The National Joint Registry 22 years on: what have we learned about knee replacements?

Abstract

The National Joint Registry (NJR) of England and Wales was established in 2002 to collect, analyse and report high-quality data about joint replacement surgery. Submission of knee arthroplasty data has been mandated since 2003 in the private sector, and 2010 in the NHS, making the NJR the largest orthopaedic registry in the world. This powerful audit tool collects information on preoperative, intraoperative and postoperative variables, enabling advancement in most of, if not every, facet of the procedure. These data can be used to inform surgeons and other healthcare providers, healthcare commissioners, industry, and patients themselves about outcomes and best practice, ensuring informed decision-making. Drawing from NJR reports, as well as work from external authors using these data, the aim of this review is to highlight what knee surgeons can learn from 22 years of the NJR. We will discuss trends in knee surgery, changes in surgical technique, surgical outcomes and future developments.

Introduction

The National Joint Registry (NJR) of England and Wales was established in 2002 to collect, analyse and report high-quality data about joint replacement surgery. It was developed to provide surgeon- and implant-specific data, but its inception was driven by the 3M Capital Hip Implant failure Report in 2001 with the aim of providing timely warnings of implant failure relating to patient safety. The Capital Hip implant was a copy of the Charnley-based system. It was withdrawn from the market in 1998, because unlike the original Charnley system it was associated with unacceptably high failure rates and significant patient morbidity. In response, the UK Government Department of Health (DoH) began to capture data about hip and knee procedures in England and Wales with the development of a NJR. For patients treated outside the NHS the DoH mandated submission of data to the NJR in 2003, and expanded this in 2010 to include patients treated within the NHS. More recently, data collection has been further expanded by including Northern Ireland, the Isle of Man and Guernsey, as well as replacements of the ankle, shoulder and elbow. With compliance rates of almost 95%, the NJR has now become the largest orthopaedic registry in the world, with approximately 250,000 new records added every year.

The NJR collects data on hospitals, surgeons and implants, creating an ever-growing and ever-more powerful audit tool. Registry data can highlight hospitals, surgeons and implants that fall above or below expected levels of performance. These data can be used to inform surgeons and other healthcare providers, healthcare commissioners, industry and patients themselves about outcomes and best practice, ensuring informed decision-making.

Regarding knee replacement surgery, the NJR collects preoperative, intraoperative and postoperative data. Preoperative data include indications for knee replacement surgery and range of motion. Intraoperative data recorded include the surgical approach, implants used, concurrent procedures, complications and use of adjuncts, such as tourniquets, and more recently, robotic devices. Postoperative thromboprophylaxis use is also collected.

The NJR now contains 22 years of data specific to knee replacement surgery. Over that time, changes have been seen in most, if not every, facet of the procedure; with new implants entering the market, surgical and rehabilitation techniques falling in and out of favour, and even changes in patient demographics. The aim of this review is to highlight what registry data has taught us up to this point.

General trends in knee replacement surgery

The NJR shows that the number of knee replacements performed in the UK, excluding Scotland, was at its peak between 2012 and 2019, at around 110,000 per annum, but fell after 2019, almost certainly due to the coronavirus disease (COVID-19) pandemic. In 2022, 103,303 primary knee procedures were undertaken.

In the latest NJR report, the NHS only accounted for 70% of pre-pandemic numbers of primary knee replacement procedures; meanwhile, the rate of knee replacement surgery within independent hospitals has continued to grow, partially filling the NHS shortfall.

Of all knee replacements, primary total knee replacements (TKR) remain the most common procedure, with around 85,000 performed each year, but the demand for tibiofemoral unicompartmental knee replacements (UKR) has never been greater, with a consistent increase every year (apart from those years affected by COVID-19). For patellofemoral replacements (PFR), the data shows the highest numbers were implanted between 2011 and 2017, with a slight decline since that time.

Population and patient trends

The rise in demand for knee replacement surgery is likely multifactorial, but data suggests that the ageing population is an important factor. Healthcare in general has improved with time, and the UK now has an average life expectancy of over 80 years, compared with 2002, when it was 78 years. The NJR tells us that since 2014, the age of those undergoing knee replacement surgery has remained similar, but as more people live into older age it is probably no coincidence that the demand for TKR has increased. The median age of patients undergoing TKR is currently 70 years old, with 71% of patients sitting within the 60–79 years age range (32% 60–69, 39% 70–79). Of the remaining 29%, slightly more (16–17%) are younger than 60, and the remainder (13–14%) older than 79. With an ageing patient demographic, it is perhaps also no surprise that the most common indication for knee replacement surgery is osteoarthritis, with 97.4% of knee replacements being attributed to this disease process.

Interestingly, the NJR also tells us that the body mass index (BMI) of those undergoing knee replacement surgery has remained remarkably constant. The mean BMI of those undergoing knee replacements between 2014 and 2022 ranged between 30.5 and 31 kg/m ² , with two-thirds of patients having a BMI between 25 and 34 kg/m ² . Contrary to these NJR data, rates of obesity in the general population have been steadily increasing, and obesity is known to be a factor in the development of osteoarthritis of the knee. These data are counter-intuitive and need further study. It may be that the rise in obesity rates within the general population are countered by an increasing reluctance amongst surgeons and commissioners to offer knee replacements to patients who exceed pre-determined BMI thresholds. This could explain increasing rates of knee replacement surgery and account for the surprising disparity between a consistent BMI within the NJR and rising rates of obesity in the general population. In other words, there are more patients in each range of increased BMI contributing to more osteoarthritis of the knee and more demand for surgery, but knee replacement is no longer being offered to patients with a very high BMI. This is an area for future studies.

With innovations in knee implant design and materials, as well as surgical techniques to improve longevity, combined with reassuring data that most knee replacements survive many years, one might hypothesize that surgeons’ reluctance to offer knee arthroplasty to young patients has reduced. This may be expected to have reduced the age at which patients undergo this surgery, but this does not seem to be the case. Up to 2020, approximately 13,000 TKR (15%) per year, were undertaken in patients aged 59 or younger. Since the COVID-19 pandemic, the total number of TKRs has reduced in this age group, and in 2022 this figure was approximately 11,500. This represents an absolute reduction in numbers, but also a proportional fall to 13% of all TKRs.

One explanation that might explain the relatively static age demographic is a progressive decrease in the physical fitness of those undergoing surgery. Within the NJR, the American Society of Anaesthesiologists (ASA) grade is used to assess a patient’s physical fitness for surgery, and comorbidity plays a key role in determining this grade. Since 2014, we have seen a proportional decline in those assigned the fittest ASA grade, with a corresponding increase in those being assigned higher (worse) ASA grades. In 2014, 10% of patients were assigned the fittest grade (Grade 1) and 16% assigned Grade 3 (severe disease that limits activity). In 2022, only 6% were assigned the Grade 1 and 20% Grade 3. There has been no change in those undergoing knee replacement surgery at higher grades (Grades 4 and 5), and this finding would not have been expected if overall comorbidity was increasing. This disparity in evidence may be because although the overall health of patients presenting with symptomatic knee arthropathy has worsened, and surgeons may be avoiding joint replacements in very high-risk patients. It is probable that co-morbidities are not increasing equally amongst all age groups, and one may expect that with increasing age, co-morbidities also increase. It is equally possible that in two different patients with ASA Grades 4 and 5, surgeons may apply different thresholds to offer surgery in older versus younger patients. Both of these factors would be expected to influence rates of knee arthroplasty within different age groups, but these data merit further investigation.

The increase in ASA is also reflected in external work drawing upon NJR data. A published study of co-morbidity trends from Bristol, in 2022, looked at over 1 million hip and knee arthroplasties between 2005 and 2018. The authors reported significant increases in the number of comorbidities recorded by patients undergoing primary arthroplasty. The key disease processes on the rise included renal failure, heart disease, diabetes and chronic obstructive pulmonary disease.

Implant trends

The most common knee surgery procedure in the NJR is, unsurprisingly, TKR. For the past 10 years, TKR has accounted for over 85% of all primary knee procedures undertaken. However, although it remains the most common knee replacement surgery, the data shows a subtle but consistent decline and a corresponding increase in UKR over the same period. UKR now represents approximately 14% of primary procedures, an increase from 8% in 2009. Interestingly, since the inception of the NJR, more UKR have been implanted in non-NHS hospitals, but since 2013, the NHS has increased its use of UKR from 6% to 12%. Overall, these data probably reflect growing evidence and confidence in UKR as well as recent changes to national guidelines. The evidence also shows that on average, younger patients and those with fewer co-morbidities undergo UKR, and this possibly explains the higher rates of these procedures outside the NHS.

The increase in popularity of UKR has also led to an increase in multicompartment knee replacements, although they still represent less than 1% of all knee replacements. PFR has consistently remained at 1%.

Knee replacements, both TKR and UKR, can be categorized depending on their type of bearing, either being mobile or fixed. In 2006, 75% of UKR were of mobile bearing design, but this has decreased, and is now only 40%, with the remaining 60% being fixed bearing. The same trend, although not as marked, is seen in TKR. In 2006, 13% of TKR had mobile bearings, and this has now declined to 4%.

In addition to bearing type, TKR can be analysed in terms of constraint. A simplified classification of TKR may be to define them as unconstrained, posterior-stabilized or hinged/linked. In 2006, unconstrained prostheses represented 73% of all TKRs implanted, and 26% were posterior-stabilized. Since 2006, there has been a small but consistent rise in unconstrained implants to 79%, with a corresponding drop in posterior-stabilized implants. Hinged or linked knee use has remained static at 1% since records began.

The NJR tells us that currently, cemented, unconstrained, fixed bearing TKR remains by far the most common type of knee replacement, at 76% of the total TKRs implanted, followed by cemented, posterior-stabilized, fixed bearing TKR, at 18%.

Trends in surgical techniques

In addition to choice of implant, surgeons need to make informed decisions regarding surgical technique. In TKR, an important consideration is whether to perform patellar resurfacing. Since 2006, the trend has been in favour of resurfacing, rather than not. Between 2006 and 2013, the patellar resurfacing rate remained relatively constant (35–38%), but the past 10 years have seen a steady rise in patellar resurfacing to the current rate of 52%. This may be in response to the known increase in revision surgery for arthritis in the remaining joint, and growing evidence that primary patellar resurfacing reduces rates of revision surgery. In 2021, a retrospective review of NJR data, encompassing 842,072 patients from 2003 until 2016, showed the cumulative risk of all-cause revision at 10 years was higher in primary TKRs without patellar resurfacing compared to those with resurfacing, at 3.54% versus 3.00%, respectively.

Once implants are chosen, decisions must be made about the method of fixation. In 2006, 7% of knee arthroplasties were uncemented, 92% cemented and 1% were hybrid. The number of hybrid knee replacements has remains unchanged, but by 2017, 96% of all knee arthroplasties were cemented.

Interestingly, despite the general trend in favour of cemented knee implants, there has been a notable increase in uncemented UKR. In 2010, only 1% were uncemented, compared to 7% in the latest NJR report.

For cemented implants, the cement itself has evolved, and now almost 100% of primary arthroplasty is performed using cement containing antibiotics.

The NJR data show that the surgical approach to a knee replacement has remained consistent, with 93% using the medial para-patellar approach. Computer-guided surgery has seen a modest increase, from <1% in 2006 to around 3% today. In 2019, it was acknowledged that there had been an increase in the use of robotic-assistance in arthroplasty surgery, and since then, robotic use has been specifically recorded on postoperative NJR forms. At present, there is a paucity of data from the NJR related to the use of this technology in knee surgery.

Outcomes and revision rates

The NJR demonstrates that over time, revision rates have progressively reduced. The 3- and 5-year revision rates have been reducing yearly since 2013, with similar trends observed at 7 and 10 years, albeit with less data. There are currently over 98,000 revision surgeries recorded in the NJR database, and in 2022, 5740 knee revision procedures were performed (5% of total knee arthroplasty volume). In the latest NJR report, the most common indication for revision surgery was recorded as aseptic loosening (35%), followed by infection (27%), progressive arthritis (19%), instability (14%), lysis (12%) and polyethylene wear (11%).

Aseptic loosening has been the most common reason for revision since the inception of the NJR, but it has been falling as a proportion of overall revisions. In 2006, this was reported as the indication for revision in 56% of cases, but this has gradually reduced to the current rate of 35%. As rates of aseptic loosening have fallen, revisions for progressive arthritis have steadily increased. This increase might well in part be explained by a matched rise in the utilization of UKRs.

Reported rates of infection, lysis and polyethylene wear as an indication for revision have all remained relatively constant.

Since the inception of the NJR, the most reported surgery for aseptic loosening is a single-stage revision. By comparison, multi-stage revisions have remained the commonest treatment for infection. More recently, rates of debridement, antibiotics and implant retention (DAIR) procedures have increased. DAIR as a specific procedure was only included within the NJR in 2018, and reported rates have progressively increased since then. The latest NJR report showed that 40% of infected knee replacements were treated with a multi-stage procedure, 40% a DAIR procedure and 20% a single-stage revision.

In terms of implants, UKR fare worse than TKR, with the overall risk of revision for UKR being double that of TKR at every timepoint. The overall revision rate for cemented UKR is × 3.1 higher than the observed rate for cemented TKR at 10 years, and × 3.6 higher at 19 years; however, there is a very broad spectrum of outcomes dependent on prosthesis type. The latest NJR report shows that the best performing UKR has a 10-year revision rate of 5.44%, and perhaps importantly, a median age of implantation of 64 years. By comparison, the worst performing prosthesis has a 10-year revision rate of 17.69% and a median age of implantation of 62. The best performing UKR is a fixed-bearing cemented implant, with a cumulative revision rate of 4% at 6 years. Uncemented or hybrid UKR have similarly low revision rates at early follow-up, but after 4 years their revision rates appear to increase at a faster rate, and this may require monitoring over a longer period.

The overall revision rate for PFR is × 5.5 higher than for cemented TKR at 10 years, with longer-term outcomes difficult to analyse because of low numbers. However, it is again important to note a very wide variation of revision rates depending on prosthesis type, with the best performing having a 10-year revision rate of 12.33%. This rate, although significantly above that of most primary TKRs, is in the context of a median age of implantation of 56 years. For comparison, the 10-year cemented primary TKR revision rate in the 55–64 age group is 4.85%.

The primary TKR implants most at risk of revision are hinged or linked prostheses, with a 6% revision rate over the first 6 years. The NJR does not suggest a reason for these implant failures, but it is likely due to confounding factors such as the complexity of the surgery in which these implants are often utilized.

The lowest revision rates are seen in monobloc polyethylene tibial and unconstrained, fixed-bearing knee prostheses, with a cumulative revision rate of around 2% at 6 years; however, it should be noted that the median age at implantation for these prostheses is 74 years (IQR 69–79). All types of mobile-bearing knee replacements have higher cumulative revision rates than their fixed-bearing contemporaries.

Since 2003, the total number of knee replacements undertaken by low-volume surgeons (less than 24 TKR per year) has remained constant, accounting for approximately 5000 procedures per year. This previously equated to 20% of all knee replacements, but there has been an increase in higher-volume surgeons’ (50 or more) output, and low-volume surgeons now only account for 4% of all TKRs implanted per annum. The latest NJR report showed that over 70% of TKRs are performed by high-volume surgeons, and the trend is similar for UKR. In the 2004 NJR report, over half of UKRs were undertaken by surgeons performing less than 12 per year, but that figure has now fallen to 15%, with 70% being performed by those doing over 24 per year.

It is known that patient outcomes tend to be better with higher-volume surgeons, and this is seen with NJR data, which demonstrate that an increase in high-volume surgeon output has coincided with a reduced rate of intraoperative complications. According to NJR data, the chance of intraoperative complications during a TKR is currently as low as 0.4%, and this figure has been stable since the upturn in high-volume surgeon output was noted in 2012. Before 2012, the chance of intra-operative complications was almost double today’s figure, but still low, at an average of 0.7%. The most common complication during TKR is fracture, and this is reported in 0.1% of primary TKRs procedures, with ligament injuries and patellar avulsions accounting for 0.05% each. The rate of intraoperative complications during UKR is lower, at 0.2%, with a similar proportion being from fracture and ligament injury.

At all timepoints, males have a higher rates of revision surgery for both TKR and UKR. Age at implantation of TKR or UKR is also a major predictor of revision risk, with younger patients being consistently at higher risk of revision, as illustrated in Figure 1 .