A systematic review of the complications of contemporary total elbow arthroplasty

Abstract

Background

Total elbow arthroplasty is a low volume procedure. We aimed to evaluate complication rates and cumulative percentages associated with the most frequently used contemporary implants and for the commonest indications.

Methods

A systematic literature search of all studies reporting complications following total elbow arthroplasty with 12-month minimum follow-up was undertaken. Quality of studies was assessed with the Methodological-Index-for-Non-Randomised-studies criteria. British NJR data identified the most common UK prostheses and indications. The complication rates for all undesirable events contributing to patient outcome were extracted and cumulative percentages were calculated.

Results

One hundred seventeen studies were screened, 12 studies included, totalling 815 procedures. Mean follow-up was 3.8 years. The overall complication cumulative percentage was 60.7%, significantly higher than that seen in other joint arthroplasty, including a 6.5% deep infection rate. Nerve injury was comparable between implants at around 4.1%. Radiographic loosening had a cumulative rate of 17.2%. Revision for symptomatic aseptic loosening was 6.3%.

Conclusions

This is the largest systematic review of the complications of total elbow arthroplasty. Surgeons should be aware of differing complications related to their implant of choice, each having its own specific complication. Trauma as an indication appears to have an increased complication rates compared to inflammatory arthropathy. There is a lack of literature regarding the independent results of osteoarthritis as a specific indication for total elbow arthroplasty.

Background

Total elbow arthroplasty (TEA) is a low volume procedure with only 430 individual cases recorded in England, Wales and Northern Ireland in 2017. ¹ This is compared to nearly 7000 shoulder replacements, over 90,000 hip replacements and over 100,000 knee replacements performed during the same period. ¹

Historically, TEA has been performed primarily for inflammatory arthropathy and post-traumatic arthritis. The outcomes for both of these indications have been good, especially so for rheumatoid arthritis and it is because of this that TEA has been used for an increasing variety of indications including osteoarthritis, haemophilic arthropathy, tumour resection and fracture non-union.^2,3 TEA has also been utilised to provide a more predictable outcome following acute intra-articular fractures around the elbow, with promising results reported.^4,5

Despite the expansion in its perceived indications for use, it is estimated that each surgeon in the UK performs between two and three total elbow replacements a year. ¹ This is notable, given the current drive to limit the number of surgeons performing low volume procedures in England, in an attempt to reduce unwarranted variation and decrease complication rates.^1,6

There are several national joint registries (NJR) that include TEA.^1,7,8 At the time of this study, the largest series held on a single registry was 2229 cases, on the UK NJR. ¹ This series involves eight different implants, although more than 95% of procedures were performed using either the Coonrad–Morrey prosthesis (Zimmer, Warsaw, IN; 57%), Discovery (DJO Global, Vista, CA) (29%) or Latitude (Wright Medical, Memphis, TN) (13%). ¹ The fourth most frequently implanted prosthesis, the GSB III (Zimmer), was used in just 2% of the cases. ¹

NIR data are important to track the performance of an implant and provide early warning of an unexpectedly high failure rate. ¹ However, NJRs currently only report survival or revision rates for implants and do not specifically report complications. This combined with the high probability that most surgeons are performing less than five cases per year, highlights the need for a study specifically investigating the complication rates reported after TEA. The complication rates related to specific indications have also not previously been published.

The aims of this study therefore were to investigate the complication rates of TEA and highlight any variations in complications seen between different contemporary implants and the indications for their use.

Methods

Study rationale

A preliminary literature search revealed a large number of small studies reporting complications from TEA performed using implants with low volume usage or for uncommon indications. To ensure that the results of this study were as relevant and transferable as possible, the decision was made to focus on investigating complications from the three most frequently used implants in the UK and for the most common indications. These are:

• Prosthesis: (i) Coonrad–Morrey

•  (ii) Latitude

•  (iii) Discovery

• Indication: Inflammatory arthropathy

•  Osteoarthritis

•  Trauma

A systematic review of the literature was then undertaken in accordance with Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines. ⁹

Search strategy

A systematic review was performed in MEDLINE, EMBASE and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) databases through OvidSP. The search strategy was designed using the Medical Subject Heading (MeSH) terms ‘Elbow', ‘Arthroplasty', ‘Replacement', ‘Osteoarthritis', ‘Fractures' and ‘Arthritis, Rheumatoid' combined with British Medical Journal systematic evidence filters for interventional trials. The final search was performed on 19 February 2018. The references for each study were checked to ensure other relevant studies not initially highlighted were included. Duplicates were checked and if two or more studies appeared to involve the same patient population only the most recent study was included.

Eligibility criteria

Studies were included if they were on human patients, were written in English, printed or published online and reported complications from TEA. For the purposes of this review, the authors used Sokol’s definition of a complication as ‘any undesirable, unintended and direct result of an operation affecting the patient, which would not have occurred had the operation gone as well as could reasonably be hoped’. ¹⁰

Studies were excluded if they had a follow-up period of less than 12 months, contained less than five patients, were systematic reviews or meta-analyses or included patients undergoing revision, tumour or instability surgery. Studies were also excluded if they presented data on a mixture of implants or pathologies or involved patients who had a previous known peri-prosthetic infection, radioulnar synostosis or needed additional reconstructive surgery.

Study selection

All titles and abstracts were reviewed and assessed by two authors using a purposely designed systematic review programme (Rayyan QCRI, Qatar Computing Research Institute, Doha, Qatar). ¹¹ Disagreements were reviewed together and settled by consensus with a third independent author. After initial screening and exclusion, the remaining full texts were retrieved and the eligibility criteria applied.

Data evaluation

The methodological quality of each study was assessed using the Methodological Index for Non-Randomised Studies (MINORS) criteria. ¹² This is a validated scoring system that allows a maximum score of 16 for non-comparative studies and 24 for comparative studies. ¹²

Results

The initial search revealed 117 studies and literature^12–24 met the eligibility criteria after screening (Figure 1). No studies reporting on the outcomes of osteoarthritis independently met the inclusion criteria. The percentage complications for each indication and implant are presented in Tables 1 and 2. Where possible the location of aseptic loosening has been presented by component affected. Aseptic loosening at any location is also presented. The assessment of study methodological quality is presented in Table 3.

Figure 1.

Flowchart of systematic review design demonstrating scientific paper identification, screening, inclusion and exclusion.

Table 1.

Implant group complications percentages and cumulative percentages.

Complication by implants	Discovery % n = 45 (References 20,22)	Latitude % n = 63 (Reference 15)	Coonrad–Morrey % n = 707 (References 13,14,16–19,21,23,24)	Overall % n = 815
Radiographic loosening anywhere	60.0	46.0	11.9	17.2
Ulna component radiographic loosening	(2.2)	(14.3)	(7.8)	(8.0)
Bipolar component radiographic loosening	(57.8)	(0.0)	(1.4)	(4.4)
Humeral component radiographic loosening	(0.0)	(6.3)	(2.7)	(2.8)
Radial component radiographic loosening	(0.0)	(25.4)	(0.0)	(2.0)
Worn bearings	0.0	0.0	18.2	15.8
Deep infections	4.4	4.8	6.8	6.5
Aseptic loosening requiring revision	2.2	3.2	6.8	6.3
Ulna neuropathy	4.4	4.8	3.4	3.6
Peri-prosthetic fracture	0.0	1.6	2.8	2.6
Rupture/Weakness of triceps	2.2	0.0	2.5	2.3
Peri-prosthetic heterotrophic ossification	11.1	1.6	1.6	2.1
Wound infection/complication	0.0	0.0	1.6	1.3
Stiffness requiring surgery	0.0	1.6	0.8	0.9
Radial neuropathy	0.0	0.0	0.6	0.5
Bursitis	0.0	0.0	0.6	0.5
Mechanical failure/locking pin	0.0	0.0	0.3	0.2
Component fracture	0.0	0.0	0.3	0.2
Instability	0.0	1.6	0.0	0.1
Ossification of brachialis	0.0	0.0	0.1	0.1
Triceps tendonitis	0.0	0.0	0.1	0.1
90-day mortality	0.0	0.0	0.1	0.1
Myocardial infarction	0.0	0.0	0.1	0.1
Pulmonary embolus	0.0	0.0	0.1	0.1
Total	84.3	65.1	58.8	60.7

Table 2.

Indication group complications percentages and cumulative percentages.

Complications with indications	Inflammatory arthritis % n = 114 (References 14,18,21,22)	Trauma % n = 40 (References 13,23,24)	Overall % n = 154
Radiographic loosening anywhere	14.9	32.5	17.2
Humeral component radiographic loosening	(7.0)	(5)	(5.7)
Ulna component radiographic loosening	(5.3)	(5)	(4.6)
Bipolar component radiographic loosening	(2.6)	(22.5)	(6.9)
Worn bearings	14.0	22.5	14.3
Aseptic loosening requiring revision	2.6	17.5	5.7
Ulna neuropathy	3.5	5	3.4
Peri-prosthetic heterotrophic ossification	0.0	15	3.4
Deep infections	3.5	2.5	2.9
Rupture/Weakness of triceps	3.5	0	2.3
Peri-prosthetic fracture	2.6	0	1.7
Wound infection/complication	1.8	2.5	1.7
Stiffness requiring surgery	1.8	0	1.1
Radial neuropathy	0.0	5	1.1
Component fracture	0.9	0	0.6
Total	35.1	70.0	38.9

Table 3.

MINORS score by indication/implant.

Indication/Implant	Number of included studies (non-comparative/ comparative)	Number of patients	Mean follow-up in years (range)	Mean MINORS score (range)	MINORS score 95% CI
Inflammatory arthritis	4 (3/1)	114	5.2 (2.3–7)	12 (6–21)	5.8–18.2
Trauma	3 (3/0)	40	6.4 (2.7–13)	7 (7–8)	6.3–7.7
Osteoarthritis	0 (NA)	N/A	N/A (NA)	N/A (NA)	N/A
Coonrad–Morrey	9 (6/3)	707	6.1 (2.3–13.0)	12 (6–21)	8.1–15.9
Discovery	2 (2/0)	45	4.9 (4.5–5.2)	8 (6–10)	4.6–11.41
Latitude	1 (1/0)	63	3.7 (N/A)	8	N/A

Discussion

This is the first study to specifically review the complications associated with the most frequently used TEA implants and the most common pathologies for which they are indicated. It is unfortunate that no studies met the inclusion criteria for osteoarthritis as this is the indication for a TEA in the UK in around a third of cases. ¹

Aseptic loosening radiographically remains the greatest problem for all TEA groups, with around 17% of all patients experiencing loosening of one or both components. The ulnar component is affected more frequently than the humeral component when looked at overall. However, when looked at for each indication, the difference between humeral or ulna loosening is minimal. The reported rate of loosening in the trauma group was over twice that reported in the inflammatory group. The Discovery group had the highest implant reported loosening and the Coonrad–Moorey the lowest. Radial head component loosening was specifically seen in the Latitude group in which non constrained implants were used. However, with a mean follow-up of less than five years for all pathologies within this review, this is a significant concern. The overall rate of revision for symptomatic aseptic loosening was at 6.3%. However, 17.5% of cases performed for trauma required revision for aseptic loosening compared to inflammatory elective cases at only 2.6%.

A high rate of deep infection was also highlighted at around 6.5% of all TEA patients in this review. This is substantially higher than in comparative joints such as the hip where peri-prosthetic infection rates of between 0.2% and 1.6% are quoted, depending on the methodology and study follow-up time. ²⁵

Ulnar neuropathy was the most common nerve injury across all groups, ranging from 3.4% to 4.4%, although this is usually a transient concern. ²⁶

The rates of component wear and mechanical failure seen in this study are also higher than that seen following other joint replacements, due in large part to the presence of bushing components that form the bearing surface in the Coonrad–Morrey implant. The presence of linking pin dissociation was demonstrated on x-rays within the articles reporting on the Coonrad–Morrey implant. ²⁴ This mechanism was described as bushing wear and not reported as a separate mode of failure.

There appears to be a similar rate of peri-prosthetic fracture overall (2.6%) when compared to the literature on other joint replacements, which range from 0.1% to 18% following total hip arthroplasty, 0.3% to 5.5% following total knee arthroplasty and 0.5% to 3% post total shoulder arthroplasty. ²⁷

When complications are stratified by implant, the discovery had the highest cumulative complication percentage overall, predominantly reflected in the much higher rate of radiographic loosening and heterotopic ossification compared to the other implants. This implant reported 100% bipolar radiographic loosening in one article. It is therefore important to note that despite this they had the lowest rate of revision for aseptic loosening at 2.2%.

The latitude elbow contained both linked (20%) and non-linked prostheses (80%) and radial components were not used in 11% of the other total elbow replacements. The complications were not analysed separately for linked and non-linked prosthesis. The authors commented on a high rate of aseptic loosening within the radial component. ¹⁵ Aseptic loosening in the humeral or ulna components was 20.6% and without the reported radial head loosening the overall complication cumulative percentage was reduced to 39.7%.

The Coonrad–Morrey had the highest rate of implant failure (predominantly due to bushing wear). It can also be seen that peri-prosthetic fracture, triceps weakness and rupture are also reported more in this group. The follow-up for these implants within the studies was longer and could potentially increase the rate of reported complications.

Trauma appears to have twice the overall complication cumulative percentage when compared to inflammatory arthritis and over five times the need for aseptic loosening revision. This may be explained by fitter patients who prior to their traumatic event had normal elbow function and have higher demands. This should be taken into account when considering elbow replacements in these higher demand patients. The longest follow-up periods were also seen in the trauma group and this may increase the likelihood of a complication being reported in these groups, especially aseptic loosening.

Implant failure and complication rates may also be affected by improvements in surgical technique (e.g. cementing technique) and also differing approaches (e.g. ‘Triceps on’ versus ‘triceps off’) used over this period. This however could not be evaluated in the constraints of this study.

Of note there is also a discrepancy in the total number of procedures included in Tables 1 and 2. This relates to some studies not reporting the complications by specific indication and hence these could not be included in Table 1. This has led to a paucity of data related to osteoarthritis patients independently, although when looking at implant complications these have been included. It is therefore not possible to draw any further conclusions regarding the complications related to osteoarthritis as an indication.

The UK Joint Registry was used to determine the most common implants and indications for this study and the authors would comment that geographical variations in populations and surgical practice may lead to some variation in extrapolating this data. Despite this however, the core indications and prosthesis type are commonly used internationally and data from across all published literature internationally were used for this systematic review.

The MINORS scoring criteria highlighted that the overall quality of the articles included in this review was low with an overall mean follow-up length of only 4.9 years. Further long-term follow-up studies and studies with higher quality methodologies would greatly assist in improving the accuracy of the estimated complication rates.

This study defined a complication as ‘any untoward event' in an effort to anchor the study to a patient centred overview, however it is recognised that reporting of all complications under this definition is not universally applied in implant surveillance studies. In view of these elements the authors feel that the cumulative complication rates may be under rather than over representative. It should also be noted that this study was unable to quantify patient focused outcomes and would echo calls for future complication reporting and survival estimates to be complimented with patient reported outcome measures. ²⁸

It should be noted that following stringently applying the exclusion criteria that only 12 studies were included for analysis, nine involving the Coonrad–Morrey, two involving the Discovery and one involving the latitude.

Conclusion

This is the largest systematic review of the complications of contemporary TEAs in the English Language. This study reflects the current use of TEA and highlights the lack of literature reporting mid- and long-term outcomes for these implants. It is also evident that the overall complication rates and cumulative percentages for TEA are high and much higher than arthroplasty in other joints. This is likely to be related to the implant design, patient selection, surgical technique and the biomechanical demands that patients put on their implants following surgery. However, the representative contribution of these elements is unknown and should remain a focus of future research. It is however, vital that surgeons are aware of the risk profile of this surgery and counsel their patients appropriately.

Trauma appears to have higher complication rates and overall cumulative percentage compared to inflammatory arthritis as an indication. There is however a paucity of literature on the complication rates associated with each indication and none in osteoarthritis independently. High aseptic loosening rates at over 10% in less than five years, regardless of the implant or indication and high rates of deep infection compared to other implants should be communicated to patients as part of preoperative consent.

Surgeons should be aware of differing complications related to their implant of choice with each having its own specific set of complications featuring more frequently with its use.

ORCID iDs

Jonathan P Evans https://orcid.org/0000-0003-0090-1012

William J White https://orcid.org/0000-0001-7682-029X