Part Three: Outcomes after joint replacement 2003 to 2016
Part Three of the 14th 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 2016.
There were 2,284,416 procedures recorded in this period, although 10% of these were excluded because there were insufficient patient details to enable linkage. This relates predominantly to the early years of the registry and was less of a feature in recent years as data quality has improved.
The numbers of primary procedures available for analysis were 890,681 total hip replacements, 975,739 knee replacements, 3,899 ankle replacements, 23,608 shoulder replacements and 2,196 elbow replacements.
Hip replacement procedures
The total number of primary hip replacements performed continues to increase with 87,733 performed in 2016, compared to 86,496 the previous year. The vast majority continue to be performed for osteoarthritis. In 2016, the ratio of women to men receiving hip replacement was 60:40 and the median age at which primary surgery was performed is 69.
Uncemented fixation is still the most common construct used by surgeons comprising 39% of the total number, compared to 30% for cemented replacements. There has been a slight decrease in both these construct types whilst hybrid fixation, mainly using a cemented stem and uncemented cup, continues to grow in popularity, with surgeons using this method in 28% of cases. Metal-on-polyethylene is still the most commonly used bearing construct across cemented, uncemented and hybrid hip replacements, but the usage of ceramic-on- polyethylene bearings continues to grow, reaching 29% of all cases. Metal-on-metal bearings including resurfacing is performed in very low numbers making up less than 1% of all cases in 2016.
In this year’s report, a total of 890,681 recorded hip replacements were available for survival analysis, with data collected over 13 years. The cumulative percentage probability of revision after primary hip replacement across all patients is 6.8% at 13 years. The lowest rate of revision continues to be seen in the all cemented construct group, with a cumulative percentage probability of revision of 4.3% at 13 years, with best results within group seen when a ceramic-on-polyethylene bearing is used (3.8%). The survival within the entire hybrid group is calculated at 5.1%, with ceramic-on-ceramic bearings providing best results of any sub-group at 3.3% at 13 years. Reassuringly the most commonly used cemented and hybrid constructs by brand all perform well.
For the uncemented construct group the pattern of failure over time is different. The revision rate is approximately double that of all cemented, calculated at 8.7%. Within group, the best survival figures are seen with a ceramic-on-polyethylene bearing, with survival rate improving to 4.5%. If metal-on-metal bearings are excluded, the commonly used constructs by brand perform similarly well.
In this year’s analysis the effect of age and gender on revision rates across the construct groups has been presented for the first time. This is particularly relevant given the increase in total numbers of younger patients undergoing joint replacement. Overall, as reported in previous annual reports, the revision rate for total hip replacement increases at a faster rate over time for younger patients. For female patients under 55 the revision rate of 13.5% at 13 years is 2.5 times greater than for women undergoing surgery between 65 and 74 years of age. However the choice of construct does affect revision rate in the younger age group and for women under 55 years a cemented ceramic-on-polyethylene construct gives the best results, with a revision rate of 3.8% at ten years. A similar trend in the relationship of age to revision rates is seen for men, although at 13 years the revision rate for the under 55 group across all bearing types is 10%, approximately 3.5% lower than for women. Again, the best performing construct for the younger patient is a cemented prosthesis, in this case using a ceramic-on-polyethylene bearing, which provides half the revision rate of cementless fixation using metal-on- polyethylene at all time points. Interestingly, for older patients all construct combinations have similarly good revision outcomes.
Presenting mortality data alongside revision outcomes provides a greater understanding of the outcome of hip replacement, particularly in the older patient. In the vast majority of patients over the age of 75 at implantation, their hip implant will remain unrevised across their remaining lifetime, with very low revision rates seen.
Our analysis confirms that choice of head size is an important factor in determining revision outcome. For both metal-on-polyethylene and ceramic-on- polyethylene bearing choices higher failure rates are seen with larger head sizes, in particular 36mm for cemented and above 36mm for hybrid and cementless. In contrast if a ceramic-on-ceramic bearing construct is used then survival is improved with larger size.
Metal-on-metal reconstructions, of either a resurfacing or stemmed variety, continue to fail at higher rates than other bearing choices, with revision rates ranging between 14% and 27% at ten years for the worst performing implant types. However the survival profile for the best performing resurfacing procedures by brand shows lower revisions rates of between 8% and 9%. Overall the net effect of higher revision rates for metal-on-metal procedures has been a dramatic and sustained reduction in their use.
The number of patients who are treated with primary hip replacement after sustaining a fractured neck of femur continues to grow with time. In 2016, 4,260 were performed, representing 4.9% of all total hip procedures. In this group of patients it is encouraging that revision rates are similar to those hip replacements performed for other indications although, as expected, mortality rates are higher.
In 2016, 7,933 revision procedures were performed, with the vast majority being single-stage procedures. The total number of revision procedures available for analysis between 2003 and 2016 is now 97,341. The most commonly recorded indication for revision continues to be aseptic loosening, followed by pain. Within the first year following primary surgery dislocation, fracture and infection are the most common indications for revision, whereas revision for aseptic loosening increases in frequency over the first ten years. The cumulative probability of hip re-revision is approximately 17% at 13 years.
Knee replacement procedures
Between 2003 and 2016 a total of 975,739 knee joint replacements were recorded and are available for analysis. Osteoarthritis remains the most common indication for knee replacement across the whole cohort (96%), with the second most common indication being inflammatory arthritis at 2%.
During 2016, 104,079 knee joint replacements were recorded in the NJR, with 98,147 primary and 5,932 revision procedures. Within primaries, the most common type of reconstruction performed was a total knee replacement, making up 89.7% of procedures. Of this group, the most widely used fixation method remains cementing (84.9%). Uncemented total knee replacement continues to decline in numbers, making up only 2% of the total number implanted. As seen in other years, within the cemented group of total knee replacements fixed bearing unconstrained (62.2%) and posterior-stabilised (19.8%) make up the vast majority of implantations performed. The proportion of unconstrained to posterior-stabilised has remained steady over the last five years, at a ratio of 3:1. Unicompartmental knee replacement (medial and lateral) makes up 9.2% of all knee replacements performed in 2016, with this percentage remaining fairly static over the last ten years. A mobile bearing construct is used in 5.1% of cases and fixed in 4.1%. Patellofemoral replacement account for 1.1% of all knee replacements and similarly this figure has remained static for the last ten years.
Patient demographics showing the trend for more women than men to undergo knee replacement continues in all types of knee replacement. The median age at which patients undergo replacement is 70 years for total knee replacement, 64 for unicondylar knee replacement and 58 for patellofemoral replacement. Over the last three years, 1,999 surgeons have undertaken total knee replacements and 820 performed unicondylar knee replacements. The median number of each performed over a three year period is 104 (IQR 26-214) for total knee replacements and 12 (IQR 3-35) for unicondylar replacements. This highlights the continuing trend for some surgeons to perform very few numbers of unicondylar replacements per year.
Survival analysis performed on the 975,739 knee replacements in the NJR was completed out to 13 years. Temporal changes over time show that the rate of change of cumulative percentage chance of revision has remained similar over the period between 2003 and 2013.
The cumulative risk of revision at 13 years for cemented total knee replacement is 4.2%, with unconstrained fixed bearing total knee replacement (the most common construct) recording 3.8% and posterior-stabilised total knee replacement 4.7%. In the cementless class, the figure reached 5.4% but interestingly for uncemented fixed bearing posterior- stabilised total knee replacements the revision rate reaches 12.1% by 13 years, demonstrating that this combination of implant choice puts patients at a greater risk of revision.
Unicondylar replacement revision rates are higher than those for total knee replacements across all times points, with a rate of 16% reached by 13 years post- surgery. The trend is the same regardless of mobile or fixed bearing choice. Patellofemoral joint replacement continues to record the highest failure rate, with the current estimate being 24.2%.
For younger patients, the risk of revision is higher with the same pattern seen for men and women. For a patient at the median age of implantation (69), the 13-year risk of revision is just over 4%. However, for total knee replacement patients under the age of 60, the risk increases with decreasing age, reaching 10% for those under 55 years old. This pattern is magnified in unicondylar replacement, with patients under the age of 55 facing a 25% chance of revision by 13 years. This has been a consistent finding across all annual reports.
In 2016, 5,932 revision knee replacements were performed, with the vast majority being single stage revisions. The total number of revisions across all years was 60,680. The most common indications recorded for first revision surgery in total knee replacement remain aseptic loosening, pain, infection and ‘other’ (excluding dislocation, lysis, periprosthetic fracture, implant fracture, instability and malalignment). Indications for first revision surgery in unicompartmental knee replacement follow a broadly similar pattern, with aseptic loosening and pain remaining as the most common specific reason, although rates are higher. Considering all knee replacements within the first year of implantation, infection remains the most common cause of revision, with aseptic loosening becoming more common in later years. The risk of subsequent re-revision is approximately 16% at 13 years across this entire group.
Ankle replacement procedures
In 2016, there were 690 primary ankle replacements entered into the NJR, compared to 602 the year before. Similar data has been collected from 2010 to 2016 and in total 3,899 primary ankle replacements are available for analysis.
From the entire series the ratio of female to male patients was 60:40 and the median age at primary surgery was 68, with a range of 17 to 92 years.
Of the 3,899 primary procedures, the vast majority (89%) of implantations have been uncemented and, with the exception of three recorded hybrid cases, the remaining are cemented.
A total of 229 consultants, working in 269 units carried out these procedures, with 44% of surgeons performing over ten procedures and 56% less than ten, over the six year period.
Between 2010 and 2016, there were 153 revision procedures, including 24 conversions to arthrodesis. The estimated rate of revision at six years was 7.7% (95% CI 5.94-8.47).
In 2016, the Infinity (30%), Box (18%) and Zenith (15%) were the most widely used brands, making up over half of all implantations.
Shoulder replacement procedures
There are now 23,608 primary shoulder replacements in the NJR with 5,944 procedures performed in 2016, with the number performed each year continuing to increase. In 2016 these procedures were performed in 338 units, with 12 as the median number per unit (IQR 5-23). The total number of consultants performing the procedures was 476, with a median per consultant of 9 (IQR 4-18).
A total of 21,570 cases were performed as part of elective care. The most common indications for surgery were osteoarthritis and cuff tear arthropathy, sometimes combined in a small proportion of patients (522). For elective cases the majority of the replacements were performed on women (70%) and the median age at the primary operation was 73 years (IQR 67-79 years), with an overall range of 17-99 years.
In 2,038 cases the indication was acute trauma. In this group of patients, 77% were female and 23% male, with a combined median age at surgery of 74 (IQR 67-80 years).
The most frequently used implant type is the reverse polarity total shoulder arthroplasty (42%), followed by total conventional shoulder arthroplasty (30%) and hemi-arthroplasty (13%). The vast majority of these primary cases were stemmed. Resurfacing humeral hemi-arthroplasty or resurfacing total shoulder arthroplasty was performed in 13% of cases.
The cumulative percentage probability of revision at four years for elective primary cases was 4.2% and 3.9% for trauma cases. In elective cases, the rate increases for patients under the age of 65 to 7.6% in men and 6.4% in women. After four years, total conventional shoulder arthroplasty and reverse shoulder arthroplasty had the lowest revision rate for elective shoulder replacement, although caution in interpretation is required as the differences seen may reflect threshold for revision and do not take account of functional outcome. In both elective and trauma procedures, the most common causes of revision were instability and cuff insufficiency.
Elbow replacement procedures
A total of 2,196 primary elbow replacement procedures have been recorded in the NJR between April 2012 and December 2016, including total, radial head and lateral resurfacing replacements. In 2016, a total of 513 procedures were performed, which is a slight decrease from the year before, although the general trend since 2012 has been an increase in overall numbers recorded.
From the entire series of 2,196 procedures, women (70%) undergo elbow replacement more often than men (30%) and the median age of patients undergoing surgery was 68 (IQR 58-77 years). Trauma accounted for 31% of all cases. In the 1,511 elective cases, the most common stated indications for elective surgery were other inflammatory arthropathy, osteoarthritis and sequelae of trauma.
Total prosthetic (63%) and radial head replacement (37%) were the prostheses used in trauma cases. In elective care the vast majority of cases were total replacement (95%), with radial head replacement performed in 4% of cases and lateral resurfacing in 1%.
In 2016, there were 210 consultants working in 160 units. They undertook primary elbow replacements with 2 (IQR 1-4) as the median number of cases performed per unit and 2 (IQR 1-3) the median per consultant.
At three years, the cumulative percentage probability of revision, across the entire group, was 4.4% (95% CI 3.3-5.8). In trauma cases the probability of revision was 2.1% (95% CI 1.1-4.2), but no radial head replacements were revised and the revision rate for total replacement was 3.2% (95% CI 1.6-6.3). This contrasts to a three- year revision rate of 5.1% (95% CI 3.7-7.0) when total replacement was performed in the elective setting. The most frequently cited causes of revision in elective care were infection and aseptic loosening.