 |
 |
 |
|
| home | CME | help | feedback | subscriptions | archive |
|
© 1996 The Journal of Bone and Joint Surgery, Inc.
Current Concepts Review - Complications of Total Shoulder-Replacement Arthroplasty* , SAN ANTONIO, TEXAS*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
Complications associated with orthopaedic prostheses or implants account for approximately 5 per cent of the more than 3.5 million hospitalizations for musculoskeletal conditions82. This is important in light of the increasing number of patients who receive prostheses and the fact that the average age of patients who have a total shoulder arthroplasty is the lowest among those for all major joint replacements28,73. Like replacement procedures in other major joints, total shoulder arthroplasty is associated with numerous complications, including prosthetic loosening, glenohumeral instability, tears of the rotator cuff, periprosthetic fracture, infection, neural injury, and dysfunction of the deltoid. In recent years, mirroring the increase in the number of total joint arthroplasties in general, the number of shoulder-replacement procedures has increased substantially28,71-73. However, despite this growth, fewer than 5000 total shoulder replacements were performed annually in the United States from 1990 through 1992, compared with 136,000 total hip and total knee arthroplasties28,71-73. Although the short-term (less than two-year) and mid-term (two to five-year) results of total shoulder arthroplasties have been encouraging, with some authors reporting good and excellent results in more than 90 per cent of shoulders, widespread experience and long-term evaluations approaching those for joint replacements in the lower extremities have not been published1,5,6,13,23,44,53,64,65,78. Our review of forty-one series involving 1858 total shoulder arthroplasties reported from 1975 through 1995 revealed an average duration of follow-up of only 3.5 years3-6,12,13,15,16,19,25,26,32,34,38,40-42,44,48,53,56-59,62,64,65,67,69,76,77,79,81,84-87,92,98,100,101. Of the twenty-one reports in which the minimum duration of follow-up was two years, only five (391 shoulders) had an average duration of follow-up of five years or more6,13,15,58,98. It has been suggested that the duration of follow-up must be sufficient to allow for the assessment of all clinically relevant outcomes, including those that may occur long after the therapeutic intervention46. An inappropriately short follow-up may result in a failure to detect a potentially important difference in prosthetic survival or the long-term rate of complications of different methods of treatment88. In the context of this concern, we agree with Neer and Kirby76 that an average duration of follow-up of three years is not sufficient to assess many of the complications associated with total shoulder arthroplasty.
Péan, a French surgeon, is credited63 with performing the first total shoulder replacement in 1893, more than twenty-five years before the first attempt at endoprosthetic replacement of the hip96. Péan implanted the device into a thirty-seven-year-old baker who had tuberculous arthritis of the shoulder. The implant functioned reasonably well for two years until recurrent infection necessitated its removal63.
Since the 1970's, several constrained total shoulder prostheses have been employed in the treatment of various disorders of the shoulder. The development of these shoulder prostheses resulted from the commonly held view that glenohumeral instability would be a natural sequela of an ineffective, attenuated, or torn rotator cuff. This incomplete understanding of the kinematics of the shoulder led to a variety of non-anatomical shoulder prostheses, which were designed both to replace the arthritic joint and to restore the stability that was presumably lost as a result of an abnormality of the rotator cuff. Most constrained shoulder implants are of a fixed-fulcrum or a semifulcrum concentric ball-and-socket design. With the fixed-fulcrum implant, constraint is maximized by the coupling of the humeral and scapular components, while the ball-and-socket prosthesis is somewhat less restricted. The basic structure of the latter design emulates a total hip prosthesis in that it provides simple static congruity and architectural stability. Unfortunately, the constrained nature of these devices in a joint that, when normal, affords a nearly unlimited range of motion is most likely the cause of a number of the complications related to total shoulder arthroplasty. Since their introduction, constrained total shoulder prostheses have had only limited clinical success and have been associated with more complications than have unconstrained implants. The rate of complications in thirteen series of constrained total shoulder arthroplasties reported on between 1975 and 1992 ranged from 8 per cent (six of seventy-one58) to 100 per cent (of twenty-six84)15,26,32,41,57,59,62,79,85-87. Moreover, the duration of follow-up was not specified or was less than one year in more than half of the studies. Only three reports included a minimum duration of follow-up of two years15,58,87. Reoperation is common after constrained total shoulder arthroplasty, with reported rates ranging from 4 per cent (three of seventy-one58) to 54 per cent (thirteen of twenty-four85)15,32,57,59,62,84. Complications have largely been the result of biomechanical considerations, and most have been attributed to three factors: mechanical loosening; instability; and failure of the implant secondary to plastic deformation, fracture, or dissociation of the components. Although these three factors have been the most common complications noted in the literature, other complications, such as infection32,59,79, neurovascular injury41,84, ankylosis58, and periprosthetic fracture15,57-59,62,79,84-86, have also been described. In 1980, Post et al.86 reported their early experience with a constrained shoulder prosthesis. A large number of the original implants had been complicated by material failure such as bending or breakage of the humeral neck. These complications led to a redesign of the prosthesis, which greatly decreased the prevalence of plastic deformation and fracture of the implant. Seven years later, Post84 reported the results of an extended follow-up of fifty implants from the original series, at an average of 103 months (range, seventy-eight to 108 months) for the prototype prostheses and an average of forty-seven months (range, two to seventy-seven months) for the redesigned prostheses. Numerous complications were noted. Material failures included breakage or plastic deformation of the humeral component (thirty-five shoulders), glenohumeral instability (nineteen shoulders), and mechanical loosening (fifteen shoulders). Failure of the implant was the most common initial complication, and its prevalence increased from 33 per cent (eight of twenty-four implants) to 48 per cent (twenty-four of fifty implants) with increased follow-up84-87. Modifications of the implant resulted in a substantial decrease in the number of prosthetic failures but led to an increase in prosthetic dissociation and aseptic loosening. In summary, our review of the literature suggests that complications, such as mechanical impingement, aseptic loosening, instability, and failure of the implant, are common after constrained shoulder replacements, and the rates of reoperation are unacceptably high. An ever-expanding knowledge of the anatomy and biomechanics of the shoulder suggests that these failures reflect not only an error in design rationale but also an underestimation of the forces involved in glenohumeral kinematics. Biomechanical studies by Inman et al.55 and by Poppen and Walker82,83 have demonstrated that reactive forces in the glenohumeral joint approximate body weight during unrestricted active elevation of the shoulder. Clearly, the forces acting across the glenohumeral joint are of substantial magnitude and must be considered in the design of shoulder prostheses if certain complications are to be avoided. Currently, most surgeons who have experience with constrained total shoulder-replacement arthroplasty use it as a salvage operation. Indications for this procedure are limited to reconstruction of the shoulder after resection of tumors of the glenoid or the proximal aspect of the humerus, degenerative joint disease associated with a capsular tear and glenohumeral instability, irreparable massive tears of the rotator cuff, and flail shoulders. However, our analysis of the overwhelming number of complications reported in the literature led us to question the efficacy of constrained total shoulder arthroplasty even as a salvage procedure.
Unconstrained total shoulder arthroplasty has been demonstrated to be a highly successful procedure, with good and excellent results reported for most of the shoulders evaluated at the time of early and mid-term follow-up1,5,6,44,53,78. While complications such as aseptic loosening, instability, infection, and periprosthetic fracture still occur despite this success, they are less common than after constrained total shoulder arthroplasty. According to Neer et al.78, four conditions must be considered at the time of unconstrained shoulder arthroplasty if these complications are to be minimized: (1) osseous deficiency of the humeral head or glenoid, (2) a defective rotator cuff, (3) a deficient deltoid muscle secondary to irreparable retraction or injury of the axillary nerve after a previous procedure or injury, and (4) chronic instability. Recognition and effective treatment of these conditions along with careful selection of patients, operative precision, and a fundamental understanding of the anatomy and kinematics of the shoulder minimize the complications associated with this procedure. Although the result after unconstrained shoulder-replacement arthroplasty with use of the system developed by Neer has been good or excellent in most shoulders74,75, long-term follow-up is lacking. Our review of the literature yielded thirty-two reports of unconstrained total arthroplasty involving a total of 1615 shoulders1,3-6,12,13,15,16,19,25,34,38,40,42,44,48,53,56,65,67,69,74,76,77,79,81,92,97,98,100,101. Fourteen reports had a minimum duration of follow-up of less than two years. Of the eighteen studies that had a minimum of two years of follow-up, only eight had an average of more than four years4,6,19,25,42,53,76,77. Although we are aware of the potential shortcomings inherent in an evaluation of complications associated with this duration of follow-up, our analysis revealed the most common complications, in order of frequency, to be loosening of the component, glenohumeral instability, a tear of the rotator cuff, periprosthetic fracture, infection, failure of the implant (including dissociation of modular prostheses), and weakness or dysfunction of the deltoid.
Symptomatic loosening of the glenoid and the humeral component after a total shoulder-replacement arthroplasty is common and represents nearly one-third of all complications that are associated with this operation3-6,12,13,15,16,19,25,34,40-42,44,48,53,56,64,65,67,69,76-79,81,92,97,98,100,101. Most of the cases of clinical and radiographic loosening have involved failure of the fixation of the glenoid component.
Loosening of the Glenoid Component Other investigators have expressed concern that the appearance or progression of radiolucent lines may herald the onset of future problems, such as symptomatic loosening of the component2,5,15,23,25,53,98,102 (Figs. 1-A, 1-B, and 1-C). The authors of one series from Sweden reported radiolucent lines around the glenoid component in twenty-five of twenty-six shoulders at an average of forty-seven months after the operation15. While no radiolucent lines were observed around the glenoid component on the radiographs made immediately postoperatively, lines developed within three years after the operation in twenty-five shoulders. The development of these radiographic findings was associated with a slight decrease in function and a mild increase in pain.
In 1984, Cofield23 reported improved motion and reliable relief of pain in most shoulders at two to six years after seventy-three Neer total shoulder arthroplasties. Although the clinical results were excellent and compared favorably with those of other studies, fifty-two shoulders had radiolucent lines at the bone-cement interface and eight had radiographic evidence of loosening of the component, as defined by a shift in the position of the component or a circumferential radiolucent line at the bone-cement interface that was at least 1.5 millimeters wide. More recently, Torchia and Cofield98 analyzed the results of eighty-nine total shoulder-replacement arthroplasties, some of which had been included in the original report by Cofield23. The clinical results were graded as excellent, satisfactory, or unsatisfactory according to the methods of Neer et al.78 and of Cofield23. An excellent result indicated that the patient had no or slight pain, had active abduction to at least 140 degrees, had external rotation to at least 45 degrees, and was satisfied with the result. A satisfactory result indicated that the patient had no or slight pain or moderate pain only with vigorous activities, had active abduction to at least 90 degrees, had external rotation to at least 20 degrees, and was satisfied with the result. An unsatisfactory result indicated that these criteria were not met. Associations and correlations between preoperative parameters and postoperative results were tested with the Wilcoxon rank-sum test, the Kruskal-Wallis test, and the Spearman correlation test. P values of less than 0.05 were considered significant. At an average of twelve years (range, five to seventeen years), radiolucent lines had developed at the bone-cement interface of seventy-five glenoid components (84 per cent), and thirty-nine components (44 per cent) demonstrated definite radiographic loosening. Comparison of the original23 and later98 reports revealed that radiographic evidence of loosening of the component had increased fourfold during the extended study period. Even more disturbing was the association between radiographic loosening of the glenoid implant and pain (p = 0.0001). With an increasing duration of follow-up, the percentage of patients who reported satisfactory relief of pain declined from 92 to 82. Concern about aseptic loosening of the glenoid component led to a variety of new innovations, including press-fit implants inserted without cement, plasma-sprayed implants, and tissue-ingrowth implants25,30,41,65,67,92. Many of the reports on these innovations were limited by incomplete radiographic review, insufficient follow-up, and inconsistent reporting methods, but some of the preliminary data were encouraging and suggested a lower prevalence of both radiolucent lines and subsequent loosening of the component. Current methods to enhance fixation and durability of the glenoid component include preservation of the subchondral plate, concentric spherical reaming that ensures optimum osseous support of the implant, the introduction of new glenoid designs and enhanced biomaterials, and mismatching of the diameters of the glenoid and humeral head27,54,95,107,109. This latter method involves using a glenoid component with a radius of curvature that is slightly larger than the corresponding radius of curvature of the humeral head. Harryman et al.52 suggested that this mismatch at the articular surface decreases the direct contact between the rim of the glenoid component and the humeral head during glenohumeral translation. This was proposed as a means to protect the fixation of the glenoid prosthesis as loading of the rim may lead to deformation or loosening of the component, or both. Despite considerable success with total shoulder arthroplasty after the independent introduction of several different glenoid components by different authors75, many problems remain unsolved. An inability to resolve some of these issues and the apparent success of hemiarthroplasty have led some investigators to question the indications for glenoid resurfacing5,10,45,102. In 1974, Neer74 reviewed the results of forty-seven shoulder hemiarthroplasties at an average of six years. Twelve patients had relief of pain, functional recovery, and no evidence of progressive degenerative changes or resorption of the glenoid fossa after more than ten years of follow-up. This led Neer to conclude that there is little reason to treat osteoarthrosis with a more extensive replacement that might increase complications and jeopardize the longevity of the prosthesis. Since that report, several studies have documented good and excellent results with hemiprosthetic replacement of the shoulder, but only a few have compared the results of hemiarthroplasty with those of total shoulder arthroplasty in a similar patient population2,12,23. As far as we know, no randomized prospective studies have been done to address this question. Studies have shown that complications are less frequent with hemiarthroplasty than with total arthroplasty, but, with the numbers available, the authors of those studies were not able to demonstrate any significant differences between the two techniques with regard to over-all relief of pain, function, or satisfaction of the patient2,6,12,19,22,56. Several authors have described an association between symptomatic loosening of the glenoid component, glenohumeral instability, and irreparable tears of the rotator cuff5,27,45,53,81. These observations, in conjunction with the finding of a high prevalence of osteolysis at the bone-cement interface of glenoid components, led us to conclude that the current indications for total shoulder arthroplasty need additional refinement.
Loosening of the Humeral Component
Glenohumeral instability has rarely been discussed in the literature in spite of a reported prevalence of 0 per cent42,64,81 to 29 per cent (twenty-six of eighty-nine98)1-3,5,6,13,19,25,41,48,53,65,69,70,76,79,92,101,107,108 and the fact that it is the second leading cause of complications associated with prosthetic arthroplasty of the shoulder. In the normal glenohumeral joint, stability is provided by a hierarchy of mechanisms that work together to ensure a substantial range of motion and optimum function. Small loads are offset by passive constraints, such as the architecture of the joint surface14,60, finite joint volume68,104, atmospheric pressures60, and the adhesive and cohesive properties of joint fluid. Moderate loads are counterbalanced by the deltoid and rotator cuff muscles, whose coordinated contractions resist displacing forces, and large loads are resisted by the capsulolabral structures and osseous architecture68,99,104,108. Unfortunately, shoulder arthroplasty can alter these complex interactions, thereby increasing the propensity for instability.
Anterior Instability In our study of operative exploration of three total shoulder replacements with anterior instability, we found decreased retroversion of the humeral component (of 20 degrees or less) in all three shoulders, with disruption of the subscapularis in two and erosion of the anterior aspect of the glenoid in one106. One of these patients had a massive irreparable tear of the rotator cuff and had had a previous resection of the distal aspect of the clavicle, a resection of the coracoacromial ligament, and an acromioplasty that had rendered the coracoacromial arch an incompetent restraint to anterosuperior glenohumeral translation. These shoulders were revised with different techniques, including restoration of normal version of the humeral component, reconstruction of the coracoacromial ligament, and transfer of the pectoralis major tendon. We have found that clinically obvious anterior subluxation or dislocation usually does not occur unless the subscapularis tendon is torn or a previous operation has compromised the stabilizing effect of an intact coracoacromial arch. Disruption of the subscapularis repair is usually attributed to poor operative technique, tissue of poor quality, inappropriate physical therapy, or the use of oversized components. A thick metal-backed glenoid component or a humeral implant with an excessively long neck displaces the geometric center of the humeral head laterally, which stresses the suture line and repair process in the subscapularis tendon.
Superior Instability Although superior migration of the humeral head is a well recognized complication of shoulder hemiarthroplasty, it does not appear to be directly related to the development of discomfort in the shoulder or to impending failure after this operation. Boyd et al.12 noted no increase in pain with proximal migration of the humeral head, and this has been our observation as well. However, the potential complication of loosening of the glenoid component as a result of progressive superior migration of the humeral component after total shoulder arthroplasty is more of a problem. In this setting, the humeral head articulates with the superior portion of the glenoid component and results in eccentrically applied glenoid compressive forces, increased stress at the bone-cement interface, and eventual loosening of the glenoid component within the glenoid fossa5,43.
Posterior Instability Our experience with posterior glenohumeral instability after shoulder arthroplasty includes seven shoulders106,109 (Figs. 2-A, 2-B, and 2-C). Objective findings in these shoulders included increased retroversion of the humeral component (of 80 degrees or more) in four shoulders, erosion of the posterior aspect of the glenoid in four shoulders, and a non-union of the greater tuberosity. These shoulders were revised by restoration of normal retroversion of the humeral component, sculpting or reaming of the glenoid to re-establish proper glenoid version, and posterior capsulorrhaphy to address the asymmetrical soft-tissue abnormality.
Inferior Instability
Postoperative tearing of the rotator cuff is the third most frequent complication of total shoulder arthroplasty, with a prevalence of 2 per cent (fifteen of 664 patients)1,5,6,13,42,53,56,101,109. Although this entity is considered common after shoulder replacement, associated substantial pain has not been reported in many series6,23,56,74. Both non-operative and operative treatment have been employed in the treatment of this complication, but the benefits of operative intervention are somewhat unclear as recurrent tearing of the cuff and little improvement in function and motion have been noted in a number of patients1,5,6,13,42,53,56,101.
Periprosthetic fractures have a reported prevalence of 3 per cent (twelve of 373)1,3,5,12,40,79 and account for approximately 20 per cent of all complications associated with total shoulder arthroplasty103. On the basis of our review of fourteen reports describing fractures of the glenoid or humerus associated with prosthetic replacement of the shoulder, most of these injuries involve the humeral shaft and occur at the time of the operation1,3,5,8,11,12,41,50,53,56,64,76,79,98.
Intraoperative Fractures Groh et al.50 reported on a series of twelve humeral fractures, eight of which had occurred intraoperatively and four of which had been the result of postoperative trauma. Of the eight intraoperative humeral fractures, six occurred during a primary shoulder arthroplasty and two, during a revision procedure. The fracture occurred during manipulation of the limb in two of the primary arthroplasties, reaming of the intramedullary canal in one, broaching of the canal in one, and insertion of the prosthesis in two. Both of the fractures that occurred during revision arthroplasties were in an area of moderate-to-severe cortical thinning. The intraoperative fractures were treated with open reduction and internal fixation with simple cerclage wiring or a long-stem prosthesis in conjunction with cerclage wiring. When we use the latter technique, we prefer to use a prosthesis that extends at least two humeral cortical diameters beyond the most distal extent of the fracture (Fig. 3). This is accomplished by extending the deltopectoral incision into an extensile anterolateral approach to the humerus35. The relatively straight, cylindrical anatomy of the humeral diaphysis is ideally suited for this method of intramedullary fixation. There are several advantages to this form of treatment compared with use of a dynamic compression plate or cerclage wiring alone. First, the need for secure purchase of the screw in bone that is often of poor quality is obviated. Second, bending and torsional loads are better tolerated, which decreases the risk of failure of the implant. Third, a rigid and biomechanically sound operative construct is usually ensured. Fourth, the extensile exposure and soft-tissue dissection needed for fixation of a plate is avoided and, finally, stress-shielding is minimized.
Intraoperative fractures of the glenoid also occur during shoulder arthroplasty. It is important to remember that stability of the glenoid component is affected by the preparation of the glenoid, soft-tissue balancing, and the availability of good-quality glenoid bone for prosthetic fixation108. A scapular fracture adjacent to a glenoid component may compromise the stability of the implant and lead to symptomatic loosening. To treat this complication, bone-grafting or a revision glenoid component that is reinforced with a wedge to accommodate the defect can be employed; however, if osseous support cannot be ensured, resurfacing of the glenoid should not be performed. In this situation, the glenoid can be shaped with a hand burr or a glenoid reamer to match the radius of curvature of the head of the humeral component.
Postoperative Fractures Current concepts regarding the treatment of periprosthetic fractures in the shoulder are divided into two seemingly divergent schools of thought. Bonutti and Hawkins8 advocated intensive treatment of these injuries, which included open reduction and internal fixation, bone-grafting, and postoperative immobilization in a spica cast for a minimum of six weeks. Similarly, Boyd et al.11 reported an increased likelihood of non-union after non-operative treatment and believed that fractures that are treated operatively have a better result. Four of the fractures in the series of twelve periprosthetic humeral fractures described by Groh et al.50 were postoperative, occurring at an average of fourteen months after the index arthroplasty. It is noteworthy that three of these four fractures extended distal to the tip of the prosthesis. All twelve fractures were treated with an Orthoplast fracture brace, isometric exercises, and early range-of-motion rehabilitation. At an average of nine weeks after the injury, all of the fractures had progressed to osseous union and clinical evaluation demonstrated an average forward elevation of 121 degrees. On the basis of this experience, we believe that a more conservative approach can yield successful results in a number of patients.
Infection after total shoulder arthroplasty is a rare but potentially devastating complication. In most cases, an increased susceptibility to infection is associated with host-related risk factors such as diabetes mellitus, rheumatoid arthritis, systemic lupus erythematosus, remote sites of infection, and previous operations on the shoulder15,20,25,41,42,48,61,69,76,79,92,98. Additionally, extrinsic factors such as immunosuppressive chemotherapy, systemic administration of corticosteroids, and multiple injections of steroids have been noted as risk factors for infection about the glenohumeral joint after total shoulder arthroplasty20,61. Although it is only speculation, the apparently increased risk of infection after local injection of steroids may be attributed to the unique bursal anatomy of the shoulder. Inadvertent or purposeful injection into the subdeltoid, subscapular, or infraspinatus bursa or the sheath of the long head of the biceps tendon may provide an avenue for intra-articular bacterial invasion47. The value of preoperative laboratory tests (such as a white blood-cell count and determination of the erythrocyte sedimentation rate and the level of C-reactive protein), radioisotope scanning, and aspiration of the joint is difficult to ascertain because of rare and inconsistent methods of reporting. In the series of Codd et al.20, eighteen patients who had an infection after a shoulder arthroplasty had an average preoperative erythrocyte sedimentation rate of forty-nine millimeters per hour and an average white blood-cell count of 11,020 per cubic millimeter (11.0 x 109 per liter). Eleven patients had radioisotope scans, of which seven were interpreted as positive, two as negative, and two as equivocal. Fluid was aspirated from seven shoulders, and the fluid from two was found to be positive on culture. Six patients have been managed at our institution for an infection around a shoulder prosthesis. The average preoperative erythrocyte sedimentation rate (seventy-five millimeters per hour) and white blood-cell count (11,980 per cubic millimeter [12.0 x 109 per liter]) were similar to those of Codd et al.20. As with other joint replacements, infections can occur early or late and optimum treatment depends on isolation of the pathogen from tissue or fluid specimens. Once the diagnosis has been made, there are several options for treatment, including antibiotic suppression, irrigation and débridement, removal of the prosthesis and reimplantation, resection arthroplasty, arthrodesis, and amputation. The type of treatment depends on several factors, such as the time from the arthroplasty to the diagnosis of the infection, the feasibility of removal of the implant as dictated by anesthetic risk, the pathogen's virulence and susceptibility to antibiotics, and the stability of the implant. The differential diagnosis of a draining wound in the early postoperative period after shoulder replacement must include early infection despite negative results on culture of aspirate or drainage and the lack of obvious symptoms such as fever. In this situation, we advocate early exploration of the wound with irrigation and débridement and judicious parenteral administration of antibiotics. In our early experience, most infections around shoulder implants were treated with resection arthroplasty because of our concern about recurrent infection and the paucity of reports substantiating the successful salvage of a functional prosthesis. More recently, we have favored retention of the prosthesis if the infection was caused by a gram-positive organism and the components are stable. This recommendation is based on the experience of some authors with the treatment of infection around total knee replacements49. For early infections with gram-negative organisms and for late deep-wound infections, we advise thorough débridement of granulation and scar tissue and the removal of all biomaterials, including cement. This is followed by a six-week course of parenterally administered antibiotics as advised by an infectious disease expert. At the end of this period, reimplantation is considered if it is not contraindicated by local tissue conditions, persistently positive results on culture, or other factors portending a poor prognosis for exchange arthroplasty. While we are not aware of any published data pertaining to infections around total shoulder prostheses to support these recommendations, they have been applied to the treatment of infections around total joint replacements in the lower extremity with acceptable results9,17,39,49,66,94.
We found reports of fourteen shoulders that had a complication involving the peripheral nerve or the brachial plexus following total shoulder-replacement arthroplasty3,5,12,22,42,44,48,64,81,98. Most of these injuries involved a neurapraxia, and non-operative treatment yielded good results. However, in two shoulders, the mechanism of neural injury involved a laceration of the axillary nerve, which occurred in a heavily scarred operative field22. While most of these complications involved the axillary nerve (six shoulders), injury of the ulnar nerve (three shoulders), the musculocutaneous nerve (two shoulders), the median nerve (one shoulder), and the brachial plexus (two shoulders) was also reported. Seven of these injuries resolved completely, two resolved incompletely, one did not resolve at all, and the status of four was not mentioned.
Occasionally, failure of a shoulder replacement is related to the implant. Dissociation of the polyethylene glenoid insert from its metal tray25,37,65, fracture of the keel or metal glenoid backing25,67, fracture of the fixation screws65,67, and subluxation or dislocation of polyethylene subacromial spacers13,19 have all been reported. Since their introduction in the late 1980's, modular total shoulder arthroplasty systems have been used widely because they permit the surgeon to more easily optimize soft-tissue balancing, maximize fit with regard to the proximal humeral metaphysis and the humeral canal, facilitate revision of the glenoid, and permit the conversion of a hemiarthroplasty to a total shoulder arthroplasty. Although some of the problems related to the implant in total shoulder replacement have been eliminated by improvements in the prosthetic design, the fixation of the components, and the biomaterials used, new problems have arisen that are unique to modular shoulder implants. Cooper and Brems29 reported on a patient who had recurrent dissociation of a modular humeral component. In addition to a similar case that was referred to our institution, we are aware of eleven more cases of dissociation of the humeral component7. All but one of these occurred within six weeks after the index arthroplasty. Unlike previous case reports involving disassembly of modular femoral components93,110, none of the dissociations of the shoulder components were associated with dislocation of the joint or subsequent reduction maneuvers. In fact, several of the patients in the report by Blevins et al.7 were unsure when the dissociation had occurred. Similarly, our patient was unaware of a specific event that led to disassembly of the humeral prosthesis but had noted discomfort in association with dysfunction of the shoulder since the time of the operation. At the time of the revision procedure, the posterior aspect of the humeral collar had subsided in the metaphyseal bone of the proximal aspect of the humerus, and this most likely had precluded adequate seating of the Morse taper components during the initial arthroplasty.
Prosthetic replacement of the shoulder joint requires familiarity with its anatomy and biomechanics. Maintenance of the origin and insertion of the deltoid muscle is critical to the final outcome of this procedure. Loss of function of the deltoid muscle as a result of a failed operative repair or of injury of the axillary nerve may result in a major loss of function of the shoulder4,51,76,78,101,105. Although optimum operative exposure is important, it is also important to note that the procedure can be performed through an extended deltopectoral approach without detachment of the origin or insertion of the deltoid muscle90. Three different exposures were used for the 194 total shoulder replacements reported on by Neer et al.78. Initially, a short deltopectoral approach was employed that involved detachment of the anterior portion of the deltoid muscle from the clavicle. The second approach was more superior and involved detachment of the middle portion of the deltoid muscle. Use of these exposures was discontinued because the first was found to weaken the anterior portion of the deltoid muscle and the second, to weaken the middle portion. The third approach involved a long deltopectoral interval and preserved the origin and insertion of the deltoid muscle. Neer and Kirby76 identified detachment of the deltoid muscle as a major cause of muscle atrophy leading to failure of the unconstrained shoulder arthroplasty (Fig. 4). This complication resulted in consistent weakening and atrophy of the deltoid muscle in all but three of thirty-seven patients. Those authors also noted denervation of the anterior aspect of the deltoid muscle after a previous operation on the shoulder in five patients. Over-all, severely compromised function of the deltoid was observed in association with thirty-four (92 per cent) of thirty-seven failed unconstrained total shoulder and humeral head arthroplasties.
The senior one of us (C. A. R., Jr.) has managed thirty-six patients who had a loss of function of the deltoid muscle after operations on the shoulder, including endoprosthetic replacement51. Of twelve patients who had endoprosthetic replacement, ten lost function of the anterior or the anterior and middle aspects of the deltoid secondary to a failed operative repair and two lost all function of the deltoid muscle as a result of denervation of the axillary nerve. All patients were substantially disabled, with an average forward elevation of only 33 degrees (range, 0 to 75 degrees). Additionally, several patients had painful anterosuperior dislocation of the glenohumeral joint with attempted elevation of the upper extremity.
The number of shoulder-replacement procedures performed in the United States has increased substantially in the past decade71-73. Although the widespread use of unconstrained total shoulder prostheses has resulted in a marked decrease in complications, difficulties necessitating a reoperation have been reported by many investigators4,12,13,15,22,25,34,41-43,48,53,56,65,67,76,81,92,97,100,101. Although many of these operations have consisted of soft-tissue procedures for postoperative complications such as tears of the rotator cuff and glenohumeral instability, nearly half have involved removal or revision of prosthetic components. In a report on forty revision arthroplasties by Neer and Kirby76, the cause of failure of the index arthroplasty was assigned to one of three groups: (1) preoperative conditions, such as neuromuscular problems, infection, or arthritis of adjacent joints; (2) operative or prosthetic complications, including detachment of the deltoid muscle, non-union of the greater tuberosity, or loosening or breakage of the component; and (3) postoperative problems, such as residual or recurrent instability or inadequate rehabilitation. The most common causes of failure included scarring and detachment of the deltoid muscle, loss of external rotation due to contracture of the subscapularis muscle, prominence or retraction of the greater tuberosity, loss of glenoid bone, and inadequate postoperative rehabilitation. Twenty-seven (68 per cent) of the forty revisions involved a prosthesis that had been originally inserted for the treatment of a displaced fracture or a fracture-dislocation of the proximal aspect of the humerus. The degree of operative difficulty in those shoulders was increased by muscle contracture and scarring, associated malunion or non-union of the greater tuberosity, and loss of bone with shortening of the humeral shaft. The results of most of these revisions were reported in another study and were found to be inferior to those of primary shoulder arthroplasty78. Caldwell et al.18 reviewed the results of thirteen revision arthroplasties at an average of thirty-six months. Two total shoulder replacements and one hemiarthroplasty were revised because of glenohumeral instability. Seven hemiarthroplasties were revised to total arthroplasties because of glenoid arthropathy, and three total shoulder arthroplasties were revised because of loosening of the glenoid component. The result was considered satisfactory in only eight shoulders, and five shoulders needed seven reoperations. Loosening of the glenoid component and incorrect version were regarded as the most frequent reasons for revision. We reviewed the data concerning thirty-eight failed unconstrained shoulder arthroplasties109 that were revised at our institution between 1977 and 1993. The indication for the initial arthroplasty had been acute trauma in nineteen shoulders, osteoarthrosis in twelve, arthropathy after reconstruction in five, and rheumatoid arthritis in two. Five patients had had eight previous attempts at revision arthroplasty. Our analysis revealed findings that were similar to those of Neer and Kirby76 in that failure was often multifactorial—that is, it was usually difficult to associate failure with one specific factor. The most common complication leading to a revision was symptomatic glenohumeral instability (42 per cent). Other causes of failure, in order of decreasing frequency, included detachment of the anterior aspect of the deltoid muscle, loosening of the glenoid component, erosion of the glenoid after hemiarthroplasty, loosening of the humeral component, detachment or malunion of the greater tuberosity, osseous or fibrous ankylosis, infection about the glenohumeral joint, and dissociation of a modular humeral component.
The validity of survivorship analysis in the evaluation of long-term clinical studies involving total hip replacements is well established31,36,80. Although non-parametric estimates of survivorship based on life tables and the Kaplan-Meier curve have proved useful for predicting the longevity of hip prostheses, the application of these instruments to studies of total shoulder arthroplasty is limited to two series13,22. Cofield performed a non-parametric estimation of the survival of 176 unconstrained total shoulder prostheses and predicated a 9.6 per cent cumulative probability of failure at five years22. The criterion for failure was defined as the need for a major reoperation, which was performed in eight shoulders (5 per cent). The indication for reoperation included early dislocation of three shoulders, loosening of the glenoid component in three shoulders, muscle transfer for paralysis of the axillary nerve in one shoulder, and resection arthroplasty for infection in one shoulder. In a more recent article by Torchia and Cofield, the cumulative probability of failure had not changed substantially at an eleven-year end point98. Brenner et al.13 analyzed the results of fifty-three unconstrained total shoulder arthroplasties with use of the Kaplan-Meier survivorship curve. Employing a more rigid definition of failure, which was the need for reoperation as well as dissatisfaction of the patient, the authors reported an eleven-year survival rate of only 73 per cent. In a large multicenter prospective study involving more than 470 unconstrained total shoulder arthroplasties, the five-year survival was estimated at 97 per cent (95 per cent confidence interval)90. A more rigid definition of failure, similar to that proposed by Brenner et al.13, was applied to a subset of patients for whom the diagnosis was restricted to osteoarthrosis. For those patients, failure was defined by one of two criteria. The first criterion, as in the studies by Cofield22 as well Brennan et al.13, was simply the need for a reoperation after the index procedure. The second criterion was based on the patient's subjective assessment of pain with use of a visual analog scale. For this analysis, failure was defined as the point in time at which the patient reported pain in the shoulder that was equal to or worse than the preoperative condition. The probability of five-year survival in the osteoarthrosis subgroup was 92 per cent with use of the more stringent criteria.
Codman was one of the earliest orthopaedists to study and report on disorders of the shoulder, but few except the most ardent of his followers are familiar with the role that he played in describing outcomes research for the shoulder21. Central to what Codman described in the early 1900's as an end-result system was the realization that complications are an inescapable aspect of an operation with a substantial impact on the patient's over-all health, function, and quality of life. Despite Codman's admonition more than sixty years ago, there is still a lack of standardized methods for the measurement of results and the reporting of complications associated with total shoulder implants. Unfortunately, the lack of a universally accepted outcome-measurement system for shoulder arthroplasty increases methodological flaws in structured reviews of the literature and often precludes meaningful retrospective or prospective comparisons between various series. More recently, the necessity for improved study designs, the definition of the important constituents of outcome measurement, and an increase in the validity of orthopaedic clinical research have been emphasized by several authors33,46,93. It has been suggested that the current emphasis of orthopaedic clinical studies should be directed toward outcome research that documents the effect of the treatment on the health of the patients and the subsequent quality of their lives46. Ultimately, this will facilitate the analysis of multicenter studies, permit validity testing of measurement tools, and provide documentation of patient outcome in terms of economics and improved quality of life.
Department of Orthopaedics, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, Texas 78284-7774.
  CiteULike  Connotea  Del.icio.us  Facebook  Technorati  Twitter What's this?
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Introduction
Historical Note
| home | CME | help | feedback | subscriptions | archive | site index |
|
Stanford University Libraries' HighWire Press (TM) assists in the publication of JBJS Online. |