Complications of Shoulder Arthroplasty

John W. Sperling

Robert H. Cofield

INTRODUCTION

Although there has been discussion in the literature concerning complications of shoulder arthroplasty, they are often addressed obliquely in the materials presented, with only a few articles concentrating on a specific complication. By collating this literature, one can evaluate the types and frequency of complications, but is less informed about the nuances of each type of complication and treatment recommendations. Therefore, we have felt it is important to not only detail the information in various patient series and display the lesser number of articles addressing a specific complication, but also present the materials developed over time at our institution that can be analyzed in-depth.

TYPES AND FREQUENCIES OF COMPLICATIONS

Literature

Previously, we identified 22-patient series published since 1980 on unconstrained total shoulder arthroplasty (TSA).²,³,⁷,⁸,⁹,¹⁴,¹⁸,¹⁹,²⁸,³⁰,⁴⁰,⁴²,⁴³,⁴⁸,⁵⁴,⁶⁴,⁷⁶,⁸³,⁸⁵,⁹⁴,⁹⁵,¹¹³ Altogether, these series have reported on 1,183 total shoulder arthroplasties. The authors identified 23 different complications. The total number of complications was 123 for a relative percentage of complications of 10.4%, assuming one complication occurred in each of the shoulders having a complication. Certainly some shoulders had two or more complications, so the number of shoulders having a complication would be somewhat less than 10%. Only rotator cuff tearing, instability, and glenoid loosening occurred in more than 1% of shoulders reported (Table 22-1).

We identified 20-patient series reporting on hemiarthroplasty of the shoulder that have been published since 1980.⁵,⁹,¹¹,¹⁴,²⁴,³³,⁴⁴,⁴⁹,⁵⁰,⁵⁵,⁶⁰,⁶¹ and ⁶²,⁷⁹,⁸⁵,⁸⁷,⁸⁹,⁹⁵,¹⁰⁸,¹²² These included reports on 498 shoulders. Nineteen different complications were identified and 78 total complications were reported for a relative frequency of complications of 15.7%, assuming only one complication occurred in each shoulder (Table 22-2). Six complications occurred at a frequency of 1% or greater. These included glenohumeral instability, painful glenoid arthritis, humeral tuberosity nonunion, rotator cuff tearing, nerve injury, and infection.

Other ways to assess complications are to identify the characteristics of unsatisfactory arthroplasties or focus on special subgroups such as younger patients, the elderly, or the obese. In reviewing 353 shoulders of patients dissatisfied with their arthroplasties, pain and reduced function were most common.⁴⁶ Major reasons for these complaints were component positioning, glenohumeral malalignment, glenoid component loosening, or glenoid bony erosion. The same investigators identified that outcomes of patients less than 50 years of age are worse seemingly due to more complex pathological conditions, including rheumatoid arthritis (RA), post-traumatic arthritis, and capsulorrhaphy arthropathy. These conditions in deference to osteoarthritis may complicate surgery, rehabilitation, and outcome.⁹⁶ As a counterpoint, the elderly can have an excellent outcome,²³ but are subject to greater perioperative morbidity, longer hospitalization,⁷⁰,⁹² and possibly earlier failure.⁶⁶

The Mayo Clinic Experience

Total Shoulder Arthroplasty

The relative incidence and type of complications at our institution have changed over time. In 1999, we reported on 419
unconstrained total shoulder arthroplasties performed by the senior author (RHC) between December 1975 and December 1989.27 The complications were defined according to time of occurrence, with early complications occurring within 90 days of surgery and late complications occurring after this period of time. Surgical complications were also classified as minor when there was no compromise of outcome and little or no treatment was required. A major complication occurred when the final result was compromised or reoperation was required.

TABLE 22-1 Complications of Total Shoulder Arthroplasty^a

Complication	No. (%)^b
Rotator cuff tear	23 (1.9)
Instability	18 (1.5)
Glenoid loosening	15 (1.3)
Intraoperative fracture	15 (0.9)
Malposition	7 (0.6)
Nerve injury	7 (0.6)
Infection	5 (0.4)
Humeral loosening	4 (0.3)
Postoperative fracture	4 (0.3)
Wound problem	4 (0.3)
Wire breakage	3 (0.3)
Impingement	3 (0.3)
Tuberosity nonunion	3 (0.3)
Pain, unexplained	3 (0.3)
Reflex dystrophy	2 (0.2)
Hematoma	2 (0.2)
Component dissociation	2 (0.2)
Extruded cement	2 (0.2)
Heterotopic ossification	1 (0.1)
Stiffness	1 (0.1)
Spacer dislocation	1 (0.1)
Intraoperative death	1 (0.1)
Pulmonary embolism	1 (0.1)
Total no. = 23	123 (10.4)
^aTwenty-two series since 1980; 1,183 shoulders studied. ^bRelative percentages of each complication, assuming one complication occurred in each of the shoulders having a complication.

Among these shoulders, 130 (31%) had a major surgical complication, with 95 of these shoulders requiring reoperation (Table 22-3). The most frequent complication requiring reoperation was joint subluxation followed in order of frequency by rotator cuff tearing, glenoid loosening, dislocation, humeral loosening, and infection. One can see that the variety and aggregate number of complications in the early experience with the shoulder arthroplasty was quite high. Therefore, we reviewed the outcome of a more contemporary group of patients to determine whether the complications have changed or the complication rates have lessened over time.

TABLE 22-2 Complications of Humeral Head Replacement^a

Complication No.	(%)^b
Instability	14 (2.8)
Glenoid arthritis	12 (2.4)
Tuberosity nonunion	9 (1.8)
Rotator cuff tear	9 (1.8)
Nerve injury	8 (1.6)
Infection	5 (1.0)
Intraoperative fracture	3 (0.6)
Humeral loosening	3 (0.6)
Wound problems	2 (0.4)
Tuberosity malposition	2 (0.4)
Hematoma	2 (0.4)
Perioperative death	2 (0.4)
Postoperative fracture	1 (0.2)
Heterotopic ossification	1 (0.2)
Impingement	1 (0.2)
Reflex dystrophy	1 (0.2)
Acromioclavicular pain	1 (0.2)
Pain, unexplained	1 (0.2)
Stiffness	1 (0.2)
Total no. = 19	78 (15.7)
^aTwenty series since 1980; 498 shoulders studied. ^bRelative percentages of each complication, assuming one complication occurred in each of the shoulders having a complication.

The complications of 431 total shoulder arthroplasties performed in patients by the senior author (RHC) between December 1990 and December 2000 were reviewed at an average of 4.2 years.²² Fifty-three surgical complications occurred in 53 patients. Thirty-two were considered to be major (7.4%), with 17 (3.9%) of these requiring reoperation (Table 22-4). Thirty-two occurred early and 21 occurred late. Rotator cuff tearing was the most common complication encountered followed in frequency by instability and periprosthetic humeral fracture. Glenoid and humeral component loosening requiring reoperation occurred in one shoulder. Of the 17 symptomatic cases of rotator cuff tearing, 8 patients had a preoperative rotator cuff tear. There were four
postoperative subscapularis tears and all four patients experienced anterior instability. When comparing the five main categories of diagnosis (osteoarthritis, rheumatoid arthritis [RA], post-traumatic arthritis, cuff tear arthritis, and osteonecrosis) there was no statistically significant difference (P = 0.088) in the complication rates observed in each of these groups—when comparing each of the last four diagnostic categories relative to the primary diagnosis of osteoarthritis. Risk factors for any complication and for specific complications were carefully assessed. The frequency of complications was not affected by age, gender, previous surgery, humeral head size, or whether the humeral component was cemented or uncemented. When comparing this with the prior study, one can clearly see that the frequency of complications has dramatically decreased as has specifically the number of major complications and the need for reoperation. In addition to the general lessening of complications, there is the noticeable striking diminution of component loosening.

TABLE 22-3 Severity of Surgical Complications

Complication	Minor^a	Major	Reoperation
Subluxation	1	11	32
Rotator cuff tear	5	11	17
Glenoid loosening		2	17
Brachial plexopathy	3	8
Dislocation			10
Humeral loosening			9
Infection (deep)			6
Impingement	5
Dysesthesias	5
Infection (superficial)	2		2
Fracture	3	1
Hematoma	1		1
Reflex dystrophy		2
Nerve laceration			1
Tuberosity nonunion	1
Long head of biceps rupture	1
Totals	27	35	95
^aSee text for definitions.

In a similar, recent review of 485 primary arthroplasties from Germany with a mean follow-up of 3.5 years, complications were classified as (1) without reoperation, (2) soft tissue revision, or (3) implant revision.¹ Of the 56 complications 34 were type (1), 11 were type (2), and only 11 were type (3). Again, as in the above study, the complication rate was relatively low with this length of follow-up.

TABLE 22-4 Severity of Complications

Complication	Minor	Major	Major with Reoperation
Rotator cuff tear	3	8	6
Fracture	7	5	1
Brachial plexopathy	8
Subluxation		1	4
Dislocation		1	3
Humeral loosening	1
Humeral and glenoid loosening			1
Infection (deep)			1
Hematoma	1		1
Long head of biceps rupture	1
Totals	21	15	17

Hemiarthroplasty

We have analyzed our hemiarthroplasties in two groups. The first group includes those with glenohumeral arthritis—either osteoarthritis or RA. Between July 1977 and March 1983, 77 shoulders were so treated in 74 patients.²⁹ Six patients with six operated shoulders were lost to follow-up and four others died before evaluations were complete. This resulted in 67 shoulders in 64 patients forming the basis of the review. There were 35 shoulders in 35 patients with osteoarthritis and 32 shoulders in 29 patients with RA. Follow-up evaluation averaged 9.3 years and ranged from 2 to 14.1 years. There were three complications: One patient developed a hematoma requiring surgical evacuation; the second had a humeral shaft fracture at the time of surgery, which was treated with internal fixation using a long-stem component and cerclage (the fracture healed); and the third complication was a brachial plexus traction injury that recovered without residual symptomatology.

The outcome of humeral head replacements performed for acute proximal humeral head fractures and chronic proximal humeral head fractures between 1979 and 1995 was also reviewed, focusing on complications.⁸⁰ Twenty-eight replacements were performed in 26 patients with acute fractures and 55 hemiarthroplasties were performed in 55 patients for problems that arose following initial treatment for a proximal humeral fracture or fracture-dislocation. Again, all patients were included to be sure that every complication was recognized during the follow-up period. In the acute group the followup averaged 56 months, ranging from 2 to 184 months, while in the chronic group the follow-up averaged 57 months, ranging from 5 to 156 months. The complications identified for these acute and chronic fracture groups are displayed in Table 22-5. One is immediately impressed that these are much
greater in number and many more types of complications arose when compared with hemiarthroplasty done for the elective treatment of shoulder arthritis.

TABLE 22-5 Complications of Hemiarthroplasty for Fractures

Complication	Acute (No. = 28)	Chronic (No. = 55)	Totals (No. = 83)
Instability	2	10	12
Glenoid arthritis	2	7	9
Rotator cuff tear	2	4	6
Infection	3	3	6
Tuberosity nonunion	2	2	4
Fracture		3	3
Implant malposition		3	3
Implant loosening	1	2	3
Tuberosity malunion		2	2
Reflex dystrophy	1		1
Totals	13	36	49
(Adapted from Muldoon MP, Cofield RH. Complications of humeral head replacements for proximal humeral fractures. Instr Course Lect 1997;46:15-24.)

Large Databases and Meta-analysis

Reporting from the Norwegian Arthroplasty Register from 1994 to 2005, 1531 hemiarthroplasties and 69 total shoulder arthroplasties were entered. For hemiarthroplasty the risks of revision were higher for the treatment of the sequelae of a fracture or for younger patients. The main reasons for revision were pain or subluxation.⁴¹ Using the California statewide discharge database of 15,288 patients, hemiarthroplasty and total shoulder arthroplasty had equal complication rates in the first 90 days after surgery.³⁸ Again fracture patients were at a higher risk for complications. Overall, the mortality was 1.3%, and pulmonary embolism occurred at a rate of 0.6%.

Total shoulder arthroplasties have been compared to total hip and total knee arthroplasties in the larger databases. Data from the Veteran’s Administration for 1999 to 2006 reviewed 793 total shoulder arthroplasties, 10,407 total hip arthroplasties, and 23,042 total knee arthroplasties. Total shoulder arthroplasties required a longer operative time, but hospital stay was less as were complication rates (2.8% versus 7.6% for THA and 6.8% for TKA) and mortality.³⁹ Similarly, using the New York State Department of Health Database, venous thromboembolism was 5 per 1000 procedures for shoulder arthroplasty versus 15.7 for hip arthroplasty and 26.9 for knee arthroplasty. Risk factors for thromboembolic events after shoulder arthroplasties were age, trauma, and cancer.⁷¹ Of note, in a single practice group using Doppler ultrasound, the prevalence of deep venous thrombosis was 13% divided almost equally between the ipsilateral upper extremity and the lower extremities. The incidence of pulmonary embolism was 3%.¹¹⁸

Meta-analyses are difficult due to wide variation in the styles of reporting. In one attempt to analyze shoulder arthroplasty, 40 studies were included with 3,584 patients.¹¹⁴ All reports showed good outcomes for both hemiarthroplasty and total shoulder arthroplasty. Diagnosis, the extent of shoulder pathology, and prosthetic specifics predicted outcomes, however, this study design constrains exploring further variations in these parameters.

SPECIFIC TYPES OF COMPLICATIONS

The six significant and most common complications will be individually presented. Material will generally be arranged in the following sequence: the recognition and evaluation of the complication with a classification system if one is applicable, reference to the frequency of the complication as presented in the literature, identification of specific literature for the complication, notation of materials included in pertinent review articles, the experience with this complication at the Mayo Clinic, methods to prevent the complication, and treatment for the complication.

Nerve Injury

The physical examination of the operated upper extremity is the key to diagnosis following surgery. It is quite practical to ask the patient to perform active movement of the hand and wrist and to test for isometric contractions of the elbow flexors and the posterior portion of the deltoid muscle. This can be accomplished on the day of surgery; however, interscalene block is now commonly used, making it sometimes necessary to perform the neurologic examination the day following surgery. Should a nerve injury be identified, the common peripheral nerve injury classification that is defined by Seddon or Sunderland is probably useful in retrospect but has less value in the acute setting.⁹⁸,¹⁰⁷ Weber et al. developed a post-hip arthroplasty nerve palsy severity scale based on symptoms, physical examination, electromyographic findings, and the compromise of postsurgical rehabilitation,¹¹⁶ and it is useful as a grading system. This scale, though, as with the more standard classification scheme, is only fully applicable over time, as the nerve injury evolves, and less helpful in the acute setting. Electromyographic testing may be useful after the initial 3 weeks, but as will be explained below, is probably more practical at 4 to 6 weeks, should neurologic recovery not occur in the interim.

In the 22-patient series encompassing 1,183 operated shoulders that were reported since 1980 and defined in Table 22-1, 7 nerve injuries (0.6%) were identified following TSA. In the 20-patient series involving 498 shoulders reported since 1980, 8 nerve injuries (1.6%) were identified following hemiarthroplasty. A slightly higher frequency following hemiarthroplasty may be attributed to this surgery, often occurring in the acute setting where some degree of nerve injury might complicate the initial fracture and yet not be fully defined because of the inability to perform a complete examination before surgery. In an article published on neurologic complications after TSA,72 the authors identified eight neurologic deficits reported in the literature. Of these five were axillary nerve palsies, only one of which completely resolved and two partially recovered. One musculocutaneous nerve palsy did
not resolve. One radial nerve palsy responded completely to removal of cement that had extruded through a humeral defect during revision surgery. The final injury produced ulnar nerve dysesthesias, which resolved. In a second article, preoperative and postoperative electromyographic evaluation was performed in 23 shoulders undergoing anatomic shoulder arthroplasty. A brachial plexus lesion occurred in one.⁶⁹

Nerve injuries have received comment in review articles of shoulder arthroplasty. Miller and Bigliani mentioned that nerve injuries are uncommon, that they most often represent a neurapraxia, and that the axillary nerve is the most likely to be injured.⁷⁸ They are of the opinion that if the initial lesion is partial and improving, observation is indicated. If there is a suspicion that the nerve was lacerated at surgery and electromyography at 6 weeks reveals a complete lesion with no improvement at 12 weeks, exploration and surgical repair are suggested. Wirth and Rockwood were able to identify 14 reported nerve injuries following total shoulder replacement.¹¹⁹ Again, they felt most of the injuries represented a neurapraxia with nonoperative treatment being appropriate. Six lesions involved the axillary nerve, three the ulnar nerve, two the musculocutaneous nerve, and one the median nerve, and two were a more general brachial plexus injury. Resolution was complete in seven, was incomplete in two, and did not occur in one, and in four the extent of recovery was not defined. Importantly, in two shoulders there was a laceration of the axillary nerve occurring in a heavily scarred operative field. Soghikian and Neviaser¹⁰⁰ presented a thorough discussion of complications of hemiarthroplasty, and similarly Muldoon and Cofield⁸⁰ presented material on complications of hemiarthroplasty for proximal humeral fractures; in neither of these reviews was nerve injury identified as a complication.

Lynch et al. reported the Mayo Clinic experience with neurologic complications after TSA.⁷² Four hundred seventeen arthroplasties were studied. Seventeen patients with 18 operated shoulders had a neurologic deficit after surgery (4.3%). All appeared to be traction injuries; 13 involved the brachial plexus. The upper and middle trunk were involved in six, the upper trunk in three, the lateral cord in two, the lower trunk in one, and all trunks in one. Interestingly, three were thought to represent the initiation of an idiopathic brachial neuritis. One patient with dysesthesias after earlier radiation therapy had an increase in the level of dysesthesias in the lower trunk and one patient developed median neuropathy at the wrist. The quality of recovery in the first 16 patients was graded as good in 11 and fair in 5. Time to recovery was less than 3 months in eight, 3 to 6 months in four, 6 to 12 months in one, and greater than 12 months in three. Numerous patient factors were studied including diagnosis, age, sex, height, weight, use of corticosteroids, the presence of diabetes mellitus, preoperative range of motion, the presence or absence of rotator cuff disease, previous surgery, and the use of interscalene block. None of these was found to be related to a nerve injury. However, exposure through the slightly more demanding deltopectoral approach (P = 0.003) and the use of methotrexate in patients with RA (P < 0.0001) were statistically associated with the development of a postoperative nerve palsy. Thus, this series of a large number of shoulder arthroplasties defines that brachial plexus stretch injuries are by far the most frequently recognized neurologic deficits following prosthetic shoulder arthroplasty, that when searched for nerve injuries are more common than has been recognized, and thankfully that recovery is the rule, without significant compromise to the arthroplasty per se.

Intraoperative nerve monitoring was used in 30 patients having shoulder arthroplasty.⁸¹ Compromise of nerve function was signaled by sustained electromyographic activity or greater than 50% amplitude attenuation of transcranial motor evoked potentials. Seventeen had episodes of nerve dysfunction. Most returned to baseline after repositioning the arm to neutral position. None returned to baseline with retractor removal. Nerve involvement by order of frequency was mixed plexopathy, musculocutaneous, axillary, ulnar, and radial. Increased nerve dysfunction was associated with previous surgery and limited passive external rotation.

As identified by the above information, prevention of nerve injuries at surgery is usually, but not always, possible. Certainly, locating the axillary nerve at the inferior aspect of the subscapularis and near the posterolateral aspect of the humerus is useful, and careful retraction of the conjoined group is important. Dissection on the undersurface of the superior and posterior aspects of the rotator cuff should not extend more than 1 cm medial to the glenoid rim, to avoid injury to the suprascapular nerve. Positioning of the arm in extension and abduction with external rotation should be limited in extent and time as much as possible. To facilitate all of these protective measures, it is important to have a dry operative field and to be especially cautious when there is distortion of anatomy such as following an old fracture or fracture-dislocation.

Concerning treatment, definition of the injury by careful physical examination is important. If there is weakness in hand, wrist, or elbow function, splinting may be necessary. Swelling should be minimized by elevation and use of compressive dressings. If active motion is not possible for the hand, wrist, or elbow, passive motion should be used. Passive motion of the shoulder should be commenced during the early postoperative period (within the limits determined at surgery) with active assisted motion initiated as the return of strength will allow.

If there is no improvement in neurologic function by 4 to 6 weeks, electromyography with nerve conduction should be performed to determine more precisely the localization and extent of the nerve injury. If the lesions are diffuse and incomplete, one would suspect a brachial plexus stretch-type lesion and conservative measures would continue. If a focal complete nerve injury is identified such as to the axillary nerve, one would be more concerned about a significant adverse intraoperative event. Quite likely in this setting, continued observation would occur. Further examination would be performed at 3 months. If there was no apparent recovery, electromyography would be repeated at that time and more serious consideration would be given to operative intervention to address the isolated nerve lesion. There is an important caveat, however: So few nerve lesions have occurred due to trauma to a specific nerve that it is hard to be concrete about the recommendations for surgical exploration, other than those indications that apply in general to focal peripheral nerve injuries associated with surgical intervention.

Periprosthetic Fractures

Fractures can occur both intraoperatively and postoperatively. Intraoperatively, the humeral shaft fractures are most common. Fractures can also involve the proximal humeral metaphysis,
the humeral tuberosities, the glenoid, and the coracoid process. Postoperatively, the humeral shaft fractures are also most common. Fractures can involve other areas, including the acromion process and the coracoid process. The recognition of a fracture intraoperatively may be obvious, but sometimes undisplaced cracks develop in the bone that are only detected on subsequent radiographs. Postoperative humeral shaft fractures have been classified by Wright and Cofield.¹²⁰ Type A extends proximally from the tip of the prosthesis and may create stem loosening (Fig. 22-1). A type B fracture is centered at the tip of the stem with minimal to no proximal extension (Fig. 22-2), and the type C fracture involves the humeral shaft distal to the tip of the prosthesis and usually includes fracturing into the distal humeral metaphysis (Fig. 22-3).

Only gold members can continue reading. Log In or Register to continue