Identification of Knee Arthrodeses
The DNPR was searched for all surgical procedure codes pertaining to knee arthrodesis (KNGG39 through KNGG99) (Table I), and information on the date and hospital of the surgical procedure was extracted. A total of 415 procedures were identified. These were linked to the merged data set by ID number, thereby identifying 195 patients who were registered as having undergone both a primary knee arthroplasty and an arthrodesis. Hospital records were obtained from the national patient administration system or from hospital archives and were reviewed. Fifteen patients were excluded because the arthrodesis was incorrectly registered. Eleven patients were excluded because the arthrodesis was performed in an extremity in which the knee arthroplasty had been performed before 1997. Four records were missing or incomplete. The remaining 165 patients were eligible for inclusion in the study (Fig. 1). Information on the hospitalizations for the primary knee arthroplasty, any subsequent knee surgery, and the arthrodesis was collected from the records.
Statistical analyses were conducted using SAS for Windows (version 9.3; SAS Institute). The frequency of arthrodesis was assessed by means of the cumulative incidence, which is the probability of an event occurring within a given period of time. The cumulative incidence of arthrodesis was estimated using a competing risk model implemented in an SAS macro18 in which arthrodesis and death were defined as competing risks. Emigration or disappearance during the study period resulted in censoring. Censoring also occurred if none of the abovementioned events had occurred by the end of the study period. Differences in cumulative incidence were compared with the Gray test19. Categorical data were analyzed with the Pearson chi-square test. A p value of <0.05 was considered significant.
A total of 92,785 primary knee arthroplasties performed in Denmark from 1997 to 2013 were identified. Of these, 165 arthroplasties (0.18%) were followed by arthrodesis. One arthrodesis was performed because of bone metastases from colon cancer. The remaining 164 arthrodeses were performed for causes related to failure of the knee arthroplasty, and these 164 patients, 86 male and 78 female, represented the study population.
The cumulative incidence of arthrodesis within 15 years after primary knee arthroplasty was 0.26% (95% confidence interval [CI], 0.21% to 0.31%). The observations were divided into 3 consecutive time periods from 1997 to 2002, 2003 to 2007, and 2008 to 2013 on the basis of when the primary knee arthroplasty was performed (Fig. 2). As a result, the observation time in the 3 periods ranged from approximately 5 to 15 years (Fig. 2). Regardless of the time period during which the primary knee arthroplasty was performed, the risk of arthrodesis was highest within the first 5 years after arthroplasty. The 5-year cumulative incidence decreased from 0.32% in the first period (upper curve in the figure) to 0.09% in the third period (lower curve). Likewise, a decrease in the 10-year cumulative incidence from 0.37% in the first period (upper curve) to 0.23% in the second period (middle curve) was observed. The observed differences in cumulative incidence were significant (p < 0.001, Gray test).
In addition, the observations were divided by region on the basis of which hospital had performed the primary procedure. The administrative regions of Denmark included the Capital Region, Region Zealand, South Region, Central Region, and North Region. The incidence of arthrodesis differed significantly across the regions (p = 0.001, Gray test). The regions with the lowest and the highest cumulative incidence are illustrated in Figure 3.
Primary Knee Arthroplasty
The mean age of the patients at the time of primary knee arthroplasty was 66.5 years (range, 22 to 92 years) (Table II). The most common underlying diagnoses were osteoarthritis (66%) and posttraumatic arthritis (13%), with the latter defined by sequelae of fracture of the patella or femoral or tibial condyles. The primary procedure was total knee arthroplasty in 149 patients (91%) and unicompartmental knee arthroplasty in 9 patients (5%), 4 of whom later underwent conversion to total knee arthroplasty. Forty-eight patients (29%) were healthy (that is, had no substantial comorbidity) at the time of primary knee arthroplasty, whereas 105 patients (64%) had 1 or more comorbid medical conditions (range, 1 to 4 conditions), most commonly hypertension in 43 patients (26%), heart disease in 27 (16%), neurologic disease in 24 (15%), and endocrine disease in 21 (13%), including diabetes in 17 (10%). There were 35 smokers (21%) and 13 alcohol abusers (8%). Thirteen patients (8%) were on immunosuppressive treatment such as glucocorticoids or methotrexate.
Subsequent Knee Surgery
A total of 153 patients (93%) underwent subsequent surgery on the affected knee prior to arthrodesis, and the remaining 11 (7%) underwent no additional procedures between the primary knee arthroplasty and the arthrodesis. The mean number of surgical procedures between the primary arthroplasty and the arthrodesis was 2.4 (range, 0 to 9), including soft-tissue surgery in 108 patients (66%) and revision arthroplasty in 118 patients (72%) prior to arthrodesis. Complications associated with failure of the primary knee arthroplasty included infection in 107 patients (65%), mechanical problems in 15 (9%), wound-healing problems in 12 (7%), extensor mechanism disruption in 9 (5%), soft-tissue deficiency in 8 (5%), aseptic loosening in 6 (4%), pain in 6 (4%), stiffness in 5 (3%), and periprosthetic fracture in 3 (2%).
The main indication for arthrodesis was periprosthetic infection, which was present in 152 knees (93%). The predominant microorganism isolated preoperatively or intraoperatively in infected knees was Staphylococcus (Table III). Fourteen infections (9%) were polymicrobial, usually a combination of gram-positive cocci and gram-negative rods. Other indications for arthrodesis included extensor mechanism disruption in 46 patients (28%), soft-tissue deficiency in 25 (15%), severe bone loss in 11 (7%), intractable pain in 9 (5%), stiffness in 6 (4%), periprosthetic fracture in 4 (2%), and chronic knee dislocation in 1 (0.6%). In 79 patients (48%), there were 2 or more indications for arthrodesis.
The surgical techniques used in the arthrodesis included external fixation in 120 patients (73%) and internal fixation in 39 (24%), of whom 32 were managed with intramedullary nailing and 7, with compression plating. One arthrodesis was performed using Kirschner wires, another was uninstrumented, and in the remaining 3 cases the surgical technique was unknown. Arthrodesis was performed as a 2-stage procedure in 89 patients (54%), with removal of the prosthesis and insertion of a spacer followed by a period of antibiotic treatment and then arthrodesis. In 71 patients (43%), a 1-stage procedure was performed. Information on staging was missing in the remaining 4 cases.
The mean duration of follow-up after arthrodesis was 1.6 years (95% CI, 1.3 to 1.9 years). Five patients died of medical complications within 2 months after arthrodesis. A solid fusion was achieved in 106 patients (65%), including 27 (84%) of 32 patients with intramedullary nail fixation and 73 (61%) of 120 patients with external fixation. The difference in fusion rate between these types of fixation was significant (p = 0.01). Only 3 (43%) of the 7 plate fixations resulted in fusion. With the numbers available, the fusion rate did not differ significantly between 1-stage and 2-stage procedures (p = 0.22). An increasing number of previous revision arthroplasties was associated with a lower rate of fusion (p = 0.02). The infection was successfully eradicated in 113 (74%) of 152 patients, including 22 (67%) of 33 with internal fixation and 88 (77%) of 114 with external fixation (p = 0.22). The success rate for eradication of infection did not differ significantly between 1-stage and 2-stage procedures (p = 0.08).
A total of 34 patients (21%) underwent repeat arthrodesis because of failure of the first attempt to obtain fusion, and successful fusion was achieved in only 19 of the 34. Twenty-three patients (14%) eventually required transfemoral amputation. Of these, 8 had a history of failed repeat arthrodesis. Causes of amputation included uncontrollable infection, failure to achieve fusion, and soft-tissue compromise.
In this nationwide study, we identified 164 arthrodeses performed after failure of knee arthroplasty performed in the period from 1997 to 2013. The 15-year cumulative incidence of arthrodesis was 0.26% compared with an overall crude incidence of 0.18% (164 of 92,785 arthroplasties). These results are consistent with previous reports in which the crude incidence has ranged from 0.21% to 1.11%3-5. During the study period, the 5-year cumulative incidence of arthrodesis decreased from 0.32% to 0.09%; the latter is lower than previously reported values. This finding may indicate that knee arthroplasties have generally improved in the last decades. However, it may also indicate that there are now better surgical techniques to treat severe complications following knee arthroplasty.
The main cause of arthrodesis was periprosthetic infection, which is consistent with the literature8. The microorganisms in infected knees were largely Staphylococcus species, consistent with other recent studies9,12,20. Other important causes of arthrodesis, which have also been previously reported7,10-12, were extensor mechanism disruption, soft-tissue deficiency, and severe bone loss. It is important to consider alternative treatment options for these complications—e.g., skin grafts, muscle flaps, and extensor mechanism allografts for the reconstruction of extensive soft-tissue deficiencies21,22. In addition, severe bone defects are now managed with modular or customized revision prostheses or even tumor prostheses23.
Although our results showed an overall decrease in the cumulative incidence of arthrodesis, we observed large regional differences in the incidence. This is probably explained by variations among surgeons with respect to the indications for arthrodesis. In other words, some surgeons may tend to treat severe complications of knee arthroplasty with more traditional methods, such as amputation, whereas others rely on repeat revisions or more experimental methods as mentioned above.
In the present study, the preferred surgical techniques used in the arthrodeses were external fixation (120 of 164 patients, 73%) and intramedullary nail fixation (32 of 164 patients, 20%). The overall fusion rate following arthrodesis was 65% (106 of 164 patients). A previous large study found a similar fusion rate of 66%4. More recent studies have reported higher rates of 75% to 85%7,10,24. Infection was successfully eradicated in 113 (74%) of 152 patients. This is also comparable with other studies, in which the success rate ranged from 67% to 80%10,25.
The surgical techniques used in the arthrodeses were evaluated retrospectively. Comparing the results of various fixation techniques would ideally have been done in a randomized controlled trial rather than a retrospective review. Despite this limitation, however, we were able to make several observations because of the large number of patients included in the study. First, we found that the fusion rate was significantly higher with intramedullary nail fixation (27 of 32 patients, 84%) than with external fixation (73 [61%] of 120 patients, p = 0.01). This finding is supported by several other studies24,26-28. Second, we found that a higher overall fusion rate was associated with a lower number of previous arthroplasty revisions, which is most likely explained by better preservation of bone stock. Other authors have made similar observations24,29. Finally, we observed that the success rate for eradicating infection was higher with external fixation (88 of 114 patients, 77%) than with internal fixation (22 of 33 patients, 67%). Although this difference was not significant, other studies have reported similar results6,24.
The main purpose of the DKR is to provide information on the epidemiology and quality of knee arthroplasty procedures. For this reason, the DKR is designed to receive reports of all knee revision procedures performed in Denmark. Any revision procedure in which the knee prosthesis is removed to perform an arthrodesis should also be reported to the DKR. However, the register only received reports of 74 arthrodeses in the period from 1997 to 2013, which is less than half of the number of arthrodeses identified in the DNPR within the same study period (n = 164)1, suggesting that orthopaedic surgeons do not systematically report this procedure to the DKR. Moreover, registers may underestimate the number of arthrodeses performed for failed knee arthroplasty. We believe that this finding is partially explained by a lack of knowledge about the reporting requirement, and that this lack of knowledge is in part due to the organizational structures of orthopaedic departments, in which arthrodeses are sometimes performed by surgeons other than those who perform the arthroplasty procedures.
To our knowledge, this is the largest study to date on arthrodesis for failed knee arthroplasty. The study included 92,785 primary knee arthroplasties identified in nationwide registers, thereby minimizing selection bias. Registration of surgical procedures in the DNPR has been validated by several studies showing high completeness (>90%)15,16. Likewise, the completeness of primary knee arthroplasty procedures in the DKR has been shown to be high17. In the present study, registration of arthrodeses in the DNPR was validated by a review of hospital records. Fifteen (8%) of 195 arthrodeses were incorrectly registered. If a similar number of arthrodeses has never been reported to the DNPR, we would still have identified >90% of all arthrodeses performed in the study period. Thus, the risk of underestimating the incidence was low. In the process of validation, 4 patients were excluded from the study because of missing or incomplete records. However, information on these patients would not have affected the incidence substantially.
A competing risk model was used to estimate the cumulative incidence. In contrast to other traditional approaches such as the Kaplan-Meier method, this model takes into account the competing risk of death30. In this study, a considerable proportion of the population died during the study period. By using a competing risk model rather than the Kaplan-Meier method, in which censoring occurs at death, the bias that results in overestimating the incidence was eliminated. Overall, it is reasonable to assume that our data are representative.
In conclusion, the cumulative incidence of arthrodesis within 15 years after failed knee arthroplasty was 0.26% in a nationwide population of 92,785 primary knee arthroplasties performed from 1997 to 2013. In the last part of the study period, the 5-year cumulative incidence was reduced to 0.09%, which is lower than previously reported. This finding suggests that orthopaedic surgeons have already made progress in the treatment of complications associated with knee arthrodesis, such as extensor mechanism disruption, soft-tissue deficiency, and severe bone loss.
NOTE: The authors thank all of the orthopaedic surgery departments in Denmark for their help in data collection. They also thank the Department of Biostatistics, Copenhagen University, Denmark, for assistance with the statistical analysis.
Investigation performed at the Department of Orthopedic Surgery, Copenhagen University Hospital Herlev-Gentofte, Hellerup, Denmark
Disclosure: The study received external funding from the Hans and Nora Buchard’s Fund, a private nonprofit foundation. The funding did not play a role in the conduct of the study or the preparation of the manuscript. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work.
2. Bernstein J, Derman P. Dramatic increase in total knee replacement utilization rates cannot be fully explained by a disproportionate increase among younger patients. Orthopedics. 2014 ;37(7):e656–9.
3. Blom AW, Brown J, Taylor AH, Pattison G, Whitehouse S, Bannister GC. Infection after total knee arthroplasty. J Bone Joint Surg Br. 2004 ;86(5):688–91.
4. Bengtson S, Knutson K. The infected knee arthroplasty. A 6-year follow-up of 357 cases. Acta Orthop Scand. 1991 ;62(4):301–11.
5. Wang CJ, Huang TW, Wang JW, Chen HS. The often poor clinical outcome of infected total knee arthroplasty. J Arthroplasty. 2002 ;17(5):608–14.
6. Iacono F, Raspugli GF, Bruni D, Lo Presti M, Sharma B, Akkawi I, Marcacci M. Arthrodesis after infected revision TKA: retrospective comparison of intramedullary nailing and external fixation. HSS J. 2013 ;9(3):229–35. Epub 2013 Aug 14.
7. Klinger HM, Spahn G, Schultz W, Baums MH. Arthrodesis of the knee after failed infected total knee arthroplasty. Knee Surg Sports Traumatol Arthrosc. 2006 ;14(5):447–53. Epub 2005 Aug 30.
8. Conway JD, Mont MA, Bezwada HP. Arthrodesis of the knee. J Bone Joint Surg Am. 2004 ;86(4):835–48.
9. Bono JV, Talmo CT, Windsor R. Arthrodesis of the knee: indications and treatment options. Tech Knee Surg. 2009 ;8(4):212–5.
10. Schwarzkopf R, Kahn TL, Succar J, Ready JE. Success of different knee arthrodesis techniques after failed total knee arthroplasty: is there a preferred technique? J Arthroplasty. 2014 ;29(5):982–8. Epub 2013 Oct 4.
11. 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 Am. 1999 ;81(10):1434–45.
12. Putman S, Kern G, Senneville E, Beltrand E, Migaud H. Knee arthrodesis using a customised modular intramedullary nail in failed infected total knee arthroplasty. Orthop Traumatol Surg Res. 2013 ;99(4):391–8. Epub 2013 Mar 17.
13. Schmidt M, Pedersen L, Sørensen HT. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol. 2014 ;29(8):541–9. Epub 2014 Jun 26.
14. Nordic Medico-Statistical Comittee. NOMESCO classification of surgical procedures, version 1.16. Nordic Centre for Classifications in Health Care 2011. http://www.nordclass.se/NCSP_1_16.pdf
. Accessed 2016 Apr 30.
15. Nickelsen TN. [Data validity and coverage in the Danish National Health Registry. A literature review]. Ugeskr Laeger. 2001 ;164(1):33–7. Danish.
16. Lass P, Lilholt J, Thomsen L, Lundbye-Christensen S, Enevoldsen H, Simonsen OH. [The quality of diagnosis and procedure coding in orthopaedic surgery Northern Jutland]. Ugeskr Laeger. 2006 ;168(48):4212–5. Danish.
17. Pedersen AB, Mehnert F, Odgaard A, Schrøder HM. Existing data sources for clinical epidemiology: the Danish Knee Arthroplasty Register. Clin Epidemiol. 2012;4:125–35. Epub 2012 May 7.
19. Gray RJ. A class of k-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988 ;16(3):1141–54.
20. Husted H, Toftgaard Jensen T. Clinical outcome after treatment of infected primary total knee arthroplasty. Acta Orthop Belg. 2002 ;68(5):500–7.
21. Burnett RS, Butler RA, Barrack RL. Extensor mechanism allograft reconstruction in TKA at a mean of 56 months. Clin Orthop Relat Res. 2006 ;452:159–65.
22. Springer BD, Della Valle CJ. Extensor mechanism allograft reconstruction after total knee arthroplasty. J Arthroplasty. 2008 ;23(7)(Suppl):35–8.
23. Somayaji HS, Tsaggerides P, Ware HE, Dowd GS. Knee arthrodesis—a review. Knee. 2008 ;15(4):247–54. Epub 2008 May 20.
24. Mabry TM, Jacofsky DJ, Haidukewych GJ, Hanssen AD. Comparison of intramedullary nailing and external fixation knee arthrodesis for the infected knee replacement. Clin Orthop Relat Res. 2007 ;464:11–5.
26. MacDonald JH, Agarwal S, Lorei MP, Johanson NA, Freiberg AA. Knee arthrodesis. J Am Acad Orthop Surg. 2006 ;14(3):154–63.
27. Parcel TW, Levering M, Polikandriotis JA, Gustke KA, Bernasek TL. Failure analysis of knee arthrodesis with the WichitaFusion Nail. Orthopedics. 2013 ;36(11):e1336–9.
28. Van Rensch PJ, Van de Pol GJ, Goosen JH, Wymenga AB, De Man FH. Arthrodesis of the knee following failed arthroplasty. Knee Surg Sports Traumatol Arthrosc. 2014 ;22(8):1940–8. Epub 2013 May 26.
29. Hak DJ, Lieberman JR, Finerman GA. Single plane and biplane external fixators for knee arthrodesis. Clin Orthop Relat Res. 1995 ;316:134–44.
Copyright 2016 by The Journal of Bone and Joint Surgery, Incorporated
30. Berry SD, Ngo L, Samelson EJ, Kiel DP. Competing risk of death: an important consideration in studies of older adults. J Am Geriatr Soc. 2010 ;58(4):783–7. Epub 2010 Mar 22.