Professor Mike Reed, Chairman, Editorial Board
Mr Tim Wilton, NJR Medical Director
It will come as no surprise to anyone that the single most pronounced factor in this year’s annual report compared to previous years is the massive impact of the COVID-19 pandemic on the volume of all joint procedures. This has meant not only that the number of cases performed in 2020 has been roughly halved across the whole spectrum of arthroplasty, but this fall in numbers occurred during the last nine months of the year. This means that actual loss of arthroplasty provision was closer to 70% during those nine months and it is clear that volume has not fully recovered during the first quarter of 2021. We can anticipate that the data for analysis will be distorted by this loss of throughput and the accompanying altered case-mix for some years to come. Our preliminary analysis suggests that simply recovering the 2020 deficit will take a decade if joint surgery can only be increased by 5% compared to 2019; and will take five years if a 10% increase can be achieved. Recovery will clearly take much longer at those rates when the further deficit in volumes that has continued into 2021 is factored in.
Readers will therefore have to interpret much of the data from all arthroplasty registries with great care over the coming years as there will have been multiple reasons why the outcome results may be different from previous years including: patients waiting longer, operation complexity changing, and alterations in readiness to perform both primary and revision procedures during the pandemic.
The pandemic has of course shaped our representation at both national and international meetings, but many have continued virtually. We supported both the BASK and BHS annual conferences this year with virtual presentations from our Medical Director Tim Wilton and Peter Howard, Chairman of our Implant and Surgical Performance sub-committees. Each session was followed by a lively open question session with delegates who were interested in hearing how the NJR is supporting the work of the orthopaedic sector.
Meanwhile the work of the NJR Editorial Board has continued. The Board develops the strategy and style of the report and all members take responsibility for producing a report that is rigorously edited, taking almost a full year to write and review. The Board brings together experts on data collection and reporting as well as generous input from a patient perspective, clinicians from specialist societies and members of the NJR Management Team. Each year the Board aims to make progress in reporting on our rich data resource, making data easily accessible to improve patient outcomes. In addition to the section on COVID-19, a key development for this report has been “volume plots” which show the number of specific procedures performed each year, but also demonstrate whether each procedure was performed by surgeons with higher or lower activity.
We hope to launch the report at the British Orthopaedic Association Congress meeting in Aberdeen in September – and at the time of going to press we are looking forward to this being a face-to-face meeting!
This year we will have a limited print run of the annual report to be issued on a first-come first-served basis at the report launch. Increasingly there is considerable additional information available online and we would encourage you to explore the NJR’s dedicated annual report website at reports.njrcentre.org.uk. The website offers a helpful interactive platform for the descriptive NJR data, with supporting appendices.
Commentary on findings
This year NJR’s Annual Report is based on 2,895,368 records entered between 1 April 2003 and 31 December 2020, and the NJR maintains its position as the largest orthopaedic registry in the world. The report presents joint replacement up to 17 years of follow-up, with data on hips, knees, shoulders, elbows and ankle replacements. Due to the pandemic, approximately half as many records were added this year. In total the following numbers of linkable primary joint replacements are available for analysis: 1,251,164 hips, 1,357,077 knees, 7,084 ankles, 50,255 shoulders and 5,043 elbow replacements. There are further linkable revisions for each joint.
Hip replacement
There are new graphic representations of the proportions of different hip operation types performed by surgeons according to their annual throughput of those cases, and these give a fascinating insight into how things have changed over the years. Bearing in mind the concerns over minimum numbers and low surgeon volumes, these graphs give useful information about the general level of surgeon experience. High proportions of most types of hip replacement can be seen to be performed by surgeons doing quite high numbers per year. More than half of unipolar hip replacements are performed by surgeons doing more than 97 such cases a year. Only a tiny number of resurfacing hips are performed by surgeons doing more than 97 such operations, but the proportion of resurfacing cases done by those surgeons is nevertheless very high. This indicates that surgeons who are performing such operations tend to be highly specialised in the procedure.
Dual mobility hips, although performed far less frequently, are on the rise with a steady climb – being almost unheard of in 2013. It is perhaps not surprising, given the more limited indications for the procedure, that even prior to the pandemic few surgeons were performing more than 25 dual mobility hip procedures per year, but most such operations are nevertheless performed by surgeons who do more than seven per annum.
For the first time, hybrid fixation has become the most popular choice for hip replacement. It is interesting to note that while for cemented and reverse hybrid fixation metal-on-polyethylene (MoP) remains the predominant bearing surface choice; in both uncemented and hybrid hips the favoured choice is ceramic-on-polyethylene (CoP). Over the years there has been little change in the choice of bearing surface for reverse hybrid hips, and a very gradual change from MoP towards CoP for cemented hips. In contrast the change to CoP has been much more marked for hybrid hips and the bearing choice in uncemented hips has seen far more pronounced variations over the years. The reasons for this greater variation for uncemented hips is unclear and deserves further research and clarification.
The temporal changes in total hip replacement (THR) revision (Figure 3.H4 (a)) indicate a deterioration in revision rates until 2008 followed by marked improvement at all time points. There may have been two phases in this recovery; a sharp improvement until 2010 followed by a more gradual improvement ever since. It is tempting to suggest the steeper improvement from 2008 to 2010 is due mainly to metal-on-metal (MoM) issues and the subsequent improvement is a secular trend which is also seen in knee revision rates. The reason for that secular trend may be multi-factorial, but coincides with the start of feedback of data on their own results to surgeons by the NJR. In Figure 3.H4 (b) the 13-year revision rates are more clearly seen to be following the improving trends of earlier time points.
Revision rates for different revision indications are shown in Figure 3.H11 and it is perhaps no surprise that rates for aseptic loosening (and lysis) are highest for MoM bearings and resurfacing. Although very low rates for aseptic loosening are seen in cemented MoP and CoP these do seem to rise significantly after ten years. The very low rates seen in hybrid MoP and CoP up to ten years do also appear to cross over such that hybrid CoC have the lower rate for aseptic loosening after ten years. There appear to be few differences in infection rates according to bearing or fixation except for a higher rate in MoM hips at most lengths of follow-up. This data is unadjusted however and two separate analyses from registry data have shown reduced infection rates in ceramic bearings.
Interesting detail is available this year on revision rates by bearing surface stratified by age groups, and also for femoral head size; and surgeons will want to delve deeply into this information to check whether more or less “tailoring” of their procedures may be desirable according to the individual patient being treated.
Revision rates (PTIRs) for different hip constructs are presented in much more detail this year and should be of particular interest to both surgeons and patients alike. It can be seen that some apparently similar constructs have differing revision rates and surgeons will wish to reassure themselves that what they may believe about the construct they are using is indeed borne out by this extensive analysis. Readers should also be aware that these revision rates for constructs are not adjusted for age and other case-mix variables, but some of the constructs may be specifically indicated in younger patients or for some specific indication, so it is necessary to look carefully at the age, gender and other factors presented.
The median age and interquartile range are often quite different between cemented, hybrid and uncemented constructs, and this is most marked when considering resurfacing constructs.
In 2020 there was a marked reduction in THR for hip fracture and this might be the group where we would have expected numbers to hold up despite the pandemic. The reasons for this will need to be examined in more detail to see if there may have been a real drop in such fractures, or whether other factors such as altered case-mix or altered threshold for certain treatments were responsible. Revision rates for THR performed for hip fracture seem to track those for THR in osteoarthritis to a remarkable extent out to 15 years, albeit with increased rates of revision for the former in the first year.
Knee replacement
There are many areas this year in which there is either new information in the report or the previous information is expanded to give much more detail. There has been controversy for years about the possibility that surgeons doing small numbers of certain operations may be systematically giving rise to higher failure rates than those performing higher numbers. This relationship seems particularly clear in the case of unicondylar knee replacements and this year’s data show that the median number of such cases performed over the past three years is 19 (per surgeon) or 49 per unit. These figures show that, on average, surgeons are still not reaching the target numbers set by the British Association for Surgery of the Knee (BASK) which will be of concern to many. Nevertheless, over recent years we can see that over two-thirds of unicondylar procedures are performed by surgeons doing 25 cases or more per year.
Although the case numbers per surgeon have been greatly distorted by the impact of COVID-19 in 2020, we can see that over recent years prior to 2020, virtually no procedures were carried out by surgeons performing fewer than seven total knee replacements (TKRs) per year and that about 75% of such cases each year were by surgeons performing more than 49 cases per year.
Table 3.K1 shows that over the life of the registry roughly 25% of all TKRs are performed using a posterior stabilised (PS) implant and these consistently show higher revision rates than operations performed with unconstrained implants. While it is sometimes argued that this could reflect PS usage by many surgeons when they encounter a particularly difficult or complicated case, the evidence from the registry seems to show that the choice is mostly based on surgeon preference. This is therefore an area where surgeons may wish to reflect on whether they are really making the safest and most appropriate choice.
The revision rates for PS and cruciate retaining (CR) knees of each implant brand are now shown in the report separately and are also stratified according to whether the patella is resurfaced or not. These data show that the above finding of higher revision rates in PS knees may in fact be seen in the majority of, but not all, implant brands. It is therefore worthwhile for surgeons to look at Figures 3.K7 to 3.K10 in detail to ascertain the precise differences between the sub-types of implant available to them in their units, and in the market in general.
Although unicondylar knees are seen to have typically involved the use of a mobile bearing over the last 18 years, this has been changing recently towards a fixed bearing. There is still controversy about the pros and cons of performing unicondylar knee procedures, and caution needs to be used in interpreting the revision data since there are clear case-mix differences between those suitable for a unicondylar and those who are not. The statistics reported here have not been adjusted for such case-mix differences, which may be of particular relevance in certain implants which have been marketed as suitable for a particular patient grouping such as younger, more active patients or a particular gender.
As predicted in the last two years, Figure 3.K3 (b) shows that the improvement in revision rates shown previously up to ten years, is now reflected at 13 and 15 years as well, starting as before with the 2008 cohort.
While revision rates for uncemented unicondylar procedures appear lower than those for cemented at many time points, it is important to view this in the context of a gradual improvement in unicondylar results in general, which has been occurring over 15 years. Thus the uncemented unicondylar implants, which have mostly been inserted over the last ten years, would be expected to have slightly better results than cemented unicondylars with similar follow-up but performed on average more than ten years ago.
Re-revision rates continue to be seen to be much higher than those revised after primary operations across all sub-groups of knee replacement. Figures for re-revision of around 16% at ten years are seen this year, which is several times higher than those for a standard primary TKR and therefore continue to be a cause for concern. The time to first revision has a huge impact on the likelihood of subsequent revision procedures, but it remains uncertain whether this is wholly, or only in part, due to the different indications for first revision which predominate in the early and late post-operative periods.
Elbow replacement
There are now over 5,000 elbow replacements available for analysis including total replacement (with or without radial head replacement), distal humeral hemiarthroplasty, lateral resurfacing and radial head replacement. Over 40% of these were performed for a trauma indication.
With the exception of 2020, the number of elbow replacements being registered has increased but the numbers of surgeons performing one to two per year has fallen, and those performing more than 13 has increased. Revision rates differ by indication, with primary total elbow replacement (with or without a radial head replacement) for acute trauma being less than 4% at eight years, with elective indications being less than 10% at eight years, although few cases have that length of follow-up.
Shoulder replacement
A rigorous review of the shoulder data has been performed. Consequently, new classifications and component attributes are now used within the report to define the primary groupings throughout the whole of the shoulder section. The report has now moved to whole construct validation, ensuring all relevant elements required to build a construct are present in every procedure being reported on in our analysis. Over 50,000 primary shoulder replacements are available for analysis. The proportion of reverse polarity total shoulder replacement continues to increase (see Table 3.S1) albeit tending towards use in somewhat older patients (Table 3.S4). The median, interquartile range, and number of procedures performed by units and consultants has remained static for the last few years, apart from 2020 due to the impact of COVID-19.
Ankle replacement
This report focuses on primary procedures performed, and also on revision and mortality, with over 7,000 procedures being available for analysis. As noted previously, all ankle replacements recorded use uncemented implants although cement was listed in the component data in less than 5% and in the context of poor bone stock and low-demand patients.
The proportion of fixed-bearing ankle replacements continue to increase, and most procedures are done by surgeons performing more than seven cases per year. In 2019, the average number of procedures performed per consultant across all ankle replacements was 6.4 dropping to 4.0 in 2020.
In 2020 the Infinity ankle replacement implant dominated at 65.2% of all total ankle replacements, although it was only introduced in 2014. It is reassuring to see that the short term implant survival has improved.
Overall revision rates appear to be just less than 9% at ten years, although we believe there is incomplete reporting of conversion to fusion which remains mandatory. Both the Star and the Infinity ankle implant are running at revision rates of less than 3% at five years, albeit running into low numbers with longer follow-up. Other implants are failing at varying, and some at concerning, rates.
Patient Reported Outcome Measures (PROMs)
Our annual report includes the failure rates of all the different brands used in hip and knee replacements, however revision surgery is not a complete marker of success. A device may well be classified as successful if it survives for 15 years, but an implanted patient may disagree should they have experienced persistent pain and disability and around 20% of patients report persistent pain following joint replacement surgery. Analysis of revision rates alone fails to identify these patients with persistent pain or disability. Therefore, there have been calls for methods to measure pain that can subsequently be used in conjunction with revision rates for accurately monitoring outcomes in hip and knee replacement surgery. See also the Independent Medicines and Medical Devices Safety Review, chaired by Baroness Julia Cumberlege, published report titled “First Do No Harm” (Cumberlege, 2020)1.
Patient reported measures of hip and knee pain and function have been collected nationally by NHS Digital since 2009 for all patients receiving a primary hip or knee replacement operation. This information, to date, has not been reported in our annual report. We aim to address this gap in implant outcome reporting by incorporating national PROMs analyses within our future annual reports.
Data will be analysed separately for hip and knee joints. Analyses will be repeated according to indication for surgery, primary and revision operations, where surgery was carried out and how it was funded. Descriptive analyses will be used to look at how the proportion of any missing PROMs data varies over time, by looking at trends for each year of data. We will further explore geographical variation in patterns of missing data by hospital trust, and operating surgeon, and use caterpillar plots to visually display hospitals and surgeons with the most and least amount of missing PROMs data.
We will present descriptive statistics to look at the association of patient characteristics according to completion of PROMs scores. We will do these analyses overall, and then repeat them for individual years of the data, as the influence of patient characteristics on missing PROMs outcomes and the quality of data may change over time. Caterpillar plots will describe the variation in PROMs outcomes within and between implant brands and constructs. Initial analyses will be descriptive about the actual variation in observed PROMs outcomes, with stratification of analyses by age and gender groups. More formal modelling methods will then be considered, for ‘within and between group’ variation between implant brands and constructs.
Having assessed data quality, our aim is to then compare and scrutinise the differing performances between the implant brands of different prostheses for associated pain and functional outcomes. By selecting a hip or knee brand with a highest improvement in pain and functional outcome as a reference group, we will perform statistical analyses to directly compare the performance of all the stem and cup combinations used in hip replacement and all the knee brands used in knee replacement against this reference. This will demonstrate if any brands are performing poorly in comparison to the best performing implants, and thus enable patients and surgeons to make better informed decisions about the relative performance, as judged by patient reported pain and functional outcome of the construct of each brand.
Concluding acknowledgements
The NJR continues to work collaboratively with our many stakeholders; the most important, of course, are the patients we serve, and whom we would like to thank for allowing us to use their data. The NJR operational collaboration is a huge team effort – this year managed almost exclusively by work performed virtually. Elaine Young, NJR Director of Operations has demonstrated the great versatility of her leadership and her team.
Many thanks also to the following without which the NJR could not function:
All members of the NJR Steering Committee
Members of the NJR sub-committees:
Executive
Data Quality
Editorial Board
Implant Scrutiny
Medical Advisory
Regional Clinical Coordinators
Research
Surgical Performance
Members of Data Access Review Group
Members of the NJR Patient Network
Other organisations:
Medicines and Healthcare products Regulatory Agency (MHRA)
Care Quality Commission (CQC)
NHS England and Improvement
NHS Digital
Getting It Right First Time (GIRFT)
British Orthopaedic Association (BOA)
British Hip Society (BHS)
British Association for Surgery of the Knee (BASK)
British Elbow and Shoulder Society (BESS)
British Orthopaedic Foot and Ankle Society (BOFAS)
European Orthopaedic Research Society (EORS)
Healthcare Quality Improvement Partnership (HQIP)
NEC Software Solutions UK Ltd (previously known as Northgate Public Services UK Ltd)
University of Bristol
University of Oxford
Confidentiality Advisory Group (CAG)
Association of British HealthTech Industries (ABHI)
We are most grateful to our contractors for their very valuable input into the NJR Annual Report, and many other functions. NEC Software Solutions, University of Bristol and University of Oxford teams help us refine and improve each year. This year’s report is the biggest and best report yet. We offer our personal thanks to Vicky McCormack, Report Project Manager and Deirdra Taylor, Associate Director of Communication and Stakeholder Engagement for the NJR, for getting the final report into shape in the face of challenging circumstances.
On a personal note, we would particularly like to thank Laurel Powers-Freeling, Chairman of the NJR. Laurel’s leadership over the last ten years has seen the NJR grow in terms of size, quality, stature, and utility. Laurel brought huge insight to the NJR from her many other areas of expertise and her guidance has enabled the organisation to grow in ways we would simply not have been able to develop without those insights. We owe her a great deal and wish her every success in her future endeavours, followed by a long and happy retirement.
1Cumberlege J. First Do No Harm: The report of the Independent Medicines and Medical Devices Safety Review. 2020 Jul 8. [https://www.immdsreview.org.uk/Report.html]