Abstract
Abstract:
Accurate prosthesis classification is critical for total joint arthroplasty surveillance and assessment of comparative effectiveness. Historically, prosthesis classification was based solely on the names of the prosthesis manufacturers. As a result, prosthesis designs changed without corresponding name changes, and other prostheses’ names changed over time without substantial design modifications. As the number of prostheses used in total joint arthroplasty on the market increased, catalog and lot numbers associated with prosthesis descriptions were introduced by manufacturers. Currently, these catalog and lot numbers are not standardized, and there is no consensus on categorization of these numbers into brands or subbrands. Classification of the attributes of a prosthesis also varies, limiting comparisons of prostheses across studies and reports. The development of a universal prosthesis classification system would standardize prosthesis classification and enhance total joint arthroplasty research collaboration worldwide. This is a current area of focus for the International Consortium of Orthopaedic Registries (ICOR).
National joint arthroplasty registries play a critical role in the detection of total joint arthroplasty revision rates, identification of patients during recalls and advisories, and comparative effectiveness research. In order to monitor and evaluate total joint arthroplasty procedures, the specific prostheses associated with the procedure must be accurately identified and classified. This is necessary to determine and report the outcome of those prostheses either alone or in combination with other prostheses and also to determine outcomes relevant to specific attributes and characteristics of the prostheses. In order to organize and store information specific to each prosthesis, registries have developed databases of prostheses reported to the registry. These have usually been developed in an ad hoc way by each registry, and there has been no standardized way of creating these databases. At the recent International Consortium of Orthopaedic Registries (ICOR) meeting, it was determined that developing a standardized approach would enable the development of a universal prosthesis database that all registries could use, which would not only ensure consistency of reporting between registries but also enhance inter-registry collaboration.
The first commercially available prostheses were often named after their inventors—for example, the Charnley low-friction arthroplasty1 or the Freeman-Swanson knee2. When registries were first established in the mid-1970s in Scandinavia, they often used this nomenclature to label and record prosthesis information. Because of the limited number of prosthesis models when registries were first established, this simplistic nomenclature was adequate for prosthesis classification. As prostheses evolved, new designs were introduced, refinement of models was undertaken, and identification of prostheses became a concern as the rudimentary system in place quickly became insufficient.
Over the course of the last thirty years, many prostheses for total joint arthroplasty have been introduced into the market, modifications of well-established prostheses have been introduced, and prosthesis names have been changed. Modifications of previous models were not necessarily associated with a name change every time. Historically, some of the modified designs were given new names, but some kept their old name or were introduced as “subbrands,” similar to what is done in the automobile industry. Keeping the same name was most common with prostheses with a good reputation and a well-established brand (Fig. 1), but eventually a prosthesis name that had initially been specific for a particular model needed additional information (e.g., Mark II and Mark III) for correct identification. Another problem was that the new models did not always directly replace the older versions but were introduced regionally over time. This led to many versions being available concurrently, without surgeons always being aware of it. In turn, this led to different models being referred to by the same generic name at different locations. In addition, prosthesis names have changed over time (e.g., DePuy's PFC changed to PFC-Sigma and then to Sigma), which also needs to be taken into account when one is reporting prosthesis utilization or studying specific prosthesis designs. Changes in prosthesis names and designs create inconsistencies in classification of prostheses for surveillance, reporting, and evaluation of performance. Misclassification bias can be introduced by the systematic error in collecting, analyzing, and interpreting data. If prostheses are not properly classified, the results from studies utilizing such data can be handicapped by either spurious associations, when differential misclassification exists, or attenuation of associations, when non-differential misclassification is present.
Accurate prosthesis classification involves several components, including individual prosthesis identifiers (catalog and lot numbers), data extraction methods (manual entry or electronic scanning), information storage, accessibility, data integrity, and reporting with use of homogeneous definitions and classification systems. The purpose of this article is to summarize the main components crucial to prosthesis identification, classification, and tracking and to highlight the current challenges in prosthesis classification.
As using names did not provide registries with sufficient specificity in prosthesis identification, catalog numbers were initiated. A catalog number is a number assigned by a company to a prosthesis so that it is specifically identified. It can be numeric, alphanumeric, or a combination of both, and it is specific for a particular prosthesis size or configuration. Any change to the design of a prosthesis necessitates a change in the catalog number.
However, there has been no worldwide consensus on the encoding of part numbers, and the effort to implement a UDI (Unique Device Indicator) has been hampered by different regulations around the world. Thus, there is no guarantee that a given number is unique. There have been instances in which different prostheses have been identified with the same catalog number and different numbers have been used for the same prosthesis, depending on where in the world it was being sold.
Even in the same country, variation in cataloguing schemes both within and between different companies is common. Some companies use very simple cataloguing schemes with as few as four digits, while others have very complicated systems with over twelve digits. While the cataloguing scheme per se is not important if it is efficient and accurate, both very short and very long numbers present obstacles to data quality. Very short numbers are associated with an increased risk of overlapping between companies or products, while very long numbers increase the risk of data entry errors. Fortunately, an increased willingness to address these logistic problems is evident in the orthopaedic community and organizations providing administrative and financial management support of orthopaedic services3.
A knee joint replacement usually involves a combination of two or more prosthesis components, and the type or class of knee replacement depends on which components are selected. Even with adequate identification of each individual component, registries need to combine these components and describe the combination as a prosthesis type. With the introduction of modularity, new types of materials and surfaces, mobile bearings, and other subtle design changes, the options have increased to the extent that, today, it can be difficult to know from registry reports and scientific papers exactly what combination of components was being studied. The types of prostheses reported by different registries often reflect the most popular brand variants used in their respective countries. There are often many different subtypes under a specific brand name. This variation is evident not only between countries but also within regions of a country, to such an extent that the named prosthesis may consist of completely different subbrands in different regions of the same country.
Furthermore, with the many combinations available, it is possible to define subbrands in many ways. This in part depends on what major design aspects are of interest. For example, the classification of the NexGen prosthesis in Sweden focuses on the type of tibial component used. In Denmark and Australia, it is classified according to whether the prosthesis is cruciate-sacrificing and/or high flex. Even when a typical brand has been defined, the use of certain additional components may alter the classification, such as when long stems or augments are used in a primary procedure. Most registries would regard this as a potentially different prosthesis, not only because of the additional components but also because the use of these components indicates that the primary arthroplasty is likely to be more complex than the usual routine case. In that situation, the prosthesis is often classified as a “revision model.”
Complexity is further increased when one major component can be used in a number of different brands. An example of this is the use of Biomet tibial baseplates with at least three different brands of knee replacements (AGC, Maxim, and Vanguard). There are a number of different types of Biomet tibial baseplates, which include a cruciform base, a single-post base, and a base allowing the attachment of a removable stem (Fig. 2). Depending on the polyethylene insert and femoral component used, the procedure may be defined as cruciate-retaining (CR), posterior stabilized (PS), or even super-stabilized (SSK).
In summary, registries have different methods of identifying prostheses and, although catalog numbers can identify the individual components, their formats as well as the information on the different components vary. Finally, there is no consensus with respect to how to convert combinations of catalog numbers into brands or subbrands. It is apparent that international collaboration is needed to resolve these issues.
Prostheses are manufactured in batches, and each batch is assigned a lot/batch number to enable identification of all prostheses manufactured in the same batch. This is done to enable batch tracking, which may be necessary if a production-related problem is identified. It is not uncommon for recalls to be made as precautionary measures involving a limited number of prostheses with minor problems. However, serious problems involving a large number of prostheses have also occurred4,5. Some, but not all, registries enter both catalog and lot numbers. When a registry does this, not only does it have the capacity to identify batch-related differences in outcome, but it also is able to assist in identifying patients who may have received a prosthesis with a batch-related issue.
The majority of prostheses, but not all, come in packages that include barcode markings. The symbol code contains information about the prosthesis such as the catalog number, but it may also contain other information such as the place of manufacture. However, similar to catalog and lot numbers, schematics or barcodes differ from company to company (Fig. 3). At present, a number of different types of barcode formats are in use, and there is as yet no international standard with respect to what information the barcode includes or in which position the catalog number should be. Reading numbers by optical scanning has the potential to reduce work and enhance accuracy. It is unfortunate that the lack of standards makes it difficult for registries to effectively use barcodes, and an international consensus on this technology would be of considerable benefit.
Registries not only undertake analyses of the outcomes of specific prostheses; they also undertake analyses related to generic characteristics, or attributes, of a group of prostheses. Examples are comparisons on the basis of whether the prosthesis had been designed to be used with or without cement as well as comparisons of different coatings, sizes, materials, and designs. However, to effectively undertake attribute-based analyses, registries have to be able to link the prosthesis characteristics to specific catalog numbers. This is time-consuming, and there are substantial differences between the registries regarding how and what attributes are registered. Most have done this linking ad hoc with use of advertisements, manuals, and brochures for the prostheses but otherwise with little help from the industry. One would think that it would be easy to acquire such information retrospectively, but, as a result of the high rate of company mergers and employee turnover, manufacturers have not been a consistent source of information on discontinued prostheses. This makes it still more important that the assignation of attributes is carried out soon after the introduction of new prostheses.
Registries devote major resources to the maintenance and updates of their individual prosthesis databases and the fact that companies provide only limited assistance makes the workload of individual registries greater than it needs to be. An example of the lack of help that companies provide registries is the almost universal failure of companies to notify registries when a new prosthesis is introduced to the market. It is usually discovered when the prosthesis is reported for the first time and a registry is unable to identify it because the catalog number is not found in their established prosthesis database. There are two reasons why a catalog number may differ from what is already in the registry database. The prosthesis may be new or the catalog number may have been entered incorrectly. For this reason, when catalog numbers are entered it is important to compare what is entered with what is in the registry database containing the numbers for all previously reported prostheses. Most registries have established systems enabling such identification, which obviously is limited to prostheses in their own register.
The modularity, complexity, and number of variants of joint replacement prostheses make analyses based on clearly defined prosthesis types and characteristics increasingly difficult. It is important that the identification and descriptors of prosthesis components and types be standardized. The ICOR initiative in this area is of paramount importance as it has the potential to ensure clarity around what prostheses and what attributes are being analyzed in the different registries. It will also greatly enhance the potential for inter-registry collaboration.
The initiative has determined that the only reasonable solution is establishing an internationally available database containing catalog numbers with adequate description of all of the different components used in joint replacement surgery. Preferably, it should also contain pictures of the parts, which often can better convey information on design aspects than can elaborate descriptions. Furthermore, a panel of experts is needed to gather information on the most commonly used combinations and to construct classifications of brands and subbrands as well as lists of catalog numbers that need to be present (or not present) for a combination to qualify as a specific brand or subbrand.
The establishment of a universal database has the potential to allow companies and registries to work collaboratively on its maintenance, enabling new prostheses to be added in a standardized and structured manner, increasing the specific attributes recorded. However, considering the previous reluctance by manufacturers to provide information, this may well be achieved only with the support of the regulating bodies. It is our hope that the ICOR project will eventually result in such a database.
Charnley
J. Total hip replacement by low-friction arthroplasty. Clin Orthop Relat Res.
1970;72:7-21.[PubMed]
Robinson
RP. The early innovators of today's resurfacing condylar knees. J Arthroplasty
. 2005;20(
1 Suppl 1):2-26.[PubMed][CrossRef]
Global Harmonization Task Force. Draft proposal on Unique Device Identification (UDI) system for medical devices. .
Zimmer. Zimmer issues update on Saint-Gobain's recalled ceramic femoral heads. .
Sulzer Orthopaedics. Sulzer recall. .