Field testing the Unified Classification System for periprosthetic fractures of the femur, tibia and patella in association with knee replacement

Transcription

1 J. M. Van der Merwe, F. S. Haddad, C. P. Duncan From University of British Columbia, Vancouver, British Columbia, Canada J. M. Van der Merwe, MBChB, FRCSC, Clinical Assistant Professor University of Saskatchewan, Department of Orthopaedics, 103 Hospital Drive, 5th Floor, Saskatoon, SK, S7N 0W7, Canada. F. S. Haddad, BSc MD (Res), FRCS (TR&Orth), Professor of Orthopaedic Surgery University College London Hospitals, 235 Euston Road, London NW1 2BU, UK. C. P. Duncan, MD, MSc, FRSCS, Professor of Orthopaedics University of British Columbia, Department of Orthopaedics, 3rd Floor, 2775 Laurel Street, Vancouver, BC, V5Z 1M9, Canada. Correspondence should be sent to Dr J. M. Van der Merwe; The British Editorial Society of Bone & Joint Surgery doi: / x.96b $2.00 Bone Joint J 2014;96-B: Received 15 March 2014; Accepted after revision 14 August 2014 ARTHROPLASTY Field testing the Unified Classification System for periprosthetic fractures of the femur, tibia and patella in association with knee replacement AN INTERNATIONAL COLLABORATION The Unified Classification System (UCS) was introduced because of a growing need to have a standardised universal classification system of periprosthetic fractures. It combines and simplifies many existing classification systems, and can be applied to any fracture around any partial or total joint replacement occurring during or after operation. Our goal was to assess the inter- and intra-observer reliability of the UCS in association with knee replacement when classifying fractures affecting one or more of the femur, tibia or patella. We used an international panel of ten orthopaedic surgeons with subspecialty fellowship training and expertise in adult hip and knee reconstruction ( experts ) and ten residents of orthopaedic surgery in the last two years of training ( pre-experts ). They each received 15 radiographs for evaluation. After six weeks they evaluated the same radiographs again but in a different order. The reliability was assessed using the Kappa and weighted Kappa values. The Kappa values for inter-observer reliability for the experts and the pre-experts were (95% confidence interval (CI) to 0.774) and (95% CI to 0.797), respectively. The weighted Kappa values for intra-observer reliability for the experts and pre-experts were (95% CI to 0.950) and (95% CI to 0.942) respectively. The UCS has substantial inter-observer reliability and near perfect intra-observer reliability when used for periprosthetic fractures in association with knee replacement in the hands of experienced and inexperienced users. Cite this article: Bone Joint J 2014;96-B: Periprosthetic fractures of the femur, tibia and patella are challenging complications following knee replacement surgery. The incidence is on the rise secondary to the increasing numbers of total knee replacements (TKRs) being performed. 1 Periprosthetic fractures are the fourth most common reason for revision surgery according to the Australian national joint registry. 1 The Unified Classification System (UCS) 2,3 has recently been developed to include patterns of periprosthetic fracture not covered by the existing systems of classification. 4-9 Several factors may be contributing to the increase in periprosthetic fractures in relation to TKR, such as increasing life expectancy, deteriorating bone health associated with ageing and progressive periprosthetic osteolysis Data from the Scottish National Database looking at TKRs over a period of 11 years estimated the risk of periprosthetic fracture around the knee to be 0.6% in the first five years after primary TKR, and 1.7% after revision TKR. 12,13 Periprosthetic fractures have an associated rate of mortality of 11%. 14 The principles of management are based on the location of the fracture, the stability of the implants and the adequacy of the remaining bone stock. Several classification systems have previously been proposed (Table I). 4-9,10,15,16 Each addresses certain aspects of the pattern and location of the fracture, as well as the stability of the implant. Rorabeck and Taylor 8 and Lewis and Rorabeck 6 looked at fractures around the femoral component. DiGioia and Rubash 17 and Chen et al 18 looked at supracondylar fractures proximal to the femoral component. The evaluation of these three systems depended on whether the fractures were displaced or not, while two other systems also assessed if the component was loose. 6,8 Felix et al 5 evaluated fractures around the tibial component, considering the location of the fracture, the stability of the component and if the VOL. 96-B, No. 12, DECEMBER

2 1670 J. M. VAN DER MERWE, F. S. HADDAD, C. P. DUNCAN Table I. Classification systems for periprosthetic fractures around the knee Supracondylar femoral fractures around total knee replacement Tibial fractures around total knee replacement Periprosthetic patellar fractures Lewis and Rorabeck 6 Felix et al 5 Hozack et al 19 Rorabeck and Taylor 8 Goldberg et al 7 Chen et al 18 DiGioia and Rubash 17 fracture occurred intra- or post-operatively. The systems addressing periprosthetic fractures involving the patella 7,19 looked at displacement of the fracture and the function of the extensor mechanism. The UCS was developed to provide a classification system that was easy and reproducible and that could be applied to any fracture around any existing component. As such, by unifying the work of previous authors, applying it in a standard fashion to all three bones (femur, tibia and patella), and following the widely accepted principles of the Vancouver Classification System 15,16,20 for fractures of the femur after hip replacement, it is our belief that the UCS will have great practical value and applicability in the clinical and research settings. By adopting it, only one classification system will be required to address all fractures around a TKR. It also includes previously unclassified fractures (i.e. UCS Type D, E and F). We are unaware of any studies assessing the reliability of any of the classification systems listed in Table I. A functional classification system should guide treatment, offer a prognosis, predict complications and permit the meaningful comparison of outcomes among different surgeons and centres. 15 It should also be reliable and valid and possess good inter-observer and intra-observer agreement. In addition, if possible, it should share the determinants of outcome with the prognostic variables of periprosthetic fractures affecting other joints. We therefore wished to determine the reliability of the UCS and whether it can be successfully used as a functional classification system for periprosthetic fractures around TKRs. The following simple mnemonic is offered to assist in the understanding and recall of the six basic types of fracture: Type A: Apophyseal. Type B: Bed of the implant or close to it. Type C: Clear of the implant bed. Type D: Dividing one bone which supports two arthroplasties. Type E: Each of two bones broken which support one arthroplasty. Type F: Facing or articulating with an implant although the bone is not itself resurfaced or replaced. Description of types of UCS and principles of management Type A is a fracture of an apophysis or protuberance to which one or more soft-tissue structures are attached, such as a ligament or a tendon. With reference to the knee in association with a TKR, this would include the proximal and distal poles of the patella, the epicondyles of the femur and the tibial tuberosity. Management would depend on the importance of the soft-tissue attachment to the health and function of the replacement. Displacement would also play a role in management. For example, a displaced avulsion fracture of the inferior pole of the patella or the tibial tuberosity would in most cases require surgical intervention. Type B is a fracture involving the bed of an implant or close to it, such as the femur or tibia around a femoral or tibial stem. Fractures of the the condyles and plateau also fall into this category. The principles of management would depend on the subtype, as originally defined in the Vancouver Classification System (VCS): 16 in B1 the implant is well fixed; in B2 it is loose and in B3 it is loose and the bone bed is substantially damaged by osteolysis, osteoporosis or comminution. This subtyping is pivotal because it defines the principles of treatment. In B1 fractures, reduction and fixation is indicated (although revision of the implant might facilitate that process). In B2 fractures, revision and fixation is required and in B3 fractures, a more complex reconstruction needs consideration. The subtypes and their management apply to the femur, tibia and patella. Type C is a fracture that is distant to the bed of the implant. It can be managed by open or closed osteosynthesis without involving the implant, unless it is elected to use a new implant to achieve intramedullary fixation. Clearly this type does not apply to the patella. Type D is a fracture of one bone, which supports two arthroplasties. Around the knee, this might involve the femur after hip and knee replacement or the tibia after knee and ankle replacement. It would not involve the patella, which only supports one arthroplasty. Management would depend on the analysis of the type of fracture within the context of each arthroplasty. The type of fracture in relation to the hip would be considered while ignoring the knee, and vice versa. Based on this exercise one, both or neither arthroplasty may need to be revised. Type E is a fracture that involves at least two bones supporting one arthroplasty. Examples around a TKR are fractures involving the femur and tibia, femur and patella, patella and tibia, or theoretically all three. The approach of considering each bone in isolation is again recommended, and treatment is based on the resulting conclusions. Type F is a fracture involving a bone which is not resurfaced or replaced, but is still considered periprosthetic because it is facing and articulating with an arthroplasty. This type would affect the patella after a TKR in which the patella was not resurfaced. The principles of management would depend on the orientation of the fracture (vertical or THE BONE & JOINT JOURNAL

3 FIELD TESTING THE UNIFIED CLASSIFICATION SYSTEM FOR PERIPROSTHETIC FRACTURES OF THE FEMUR, TIBIA AND PATELLA 1671 Fig. 1 A radiograph of periprosthetic fractures involving a total knee replacement which is classified by the Unified Classification System as a type E fracture. Each bone should then be assessed in isolation and sub categorised as either a type B1, 2, 3 or Type C fracture in order to determine a proper treatment strategy. transverse), the displacement and the integrity of the extensor mechanism. In most cases, reduction and fixation would be required, with or without simultaneous resurfacing. Patients and Methods After obtaining ethical approval, our database was used to identify suitable patients with a periprosthetic fracture involving the femur, tibia or patella after TKR between the years 2000 and These radiographs were screened using WebDI imaging storing system (Integrated Medical Imaging, Vancouver, Canada). Examples with good imaging quality, which represented the types of periprosthetic fracture, were retained and 13 radiographs were chosen to include in the survey. Two radiographs were used twice with different aspects being used to formulate a new question. For example, in the first question, the panel was asked to classify the fracture seen in Figure 1 in relation to a TKR (i.e. UCS Type E: a fracture of two bones supported by one arthroplasty). In the second question the panel was asked to classify the fracture using the UCS only in relation to the tibial component (i.e. UCS Type C: fracture distant to the bed of the implant). This demonstrates that two distinct aspects of the same radiograph were enquired upon. This resulted in a total of 15 separate questions. After the 15 questions were formulated, we invited an international panel of ten subspecialised adult reconstruction surgeons, as well as ten orthopaedic residents in their final two years of training at the University of British Columbia. They were used to test different aspects of the UCS, including the inter-observer and intra-observer reliability. The international panel of ten surgeons, each with subspecialty fellowship training and practice in adult knee reconstruction, collaborated individually in the survey, without inter-rater communication. These were designated as experts. The panel of ten orthopaedic residents in the final two years of training also collaborated individually, without inter-rater communication. These were designated as pre-experts. This latter group was chosen to assess the use of the UCS among those in training, for whom exposure to general orthopaedics had been undertaken but subspecialty fellowship experience had yet to be acquired. A simple preparative package containing an outline of the UCS pertaining to the knee was sent to each observer to review. One week later a multiple-choice questionnaire was sent to each using a web survey development cloud-based company, SurveyMonkey 21 There were 15 questions based on 13 cases, using the 13 images. Each question contained a brief clinical statement, a radiograph and an eight-part multiple-choice questionnaire. An example of one such question: Case 3: Following a knee replacement 12 years previously this patient was hit by a car while crossing the street. This and other radiographs reveal a fracture of the femur and tibia. Disregard the femur and only focus on the tibia (so-called block-out analysis). Please see figure 1. How would you classify the periprosthetic fracture of the tibia associated with the knee replacement by using the UCS? - Type A - Type B1 - Type B2 - Type B3 - Type C - Type D - Type E - Type F Six weeks later they received the same package, but with the radiographs and related questions in a different order. This was designed to assess intra-observer reliability. Statistical analysis. All collected data were analysed using the Kappa statistic 22 to measure the inter-rater reliability at baseline within the experts and pre-experts. The weighted Kappa was calculated to measure the intra-observer reliability within the same groups. Kappa values of <= 0 indicated poor agreement; 0.00 to 0.20 indicated slight agreement, 0.21 to 0.40 fair; 0.41 to 0.60 moderate; 0.61 to 0.80 substantial and 0.81 to 0.99 almost perfect agreement. The level of observer reliability that should be demonstrated before adoption of a classification system has yet to be defined, but clearly this should be high, ideally in the substantial range or higher, as defined by Landis and Koch. 22 An independent statistician was employed to perform all the statistical analysis. The 95% confidence intervals (CI) were determined. Because CI assume that there is complete independence between subjects, and in two sets of questions we used the same radiograph for interpretation, that assumption might be questioned. We therefore did a bootstrap-based analysis to account for any possible dependence between these sets of readings. However, we used different aspects of these radiographs to compose a VOL. 96-B, No. 12, DECEMBER 2014

4 1672 J. M. VAN DER MERWE, F. S. HADDAD, C. P. DUNCAN completely different question and therefore the 95% CI could be applied assertively. We still included the bootstrapbased results at least as a sensitivity analysis. Results The inter-observer reliability for the experts was substantial with a Kappa value of (95% CI to 0.774). The pre-experts inter-observer reliability also demonstrated a substantial Kappa value of (95% CI to 0.797). The bootstrap-based analysis that was performed to account for the possible dependence between the sets of readings showed a larger CI as expected, but still remained substantial. For the experts it was calculated as 95% CI to 0.838; and for the pre-experts as 95% CI to The intra-observer reliability demonstrated almost perfect agreement for the experts at (95% CI to 0.950). The pre-experts had similar impressive findings with a value of (95% CI to 0.942). The bootstrap based analysis for the experts was 95% CI to 0.958; and for the pre-experts 95% CI to Discussion An accepted global classification system would encourage logical standardised treatment, leading to better patient care as well as facilitating effective communication between surgeons. Such a system should be reliable as well as valid. Our focus was on the assessment of the reliability of the UCS classification system in relation to fractures around TKRs. We have confirmed that both adult reconstruction fellowship trained orthopaedic surgeons and non-fellowship trained general orthopaedic surgeons can apply the system around TKRs with periprosthetic fractures. It was not our intention to evaluate the outcome of periprosthetic fractures treated using the UCS, nor was it intended to determine the prognosis of each fracture defined in the UCS. We realise the difficulty in achieving reliability and validity with a classification system where distinguishing between loose and well-fixed components plays an integral part. Previous studies have shown the difficulties in obtaining agreement between consultants on the distinction between loose and well-fixed components. 20,23,24 The VCS, however, which was initially described in 1995, remains reliable and valid as shown in many studies. 15,20,22,24 The UCS has been developed from the VCS to address patterns of fracture which have not previously been covered, and it can be applied with minor adaptation to any periprosthetic fracture. By using only one classification system for periprosthetic fractures, surgeons and non-surgeons should quickly become familiar with classifying the fractures, and treatment can become more standardised. The inter- and intra-observer reliability was high for experts and more general (and less experienced) orthopaedic surgeons. Even by using the lower values of the CI the inter-observer reliability of both groups remained substantial. The lower values of the intra-observer reliability were considered almost perfect in both groups. The 95% bootstrap-percentile confidence intervals, which were included because the same set of radiographs was used twice, are wider and thus less impressive than the asymptotic confidence intervals, as expected. The questions, which used similar radiographs, assessed reasonably distinct aspects of the radiographs, and therefore it might not be unreasonable to refer only to the asymptotic limits rather than the bootstrap limits. The bootstrap limits, however, can be used at least as a sensitivity analysis. The increased complexity of a classification system tends to decrease the reliability between users. 24 One would anticipate that this might be a similar finding with the UCS, being more complex than its predecessors. 4-9,17-19 However, this was not seen with the UCS, showing substantial, and almost perfect inter- -and intra-observer reliability. The study has some limitations. In some of the survey questions, the stability of the components was revealed. This was done, not to reproduce previous studies findings of accuracy in distinguishing between B1 and B2 fractures, 20,22 but only to assess the inter- and intra-observer reliability of the UCS. Also, it was not our goal to assess the validity of the UCS in the current study, but we recognise that this warrants further investigation. This could be accomplished by looking at the agreement between the preoperative radiological classification and the actual intraoperative findings. Another limitation is the small number of cases we used. However, we tried to reproduce the relative frequency in which they occur in clinical practice. We therefore had at least one type of periprosthetic fracture from the classification system represented in the questionnaire, and up to four of the more frequently occurring fracture patterns. We acknowledge this is a low number, but we tried to accommodate the experts who kindly agreed to participate in the study. Another limitation was not comparing the reliability of the UCS with one of the existing classification systems of fractures around a TKR. This would have been difficult as we addressed many fractures that currently are considered unclassifiable with the existing classification systems. Also we are not aware of any studies looking at the reliability of the classification systems listed in Table I. In addition, all the pre-experts are from a training program located in Vancouver, which could theoretically cause bias toward the results due to local familiarity with its use. However, the system had not been introduced to practice locally, the preexperts were not exposed to its application prior to this study, neither were they involved in the development of the UCS. The experts come from around the world, and similar findings were seen between the experts and the pre-experts. Thus, it is unlikely that the training location of the preexperts would have played a significant role in the final results of the study. In conclusion, the UCS is a reliable tool that can be applied with confidence by fellowship and non-fellowship trained orthopaedic surgeons to any periprosthetic fracture THE BONE & JOINT JOURNAL

5 FIELD TESTING THE UNIFIED CLASSIFICATION SYSTEM FOR PERIPROSTHETIC FRACTURES OF THE FEMUR, TIBIA AND PATELLA 1673 around a TKR, whether the fracture is sustained intra- or post-operatively. Supplementary material Two tables showing all the specialists asked to review the questions and one table demonstrating the number of questions represented by each subtype of the UCS are available alongside the online version of this article at We would like to thank Dr. E. C. Sayre for his analysis of the data in the UCS study. 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. This article was primary edited by G. Scott and first proof edited by J. Scott. References 1. No authors listed. Australian Orthopaedic Association National Joint Replacement Registry. (date last accessed 8 October 2014). 2. Duncan CP, Haddad FS. Periprosthetic Fractures After Joint Replacement: A Unified Classification System. In AO Manual of Management of Periprosthetic Fractures. Ruedi T, Perka C, Schuetz M. Ed. AO Publishing Foundation, Dubendorf, Switzerland Duncan CP, Haddad FS. The Unified Classification System (UCS): Improving our understanding of periprosthetic fractures. Bone Joint J 2014;96-B: Su ET, DeWal H, Di Cesare PE. Periprosthetic femoral fractures above total knee replacements. J Am Acad Orthop Surg 2004;12: Felix NA, Stuart MJ, Hanssen AD. Periprosthetic fractures of the tibia associated with total knee arthroplasty. Clin Orthop Relat Res 1997;345: Lewis PL, Rorabeck CH. Periprosthetic fractures. In Engh GA, Rorabeck CH eds. Revision Total Knee Arthroplasty. Baltimore, Williams & Wilkins, Goldberg VM, Figgie HE 3rd, Inglis AE, et al. Patellar fracture type and prognosis in condylar total knee arthroplasty. Clin Orthop Relat Res 1988;236: Rorabeck CH, Taylor JW. Classification of periprosthetic fractures complicating total knee arthroplasty. Orthop Clin North Am 1999;30: Ortiguera CJ, Berry DJ. Patellar fracture after total knee arthroplasty. J Bone Joint Surg [Am] 2002;84-A: Parvizi J, Jain N, Schmidt AH. Periprosthetic knee fractures. J Orthop Trauma 2008;22: Small SR, Ritter MA, Merchun JG, Davis KE, Rogge RD. Changes in tibial bone density measured from standard radiographs in cemented and uncemented total knee replacements after ten years' follow-up. Bone Joint J 2013;95-B: Liddle AD, Pandit H, O Brien S, et al. Cementless fixation in Oxford unicompartmental knee replacement: A multicentre study of 1000 knees. Bone Joint J 2013;95- B: Sarmah SS, Patel S, Reading G, et al. Periprosthetic fractures around total knee arthroplasty. Ann R Coll Surg Engl 2012;94: Bhattacharyya T, Chang D, Meigs JB, Estok DM 2nd, Malchau H. Mortality after periprosthetic fracture of the femur. J Bone Joint Surg [Am] 2007;89-A: Brady OH, Garbuz DS, Masri BA, Duncan CP. The reliability and validity of the Vancouver classification of femoral fractures after hip replacement. J Arthroplasty 2000;15: Duncan CP, Masri BA. Fractures of the femur after hip replacement. Instr Course Lect 1995;44: DiGioia AM 3rd, Rubash HE. Periprosthetic fractures of the femur after total knee arthroplasty. A literature review and treatment algorithm. Clin Orthop Relat Res 1991;271: Chen F, Mont MA, Bachner RS. Management of ipsilateral supracondylar femur fractures following total knee arthroplasty. J Arthroplasty 1994;9: Hozack WJ, Goll SR, Lotke PA, Rothman RH, Booth RE Jr. The treatment of patellar fractures after total knee arthroplasty. Clin Orthop Relat Res 1988;236: Rayan F, Dodd M, Haddad FS. European validation of the Vancouver classification of periprosthetic proximal femoral fractures. J Bone Joint Surg [Br] 2008;90-B: No authors listed. Survery Monkey. (date last accessed 8 October 2014). 22. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33: Lindahl H, Malchau H, Odén A, Garellick G. Risk factors for failure after treatment of a periprosthetic fracture of the femur. J Bone Joint Surg [Br] 2006;88-B: Gozzard C, Blom A, Taylor A, Smith E, Learmonth I. A comparison of the reliability and validity of bone stock loss classification systems used for revision hip surgery. J Arthroplasty 2003;18: VOL. 96-B, No. 12, DECEMBER 2014