REGISTRIES....What have we all learned and what of the future? Our experiences with NJR

Introduction

Nowadays joint registries are accepted as part of everyday orthopaedic practice. Not only do we have registries for total joint replacement but there are now registries of one type or another in many other orthopaedic specialities. Arguably it was the joint registries, particularly the National Joint registries that led the way. They have given orthopaedics a massive profile and many other specialities are now trying to emulate us. They have initiated the whole concept of “Observational Studies” and completely changed the mind set of most orthopaedic surgeons and given us “EVIDENCED BASED DECISION MAKING”. It was not that long ago they started and like so many culture changing initiatives it took quite a while to get them started and before their use fully appreciated.

The theme for this EFORT meeting is Registries and it is an opportunity for us to celebrate the contribution that all registries have brought to orthopaedics, particularly joint replacement, over the past few years. Europe has led the world with the development of registries and we have shown how the data collected in registries can influence practice for the benefit of all or patients. We have both the oldest and the biggest registries, the Swedish and the NJR. I hope you find my views on registries and allied systems interesting and that you enjoy EFORT 2019.

Disasters give registries a kisk start

We all know that the two men, the original pioneers of joint replacement, were Sir John Charnley from Wrightington and Ken McKee from Norwich. There is no doubt they were competitors but were also friends and both kept impeccable records of their cases (Fig 1).

Perceptively, Sir John wrote in 1972: “Serious consideration should be given to establishing a Central Register to keep a finger on the pulse of total implant surgery on a nation-wide basis” (JC Internal Publication No. 39 July 1972). The late Robin Ling, whose Exeter unit are renowned for their record keeping, also made the case for a registry in the early 1990s.

In the UK, it took the 3M capital hip disaster (1) to mobilize the British government of the time, to support the British Orthopaedic Association (BOA) in their case for a National Joint Registry (NJR). The 3M Capital Hip was a “look-a-like” Charnley in appearance but not in performance (Fig 2). They were cheaper than the Charnley and the NHS encouraged surgeons to use them. Perhaps it was the fact that the government held the smoking gun and that they had to “do” something that we at last got our registry!

LESSON! AN APPARENTLY MINOR MODIFICATION CAN LEAD TO A MASSIVE CHANGE IN PERFORMANCE
The 3 M Capital hip disaster brought to light several other failings besides the uptake of a poor design. In the UK, we had little idea of which patients had received these hips and no easy way of contacting them. The 3 M capital hip was the trigger for the NJR and ODEP for hips, but we all were aware of several types of joint replacement that not lived up to expectation and as surgeons there were many of us who had been demanding a registry for a considerable time (Fig 3).

The late Mike Freeman’s mark 1 knee worked well. The addition of a metal plate underneath the HDP Tibial component showed the unexpected consequence at 1 year.

In England, before NJR we had the Trent Regional Artroplasty Register which was pioneered by Prof Paul Gregg in Leicester. Paul’s vision and experience was pivotal in commissioning, promoting and developing the NJR. Tim Wilton has recently become the Medical Director of the NJR, having succeeded Mr Martyn Porter. Tim previously served on the Steering Committee besides having been President of BASK and the British Orthopaedic Association besides being Chairman of the Bone and Joint Journal (BJJ). Tim is a member of ODEP and Beyond Compliance (Fig 4).

Other registries have not needed such a sad reason for getting under way; the father of them all, the Swedish Knee Registry (SKAR) was started by Goran Bauer in 1975 who is quoted as saying that Total Knee Replacement was an experiment (2). He also reported a case where a patient received a sided TKR in the wrong leg. He felt that careful records should be kept of all joint replacements (Fig 5, 6).

The Swedish Hip Registry (SHAR) was next 1979, started by Peter Herberts and Lennart Ahnfelt. For many years, and to this day Henrik Malchau and Göran Garellick have had a massive and positive influence on all our registries (Fig 7-9).

Many other registries have started because the local surgeons had heard what problems had occurred in another country and wanted to make sure it was not repeated in theirs. The Australian registry has been pre-eminent. Steve Graves’ name will be forever associated with it and he kindly wrote to describe its origins: The instigation of the AOANJRR commenced in 1993 when the AOA Arthroplasty Society asked the AOA to look at establishing a National Registry. The reason for this was there was uncertainty of the outcomes and it was not even clear how many joint replacement procedures where being done. It was also based on the success of the Swedish registries which had been demonstrated to improve the outcomes of joint replacement. There was no concern over a particular device but it was thought likely that there was variability of results based on the type of prosthesis used. I was first approached by the AOA to look at implementing the registry in 1996. I lead a pilot study to determine the feasibility and the success of this led to federal Government funding being provided in 1998. The AOANJRR commenced data collection in 1999 (hips and knees) with complete national implementation being completed by mid 2002. Other joints shoulders elbows wrists, ankles and spinal discs commenced in 2007 and osteotomies 2015.

What is needed for a registry to function properly ?

The best and easiest way to produce a registry is where an implant is used. The data collection can be built around the event, using the bar code on the device.

PATIENT DATA + HOSPITAL DATA + IMPLANT BARCODE = REGISTER
REGISTER + FILTERS = REGISTRY
REGISTRY + INTELLIGENCE = PATIENT SAFETY

In these days of patients having electronic hospital records their dated demographic data, the surgeon’s name and the barcode can easily be put together by a data clerk and uploaded into a computer and exported to a registry. In the UK a patient’s signed consent is needed for this system to work most effectively (Fig 10).

The barcode usually contains the UDI (Unique Device Identification) and the product code. Some barcodes include the Global Medical Device Nomenclature (GMDN). Behind the product code are details of the implant. Nowadays all the major registries are computer based and all this data is collected in their main frame. For the registries, like NJR who are set up to evaluate implant performance, manufacturers will inform the registries of the attributes associated with an individual product code. The registries will have a library of attributes that a manufacturer will be asked to ascribe to a product. AN ATTIBUTE IS JUST ONE FEATURE OF AN IMPLANT such as size, shape, substrate material, bone facing surface, bearing surface, etc. The example below just relates to a femoral component of a knee replacement. You will notice that this classification is common to both NJR and EPRD. In the combined NJR / EPRD registry there are >560 Hip and Knee Attributes (Fig 11).

The architecture of the database / library... "Camouflaging"

A few years ago, NJR realised that their database architecture had not kept pace with our aspiration of detecting outlier implant performance. The small beginnings of the NJR had steadily developed into “BIG DATA” and we were running into the problems of Big Data camouflaging poor performance of a variant of a brand. The first example of camouflaging that we eventually identified was a very painful one for all of us at NJR. We failed to pick the early failures of “Stemmed Metal on Metal” hips which lead to so many patients being severely damaged. Until then we had been collecting cemented hips, uncemented hips, hybrids (cemented stems and uncemented cups) and resurfacings.

By the time of large head stemmed MoM THRs both the hybrid and uncemented silos were very large (> 10,000 hips). These silos were mainly made up from a large number of metal on HDP hips with modular uncemented acetabular components. The PTIR (Patient Time Incident Rate) for the whole “Hybrid Group” was extremely low and (initially) the small number of failed MoM hips were not enough to trigger outlier status for the group although the PTIR for the MoM subset was grossly raised. It was reports to the MHRA (Medicines and Healthcare products Regulatory Agency, the UK’s Competent Authority) that triggered us looking at the MoM group where to our horror we found the truth. We also found extensive use of “Mix and Match” where a surgeon had used components from more than one manufacturer (3).

The answer

What we needed was more granularity, ie many more attributes in the database and a new architecture. Before we risked “redesigning the wheel” we looked around other registries and decided the layout and granularity of the EPRD (Endoprothesen Register Deutschland) was what we needed to at least emulate. We needed to have database that would evolve and which we could modify with the introduction of new attributes be major ones (a completely new material) or ones which were a lot more minor. For an attribute to be accepted into the database we had to think it was relevant. The test we applied to all the attributes before they were adopted was “Could a modification to this attribute possibly lead to an increase in risk?” The database has also been designed so that an investigation could be conducted vertically down through one brand or “horizontally” across a certain attribute that is common to several implant brands.

It was important that manufacturers were continuously involved in its design; there are two representatives from the implant industry on the NJR steering committee and we hold regular meetings with Industry several times a year. Recently we dropped 3 relatively unimportant attributes from the list on account of the difficulties industry was having with them and their rather doubtful use (Fig 12).

Thus, NJR and EPRD have a common database which is regularly reviewed. RIAP (Registro Italiano di Artro Protesi) are also adopting the data base and there has been considerable interest from other registries. This linkage will increasingly give us the ability to simultaneously check on the performance of an implant or an attribute in one very large registry (NJR) and one that will be eventually be much bigger (EPRD) in the future. It is anticipated that “Beyond Compliance” (vide infra) will extend to Germany at some stage.

Others have asked NJR assistance

Enthusiasts from several other counties / registries have come to NJR for advice, to set up their joint replacement registry on the basis of NJR experience. They include Japan, Singapore, India, the Republic of Ireland, Zimbabwe, Brazil and several others. We have always been pleased to help.

Several non-arthroplasty registries have built on our experience and we are or have been working with Spine Tango, the British Spine Registry, the Brain pacemaker registry and several others besides TORUS (Trauma & Orthopaedic Registries Unifying Structure) Registries that do not have an implant around which data collection can be built have other challenges, particularly if “Revision” is not a reliable metric. Enquiries are always welcome. We believe in what we do and are always delighted to help others.

A REGISTRY CANNOT SIMPLY BE A MASS OF COLLECTED DATA….. IT IS A WHOLE VITAL, EVOLVING ORGANISATION WHICH ADDS INTELLIGENCE TO THE DATA

What makes up NJR?

1. A very advanced IT company (Northgate Public Services) who are responsible for the data collection from hospitals, data housing and making data available to industry (Supplier feedback) and the analysts….IT is critical to the success of any registry.
2. The Analysts, “The University of Bristol” who are responsible for making sense of the data principally for
3. The Implant Performance Committee
4. The Surgeon Performance Committee
5. Research Committee
6. The organisation behind it all, HQIP (Healthcare Quality Improvement Partnership) with the Steering committee and the medical advisory committee.

In my view, Registries must be “inclusive” if they have any chance of succeeding. On the main committee of the NJR, the “Steering committee” all our stakeholders are fully represented. It is chaired by a lay person, at present Laurel Powers-Freeling . There are surgeon representatives, patient representatives, representatives from the joint manufacturers, procurement, NHS management, allied professions and of course its own management, headed by Elaine Young, the Director of Operations, on behalf of HQIP. There are several committees most of which are attended by the Medical Director, Mr. Tim Wilton who is a highly respected orthopaedic surgeon. The previous medical director was Martyn Porter who was in at the beginning of NJR.

Working with others

NJR, works closely with the MHRA (The Medicines and Healthcare products Regulatory Agency ), ODEP (The Orthopaedic Data Evaluation Panel) and Beyond Compliance besides EPRD and other registries and organisations. All the world’s major registries link up with ISAR (International Society of Arthroplasty Registries). ISAR had its 3 day 8th Annual International Congress last weekend in Leiden. Some of the member registries are National Registries and those who play a big part, include the Scandinavian Registries, the Australian and New Zealand Registries, the Netherlands and the German Registries besides the relatively small but very important American Registry. Other registries are regional such as some of the Italian Registries (who come together under RIAP), the Kaiser Permanente Registry and many registries in the USA. And then there are local registries such as the very excellent Geneva Registry. They all play their part.

Completeness, compliance and linkability

ISAR has drawn up guidelines for the quality assessment of Registries. In NJR we have well over 99% coverage and 97% compliance. Compliance is the ratio of operations performed and records being uploaded to NJR. The link between the index operation and a revision is critical and this is what is meant by linkability. Linkability is pivotal to the monitoring of surgeons, hospitals and implants with any reliability.(Fig 13-14)

Whilst ISAR is an international organisation, the Network of Registries of Europe (NORE) was formed about 4 years ago to be a forum for European Registries and is chaired by Prof Rob Nelissen. NORE advises EFORT on Registry matters and importantly is available to advice the European Commission should the commission require advice on joint replacement. It is also a forum for members of registries to communicate and also to produce joint papers such as the recent publication on the Surveillance of Metal on metal hip replacements in Europe (Fig 15).

THE MOST IMPORTANT QUESTION!
WHAT DO WE USE REGISTY DATA FOR?
THE ANSWER “FEEDBACK”
TO DISCRIMINATE BETWEEN THE GOOD AND THE BAD
THERE IS NO POINT IN COLLECTING DATA IF IT IS NOT FED BACK!

There are limitations to "revision"

Revision can be a rather “blunt tool” and it is important to understand its limitations (Table 1).

An example of where we hope PROMS data will be particularly useful in benchmarking shoulder replacements where revision can be a very week metric. Unless a shoulder prosthesis is infected or very painful it might not be removed (which would trigger the revision word) if it was thought that surgery would not be helpful. The implant is not revised but is pretty useless. (Fig 16)

In the NJR, the Oxford Hip, Knee and shoulder PROMS together with WOMAC, are collected by NJR pre operatively and then postoperatively at 3 & 5 years. PROMS mean a patient has to fill in a form and in the NJR, the post-operative PROMS scores are completed by between 60-80% of patients. These are linked to the pre-operative PROMS score collected pre-operatively. Paper PROMS are extremely costly and when patients give us an e mail address it is very much cheaper.

Example of uses of registry data

1. RECALL
It goes without saying the most important function of registry is to be able to contact patients if something has been noticed to go wrong with their implant. Within 2 weeks of the recall notice for the ASR NJR had contacted all patients who were recorded in the registry with this implant.

2. INFORMATION FOR REFLECTION AND AUDIT
Nowadays, in the Annual reports of all the major registries there are excellent data showing usage, trends and outcomes associated with total joint replacements. The NJR Annual report focuses very much on outcomes. (Fig. 17)

The steady improvement in completeness, compliance and linkability since the start of NJR.
A reassuring message to any registry that has a period of despondency in its early days! (Fig. 18)

3. TRENDS
NJR has reported annually for 15 years and many observations have been made. Trends have been followed such as the trend in the use of different types of THR. (Fig. 19)

One of the most startling outcomes we have come across was with Metal on Metal (MoM) particularly in respect of the ASR and Stemmed Metal on Metal (Fig 20).

4. REVISIONS
Fig 20. The graph for revision hip shows a steady improvement rate in hip revision over the years except 2007 -9 when the results were compromised with MoM (Fig 21)

More recently NJR has been tracking Total Shoulder Replacement. Interestingly this graph shows how the performance of TSR in trauma is tracking the performance of TSR in elective patients. (The time frame is very short)(Fig 22)

5. FEEDBACK
Feedback is THE KEY TO “BUY IN” AND SUCCESS FOR A REGISTRY.

a) NJR CLINICIAN FEEDBACK
NJR clinician feedback is a resource available to all orthopaedic surgeons in England, Wales, Northern Ireland and Isle of Man who undertake hip, knee, ankle, elbow and shoulder replacement surgery and are registered with the National Joint Registry (NJR). It gives them all the data about any operation that NJR has recorded against their name and also includes details of procedures undertaken by other members of their team. Their data will also include their funnel plots, arranged in such a way that they will be able to detect any trend that is developing in their practice”.

www.clinicianfeedback.org.uk is therefore the route all our surgeons take to check on their practice and many of them will use the data for their annual appraisal. Best practice is to share their results with colleagues at a 6 monthly department meeting. Below is the home page for my clinician feedback. (Fig. 23)

b) SUPPLIER FEEDBACK
Supplier feedback is very popular with manufacturers and is unique to NJR. (Fig. 24, 25)

• access to non-identifiable record level data, tracking usage and outcomes of their products
• data for post market surveillance purposes only.
• data for submission to ODEP and to MHRA.

6. NJR VIGILANCE
The NJR monitors the performance of

• All Surgeons
• All Implants
• All Hospitals

SURGEON PEFORMANCE COMMITTEE
Since 2008, when NJR had reached a degree if maturity, the surgeon performance committee, chaired by Peter Howard (Derby) meets twice yearly to review all the NJR listed surgeon’s performance. Their performance is measured against the norm using the SRR (Surgeon Revision Rate) as the metric. When a surgeon's position on the funnel plot crosses the 95% control limit (alert level), they are contacted to advise them of this, and recommending they examine their data and practice. When they cross the 99.8% line (alarm level, or "outlier") they are contacted again, and the Medical Director of their institute is notified. Should they wish to discuss their status with an NJR surgeon (with or without declaring their name) they can.

The performance of hospitals/Trusts/units is similarly monitored and reported upon. Units with persistently poor performance are recommended for an external review process with the British Orthopaedic Association. (Fig. 26, 27)

Peter Howard has spent a lot of time looking at the data that is typically associated with outlier surgeons and has come up with some very interesting generalisations which he has presented in numerous meetings: (Fig. 28)

THE IMPLANT PERFORMANCE COMMITTEE
In NJR, implants are monitored in a similar way by surgeons. Since 2009, the Implant Performance Committee meets twice yearly and reviews the performance of all implants in the registry. Electronic filters are applied to the raw data by the team at Bristol. PTIR (Patient Time Incident Rate) is used as the main detector of poor performance. An implant with a PTIR of more than twice the group average and has > 100 usages are described as a potential outlier and the MHRA is informed together with the manufacturer. If the PTIR is between 1.5 and 2 times the group average or > 2 but in a cohort < 100 usages, the manufacturer is informed. We also use “Word on the street” as a trigger for a drill down and we look to see if other registries have identified a particular implant that is performing poorly. “Word on the Street” comes from surgeons at meetings and other informal gatherings. It can be a pointer to more serious findings.

As of January 2019, 163 brands of knee implant, have been monitored with the identification of 4 Level 1 outliers and 12 Level 2 outliers. Some 2,806 different combinations of hip stem/cup implants have been monitored with 17 Level 1 outliers and 40 Level 2 outliers.

7) NJR RESEARCH
The massive amount of data that is stored in a mature registry lends itself to research and audits. NJR research is conducted through the research committee chaired by Prof Mark Wilkinson (Sheffield). Some of the research is undertaken “Internally” by the NJR department in the University of Bristol (UOB) under the supervision of Prof. Ashley Blom or his senior lecturer, Mike Whitehouse whilst many other papers have been written externally.

To date, 91 papers have been published using NJR data.

RESEARCH FELLOWS
The NJR has on its staff one or two research fellows. To date these have always been “out of programme” senior trainees (senior residents) who take a year out from their adopted training programme and undertake audits or research under the direction of a UK academic orthopaedic unit. They are appointed by the Royal college of Surgeons of England through open competition. Their work usually leads to a MD or occasionally a PhD. NJR research fellows have had 28 peer reviewed papers accepted over the past 8 years.

Registries have given Orthopaedic Surgeons, for the first time, the ability to do observational studies using “Big Data”. Orthopaedic is way ahead of all other specialities, certainly in the UK, in collecting large quantities of data to assess the value of a product or any of its attributes. (Table 2)(Fig. 29)

8) REPORTING
Most Registries report annually electronically or on paper or both. Some annual reports tend to look like a large telephone directory! (Fig. 30)

ACTA Scandinavia
It has been a great boost for all registries that the editors of ACTA have recognised that research papers using registry data need a home. Most journals have now come round to accpeting articles quoting registry data and have got their heads around “Observational Studies”. There was a time when several articles were pushed back by reviewers because some of the patients in a cohort had only follow up of a year or so. The reviewers failed to understand that a cohort from a registry will always have patients in who the joint has been in place for say 10 years and some patients will have only reccently had their operation. The fact that surgeons are still using that prosthesis is information in itself.

9) BENCHMARKING
Quite rightly, Patients, Surgeons and hospital mangers like to know that an implant has been independently assessed as being fit for purpose. Benchmarking is an accepted method for achieving this end. Three countries provide a benchmarking service for joint replacements, Australia, the Netherlands and the UK. ODEP, the Orthopaedic Data Evaluation Panel is based in the UK and has an international profile.
REGISTY DATA NOW FORMS THE BASIS FOR BENCHMARKING FOR ODEP.

ORTHOPEDIC DATA EVALUATION PANEL (ODEP)
ODEP for Hips was introduced in 2003 and at that time there was a paucity of registry data. Only the Scandinavian registries had any maturity; the NJR had not started and for most implants there was little data outside often very limited clinical trials etc. Nowadays, although validated data from RCTs and other trials is welcome by ODEP, the majority of data that manufacturers use in their ODEP submissions is derived from registries. ODEP accepts data from registries where there is close linkability between primary and revision surgery.

ODEP for knees was introduced in 2014 and since 2016 there has been a staged introduction for ODEP for shoulders.

The panel consists of experienced orthopaedic surgeons from the UK and the Netherlands. The Netherlands merged their own benchmarking system with that of the UK about 3 years ago. (Fig. 31, 32)

ODEP has become an international brand. We benchmark implants from all around the world. ODEP seems to be very popular. Manufacturers use their ratings in their advertising. ODEP promotes good implants whilst making it difficult for poor implants to be marketed. (Fig. 33, 34)

"BEYOND COPLIANCE"…… the gap between the CE mark and ODEP 3
NJR was the registry that confirmed that the ASR surface replacement was doing very poorly and that “Stemmed Metal on Metal” was running into serious trouble. The Australian Registry had picked up the poor results from ASR before NJR data was statistically sufficiently strong to make the final call which lead to the MHRA field safety alert. A lot of us felt that we should have picked up these problems earlier. All these implants were “COMPLIANT”, they had a CE mark and had passed the FDA 510K assessment. We felt that there had not been nearly enough emphasis on diligent post market surveillance.

What all this meant was that unless we changed the CE mark rules, which was clearly impossible, we would continue to have implants in the British market which were not adequately monitored.

So, provided it was voluntary and did not break the rules around the CE mark, why could we not go “BEYOND COMPLIANCE” (fig. 35)

With NJR, we were in an excellent position to introduce such a system. Manufacturers would meet with the Beyond Compliance Advisory group so that a risk assessment could be performed which would lead to a discussion as to the rate of introduction for that particular implant and then it would be very carefully monitored. This system has been working since 2012 and the figure below give an idea as to where we are now. Manufacturers have to confirm that any surgeon using one of these new implants has been appropriately trained.

The upload into the NJR is as for any upload but when the data hits the NJR main frame the product is recognised by its bar code details and all the data is copied into a separate dedicated electronic file prepared for that implant as in the figure below. You will see how other information is added to this separate repository to generate a rich information profile for the implant. (Fig. 36)

At present there are about 43 implants at various stages in “Beyond Compliance” and 1,353 are registered as having been trained in the use of one or more BC products. Several implants have moved on into “Safeguard” which is a less stringent system but means that the implant is still kept under review.

After the initial assessment there are BC review meetings and “User Group meetings” when we share the data we have with surgeons using the implant and the manufacturer, whilst at the same time taking note of any problems that users might have experienced, whilst using the new implant.

ODEP and Beyond Compliance are developing an App so that we can develop a personal link with the patient and we are working closely with the company “My Recovery”.

We have no doubt that Apps are the way forward in data collection, particularly improved PROMS. We are investigating how machine learning might help us analyse all the data that will be produced in the future. (Fig. 37)

9) VALIDATION OF REGISTRY DATA
This means validation of compliance and particularly compliance with uploading revisions. It has been suggested that surgeons and hospitals might be reluctant to upload their revisions for what some would say were obvious reasons. Maybe patients going through emergency operating theatres could get missed out by a local system based on ORs dedicated to joint replacement. NJR takes validation of its data very seriously. In 2014, the NJR started a review of its current understanding of data quality in the registry and how to increase data completeness, timely submission and data accuracy further. This work is summarised in the NJR's Supporting data quality strategy document and covers all of the areas that affect the quality of data including existing activities and new initiatives that was implemented over the subsequent years. NJR are confident that the data they issue is extremely reliable, particularly since 2012.

10) ACCEPTABILITY….. The Attitude of Surgeons in the UK
This figure demonstrates my view about the change of attitude of surgeons to the NJR since 2003. (Table 3)

Surgeons do get very concerned when they think that their data could be made public when doubtless the press would pick up on it but basically it is “Not Cool” not to co-operate with NJR.

Costs

The NJR is funded in two ways. Firstly, there is a levy of circa £13 attached to each joint replacement and then manufacturers pay for the NJR data they require for assessing the performance of their implants, ODEP submissions and regulatory reasons etc. The annual cost of running NJR amounts to about £2 million.

REGISTIES AND THE NEW MDR (MEDICAL DEVICE REGULATIONS)
To obtain a CE mark with the new MDR, manufacturers are going to have to undertake a clinical investigation for any of their new implants, unless they have convincing evidence that they have a closely equivalent implant in their portfolio. No longer will they be able to cite “Equivalence” to a product from another manufacturer with the new rules. Obviously, supporting systems such as “Beyond Compliance” could be very helpful in this respect. Likewise, with legacy products, clinical investigations are going to be needed for manufacturers to support their continuing use of their CE mark. ODEP would provide externally validated clinical data for this purpose. It looks certain that surgeons will need to realise they will have to co-operate with manufacturers to produce clinical evidence to satisfy the Notified bodies so that their favourite joint replacement can stay in the market.

WHAT JOINT REGISTRIES HAVE WE ACHIEVED
Many would argue that our joint replacement registries have set a new standard for information backed decision making across the whole of medicine. (Table 4)

The future

For us the future is exciting. Bigger and bigger data will probably generate some unexpected issues. It seems that Artificial Intelligence (AI) or at least machine learning will have a part to play in data analysis if the experts are proved correct. Apps for patients so that we really do know how they are performing will be the norm and that will certainly need some very smart analytic systems if only for the massive volume of data traffic. There can be little doubt that patients and surgeons will only want to undertake treatments that are evidenced based. To all those who are interested in this work now and in the future, I send my very best wishes.

Acknowledgments

My thanks go to numerous people who have helped supply material for this article including Tim Wilton, Peter Howard, Peter Kay, Otto Robertson, Henrik Malchau, Liz Paxton, Paul Gregg, Richard Armstrong, Olga Taylor, Edd Caton, Mike Swanson and Chris Boulton

All our thanks goes to all the patients who allow their data to be recorded in registries

I would like to record a special thanks to Prof Siegfried Hoffmann for his editing and encouragement

The views expressed in this article are those of the author, based on his experience which includes, helping to set up NJR, serving on the steering committee, chairing the Implant Performance committee and being a member of several other NJR committees. He continues to be involved with the Implant Performance committee and the linkage with EPRD and other registries.