Secondary Logo

Journal Logo

Why Are Total Knee Arthroplasties Failing Today?

Sharkey, Peter F., MD*,**; Hozack, William J., MD*,**; Rothman, Richard H., MD, PhD*,**; Shastri, Shani, MD*,†; Jacoby, Sidney M., BA

Clinical Orthopaedics and Related Research®: November 2002 - Volume 404 - Issue - p 7-13
SECTION I SYMPOSIUM: The Papers Presented at the Knee Society Meeting 2002: Insall Award Paper

The incidence of failure after knee replacement is low, yet it has been reported that more than 22,000 knee replacements are revised yearly. The purpose of the current study was to determine current mechanisms of failure of total knee arthroplasties. A retrospective review was done on all patients who had revision total knee arthroplasty during a 3-year period (September 1997–October 2000) at one institution. The preoperative evaluation in conjunction with radiographs, laboratory data, and intraoperative findings were used to determine causes of failure. Two hundred twelve surgeries were done on 203 patients (nine patients had bilateral surgeries). The reasons for failure listed in order of prevalence among the patients in this study include polyethylene wear, aseptic loosening, instability, infection, arthrofibrosis, malalignment or malposition, deficient extensor mechanism, avascular necrosis in the patella, periprosthetic fracture, and isolated patellar resurfacing. The cases reviewed included patients who had revision surgery within 9 days to 28 years (average, 3.7 years) after the previous surgery. More than half of the revisions in this group of patients were done less than 2 years after the index operation. Fifty percent of early revision total knee arthroplasties in this series were related to instability, malalignment or malposition, and failure of fixation.

From *The Rothman Institute, Philadelphia, PA; **Orthopaedic Surgery, Thomas Jefferson University Hospital, Philadelphia, PA; Thomas Jefferson University Hospital, Philadelphia, PA; and Jefferson Medical College, Thomas Jefferson University, Philadelphia, PA.

Reprint requests to Peter F. Sharkey, MD, Rothman Institute, 925 Chestnut Street, Philadelphia, PA 19107. E-mail:

DOI: 10.1097/01.blo.0000036002.13841.32

Total knee arthroplasty is a successful operation when measured by pain relief and functional improvement. Several authors reported patient satisfaction rates of 90% to 95%. 8,10,22 Additionally, the results of surgery are durable with 10- to 15-year implant survival of greater than 90%. 15,27 Nonetheless, some patients achieve a poor early result after surgery or the implant fails prematurely and a revision operation is required. In 1982, Cameron and Hunter 6 and Rand and Bryan 23 independently reported their experience with total knee replacement failure and revision surgery. The most common indications for revision in these series were infection, loosening, instability, and patellofemoral complications. Fehring et al 12 reported on 440 revision total knee arthroplasties done between 1986 and 1999. The primary cause of failure in this series was infection. Other common causes of failure were instability, loosening, and patellofemoral problems. Interestingly, 64% of the patients had revision surgery within 5 years of their index procedure. Other authors have described failure mechanisms that seem to be design-specific, such as metal tibial tray fracture, 1 accelerated polyethylene wear, 2,14 and metal-backed patellar component failure. 7

Other than the recent report of Fehring et al, 12 there has been scant literature on failure mechanisms of total knee arthroplasties. In terms of understanding the pathogenesis of contemporary total knee arthroplasty failure, prior investigations are dated and have deficiencies. In most series, the emphasis was on the results of knee revision surgery, not the mechanism of failure of the primary knee replacement. In previous studies, minimal demographic data are provided making epidemiologic conclusions difficult. Furthermore, in each of these series, the revisions were done during a relatively long interval and failure occurred for various reasons. In addition, the mechanisms of failure may not represent contemporary causes for implant failure. Knowledge of the current mechanisms by which total knee arthroplasties fail would allow investigators to focus their efforts on pertinent problematic issues. The purpose of the current study was to determine why total knee arthroplasties are failing today.

Back to Top | Article Outline


All total knee revision arthroplasties done at one institution from September 23, 1997 to October 4, 2000 were studied retrospectively. Patient demographics, age, gender, height, and weight were documented. The interval from index total knee arthroplasty to revision surgery was recorded. The preoperative history, physical examination, and radiographic findings were considered when making an assessment of the failure mode. The cause of failure also was determined at surgery by the attending surgeon. The determination of failure mechanism was based on analysis of intraoperative findings that included an examination with the patient under anesthesia, gross inspection of the components, and fluid and tissue culture. In some patients, more than one cause of failure was identified and noted. Failures were subdivided into early (revision surgery done < 2 years after index operation) and late (revision surgery done > 2 years after the index operation) groups.

Back to Top | Article Outline


Two hundred twelve consecutive total knee arthroplasty revisions in 203 patients (nine patients had bilateral surgeries) were included in this study. One hundred twenty-eight patients were women and 75 patients were men. The average age of the men was 68.7 years (range, 31.6–88.7 years), and the average age of the women was 68.4 years (range, 36.1–96.9 years). The average height and weight of the men was 70 inches (range, 61–76 inches) and 211.5 lb (range, 134–348 lb), respectively. The average height and weight of women was 63.7 inches (range, 48–69 inches) and 192.1 lb (range, 105–389 lb), respectively.

The average time to revision surgery for the patients in the early failure group was 1.1 years (range, 9 days–2 years). In the patients in the late failure group, the average time to failure was 7 years (range, 2.2–28 years). Figure 1 shows the percentage of patients with each failure mechanism, stratified into early and late subgroups. Of the revisions, 55.6% were done less than 2 years after the index total knee arthroplasty. The remaining 42.4% of revisions were done more than 2 years after the index operation and were classified as late failures.

Fig 1.

Fig 1.

The most prevalent mechanism of failure was polyethylene wear. Significant wear was seen in 25% of all revisions. This failure mechanism was more common for the patients who had revision surgery more than 2 years after the index procedure, accounting for 44.4% of all revisions in the late failure group. Although polyethylene wear was not the primary mode of failure for any patient in the early failure group, 11.8% of patients who had revision surgery early had evidence of significant polyethylene damage. Most of these patients had revision surgery primarily for instability or malalignment.

In the early failure group, the most common failure mode was infection, responsible for 25.4% of all early revisions. In the late failure group, infection occurred in only 7.8% of patients.

Component loosening also was a common cause of early and late knee arthroplasty failure, 16.9% and 34.4%, respectively. Component loosening was associated with use of cemented and uncemented components at the time of index arthroplasty. Early loosening was associated with uncemented components. Additionally, surface cementing of the tibial component, without cement around the implant keel also was a cause of early failure. Instability was seen in 21.2% of early and 22.2% of late revision knee arthroplasty failures. Of the knee arthroplasties that required revision for instability, 37% were cruciate-sparing implants and the remainder were cruciate-substituting implants.

Significant stiffness associated with arthrofibrosis after total knee arthroplasty was present in 16.9% of patients requiring early revision surgery and 12.2% of patients who had late revision surgery.

Complications related to the extensor mechanism also were an important cause of knee arthroplasty failure. Patients had revision surgery for various extensor mechanism related complications including extensor mechanism deficiency (significant extensor lag [6.6%], avascular necrosis of the patella [4.2%], and revision of the unresurfaced patella [0.9%]).

Component malalignment or malposition was present overall in 11.8% of revisions. Malalignment or malposition of the components was present in 11.9% of early revision knee arthroplasties and 12.2% of late revision knee arthroplasties. Revision after periprosthetic fracture accounted for 2.8% of all revisions.

Back to Top | Article Outline


The percentage of patients who need early or late revision surgery after total knee arthroplasty is relatively small compared with the rates of failure for other medical interventions. Heck et al 17 reviewed global data and showed less than a 3% revision rate in the first 2 years after index total knee arthroplasty. Nonetheless, when the large number of total knee arthroplasties done annually is considered, this small percentage of failures translates into a significant absolute number. The financial cost of total knee arthroplasty failure also is a value that must be considered. It is conservatively estimated that the cost of a revision knee surgery is $11,922. 19 In 1999, 22,000 knee revision operations were done in the United States at an estimated expense of $262 million. 19 Knee revision surgery also causes significant emotional distress for the patient and the morbidity of the procedure is not inconsequential. Therefore, investigative efforts should be focused on determining the mechanisms for knee replacement failure and attention should be directed toward correcting these causes. Before this study, demographic data regarding total knee arthroplasty failure had not been reported, prohibiting epidemiologic analysis. Other studies 6,23 evaluating the mode of failure of total knee arthroplasty are dated and may not be pertinent to current failure mechanisms. In 1982, Cameron and Hunter 6 reported on 94 revision knee arthroplasties. Infection, polyethylene wear, instability, and loosening were reported to be common reasons for failure. Likewise, in 1982, Rand and Bryan 23 reported on 142 failed knee arthroplasties listing loosening, instability, and malalignment as the most frequent indication for revision surgery. More recently, Fehring et al 12 reported on 440 revision arthroplasties done between 1986 and 1999. They reported that failure of total knee arthroplasties frequently occurred early (< 5 years) after the index procedure and often was related to infection or surgical error. This study evaluated knee arthroplasties that failed as early as 1986 and reviewed data through a 13-year interval. The current series reviews revision surgeries done between September 23, 1997 and October 4, 2000 providing knowledge of contemporary causes of total knee arthroplasty failure.

In the current series, the primary causes of total knee arthroplasty failure were polyethylene wear (25%), aseptic loosening (24.1%), instability (21.2%) infection (17.5%), arthrofibrosis (14.6%), malalignment or malposition (11.8%), extensor mechanism deficiency (6.6%), avascular necrosis in the patella (4.2%), periprosthetic fracture (2.8%), and resurfacing of the unresurfaced patella (0.9%) (Table 1). In many patients, more than one failure mechanism was seen. Some of these failure modes could be minimized by design and material improvements of the implants. Other reasons for failure were related to surgical technique and the burden of lowering the incidence of these problems rests with the surgeon.



Infection rates in most series at high volume centers occurs in less than 1% of cases. If an institution achieves an infection rate of 1% or less, then reduction in the incidence of this complication is hard to improve. Infection after knee arthroplasty can occur either early or late. Early infections probably are perioperative in nature whereas late infections most likely result from hematogenous seeding. In the current series, 30 of 37 knee arthroplasties revised for infection had early or perioperative infections and the remainder most likely were the result of hematogenous spread. Postoperative wound problems such as persistent drainage, delayed healing, or hematoma, have been shown by Bliss and McBride 5 to lead to an increased risk for infection. Hartman et al 16 showed that aggressive treatment of postoperative wound problems could lower the incidence of chronic infection and enhance the likelihood of implant retention. Likewise, Segawa et al 25 showed that perioperative infection, recognized early, and treated with irrigation and debridement could lead to implant retention in 50% of cases. Based on this information, revision rates attributable to infection could be lowered further with early recognition and aggressive intervention.

Late hematogenous infection was a cause of failure seven times in the current series. There has been recent controversy about the use of prophylactic antibiotics after joint replacement. 9 Although hematogenous infection is a relatively uncommon cause of knee arthroplasty failure, for the patient who suffers this complication, it is a devastating event. The authors’ institution continues to recommend dental and procedural prophylactic antibiotics for their patients.

Twenty-five percent of implants in the current series had significant polyethylene wear. The average interval that these implants had been in place before failure was 7 years. Polyethylene wear is not considered as significant a problem for knee implants when compared with hip replacement components. Certain knee implant designs, however, have been associated with accelerated polyethylene wear and osteolysis. Feng et al 14 showed a very high polyethylene failure rate with use of a flat polyethylene articular surface. The number of failures in the current series attributable to wear was not inconsequential. Failure attributable to polyethylene wear in the current series was noted if gross polyethylene wear or damage was present. Hirakawa et al 18 characterized wear debris of failed knee arthroplasties. They reported that when compared with total hip arthroplasty, a larger range of particle sizes adjacent to failed knee arthroplasties indicating a higher failure rate attributable to delamination and fragmentation in the knee. Wasielewski et al 26 and Robinson et al 24 have shown that microscopic backside wear of modular tibial components can generate significant particles and lead to synovitis and osteolysis similar to the hip. Microscopic wear was not considered in the current series and all components were inspected visually. Admittedly, this type of wear is difficult to quantitate and the contribution of microscopic wear to component failure was not estimated.

Instability and component malalignment or malposition are complications that are controlled by surgical technique. Most clinicians agree that the instrumentation currently available for the insertion of total knee arthroplasty components is relatively precise and accurate. However, in certain situations, the available instrumentation does not guarantee the implantation of a perfectly aligned, positioned and stable component. Fehring and Valadie 13 reported on 25 revision knee arthroplasties done for instability. Inadequate prosthetic design and surgical error were shown to be common reasons for instability. Mitts et al 21 studied results of knee arthroplasty using a cruciate-retaining prosthesis. Overall the results were good except in a subgroup of patients with instability. They concluded that certain patients are not well suited for this implant design. Failure possibly could have been avoided with selection of a cruciate-substituting implant. Instability after knee replacement can be minimized by having implants available with various levels of constraint. The surgeon should have implants with varying levels of constraint available during knee arthroplasty. Implant manufacturers should continue efforts to develop instrumentation, some of which may be computer-assisted, to improve predictability and ease of component insertion.

For patients reported on in the current study, loosening was a significant mode of implant failure. The average interval to failure for this failure mode was 4 years (range, 1 month–28 years). A large number of aseptically loose implants were uncemented (21%) or were implanted by a technique referred to as surface cementing of the tibial component (10.5%). Fehring and Griffin 11 reported in their series of revisions done for aseptic loosening that cementless implants loosened earlier than cemented implants and led to the need for revision much sooner. They concluded cementless knee arthroplasty should be abandoned. Bert and Kelly reported that surface cement technique for the tibial component is associated with a relatively high failure rate. With this cement method, the undersurface of the tibial base plate is cemented but the keel is not cemented (Bert JM, Kelly FB: The incidence of tibial component loosening in cemented total knee arthroplasty when the tibial stem is not cemented. Presented at the American Academy of Orthopaedic Surgery Meeting, Orlando 2000). In the series of Bert and Kelly, 6.2% of patients who had surface cement technique of the tibial component had aseptic loosening after a mean followup of 3.2 years. In the current series, a significant number of implants associated with early failure were implanted by surface cementing. This technique should be abandoned. Malalignment of components also can lead to premature aseptic loosening. In the current series, malalignment most likely led to loosening in 44% of patients with aseptic loosening.

Extensor mechanism problems were shown to be an important cause of failure in the current series. The primary extensor complication seen was mechanism deficiency including extensor lag attributable to patellar baja, extensor mechanism rupture, or prior patellectomy. Avascular necrosis, when it occurs, is a serious complication frequently associated with pain and functional problems such as weakness and an extensor lag. Barrack et al 3 reported that avascular necrosis is more common with patellar resurfacing. Additional investigation into selective patellar resurfacing is necessary. In addition, the incidence of avascular necrosis associated with total knee arthroplasty when the patellar is not resurfaced needs to be investigated further. Peripatellar fibrosis, resulting in anterior knee pain and crepitation, has been reported after total knee arthroplasty. 4 Occasionally, the fibrous tissue will result in altered knee mechanics and locking related to entrapment of fibrous nodules, patellar clunk syndrome. 4 These problems have been shown to be more common with certain implant designs. Recently, a new generation of patellar enhancing implants has been introduced with design features intended to enhance patellar mechanics. 20 The long-term results with these implants remains to be determined.

Ayers et al reported that the preoperative mental health score, ascertained from a Short Form-36, closely correlates with surgical outcome (Ayers DC, Jain RK, Rogers M, et al: How patient outcome measures in total knee replacement can be useful in your practice. Presented at the American Association of Hip and Knee Surgeons Annual Meeting, Dallas, 2000). Patients with a low preoperative mental health score have a 10.1 time increased risk of a poor result from knee replacement surgery. Patients without a radiographically or clinically determinable cause for a poor outcome after total knee arthroplasty have a poor chance of a good result from revision knee surgery. With this considered, patients with unidentifiable causes of pain were excluded from this series and were not offered knee revision as an option. The percentage of failures, related to poor patient selection, was not determined by this study. Proper patient selection is essential in achieving a good result after total knee replacement.

Although knee replacement remains one of the most successful operations done today, failure sometimes does occur and when it happens, the impact is significant. Logically, failure rates could be reduced if attention is focused on the most common failure modes. Improvements on the manufacturing side with improved polyethylene or alternative-bearing surfaces certainly could diminish the rate of failure after total knee arthroplasty. In the current series, polyethylene failure was the most common cause of failure particularly when late revision surgery is considered. Efforts to alter polyethylene in total knee arthroplasty must be done through well-controlled clinical studies only after new polyethylene has been tested extensively in the laboratory.

Early failure after total knee arthroplasty frequently is related to infection. Institutional sepsis rates should be monitored continually and compared with national standards. Many early revisions also were related to instability and malalignment or malposition of the knee components. Investigators must maintain efforts at improving instrumentation to allow for more predictable component placement and implant stability. Although the current instrumentation for most cases is adequate it does not work universally. Loosening in this series was significantly associated with the technique used for component fixation. Orthopaedic surgeons should abandon surface cementing of tibial components, which also would diminish early failure rates. The widespread use of uncemented implants should be considered carefully in light of the excellent results reported with well-cemented devices.

Back to Top | Article Outline


1. Abernethy PJ, Robinson CM, Fowler RM: Fracture of the metal tibial tray after Kinematic total knee replacement: A common cause of early aseptic failure. J Bone Joint Surg 78B:220–225, 1996.
2. Ahn NU, Nallamshetty L, Ahn UM, et al: Early failure associated with the use of Hylamer-M spacers in three primary AMK total knee arthroplasties. J Arthroplasty 16:136–139, 2001.
3. Barrack RL, Bertot AJ, Wolfe MW, et al: Patellar resurfacing in total knee arthroplasty: A prospective, randomized, double blind study with five to seven years of follow-up. J Bone Joint Surg 83A:1376–1381, 2001.
4. Beight JL, Yao B, Hozack WJH, Hearn SL, Booth RE: The patellar “clunk” syndrome after posterior stabilized total knee arthroplasty. Clin Orthop 299:139–142, 1994.
5. Bliss DG, McBride GG: Infected total knee arthroplasties. Clin Orthop 199:207–214, 1985.
6. Cameron HU, Hunter GA: Failure in total knee arthroplasty: Mechanism, revisions, and results. Clin Orthop 170:141–146, 1982.
7. Chan WH, Yeo SJ, Lee BP, Tay BK, Tan SK: Isolated metal-backed patellar component revision following total knee arthroplasty. Singapore Med J 39:303–305, 1998.
8. Colizza WA, Insall JN, Scuderi GR: The posterior-stabilized knee prosthesis: Assessment of polyethylene damage and osteolysis after a ten-year minimum follow-up. J Bone Joint Surg 77A:1713–1720, 1995.
9. Deacon JM, Pagliaro AJ, Zelicof SB, Horowitz HW: Prophylactic use of antibiotics for procedures after total joint replacement. J Bone Joint Surg 78A:1755–1755, 1996.
10. Emmerson KP, Moran CG, Pinder IM: Survivorship analysis of the kinematic stabilizer total knee replacement: A 10–14 year follow-up. J Bone Joint Surg 78B:441–445, 1996.
11. Fehring TK, Griffin WL: Revision of failed cementless total knee implants with cement. Clin Orthop 356:34–38, 1998.
12. Fehring TK, Odum S, Griffin WL, Mason JB, Naduad M: Early failures in total knee arthroplasty. Clin Orthop 392:315–318, 2001.
13. Fehring TK, Valadie AL: Knee instability after total knee arthroplasty. Clin Orthop 299:157–162, 1994.
14. Feng EL, Stulberg SD, Wixson RL: Progressive subluxation and polyethylene wear in total knee replacements with flat articular surfaces. Clin Orthop 299:60–71, 1994.
15. Fort-Rodriquez DE, Scuderi GR, Insall JN: Survivorship of cemented total knee arthroplasty. Clin Orthop 345:79–86, 1997.
16. Hartman MB, Fehring TK, Jordan L, Norton HJ: Periprosthetic knee sepsis: The role of irrigation and debridement. Clin Orthop 273:113–118, 1991.
17. Heck DA, Melfi CA, Mamlin LA, et al: Revision rates after knee replacement in the United States. Med Care 36:661–669, 1998.
18. Hirakawa K, Bauer TW, Stulberg BN, et al: Comparison and quantitation of wear debris of failed total hip and total knee arthroplasty. J Biomed Mater Res 31:257–263, 1996.
19. Ingenix: Data Analyst Group. Columbus, OH, Ingenix 1999.
20. Ip D, Wu WC, Tsang WL: Comparison of two total knee prostheses on the incidence of patella clunk syndrome. Int Orthop 26:48–51, 2002.
21. Mitts K, Muldoon MP, Gladden Jr M, Padgett DE: Instability after total knee arthroplasty with the Miller-Galante II total knee: 5 to 7 year follow-up. J Arthroplasty 16:422–427, 2001.
22. Ranawat CS, Luessenhop CP, Rodriquez JA: The press-fit condylar modular total knee system: Four to six year results with a posterior-cruciate-substituting design. J Bone Joint Surg 79A:342–348, 1997.
23. Rand JA, Bryan RS: Revision after total knee arthroplasty. Orthop Clin North Am 13:201–212, 1982.
24. Robinson EJ, Mulliken BD, Bourne RB, Rorabeck CH, Alvarez C: Catastrophic osteolysis in total knee replacement: A report of 17 cases. Clin Orthop 321:98–105, 1995.
25. Segawa H, Tsukayama DT, Kyle RF, Becker DA, Gustilo RB: Infection after total knee arthroplasty: A retrospective study of the treatment of eighty-one infections. J Bone Joint Surg 81A:1434–1445, 1999.
26. Wasielewski RC, Parks N, Williams I, et al: Tibial insert undersurface as a contributing source of polyethylene wear debris. Clin Orthop 345:53–59, 1997.
27. Weir DJ, Moran CG, Pinder IM: Kinematic condylar total knee arthroplasty: 14-year survivorship analysis of 208 consecutive cases. J Bone Joint Surg 78B:907–911, 1996.

Section Description

Richard S. Laskin, MD—Guest Editor

© 2002 Lippincott Williams & Wilkins, Inc.