Executive summary (across all report areas)

Part Four: Implant- and unit-level activity and outcomes 2017

Part Four of the annual report gives performance and data entry quality indicators for Trusts, Local Health Boards (many of whom comprise more than one hospital) and independent (private) providers in England, Wales, Northern Ireland and the Isle of Man for the 2017 calendar year. Outcomes analysis after hip and knee replacement surgery is also provided for the period 2003 to 2017.

This section now also provides data for implant outliers since 2003 and further information on notification and last usage date.

The full analysis for units can be found in the Part Four online document which is available in the downloads section at www.njrreports.org.uk

4.1 Implant performance

The Implant Scrutiny Committee reports Level 1 outlier implants to the MHRA. Since the formation of the committee in 2009, there have been four hip stems, seven hip acetabular (cup) components and 23 hip stem/cup combinations reported. Nine knee brands have been notified as level 1 outliers.

An implant is considered to be a Level 1 outlier when its Prosthesis Time Incident Rate (PTIR) is more than twice the PTIR of the group, allowing for confidence intervals. These are shown as the number of revisions per 100 patient-years. As of March 2015, we have started to identify the best performing implants, these would have a PTIR less than half that of their group, allowing for confidence intervals. To date no implants have reached that level.

Hip implant performance

Table 1 Level 1 outlier stems/femoral components reported to MHRA.

Table 2 Level 1 outlier acetabular components reported to MHRA.

Table 3 Level 1 outlier stem/cup combinations.

Best performing hip implants

There are no hip implants or combinations performing statistically less than half their expected PTIR.

Knee implant performance

Table 4 Level 1 outlier implants reported to MHRA

Best performing knee implants

There are no knee implants performing statistically less than half their expected PTIR.

4.2 Clinical activity

Overall in 2017, 148 NHS Trusts and Local Health Boards (comprising 248 separate hospitals) and 179 independent hospitals were open and eligible to report patient procedures to the NJR. All units except for three trauma units submitted data in 2017. The proportion of all hip and knee joint replacements entered in to the NJR against those carried out (compliance) is only available by NHS Trust and Local Health Board. No data on this is currently available from private providers and figures would also exclude units in Northern Ireland as compliance data is not available.

• 59% of NHS providers reported 95% or more of the joint replacements they undertook

• 30% of NHS providers reported between 80% and 95%

• 10% of NHS providers reported less than 80%

Of those hospitals submitting data, the proportion of patients who gave permission (consent) for their details to be entered into the NJR were:

NHS hospitals

• 46% of NHS hospitals achieved a consent rate of greater than 95%

• 34% achieved a consent rate of 80% to 95%

• 20% recorded a consent rate of less than 80%

Independent hospitals

• 69% of independent hospitals achieved a consent rate greater than 95%

• 24% achieved a consent rate of 80% to 95%

• 7% recorded a consent rate of less than 80%

Similarly, the proportion of entries in which data is available to enable the patient to be linked to an NHS number (linkability) are listed below:

NHS hospitals

• 88% achieved a proportion of patients with a linkable NHS number greater than 95%

• 10% achieved a proportion of 80% to 95%

• 2% recorded a proportion of linkable records of less than 80%

Independent hospitals

• 81% achieved a proportion of patients with a linkable NHS number greater than 95%

• 14% achieved a proportion of 80% to 95%

• 5% recorded a proportion of linkable records of less than 80%

Note: Independent hospitals might be expected to have lower linkability rates than NHS hospitals, as a proportion of their patients may come from abroad and not have an NHS number. Linkability figures are not currently available for Northern Ireland.

4.3 Outlier units for 90-day mortality and revision rates for the period 2003 to 2017

The observed numbers of revisions of hip and knee replacements for each hospital were compared to the numbers expected given the unit’s case-mix in respect of age, gender and reason for primary surgery. Hospitals with a much higher than expected revision rate for hip and knee replacement have been identified. These hospitals had a revision rate that was above the upper of the 99.8% control limits (these limits approximate to +/-3 standard deviations). We would expect 0.2% (i.e. one in 500) to lie outside the control limits by chance, with approximately half of these (one in 1,000) to be above the upper limit.

When examined over the life of the registry, a total of 29 hospitals reported higher than expected rates of revision for knee replacement and 45 hospitals had higher than expected rates of revision for hip surgery.

However, revisions taken only from the last five years of the registry showed only ten hospitals reporting higher than expected rates for knees, and seven for hips.

The 90-day mortality for hip and knee replacement was calculated using the last five years of data for all hospitals by plotting standardised mortality ratios for each hospital against the expected number of deaths. No hospitals had higher than expected mortality rates for either hip or knee replacement.

Note: The case mix adjustment for mortality includes age, gender and ASA grade. Trauma cases have been excluded from both the hip and knee mortality analyses together with hips implanted for failed hemiarthroplasty or for metastatic cancer (the latter only from November 2014 when recording of this reason began). Also, where both left and right side joints were implanted on the same day, only one side was included in the analysis.

Note: Any units identified as potential outliers in Part Four have been notified. All units are provided with an Annual Clinical Report and additionally have access to an online NJR Management Feedback system.

Important note about the outlier hospitals listed

In earlier annual reports, the NJR reported outlying hospitals based on all cases submitted to the NJR since 1 April 2003. To reflect changes in hospital practices and component use, the NJR now also reports outlying hospitals based on the last five years of data (1 March 2013 to 1 March 2018 inclusive, the latter date being when the dataset was cut). This five year cut of data excludes from the analysis the majority of withdrawn outlier implants, and metal-on-metal total hip replacements, and thus better represents contemporary practice.

Note: 1 Date range 1 April 2003 to 1 March 2018 inclusive. 2 Date range 1

March 2013 to 1 March 2018 inclusive.

Outliers for Hip mortality rates since 2013

None identified

Outliers for Knee mortality rates since 2013

None identified

Outliers for Hip revision rates, all linked primaries from 2003

Ashtead Hospital (Surrey)

Basingstoke and North Hampshire Hospital

BMI Gisburne Park Hospital (Lancashire)

BMI Sarum Road Hospital (Hampshire)

BMI The Somerfield Hospital (Kent)

Cheltenham General Hospital

Clifton Park Hospital (North Yorkshire)

Conquest Hospital

Homerton University Hospital

KIMS Hospital (Kent)

Llandough Hospital

Maidstone District General Hospital

Medway Maritime Hospital

Musgrove Park Hospital

Nevill Hall Hospital

New Hall Hospital (Wiltshire)

North Tyneside General Hospital

Northampton General Hospital (Acute)

Nuffield Health Brighton Hospital (East Sussex)

Nuffield Health Cheltenham Hospital (Gloucestershire)

Nuffield Health Haywards Heath Hospital (West Sussex)

Nuffield Health Tees Hospital (County Durham)

Nuffield Health Wessex Hospital (Hampshire)

Nuffield Health York Hospital (North Yorkshire)

Pilgrim Hospital

Prince Charles Hospital

Rotherham District General Hospital

Royal Cornwall Hospital (Treliske)

Salisbury District Hospital

Shepton Mallet Treatment Centre (Somerset)

Spire Alexandra Hospital (Kent)

Spire Cardiff Hospital (Glamorgan)

Spire Dunedin Hospital (Berkshire)

Spire Tunbridge Wells Hospital (Kent)

St Albans City Hospital

St Michael's Hospital

Sussex Orthopaedic NHS Treatment Centre

The Berkshire Independent Hospital (Berkshire)

The Royal London Hospital

University Hospital (Coventry)

University Hospital of Hartlepool

University Hospital of North Tees

Wansbeck Hospital

Watford General Hospital

York Hospital

Outliers for Hip revision rates, all linked primaries from 2013

Homerton University Hospital

KIMS Hospital (Kent)

Salisbury District Hospital

Southampton General Hospital

St Richard's Hospital

Wansbeck Hospital

Weston General Hospital

Outliers for Knee revision rates, all linked primaries from 2003

Bradford Royal Infirmary

BMI Bishops Wood Hospital (Middlesex)

BMI Goring Hall Hospital (West Sussex)

BMI The London Independent Hospital (Greater London)

BMI The Meriden Hospital (West Midlands)

Broadgreen Hospital

Castle Hill Hospital

Charing Cross Hospital

Conquest Hospital

County Hospital Louth

Good Hope Hospital

Guy's Hospital

Hinchingbrooke Hospital

Homerton University Hospital

Horton NHS Treatment Centre (Oxfordshire)

Hospital of St Cross

King Edward VII Hospital Sister Agnes (Greater London)

Kings Mill Hospital

Llandough Hospital

New Hall Hospital (Wiltshire)

Southampton General Hospital

Southmead Hospital

Spire Alexandra Hospital (Kent)

Spire Clare Park Hospital (Surrey)

Spire Southampton Hospital (Hampshire)

St Albans City Hospital

St Richard's Hospital

University College Hospital

Withybush General Hospital

Outliers for Knee revision rates, all linked primaries from 2013

BMI The London Independent Hospital (Greater London)

BMI The Meriden Hospital (West Midlands)

Broadgreen Hospital

Ealing Hospital

Guy's Hospital

Heatherwood Hospital

King Edward VII Hospital Sister Agnes (Greater London)

Leighton Hospital

Nevill Hall Hospital

St Richard's Hospital

4.4 Better than expected performance

This year we have again listed hospitals where revision rates are statistically better than expected. The list below shows units that lie below the 99.8% control limit which also achieved greater than 90% compliance in the 2014/15 NJR data quality audit (NHS hospitals only).

Better than expected for Hip revision rates, all linked primaries from 2003

Alexandra Hospital

Bedford Hospital South Wing

Bishop Auckland Hospital

Chapel Allerton Hospital

Harrogate District Hospital

Hereford County Hospital

Ipswich Hospital

Kidderminster Treatment Centre

Manor Hospital

New Cross Hospital

Northern General Hospital

Prince Philip Hospital

Princess Alexandra Hospital

Queen Alexandra Hospital

Queen’s Hospital Burton Upon Trent

Queen’s Medical Centre Nottingham University Hospital

Royal Derby Hospital

Royal Devon & Exeter Hospital (Wonford)

Royal Surrey County Hospital

St Mary's Hospital

Sunderland Royal Hospital

The Great Western Hospital

The Princess Royal Hospital

Wrightington Hospital

Better than expected for Hip revision rates, all linked primaries from 2013

Ipswich Hospital

Royal Devon & Exeter Hospital (Wonford)

Royal Surrey County Hospital

Better than expected for Knee revision rates, all linked

primaries from 2003

Bishop Auckland Hospital

Bronglais General Hospital

Darlington Memorial Hospital

Ipswich Hospital

New Cross Hospital

Princess Alexandra Hospital

Queen’s Hospital Burton Upon Trent

Royal Bournemouth Hospital

Royal Derby Hospital

Russells Hall Hospital

Sandwell General Hospital

St Woolos Hospital

Stepping Hill Hospital

Whipps Cross University Hospital

Worcestershire Royal Hospital

Wrightington Hospital

Better than expected for Knee revision rates, all linked primaries from 2013

None identified

 

Part Three: Outcomes after joint replacement 2003 to 2017

Part Three of the 15th Annual Report provides outcome data in relation to hip, knee, ankle, shoulder and elbow replacements. It describes activity between 1 April 2003 and 31 December 2017.

There were 2,526,601 procedures entered in to the NJR across all joint types. After removing procedures without linkage identifiers (217,484) and those with data errors that hindered linkage (167), 2,308,950 remained. The majority of these exclusions are associated with the early years of the registry, where insufficient patient details were collected to enable linkage.

There were 992,090 primary total hip replacements, 1,087,611 knee replacements, 4,687 ankle replacements, 30,720 shoulder replacements and 2,872 elbow replacements available for analysis.

Hip replacement procedures

The total number of hip replacements recorded in the NJR continues to increase totalling just under one million replacements since data was first collected in 2003. More women (60%) have undergone surgery than men (40%). The mean age at implantation across the whole group is 68.0, but varies with fixation method with all uncemented as the most common fixation method recorded, having a mean age of 64.6, hybrid 69.0 and all cemented 72.8.                                                                                   

During 2017, 91,698 hip replacements were entered into the NJR which was a slight reduction on the number performed in 2016. This may represent delayed data entry for the period by some units, a phenomenon that has also been observed in previous years.

The most commonly performed hip replacement continues to be all uncemented , but in 2017, for the first time, the total number of hybrid hip replacements implanted was greater than cemented hip replacements.

The number of resurfacing metal-on-metal hip replacements continues to decrease, contributing only 0.6% of the total number of hip replacements performed in 2017. Metal-on-polyethylene bearings were still the most widely used but the trend for an increasing number of ceramic-on-polyethylene implants continues.                                                                                   

Within the entire primary hip dataset available for analysis of 992,090, the commonest indications for revision remain aseptic loosening, dislocation (instability), adverse reaction to particulate debris, pain, infection and periprosthetic fracture. The data now available in the NJR allows for estimation of cumulative revision rates out to 14 years. For the entire cohort the revision rate at 14 years is 7.27%.                                                       

Over the same timeframe the revision rate for all cemented constructs is 4.88%, 8.94% for all uncemented and 5.38% for hybrid. These findings are influenced by inclusion of patients with metal-on-metal bearings, which for all categories of fixation have a much higher revision rate (19.06% – 22.21%).

However, it is striking that different combinations of fixation and bearing surface can produce very low failure rates at 14 years, with cemented ceramic-on-polyethylene bearing failure rates reported as 3.77% and hybrid ceramic-on- ceramic 3.52% (with limited data). The most common construct recorded in the NJR is cemented metal-on- polyethylene, which records a reassuringly low failure rate of 4.87% at 14 years.                                        

Metal-on-metal resurfacings have a revision rate at 14 years of 14.76% and their use, as with all metal-on- metal bearings, is now very limited.

Some of the most important data presented in this year’s hip report relate to patient age at implantation and its relationship to the outcome of different combinations of fixation and bearing type.

This year’s analysis confirms that across the whole cohort revision rates increase as age of implantation decreases, with the effect more pronounced in women.

In men under the age of 55, hybrid ceramic- on-polyethylene and ceramic-on-ceramic hip replacements have revision rates of under 4% at 12 years, less than half the 9.98% seen with cemented metal-on-polyethylene constructs although, again, limited data suggests some caution should be employed in interpreting this finding. In women under 55 years, hybrid ceramic-on-ceramic constructs give the lowest revision rate of 4.11% at 12 years and cemented ceramic–on-polyethylene gives good results with failure rates of nearly 4.93% at the same time point. However changing the bearing type in all cemented fixation to metal- on-polyethylene increases failure rate to 8.16% in women.

These data represent important results and can help decision making when selecting constructs for the younger patient. However, it is important to recognize the possibility of selection bias with existing constructs that may confound this relationship. 

For patients over the age of 75, the situation is less complex as all combinations, other than metal-on- metal bearings, have similarly reliable outcomes. The lowest failure rates are seen with cemented or hybrid ceramic-on-polyethylene constructs.

In addition, combining revision and mortality has identified that for the majority of patients undergoing hip replacement over the age of 75, the replacement will last them for the rest of their lives.

The analysis of head size (bearing diameter) provides a mixed picture with respect to performance. The largest head sizes in each class of cemented metal-on- polyethylene, of uncemented metal-on-polyethylene, and cemented ceramic-on-polyethylene tend to perform poorly. However, the absolute effect of head size is less consistent e.g. a 36mm cemented metal- on-polyethylene head is the worst in class whereas a 36mm uncemented metal-on-polyethylene head has comparative performance to the most widely used bearing size of 28mm. Conversely, a 28mm ceramic- on-ceramic bearing tends to have worse performance than larger bearings therefore, the choice of bearing size needs to be carefully considered.                               

The number of hip replacements recorded in the NJR which are performed in patients with fractured neck of femur continues to increase, with 4,445 entered in 2017. The yearly revision rate of this cohort remains very similar to those performed for osteoarthritis after the first three months following surgery, with only minor elevation in absolute risk resulting from an increased risk of failure in the first three months after surgery, despite higher mortality rates.                                                                                                   

To study re-revision of hip replacements, 92,348 records of first revision were available for analysis leading to 8,528 subsequent re-revisions. The risk of revision was strongly related to the time to first revision surgery and 11.66% of hips revised within a year of the primary procedure are re-revised within three years.                                                                                   

Knee replacement procedures

Within the whole registry there were 1,087,611 verifiable primary knee replacements recorded between 1 April 2003 and 31 December 2017, with maximum follow-up of 14.75 years. Across the whole registry the overwhelming majority (97%) of procedures have been performed for osteoarthritis, with more women undergoing surgery then men and a mean age at implantation of 68.9 (SD 9.6).

In 2017, 102,177 primary knee procedures were entered into the NJR, a marginal decrease on the number performed in 2016. There may be delayed data entry for the period by some units, a phenomenon that has also been observed in previous years. Within this group, 89% were total knee replacements, 10% unicondylar procedures and 1% patellofemoral.

Since 2003, cemented TKR has remained the most widely used knee replacement construct in the UK, representing 85% of all primary prostheses recorded in the registry.

Over time the total numbers of uncemented or hybrid total knee implants used has declined, reaching a current level of 2.2% of all primary knee replacements.
 
In 2017 of all knee replacements, 63% were cemented unconstrained prostheses and 23% cemented posterior stabilised, with mobile bearing prosthesis less commonly used.

With a maximum follow-up of nearly 15 years the registry now provides an insight into knee implant survival into the second decade. The Kaplan-Meier estimate of revision rate at 14 years for all cemented TKRs is 4.47%. The revision rate for unconstrained fixed bearing devices (4.07%) was slightly lower than that recorded for posterior stabilised components (5.39%). The lowest revision rates for cemented TKR are seen when monobloc tibial components are used, although again limited data suggests some caution should be employed in interpreting this finding.                                             

Over the last five years there has been an increase in the number of tibio-femoral unicondylar prostheses entered into the registry, reaching 10% for the first time in 2017. Over the last three years, 809 surgeons in 358 units now perform this operation. However revision rates for unicompartmental knee replacement remain higher when compared to TKR reaching 16.9% at 14 years.                                                                                                  

The use of patellofemoral prostheses represented 1% of all primary knee replacements recorded in 2017. This type of knee replacement records the highest revision rate at 14 years, over 24.4%, which may explain the trend for reduced usage over the last five years.

Age continues to have a significant effect on outcome after primary knee replacement, where younger patients experience a higher revision rate. This important observation is most vividly seen in patients under the age of 55 and in those undergoing unicompartmental knee replacement.

However it is reassuring to note that revision is still relatively uncommon and for patients around the average age of undergoing surgery, their knee replacement is likely to last the rest of their life without the need for revision.

Over the last four years, the total number of new revision operations recorded in the registry has remained fairly static, with 6,289 entered into the registry for 2017. Over 80% of these cases were single stage procedures, with the most common indications for surgery recorded as aseptic loosening, other indication, instability and pain.                                                                    
The collection of data over a 14 year time frame allows the results of re-revision to be recorded. The 2017 NJR analysis demonstrates that the risk of re-revision following a first revision is higher than the risk of revision after a primary procedure. In addition the rate of re-revision after unicompartmental knee replacement is broadly similar to re-revision after total knee replacement.

Ankle replacement procedures

A total of 4,687 primary ankle replacements have been recorded on the NJR between 1 April 2010 and 31 December 2017.                                              

Within each year of data collection there has been an increase in the number of primary procedures performed and this trend continued with 734 recorded in 2017. They were performed by 135 consultants working in 139 units with a median number of 3 performed per surgeon (IQR 1-8) and per unit (IQR 1-6). Over the eight years data has been collected, a total of 69% of consultant surgeons and 74% of units have submitted less than 20 procedures in total.

The median age at primary surgery remains at 68 (IQR 61-74) years and 59% of procedures were carried out in men.

Uncemented fixation has been used in the majority of implantations recorded (98.2%). Infinity has rapidly become the most common prosthesis used in ankle replacement since its introduction in 2014.                                          

The 7-year cumulative revision rate for the entire cohort is 8.70%. However, in younger patients under 65 years, the revision rate is greater than 10%, although again, limited data suggests some caution should be employed in interpreting this finding.

Shoulder  replacement procedures

Primary shoulder replacements have been recorded in the NJR from 1 April 2012 and up to 31 December 2017 a total of 30,720 were available for analysis.

Since 2012 the total number of primary procedures performed each year has increased and the same trend was seen in 2017 with 6,526 implantations recorded in the NJR.

The number of units performing shoulder replacements in 2017 was 354, a slight increase from 2016, with 14 (IQR 5-26) as the median number of procedures per unit. Over the last three years the number of consultants carrying out this procedure has remained just over 480, with a median number per consultant of 10 (IQR 4-19) in 2017.

Within the entire cohort, elective procedures account for 27,990 cases, with osteoarthritis (17,297) and cuff tear arthropathy (7,822) the commonest indications for surgery.

The median age of patients undergoing elective shoulder replacement was 73 (IQR 67-79), with 70% of all cases performed in women.

Primary shoulder replacement was performed for trauma in 2,730 cases, with a median age of 74 (IQR 67-80) and again predominantly in women (77%).

Looking at the entire cohort, reverse polarity total shoulder arthroplasty has been the most widely used prosthesis for elective indications, followed by total shoulder arthroplasty and humeral hemiarthroplasty. For use in trauma, reverse polarity total shoulder arthroplasty is also the most commonly recorded implant, with humeral hemiarthroplasty also
widely used.

Within the entire cohort, there have been 794 revisions of elective procedures and 66 revisions of trauma procedures.

For both elective and trauma surgery instability, cuff insufficiency and conversion of hemiarthroplasty to total shoulder arthroplasty are the most frequently recorded indications for revision.

The cumulative revision rate at five years for all elective procedures is 5.0%, and 3.7% for acute trauma cases. In elective surgery, total shoulder arthroplasty and reverse polarity total shoulder arthroplasty have 5-year revision rates of 3.3% and 3.5% respectively, compared to 7.4% seen with humeral hemiarthroplasty.                                                                          

Further analysis of the Oxford Shoulder Scores collected pre-surgery (Q1) and six months post- surgery (Q2) has focused on the 275 (8%) of patients who were identified in the 13th NJR Annual Report (2016) as having worse scores after surgery. This poor outcome is seen across all implant types and by the end of 2017 a total of 33 (12%) had been revised.

Elbow replacement procedures

Data on primary elbow procedures has been collected since 2012 and by 31 December 2017 a total of 2,872 procedures had been recorded in the NJR. The majority of cases entered are total prosthetic replacements (77.6%), isolated radial head replacement (17.0%) or humeral hemiarthroplasty (3.2%).

Of the 2,872 elbow replacements within the NJR, 1,938 have been used for elective care. In this group, total prosthetic replacement has been used in 1,772 (91.4%) cases, treating mainly inflammatory arthritis (50.6%), osteoarthritis (34.2%) or the sequelae of trauma (14.6%). In 934 cases elbow replacement has been used to treat acute trauma, with total prosthetic replacement (48.9%), radial head replacement (40.9%) or humeral hemiarthroplasty (8.6%) the three most commonly used procedure types.

In 2017 there were 612 primary procedures recorded, an increase of approximately 10% on the number in 2016. Total prosthetic replacement (70.3%), isolated radial head replacement (24.8%) and humeral hemiarthroplasty (4.1%) remain the most commonly used procedure types.

These procedures were performed in 161 different units by 222 consultant surgeons, with a median of 2 (IQR 1-5) procedures per unit and a median of 2 (IQR 1-3) per consultant.

The cumulative probability of revision of all primary elbow replacements at four years is 4.7%, with 1.7% recorded for trauma cases and 5.78% for all combined elective cases. The main indications for revision were infection and aseptic loosening.

Part Two - Clinial Activity

Part Two of the NJR’s 15th Annual Report can now to be found online via the registry’s dedicated NJR Reports website at: www.njrreports.org.uk.

Part Two presents data on clinical activity during the 2017 calendar year. This includes information on the volumes and surgical techniques in relation to procedures submitted to the NJR, with the most recent data being for the period 1 January 2017 to 31 December 2017. To be included in the report all procedures must have been entered into the NJR by 28 February 2018.

The information in Part Two now includes historical data, going back to 2005 in most cases. Using the dedicated website, readers are able to use interactive, filterable graphs to identify the key information and trends associated with the following reports for hip, knee, ankle, elbow and shoulder data (where sufficient data are available):

• Total number of hospitals and treatment centres in England, Wales and Northern Ireland able to participate in the NJR and the proportion actually participating

• Number of participating hospitals and the number and type of procedures performed

• Number of procedures undertaken as a proportion of all procedures submitted annually

• Procedure details by type of provider

• Primary procedure details by type of provider

• Types of primary replacements undertaken

• Patient characteristics for primary replacement procedures, according to procedure type

• Age and gender for primary replacement patients

• Patients’ physical status classification (ASA grades) for primary replacement procedures

• Body Mass Index (BMI) for primary replacement patients

• Indications for primary procedure based on age groups

• Age of patients undergoing primary joint replacement

• Surgical technique for primary replacement patients

• Thromboprophylaxis regime for primary replacement patients, prescribed at time of operation

• Reported untoward intra-operative events for primary replacement patients, according to procedure type

• Patient characteristics for revision procedures, according to procedure type

• Indication for surgery for revision procedures

• Trends in use of the most commonly used brands

For hips specifically

• Components removed during hip revision procedures

• Components used during single-stage hip revision procedures

• Trends in femoral head size and hip articulation

For knees specifically

• Implant constraint for primary procedures

• Bearing type for primary procedures

1.1 Annual Report Introduction

The 15th Annual Report of the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man (NJR) is the formal public report for the period 1 April 2017 to 31 March 2018. The report consists of a number of parts which are outlined in the summary table opposite.

As part of the continued approach to sharing information about NJR progress, clinical activity and hospital and implant activity, the NJR has updated the data on its dedicated annual report website, ‘NJR Reports’, to showcase annual report data and information.

Some of these data can also be found in this printed report – in particular, the summaries and the full detailed, statistical analysis of outcomes following joint replacement surgery.

A short summary of the NJR’s progress over 2017/18 is included below and in both the Chairman’s Foreword and Annual Report Executive Summary.

Additional information and reports are available online via ‘NJR Reports’ at www.njrreports.org.uk

Progress

As at 31st March 2018, the total number of procedures submitted to the NJR was approximately 2.52 million. In the financial year 2017/18, a total of 252,251 records were submitted, which is an increase of 9,632 over the previous year. This is despite concern that the overall number of joint replacement procedures being undertaken was decreasing. Overall key performance indicators demonstrated:

• Patient consent (to allow the recording of their personal details in the NJR) was recorded as 92.4%, an increase of 0.3% from the previous year. However, the consent rate for Northern Ireland decreased to 95.7% from 96.3% in the previous year, while the overall consent rate for England, Wales and the Isle of Man increased by 0.2%.

• Linkability (the ability to link a patient’s primary procedure to a revision procedure) was recorded as 94.0%, a drop of 1.4% on the previous year.

Whilst a comparison of successive years will show variation, the drop in the rates of the key indicators of consent and linkability may be attributable to the outcomes of the data quality audits that have taken place this year. This has resulted in the retrospective submission of missing procedures for which some will not have had patient consent recorded. Linkability is dependent on the submission rate of NHS and, in Northern Ireland, HCN numbers. Please see the data completeness and quality indicators section online for further detail.

The primary focus for the NJR in 2017/18 continued to be data quality, with the third year of our data quality audit across NHS and independent sector units. Following the submission of the missing entries for 2014/15 identified by the audit, the registry now has a patient entry rate of more than 99% for the year. Timely submission of revision procedures to the NJR continues to be a concern, although engagement with units has shown an improvement in retrospective entry. Please visit www.njrreports.org.uk for further details of the audit.

Significant changes have also been made to the NJR Minimum Data Set with the new update released in Summer 2018. These changes reflect advancements in clinical practice and technologies with the addition of further procedure types and surgical approaches. Shoulder data collection in particular will be enhanced by new procedure categories and an increased number of indications for surgery.

Further progress and updates will be available at www.njrreports.org.uk and also via the main NJR website at www.njrcentre.org.uk.

Executive summary

Mr Martyn Porter, NJR Medical Director and Chairman, Editorial Board 

On 31 March 2018, the NJR celebrated 15 years of collecting data when 2,520,000 procedures had been registered. Quite an achievement!

Main headlines and achievements for 2017/18

Data quality has been a major focus for the NJR over the last three years and continues to be so. Only by providing and analysing high quality data that is complete and accurate will the NJR maintain credibility with its various stakeholders.

The data quality audit has now expanded to include the independent sector. Compliance with the audit has been excellent and plans are underway to incorporate a significant amount of automation in to this process to reduce the burden on the hospitals.

The audit itself identifies cases that have not been registered with the NJR but were recorded on hospital systems, further improving compliance with the NJR. Precise calculations of compliance are complex but at the conclusion of the 14/15 audit, approximately 95% of all primary hip and knee operations had been registered and 90% of all revisions.

Within the year, the NJR has worked hard to develop its Accountability and Transparency Model. Twice a year, the NJR carries out a structured analysis of revision and post-operative mortality to identify unexpected variations of both individual surgeons and hospitals. The NJR is now working with the BOA, CQC and NHS Improvement to ensure that the appropriate governance actions are carried out in a timely way. The NJR’s symposium at the 2018 BOA Congress entitled “How to use NJR data responsibly to improve care”, includes the use of data to potentially improve outcomes as well as identifying causes for concern.

This is further supported by enhancements to the consultant level report available from the clinician feedback portal. By providing more detail and early warnings to surgeons regarding potential adverse performance, hopefully any issues can be addressed before they become problematic. The ethos of the registry is to enable surgeons to practice safely rather than to alert them to matters that may be at variance with expected outcomes.

The minimum dataset has also been updated this year. One of the changes will now allow for the registration of Debridement and Implant Retention (DAIR) procedures used in treating infected joint replacements in the early periods following surgery as a separate procedure type, distinct from single stage revision under which they were previously recorded.. To counter the argument that this may encourage surgeons not to carry out exchange of the modular components, simple debridement and washout will also be recorded. These changes will allow more accurate and detailed assessment of these procedures in relation to later revision surgery that may be required for deep infection.

The NJR component database is in the process of being redesigned to allow for more detailed analysis of the types of components that are implanted. This should lead to a better understanding of how the attributes of devices may affect their performance.

The NJR continues to work closely and has strengthened its relationships with GIRFT, ODEP and Beyond Compliance. The NJR also collaborates with other international joint replacement registries and is part of ISAR - the International Society for Arthroplasty Registers. The 2018 symposium was held in Reykjavik, Iceland in May 2018. The NJR was also represented at the EFORT meeting (European Federation of Orthopaedics and Traumatology) in Barcelona in June 2018. The focus of this international work has been on collaboration, sharing registry research findings, developing new methodologies, discussing the utility of PROMs (Patient Reported Outcome Measures), early signal detection and benchmarking of devices.

Main headlines for 2017: Outcomes

The data builds on last year’s findings and the main message of the report is that revision estimates following joint replacement surgery remain low.

The data is structured to show the effect of patient and implant factors on revision estimates. For example, patient factors include gender and age at time of surgery, while implant factors include type of fixation, brand, bearing and head size. The debate has moved on from a comparison of fixation method to one of comparison of joint replacement ‘constructs’. In a hip replacement, for example, a construct would include a description of the brand of socket, femoral stem and bearing material. It is the ‘whole replacement’ that is important to the patient rather than the individual components or method of fixation.

Some hip constructs have revision estimates as low as 2% at 14 years and some knee brands have revision estimates as low as 3% at 14 years. These results are quite remarkable in terms of outcomes for patients and value for the taxpayer.

Although there is variation in revision estimates for implants, the patient continues to be a major factor in outcomes. In particular, it is the young patient who has a much higher revision risk compared to more elderly patients. For example, for hip replacement in male patients under the age of 55 years at the time of surgery, overall, the revision estimate is nearly 16% at 14 years compared to just 2.5% in male patients over 75 years. A similar pattern is evident in knee replacement with revision estimates of nearly 15% at 14 years in female patients less than 55 years at the time of surgery, compared to just 2.5% in female patients over 75 years.

This does not mean that young patients should be denied joint replacement surgery just because they are young and have a high risk of revision, but it does mean that both surgeons and patients should be aware of these facts in order to facilitate ‘shared decision-making’. In this regard the development of an NJR ‘risk calculator’ by Mark Wilkinson and his research team in Sheffield may prove to be timely. Risk calculators are not a new concept but the NJR model is nearing the end of its development and should be available for clinical testing early in 2019. The aim is to be able to enter individual patient information such as age, gender, BMI etc. and for the tool to predict the likelihood for revision, mortality or patient reported outcome measures (PROMs) in a ‘patient like them’. The science and methodology behind this is complex and the current indications are that the predictive power is at the best modest but certainly an improvement on the current general advice on risk given to patients. It is envisaged that these tools will be used in the outpatient setting by the surgeon and patient to facilitate ‘shared decision-making’.

In relation to knee replacement surgery, revision rates are higher for partial knee replacement, compared to total knee replacement. However, other data demonstrates that partial replacements have lower mortality rates and fewer complications than total knee replacement. The effect of surgeon volume is also important. The NJR’s role is not to discourage partial knee replacement surgery, but it would be advisable to consider the implications of these other factors and perhaps limit surgical activity to surgeons who have sufficient volume and can achieve low revision rates.

Much work has been carried out on the classification of components used in shoulder replacement surgery. This is not a straightforward matter as the plethora of implants available and the modes in which they combine are complex. The section on shoulder replacement is very impressive thanks to this classification work. Revision estimates for all elective shoulder replacements in male patients over 75 years are approximately double those for female patients in the same age group.

There are over 700 ankle replacement procedures carried out per year and revision estimates are just under 9% at seven years for all cases, but as for hips and knees, there is considerable variation according to the gender and age of the patient.

Concluding acknowledgements

As mentioned earlier there is considerable additional information available online and I would encourage you to explore the NJR’s dedicated annual report website at www.njrreports.org.uk. The website offers a helpful interactive platform for Part Two of the report, which is the descriptive NJR data; supporting appendices; and, when published, the latest NJR Patient and Public Guides to the annual report.

The NJR continues to work with many stakeholders; the most important, of course, are the patients whom I would like thank for allowing the NJR to use their data.

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:
Data Quality Editorial Board Executive
Implant Performance Implant Scrutiny Medical Advisory
Regional Clinical Coordinators Research
Surgical Performance
Members of Data Access Review Group Members of the NJR Patient Network
Other organisations: MHRA
CQC
NHS England NHS Improvement
British Orthopaedic Association BHS
BASK BESS BOFAS HQIP
Northgate Public Services (UK) Ltd University of Bristol
University of Oxford

On a personal note I would particularly like to thank Laurel Powers-Freeling, Chairman of the NJR and Elaine Young, NJR Director of Operations.
Northgate Public Services and University of Bristol teams have, yet again, done an exceptional job.

Finally, having served on the NJR Steering Committee for over 15 years I will be standing down from the NJR later this year. I have very much enjoyed and learnt from the experience. 

Thank you,

Martyn Porter