentially important implications. They demonstrate that the current approach to the development and regulation of new hip and knee prostheses may need to be reviewed. Orthopaedic manufacturers must carefully consider the reasons for introducing new technology as well as the pathway to development and assessment of new hip and knee replacement prostheses. Currently, it appears that most new technology is rarely used, and when it is used it is no better than what is already available. The finding that a substantial number of new prostheses are associated with a potential for a less satisfactory outcome indicates that premarket assessment may be inadequate.
In order to prevent the introduction of these unsatisfactory prostheses, a clear strategy should be in place to ensure that they never reach the market. Currently, there are no standards that define what is required for premarket assessment of hip and knee replacement prostheses and the approach to the assessment by both manufacturers and regulators is ad hoc. Being that a large number of hip and knee prostheses are already available on the market and have had excellent long-term clinical outcomes, it is imperative that new prostheses be introduced only if there is sufficient evidence to justify their introduction. Therefore, the type and amount of information that are required to undertake a premarket assessment need to be reevaluated and clarified. Also, once prostheses are implanted, they should be carefully monitored. No amount of stringent regulation of premarketing could definitely predict all possible failures. Therefore, regulators must develop more stringent approaches to both product approval and postmarket surveillance.
It could be argued that changing the regulatory classification of hip and knee prostheses may negatively impact the development of new technology in this area. As a consequence, patients may be denied beneficial innovation or beneficial innovation in a timely fashion. However, the whole purpose of the introduction of new prostheses is to improve the patient's outcome. One needs to be mindful that the value of an intervention does not reside in its newness or unique mechanism of action; rather, it lies in its ability to improve the patient's outcome. Identifying prostheses that are prone to an increased risk of revision will prevent the introduction of these prostheses and will ultimately prevent iatrogenic harm to patients. Not only would there be a reduction in secondary surgical procedures, arthroplasty revisions, and the complications associated with each, but there would also be cost savings.
Finally, it is also clear that orthopaedic surgeons need to be more discriminatory and evidence-based in their approach to their choice of prosthesis. However, this poses considerable difficulties for surgeons as the available literature is limited in its ability to identify and recommend specific prostheses. Caution must be exercised when assessing advertisements and industrial claims. It seems sensible to change to a new prosthesis only if there is sufficient evidence that it is likely to perform better than what the surgeon is currently using.