Complications following shoulder arthroplasty: A review of the recent literature

1. Introduction

Glenohumeral joint arthroplasty is a safe and effective treatment with osteoarthritis being the most common indication. The incidence of shoulder arthroplasty has increased over twofold in the last ten years¹ but increasing reoperation rates are also noted.² Revision surgery is associated with higher complication rates and poorer patient outcomes.²

The two main primary modalities include reverse shoulder arthroplasty (rTSA) and total/anatomic shoulder arthroplasty (TSA). Hemiarthroplasty of the shoulder is reducing in prevalence in the recent literature.

The indications for reverse shoulder arthroplasty have widened recently. Introduced by Grammont in the late 1980s, the implant was designed to inferiorise and distalise the humerus and medialise the centre of rotation, thereby achieving tension of the deltoid muscle.³ The success of the modality has resulted in widened indications such as severe primary cuff deficiency, rheumatoid arthritis, tumours, pseudo-paresis and revision after failed TSA.4, 5, 6

Indications for total shoulder arthroplasty by prevalence include osteoarthritis, rheumatoid arthritis and post-fracture sequelae,⁷ but an intact rotator cuff is vital.⁸

Complications are an undesirable outcome and can be classified by cause including general surgical and implant-related complications. These can lead to patient morbidity and poor outcomes, may result in revision surgery and can cause significant economic impact.

Certain patient groups exhibit higher prevalence such as: elderly, smokers, high body mass index (BMI), medical comorbidities (such as diabetes mellitus, renal failure and immunosuppression) and drug use (such as steroids⁹).

We consider pan-modality complications such as infection or chronic stiffness and more specific complications such as intraoperative or periprosthetic fracture, glenoid loosening, humeral loosening or dislocation/instability.

2. Aims

We aim to review the recent literature on complications following total shoulder arthroplasty and reverse shoulder arthroplasty and evaluate and amalgamate the recent evidence.

We intend to present the epidemiology and risk factors for complications as well as measures to reduce the risk of adverse outcomes and outline the presentation, investigation and management of these patients.

We aim to review recent advances in techniques and technologies and aim to highlight areas of further research in the field.

3. Methods

A literature review was performed using PubMed and Google Scholar databases. The search strategy used keywords: Complications AND ‘shoulder arthroplasty’ OR (‘total shoulder arthroplasty) OR TSA OR reverse shoulder arthroplasty’ OR (RSA)) OR (rTSA). This initial search resulted in 205 articles. Duplicates were removed, articles were screened and resulting full texts were analysed and further excluded, leaving 57 articles suitable for inclusion.

Exclusion criteria applied: studies not involving patients, studies not including primary arthroplasty and papers greater than 15 years old (most papers were published within the last 10 years).

A PiCO search strategy was used (see Table 1).

Table 1.

PICO search strategy.

PICO element	Keywords	Search Terms	Search Strategy
Patient/Population	Patients undergoing anatomic total shoulder replacement or reverse total shoulder replacement	Complication (s) shoulder arthroplasty	Complication(s) AND ‘shoulder arthroplasty’ OR ‘total shoulder arthroplasty OR ‘TSA’ OR ‘reverse shoulder arthroplasty’ OR ‘RSA’ OR ‘rTSA’.
Intervention	1.New evidence surrounding general complications and specific complications of each arthroplasty modality. 2.New patterns of complications due to new implant availability
Comparison	1.Traditional pattern of complications and investigations available and their evidence base 2.Historical evidence for the treatment options
Outcome	Up to date evidence surrounding epidemiology, diagnosis, management options of common complications and revision rates and patient-reported outcome measures

These studies were then analysed and data extracted based upon common complications: infection, specific complications of total shoulder arthroplasty (rotator cuff failure, instability, periprosthetic fracture, implant loosening, neurological injury) and reverse shoulder arthroplasty, with some consideration given to prevalence and chronicity of the complication where possible.

4. Results and interpretation

4.1. Common complications of shoulder arthroplasty

4.1.1. Infection

Infection prevalence is reported in up to 4 % of TSA cases¹ and 2.4 % in rTSA.⁶ Infection rates vary by indication: 2.4 % in irreparable or massive rotator cuff tear, 0.9 % in acute fracture and 3.7 % in fracture sequelae.⁶ This recent review reports an overall reduction in rTSA infection rates compared to those reported in 2011 by Zumstein et al..³

Infection can vary from simple wound discharge to frank sepsis requiring emergency washout.¹⁰ Risk factors for infection include diabetes mellitus, smoking, chronic rheumatological disease and intra-articular steroid injections or corticosteroid therapy (OR 2.0 at three and six months)¹⁰ as well as preceding arthroscopic procedures.¹¹

Cutebacterium acnes (C.acnes) is the most common organism (38.9 % of cases) and has an indolent and insidious nature. It often triggers a minimal inflammatory response and causes mild or minimal clinical symptoms, therefore a high index of suspicion is required.¹⁰ In addition, there may be little or no rise in traditional biochemical markers (such as C-reactive protein), which presents a diagnostic challenge. Other causative organisms include Staphylococcus aureus and staphylococcus epidermidis.¹

General principles of investigation include thorough clinical history and examination, biochemical and microbiological tests including C-reactive protein and blood cultures, synovial aspirate, including microscopy, culture and sensitivity and radiological investigation such as plain radiograph. Synovial alpha-defensin or interleukin based testing media have been described in the literature, with varying success rates.¹⁰ Arthroscopic tissue biopsy has been described as a useful tool to guiding management of the infected arthroplasty. Procuring tissue samples in this manner for microbiology analysis has been shown to be highly sensitive and specific and superior to fluoroscopic-guided tissue synovial aspirate.¹² Antibiotic therapy should be targeted to the patient, depending on what goals of treatment are utilized. If revision surgery is planned, antibiotics should ideally not be started until first stage revision is completed. If the patient is showing signs of systemic infection, all efforts to procure tissue samples should be made before commencing antibiotic therapy, if appropriate to the patient's condition.¹⁰

Management can vary from antibiotic suppression therapy (usually if medically unfit for revision procedure) to one or two-stage revision. If the organism is known, recent evidence has shown good patient-reported outcomes and low reoperation rate in one-stage revision,¹⁰ however the higher-demand patients frequently benefit from a two-stage revision including temporary antibiotic spacer. Salvage options including resection arthroplasty or definitive spacer can be used in a low-demand or surgically unfit patient.¹⁰

4.2. Specific complications of total shoulder arthroplasty

The overall complication rate following TSA is 5.6%–6.36 %.^9,13 Predictive factors for complications include preoperative function (OR 1.8 in dependent patients), American Society of anaesthesiologist (ASA) score 3 (OR 3.6) or 4 (OR 8.5), age over 80 years (OR 2.3)¹³ and medical comorbidities such as diabetes mellitus, renal failure or transplant patients.¹⁴ Therefore, accurate patient selection and thorough preoperative workup are suggested.^1,15

Specific complications are now outlined. A suggested investigation algorithm for the painful shoulder arthroplasty is suggested in Fig. 1. A suggested management protocol for periprosthetic fracture is outlined in Fig. 2 and a management algorithm based on the recent literature for the painful TSA is then provided in Fig. 3.

Fig. 1.

Suggested Investigation Algorithm for painful shoulder arthroplasty.

Fig. 2.

The Wright and Cofield classification for periprosthetic fracture and suggested management principles.

Fig. 3.

Suggested Management Algorithm for the painful TSA.

4.2.1. Implant loosening

Implant loosening has been considered separately in TSA and rTSA. Component loosening often presents with pain. Infection should always be considered until proven otherwise.^1,16 Investigations include plain-film radiography and CT scans. CT can show nonspecific radiolucent lines around the glenoid component, either as a precursor or as a sign of loosening, with up 15–84 % prevalence reported.^1,16

Glenoid loosening is more common than humeral, with the former having a prevalence of 6 %.¹ The risk factors include: progressive rotator cuff failure, glenoid morphology and implant design.¹

Glenoid morphology assessment is vital in preoperative planning.¹⁷ The “B2” glenoid features a normal anterior glenoid (with up to the anterior 50 % being unaffected¹⁷) with varying degrees of posterior bone loss, resulting in a biconcavity. The humeral head then migrates posteriorly to articulate with this “neoglenoid”.¹⁸ This is associated with higher complication rates, including glenoid loosening in up to 20.6 %¹⁸ with rates of failure proportional to bone loss. If concurrent with rotator cuff failure, the “rocking horse” phenomenon contributes to eccentric loading¹ and accelerated wear of polyethylene..¹ Described preoperative addressing measures include accurate preoperative imaging (plain radiography and CT scanning formatted along the plane of the scapula¹⁷). Intraoperatively, if less than 20 % bone loss is noted, use of the anterior glenoid as a reference point may be possible but adequate clearance of osteophyte is key.¹⁷ In such cases, asymmetric reaming has been described as a useful technique to remove the biconcavity. Traditionally, in severe bone loss cases, bone grafting or augmented glenoid components may be used, or rTSA considered to reduce the risk of failure.¹⁷ However, a recent study has compared bone graft versus augmented baseplates and reported superior outcomes in terms of range of motion and patient satisfaction in the augmented baseplate group¹⁹ in addition to a shorter intraoperative time and reduced risk of scapula notching and revision. Other surgical techniques include glenoid bone stock preservation, minimal reaming and good cementation techniques.¹

Thorough preoperative imaging workup is vital to optimise implant positioning. Starting with plain radiography, the critical shoulder (CSA)²⁰ is defined as “the angle formed between the plane of the glenoid and the line connecting the most lateral border of the acromion process” on the true AP radiograph.²¹ The average CSA is around 32° and CSA >36° can increase superior glenoid component wear (Loosening OR - 1.2 per degree above 36°).²⁰ In addition, many studies advocate a preoperative CT scan^22,23 and many surgeons utilise this as standard practice in shoulder arthroplasty. Specific uses include evaluation of glenoid morphology and bone stock available.²⁴ Advancements in the quality and formatting of 3D images allow images to be centred along the plane of the scapula, allowing for accurate measurement of the glenoid version.²⁴ Further recent developments in the field of TSA include 3D CT-guided templating software as well as patient-specific instrumentation.²⁴ This allows the CT scan to be uploaded to software that will calculate critical measurements and allow simulated placement of implants. A definitive consensus is not yet available as to whether this is superior to classic techniques, however some studies report on higher glenoid implant placement error rates with classic instrumentation alone.²⁵ We highlight analysis of 3D planning software and patient-specific instrumentation outcomes as an area for further research.

Navigation is another recent advance in the field, which is increasing in prevalence and more studies are reporting its efficacy. Certain operative measurements which are superior in navigated cases include increased glenoid implant screw siting and length (if used) and reduced risk of inaccurate glenoid implant placement and cage perforation.^22,23 However, navigation is associated with longer operative times and we highlight this as an area for further research to ascertain if outcomes are superior to traditional instrumentation.²²

Patient-specific instrumentation is another recent development in total shoulder arthroplasty. This involves workup using CT scan and 3D planning software, as previously described, in addition to creation of patient-specific guides.²⁶ It's efficacy is debated in the literature, with some studies reporting superior outcomes²⁷ and other studies highlighting its extra cost and time consumption, with no significant difference in outcomes.^26,28 Therefore we advocate for further research to ascertain whether outcomes are favourable compared to traditional instrumentation.

Implant material design is crucial. Traditional metal-backed glenoid components are prone to loosening, due to polyethylene liner wear and failure. All-polyethylene cemented components have also demonstrated high revision rates.²⁹ These have been superseded by newer porous-coated bone-implant interface, which have shown reduced rates of glenoid loosening,¹ however higher revision rates (OR 5.0) have been reported.³⁰

Recent development in implant design include trabecular metal implants, which are increasing in prevalence.³¹ No statistically significant difference in clinical outcomes has been reported between these cemented and cementless groups and 100 % implant survival rate at 5 years is described.³¹ When comparing implants to each other, there is a paucity of literature surrounding these newer implants and we highlight this as an area for further research. At present, it is suggested that clinicians consider implants that are most appropriate for specific patients and within their skillset and familiarity.³²

Revision arthroplasty is an accepted treatment, but challenges arise, such as significant bone loss and increased risk of infection. Outcomes are variable, with one study reporting 91 % satisfactory PROMs at 5 years,¹ however other studies report 67 % of patients suffering recurrent loosening, and further revision procedures required in 17 %.²⁹

Humeral loosening is rare (less than 1 %)¹ and implant tilt or subsidence is reported as being one of the modes of failure. Revision of the implant may be indicated; however, this may be difficult if the stem is well-fixed distally and may require a humeral osteotomy.

4.2.2. Instability

Instability is a rare complication of TSA. It can be classified by direction and by failing structure.¹ Accurate glenoid placement and sizing, especially in cases of B2 or retroverted glenoid, is crucial. Superior migration of the humeral implant can be noted in infraspinatus failure and anterior instability can be noted in subscapularis failure. Repair of the failing rotator cuff can be attempted but one study has reported poor outcomes (80 % failure rate)¹ therefore revision to rTSA remains a reliable option.¹

4.2.3. Rotator cuff failure

Progressive rotator cuff failure has a reported incidence of 1.3 %,^1,33 most commonly in subscapularis. Causes include tendon insufficiency, "overstuffing" the joint or multiple preceding operations. This can result in eccentric loading of the humeral head on the glenoid component in a “rocking horse” fashion, resulting in loosening.¹ In addition, proximal migration of the humeral head can be noted, in similar fashion to the native cuff-deficient shoulder. One study reports the rate of cuff dysfunction as 16.8 % at 103 months³³ and patients may also experience poorer PROM scores, reduced range of motion and increased radiolucent lines on imaging.³³

This highlights the importance of appropriate patient selection for TSA and may contribute to the increasing number of rTSA performed in the recent literature. Primary repair or pectoralis major tendon transfer can be considered in anterior subluxation¹ as well as revision to rTSA.

4.2.4. Periprosthetic fracture

The incidence of periprosthetic fracture is 1.2–19.4 %.^1,34 Humeral fractures represent the majority and risk factors for intraoperative fracture include: uncemented press-fit component (RR, 2.9), revision arthroplasty (RR, 2.8), history of instability (OR, 2.65), female sex (OR, 4.19) and posttraumatic arthritis (RR, 1.9). Risk factors for a postoperative humeral fracture include osteonecrosis and increased medical comorbidity index (OR, 1.27).³⁴ The Wright & Cofield classification³⁵ stratifies the fracture based on its location and the integrity of the fixation of the stem. It is outlined in Fig. 2 along with a suggested management plan for each fracture type.^1,35

Management of these fractures can be challenging, as one study reports up to 50 % of stems could be loose in periprosthetic fracture³⁶ with a high reoperation rate.

4.2.5. Neurological injury

The prevalence of neurological injuries is 1–4.3 %.¹ The most common nerves affected are the axillary nerve (direct causes include surgical dissection risk and stretch injury from retractors and indirect causes include postoperative haematoma, excessive arm traction or cement extrusion) and the musculocutaneous nerve during placement of the conjoint tendon retractor.^1,37

Brachial plexus neuropathy is rare¹ but can occur as a result of anaesthetic interscalene blocks.^38,39 Although most neurologic complications manifest as transient neuropraxias, nerve injuries can lead to persistent pain and functional impairment.⁴⁰

Investigation includes thorough clinical history and examination, radiographic investigations such as MRI scanning and EMG testing (minimum 6 weeks after injury). Most neuropathies will resolve spontaneously within 12 months, however direct exploration may be indicated in cases such as space-occupying lesion or haematoma.¹

4.3. Specific complications of reverse shoulder arthroplasty

A general higher complication rates in rTSA is accepted compared to TSA. The odds ratio in rTSA compared to TSA are: 1.66 for infection, 2.1 for dislocation, 4.01 for periprosthetic fracture and 3.58 local fracture.⁴¹ Another study reports a 15%–24 %⁵ incidence of complications in primary rTSA, with up to 40 % in revision rTSA.⁴²

rTSA is often performed in the fracture setting and in older or frail patients, which may contribute to these findings. Osteoporosis was prevalent in 26 % of patients and a higher risk of periprosthetic fracture (OR 1.86) and revision arthroplasty (OR 1.42)⁴³ is noted. This highlights the importance of patient selection and optimisation.

Infection has already been discussed above, with principles applying to rTSA.

4.3.1. Scapular notching

Scapula notching refers to erosion of the inferior scapula neck region due to impingement of the humeral component during adduction.⁴⁴

A recent study has reported scapula notching in up to 29%–43 % of rTSA cases.^6,45 Severe notching can lead to baseplate loosening and implant failure.⁶ The literature presents a debate on the clinical significance, with some papers reporting increased pain and reduced Constant-Murley scores in the notching patients, but others reporting no statistically significant differences.⁶

Causes described include implant design and surgical technique: including positioning of the glenosphere, neck-shaft angle and offset and patient factors such as natural scapula anatomy.^6,44 Implant design is a crucial factor in scapula notching. 3 main implant designs include the: medial glenoid/medial humerus (MGMH), medial glenoid/lateral humerus (MGLH), and lateral glenoid/medial humerus (LGMH). Scapula notching in the MGMH designs were significantly higher (up to 52 % of cases).⁴⁶ The initial Grammont design has also been associated with high rates of notching⁶ compared to all other designs (42 % vs 12.3 %).⁶ Recent systematic reviews have reported overall reduced rates of scapula notching compared to older reviews, perhaps due to improved implant design and surgical techniques.^3,6

Techniques which can be employed to reduce the occurrence of notching include lateralising offset, inferior baseplate overhang and thorough preoperative investigation and planning, with particular attention to glenoid morphology.⁴⁴

4.3.2. Implant loosening

As previously discussed, radiolucent lines on radiological imaging may correlate with implant loosening. The prevalence varies from 1 to 60 % on the glenoid (mean 7.7 %) and mean 12 % on the humerus.⁶ This correlates with a 2.3 % and 1.4 % revision rate respectively.⁶

4.3.3. Humerus

Short humeral stems, with a metaphyseal-fixed design are a recent increasing trend. However, the complication rate was significantly higher in the rTSA group (11 % revision rate compared to 7.8 % in TSA).⁴⁷ There has been no significant clinical or radiological difference in outcomes reported between cemented and uncemented humeral stems.⁴⁸ However, there is a higher preference towards uncemented stems used in primary rTSA,⁶ with metaphyseal fixation as the biomechanical principle. This design benefits from good metaphyseal blood supply and reduced stress-shielding with easier stem removal if required.⁶

4.3.4. Glenoid

High forces are transmitted through the glenoid baseplate and component as a result of the design of rTSA and this may contribute to reduced stability due to disrupted bony ingrowth.^3,6 There has been an observed reduction in glenoid loosening rates recently, possibly due to improvements in biomaterials, implant design and surgical technique.⁶ For example, the MGMH design was also linked with higher rates of glenoid loosening, in up to 6 %.⁴⁶ Whilst the lateralising designs are gaining in popularity, those increased mechanical forces which may contribute to a risk of loosening are balanced by hydroxyapatite coating and a wide range of variable-angle, higher diameter locking screws in the newer designs which can target good bone stock, improving fixation.⁶

4.3.5. Neurological injury

The rTSA demonstrates a higher rate of peripheral nerve injuries than the TSA (RR 10.9), likely due to the relative lengthening of the arm, with the axillary nerve most commonly affected.^1,37 Ladermann et al. attributed a nearly eleven-fold increased risk of axillary nerve injury correlated with arm lengthening, described as up to 2.7 cm compared to the contralateral side. Patients with subclinical changes on electrodiagnostic studies indicating axillary nerve injury exhibited greater arm lengthening compared to those without such changes (4.2 cm vs. 2.6 cm).³⁷

The axillary nerve course runs average 8.1 mm from the humeral metaphysis and 13.6 mm from the inferior glenoid rim.⁴⁹ The incidence can vary between 0.6 and 8 %,^6,50 with a higher neurological injury risk in revision rTSA.⁶ The vast majority of axillary nerve palsies are transient and will resolve spontaneously⁵⁰ but neuromuscular monitoring can be useful to document progress.⁶

Iatrogenic axillary nerve injury often leads to deltoid weakness. Additional presenting symptoms may include shoulder instability, pseudo paralysis, and persistent pain or paraesthesia.³⁷ Kohan et al. identified axillary nerve palsy as a significant factor in recurrent dislocations.⁵¹ Strategies to mitigate the risk of nerve injury, include the use of larger glenospheres or employing superolateral approaches to minimize lengthening without increasing the risk of instability.³⁷

The traditional Grammont rTSA results in a centre of rotation that is both medialized and distalised⁵² and this leads to lengthening of the moment arm, resulting in increased tension on the brachial plexus and its associated peripheral nerves.⁵³

Parisien et al. utilized intraoperative neuromonitoring and observed a 5 times higher incidence of nerve alerts during the reduction phase in the rTSA group compared to the TSA group.⁵⁴

Kim et al. investigated the effect of humeral distalization on nerve injury in and found that patients who sustained a nerve injury had a mean increase in the acromion-humeral distance of (24.5 mm v 20.5 mm in controls).⁵

Lowe et al. found an average arm lengthening of 13.7 mm in their rTSA group, which is lower than reported in other studies, and only 1 out of 30 patients showed evidence of nerve injury post-operatively. They used a lateralized prosthesis with a lower neck-shaft angle and reported that this can lead to decreased arm lengthening and, consequently, lower the risk of nerve injury.⁵⁵ Although the literature consistently indicates an increased risk of nerve injury in rTSA primarily due to arm lengthening, further research is needed to explore the specific prosthesis design which may impact nerve injuries.

The suprascapular nerve can also be at risk during posterior and superior drilling for screws in the glenoid baseplate⁶ and should a clinical significance be determined, removal of the offending screws may be pertinent.

4.4. Hemiarthroplasty use

Shoulder hemiarthroplasty is mainly currently used for the treatment of proximal humerus fractures in the elderly^56,57 and in some cases in the young patient to avoid involvement of the glenoid. It boasts shorter surgical time and good retention of shoulder range of movement postoperatively⁵⁷ however, this is largely dependent on healing of the tuberosities. In addition, malposition of the implant can cause significantly poor outcomes. For this reason, amongst others, this has resulted in a 406 % increase in uptake in rTSA for proximal humerus fracture.⁵⁷ One study quoted a 14.7 % rate of complications with hemiarthroplasty use, but it is worth noting that this is not significantly higher than rTSA in this setting.⁵⁷ It is important to highlight that these injuries can often lead to reduced function in patients regardless of treatment modality used.

We acknowledge the limitations of our review, which include a variety of papers of differing levels of evidence. Many of the studies are retrospective with differing patient sample sizes, study designs and length of follow up, as well as potential reporting bias. In addition, many of the papers are from different centres (high versus low-volume) which may produce differing outcomes.

5. Conclusion

In conclusion, shoulder arthroplasty, primarily TSA and rTSA, is increasing in popularity and there are a plethora of new implants, techniques and indications (primarily for rTSA) in the literature. With the increasing prevalence, there is the potential for increasing complications and these complications can lead to patient morbidity, revision surgery and can have adverse economic effects. In TSA, accurate patient selection and optimisation, and meticulous preoperative workup as well as surgical technique can reduce risks of cuff failure or implant loosening. In rTSA, an appreciation of the patient demographic, optimisation and preoperative workup, as well as meticulous surgical technique and implant design choice, may reduce risk of scapula notching, implant loosening and neurological complications. Whilst we have presented evidence from the recent literature, there are many areas for further research to improve and enhance patient outcomes further in the field of shoulder arthroplasty.

Credit statement

Giles Faria: Conceptualization, Methodology, Literature search, Writing- Original draft preparation Zaid Ali.: Data synthesis, Manuscript review MA Rasheed: Data synthesis and manuscript review, Ali Abdelwahab: Writing and manuscript review Hariharan Mohan: Manuscript review: Nik Bakti: Manuscript review Writing- Reviewing and Editing, Bijayendra Singh: Manuscript review.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.