Part Three: Outcomes after joint replacement 2003 to 2014
Part Three of the 12th Annual Report provides outcome data in relation to hip, knee, shoulder, elbow and ankle replacements. It describes activity between 1 April 2003 and 31 December 2014.
There were 1,837,781 procedures recorded in this period, though 11% of these were excluded because there were insufficient patient details to enable linkage. This relates predominantly to the early years of the registry and less so as data quality has improved. There were 708,311 primary total hip replacements, 772,818 knee replacements, 2,554 ankle replacements, 11,399 shoulder replacements and 1,079 elbow replacements.
Hip replacement procedures
The potential follow up for hip procedures was 11.75 years. Osteoarthritis was the predominant diagnosis in 93% of cases. A total of 60% of procedures were carried out on women and the median age at primary was 69 years. In terms of fixation there has been a trend away from cemented hip replacements which was 60.4% in 2003 and in 2014 represented just 31.8%. The most common form of fixation is uncemented but this has slightly declined from 2009 to 41.2% in 2014 with a slight rise of hybrid fixation which is now 23.1%. Hip resurfacing now represents less than 1%. When hips were cemented the most common articulation was metal-on-polyethylene and in uncemented procedures ceramic-on-ceramic was also favoured. With hybrid fixation there has been an increase in the use of ceramic-on-polyethylene.
The Kaplan-Meier cumulative revision risk estimates at eleven years were lowest in the cemented fixation group at 3.63% (95% Confidence Interval 3.43-3.83) and higher in the uncemented group at 8.25% (7.90- 8.62). However, the uncemented group contained the majority of metal-on-metal articulations and when uncemented fixation was used with ceramic-on-polyethylene bearing, the eleven-year revision estimate was 3.62% (3.24-4.05). Low revision risk was generally seen with the ceramic-on-polyethylene bearing, the revision probability being 2.98% with the cemented fixation and 2.15% with the hybrid fixation. It would appear that the articulation rather than the fixation has a major influence on survivorship.
The effect of gender and age are presented by fixation and bearing and once again there is a significant increase in revision risk in younger patients. For example, in male patients less than 55, the ten-year revision risk estimate was 7.26% (5.79-9.09) with the cemented hip replacement. In comparison this was just 2.83% (2.46-3.26) in patients over 75 years.
The common stem brand combinations are reported in terms of revision risk. Their numbers are large enough to be further sub-divided into bearing type. Several brands had low revision risk at ten years and were essentially comparable, for example the cemented Exeter V40 with a Contemporary cup with ceramic-on-polyethylene bearing had a ten-year revision estimate of 2.70% (1.72-4.21). The uncemented Corail Pinnacle with ceramic-on-polyethylene bearing had a ten-year revision risk of 2.19% (1.40-3.41). The ASR resurfacing had a revision estimate of 28.28% at ten years (26.21- 30.48). It is important to note that the figures at ten years in this paragraph are approximate as at this time point fewer than 250 cases remain at risk.
Revisions for different causes after primary hip replacement identified different trends with different methods of fixation and bearing. With a cemented metal-on-polythene hip replacement, for example, the incidence of aseptic loosening increased with time whereas dislocation and infection decreased. Revisions for adverse reaction to particulate debris were more common in metal-on-metal and the incidence increased with time.
This year the risk of dissociation of liner from the acetabular component was investigated as a reason for revision in uncemented cups with liners. This was generally a rare event with no particular trends between brands and articulation combinations.
The cumulative mortality was examined up to eleven years following primary surgery and as expected increased with age so for example this was low in men under 55 years of age, 5.48% (4.95-6.05) but rose to 86.20% (83.10-88.92) in men over 85 years.
Primary hip replacement for fractured neck of femur has been investigated in more detail with a cohort of 15,786 primary hip replacements identified. The revision risk was slightly higher than the non-fracture group and the mortality up to eleven years significantly higher approaching 50%, compared to just under 30% for the comparator group.
The revision total hip replacement has been studied. There were a total of 79,859 revisions of which 17,916 were revisions of primary operations identified in the registry and the remaining 52,780 related to unrecorded primaries, (either pre-dating 2003 or the primary had not been captured in the NJR). A total of 87.2% were single-stage revisions, 5.8% were stage one of two-stage procedures and 6.9% were stage two of two-stage procedures. The ten-year re-revision risk was 15.30% (14.72-15.89) which is nearly an order of magnitude greater than in the primary group. It was interesting that the ten-year re-revision risk estimate was 22.67% when the primary was recorded in the NJR compared to 13.9% when the primary was not recorded in the NJR. This probably relates to the fact that the revisions recorded in the NJR relate to infection, dislocation and adverse metal reaction compared to the other group.
Knee replacement procedures
Of the 772,818 primary knee replacements, osteoarthritis was the sole stated indication for surgery in 96% of cases. A total of 43% of primary knee replacement surgeries were performed on men and the median age for a male patient undergoing primary surgery was 70 years. Of all primary knee replacements, 84.3% were cemented total knee replacements, the majority of which were unconstrained fixed bearing knees, 4.7% were uncemented and 1.1% were hybrid. Unicondylar knee replacements were used in 8.7% of procedures and patellofemoral replacement made up 1.3% of all procedures.
In comparison to hip replacement there has been little temporal change between 2003 and 2014 in terms of implant selection. All cemented total knee replacement has risen from 85.1% of all recorded surgeries in 2003 to 87.5% in 2014 and there has been a decline in uncemented total knee replacements from 6.7% to 2.5% over the same time period. Unicondylar replacements have formed between 8% and 9% of all primaries each year over the eleven-year period and patellofemoral replacements have continued to form 1% to 1.5% of surgeries year on year. The Kaplan-Meier cumulative revision risk estimates at eleven years were 3.62% (3.51-3.75) for cemented total knee replacement, 4.91% (4.38-5.50) for uncemented total knee replacement and 3.57% (3.06-4.16) for hybrid total knee replacement.
As reported in previous years the corresponding eleven-year revision estimate for unicondylar replacements were higher than total knee replacements at 14.29% (13.44- 15.18) and for patellofemoral replacement the revision risk was 20.22% (18.05-22.61). Revision estimates have been broken down according to level of constraint, for example the eleven-year estimate for cemented total knee replacement with an unconstrained, fixed bearing was 3.35% (3.20-3.50) and the posterior-stabilised fixed bearing was 4.02% (3.77-4.27). Further detailed breakdown in relation to fixation, bearing, constraint, gender and age show marked differences in outcomes. For example, when a cemented, unconstrained, fixed bearing total knee replacement was used in women over 75 years of age, the risk of revision at eleven years after the primary was just 1.53% (1.36-1.71). In comparison, in women aged under 55, the revision risk estimate was 6.94% (6.09-7.90).
The detailed breakdown of brands with a sub-division of fixation, bearing and constraint within brand continues to show that the ten-year revision estimates are low (less than 4% for many brands). The chance of first revision at ten years after the primary surgery is similar across the majority of implant brands used in total knee replacements. For example, at ten years the revision estimate was 3.56% (3.33-3.81) for the Nexgen knee and the AGC had a comparable revision risk estimate of 3.55% (3.29-3.82).
Within the unicondylar brand group, the Zimmer unicompartmental implant shows a low revision estimate at seven years of 5.89% (4.93-7.07) compared to the Preservation at 14.38% (12.68-16.32). The cumulative mortality at eleven years after the primary knee replacement was similar to that observed in hip replacement and for men under 55 years of age this was 6.38% (5.40-7.54) but rose to 85.01% (81.51- 88.17) in men over 85.
Outcomes of revision knee replacement surgery are also reported. There were 47,829 revision operations recorded of which 17,649 were linked to a primary record. In 2014, 78% of revisions were single-stage; 11.2% were stage one of two-stage and 10.8% stage two of two-stage. The ten-year cumulative percentage probability of re-revision for the whole group was 14.32% (13.60-15.07) and, similar to hip replacement, the re-revision risks were higher when the primary was recorded in the NJR 17.06% (15.32-18.97), compared to 12.43% (11.64-13.27) when the primary was not recorded in the NJR.
A detailed analysis of pre- and post-operative Patient Reported Outcome Measures has been undertaken of a sample of patients who had an elective primary knee replacement in 2010 and had returned pre-operative and 6-month National PROMs. These patients were sent further PROMs at one and three years after their primary. Of the 32,147 invited participants, 20,721 and 17,485 respectively responded at one and three years.
The median pre-operative Oxford Knee Score (OKS) was 19, rising to 37 at six months, 38 at one year and 39 at three years after the primary. The corresponding median EQ-5D Index values at the same points in time were 0.587, 0.760, 0.760 and 0.796.
In relation to OKS, the pre-operative distribution was symmetrical but distributions were highly left skewed in post-operative periods. The EQ-5D Index distribution was bimodal in shape prior to the primary operation while all three post-operative EQ-5D Index distributions featured distinct clusters of Index values. Breakdowns of the OKS distribution are given by age, gender, area deprivation, ethnicity, BMI, ASA grade, living arrangements and co-existing diseases prior to surgery and prosthesis fixation.
Ankle replacement procedures
A total of 2,554 primary ankle replacements have been recorded on the NJR between 1 April 2010 and 31 December 2014. These were carried out by a total of 201 consultants in 217 hospitals. A total of 61% of consultants entered ten or more procedures and the maximum number carried out by any one hospital was 186. The median age at primary surgery was 68 years and 58% of procedures were carried out in men. A total of 98% of the procedures were uncemented.
The Mobility was the most commonly used brand of replacement until 2013. This has now been overtaken by the use of the Zenith. The Mobility was withdrawn from the market in 2014. There have been 49 revisions submitted to the NJR which may represent under-reporting. The four-year cumulative revision risk was 3.28% (2.37-4.55).
Shoulder replacement procedures
A total of 11,399 primary shoulder replacements were recorded on the NJR from 1 April 2012 until 31 December 2014. These were carried out by a total of 553 surgeons in 335 hospitals. The median age of primary surgery was 73 years and 71.6% of procedures were carried out in women.
Over the last two years there has been a slight decrease in the use of resurfacing arthroplasty and an increase of the reverse polarity total shoulder replacement which in 2014 represented nearly 40% of cases.
In terms of brand the market leader is the Delta Xtend. There were 165 shoulder revisions overall and the cumulative revision estimate at 2.5 years was 2.8% (2.35-3.42). The relatively small numbers and short follow-up did not allow for detailed breakdown of causes of revision or differences between the brands. The 90-day mortality following surgery, as expected, differed by indication for acute trauma and elective, 90- day mortality following trauma was 2.2% (1.4-3.4) and for elective surgery was 0.3% (0.2-0.4).
A NOTE ON NATIONAL PATIENT REPORTED OUTCOME MEASURES (PROMs)
In 2013, the NJR reported PROMs for the first time. In 2014 and 2015, no data from the NHS England PROMs programme was available to the NJR from the NHS Health and Social Care Information Centre.