Abstract
Background: Currently, there is little information available on the treatment and outcome of intraoperative periprosthetic humeral fractures that occur during shoulder arthroplasty. The purpose of this study was to report on the incidence, treatment, and outcome of, as well as the risk factors for, intraoperative periprosthetic humeral fractures.
Methods: Between 1980 and 2002, forty-five intraoperative periprosthetic humeral fractures occurred during shoulder arthroplasty at our institution. Twenty-eight fractures occurred during primary total shoulder arthroplasty, three occurred during primary hemiarthroplasty, and fourteen occurred during revision arthroplasty. Nineteen fractures involved the greater tuberosity, sixteen involved the humeral shaft, six involved the metaphysis, three involved the greater tuberosity and the humeral shaft, and one involved both the greater and lesser tuberosities. All patients were followed for a minimum of two years. At the time of the latest follow-up, outcomes were assessed, radiographs were examined, and relative risks were calculated.
Results: Over the twenty-two-year study period, the rate of intraoperative humeral fractures at our institution was 1.5%. All fractures healed at a mean of seventeen weeks. In the primary arthroplasty group (thirty-one patients), range of motion and pain scores improved significantly (p < 0.05) at the time of follow-up. In the revision arthroplasty group (fourteen patients), range of motion remained unchanged whereas pain scores improved significantly (p < 0.005). Transient nerve injuries occurred in six patients. Four fractures displaced postoperatively and were then treated nonoperatively; all four healed. Significant relative risks for intraoperative fracture were female sex, revision surgery, and press-fit implants (p < 0.05).
Conclusions: The data from the present study suggest that although intraoperative humeral fractures are associated with a high rate of healing, there was a substantial rate of associated complications, including transient nerve injuries and fracture displacement. Significant risk factors for intraoperative fractures include female sex, revision surgery, and press-fit humeral implants.
Level of Evidence: Therapeutic Level IV. See Instructions to Authors for a complete description of levels of evidence.
Intraoperative periprosthetic fractures of the humerus are uncommon. The literature provides little direction regarding the incidence, treatment, and outcomes of this complication. When these fractures occur, operative time is increased, a change of implant or the addition of internal fixation or bone graft may be required, and postoperative rehabilitation usually is affected. The literature has also indicated that fracture-healing may be difficult1-5.
The literature includes several reports on periprosthetic humeral fractures; however, most reports have dealt exclusively with postoperative fractures1,5-7. One report dealt with mixed intraoperative and postoperative periprosthetic fractures3, but no report has dealt exclusively with isolated intraoperative fractures.
The purpose of the present study was to review the incidence, etiology, treatment, and outcomes of, as well as the risk factors for, intraoperative humeral fractures during shoulder arthroplasty.
Patients
Between 1980 and 2002, 3088 shoulder arthroplasties (including 2666 primary procedures and 422 revision procedures) were performed at our institution. The average age of the patients at the time of arthroplasty was sixty-five years (range, thirty-six to eighty-one years).
A cohort of forty-five patients with forty-five intraoperative periprosthetic humeral fractures was identified. The inclusion criteria were the documentation of an intraoperative humeral fracture during shoulder arthroplasty, skeletal maturity, and the ability of the patient to comprehend and complete outcome questionnaires. The exclusion criteria were the occurrence of a fracture during a shoulder arthroplasty that was performed because of acute trauma or tumor and an unwillingness of the patient to be followed through the arthroplasty database. The study group included thirty-five women and ten men. Twenty-eight fractures occurred during primary total shoulder arthroplasty, three occurred during primary hemiarthroplasty, and fourteen occurred during revision shoulder arthroplasty. Nineteen fractures involved the greater tuberosity, sixteen involved the humeral shaft, six involved the metaphysis, three involved the greater tuberosity with an extension into the humeral shaft, and one involved both the greater and lesser tuberosities. The surgical procedures during which the intraoperative fractures occurred were carried out by six different orthopaedic surgeons (including two of the authors [J.W.S. and R.H.C.]), and forty-two (93%) of the fractures occurred during procedures performed by experienced shoulder arthroplasty surgeons.
The diagnoses in the primary arthroplasty group included osteoarthritis (fifteen patients), posttraumatic arthritis (seven patients), rheumatoid arthritis (six patients), and osteonecrosis (three patients). The reasons for surgery in the revision arthroplasty group included joint instability (four patients), aseptic loosening (four patients, including two with humeral component loosening, one with glenoid component loosening, and one with both humeral and glenoid component loosening), failed hemiarthroplasty for a proximal humeral fracture (three patients), glenoid arthritis after hemiarthroplasty for osteoarthritis (two patients), and second-stage replantation after a resection for implant-related infection (one patient). Comorbidities included diabetes mellitus (seven patients), long-term steroid therapy (six patients), preexisting brachial plexus palsies (three patients), renal failure (two patients), and Parkinson disease (one patient).
At the time of surgery, all shoulders were exposed through a deltopectoral approach. The subscapularis tendon was sharply released from the lesser tuberosity in twenty-five patients, and an intrasubstance subscapularis tenotomy was performed in twenty patients. There were no lesser tuberosity osteotomies. During total shoulder arthroplasty, the humerus was prepared first and the trial implant was left within the humeral canal during glenoid exposure.
Greater Tuberosity Fractures
Of the nineteen greater tuberosity fractures, five occurred during insertion of the humeral component, three occurred during removal of a well-fixed humeral component, and one each occurred during initial dislocation of the humerus, retractor placement for glenoid exposure, broaching of the humeral shaft, reaming of the humeral shaft, trial reduction, and dislocation of the humerus after implantation of the glenoid component (when the greater tuberosity became caught on the posterior glenoid rim). In five cases, the cause of fracture was unknown as the fracture was identified during closure (two), after glenoid implantation (one), during cement placement for the humeral component (one), and during trial humeral component insertion (one).
Of the nineteen greater tuberosity fractures, eleven (including six in the primary arthroplasty group and five in the revision arthroplasty group) were described as displaced; all eleven were treated with reduction and suture fixation, with three undergoing additional bone-grafting. The other eight greater tuberosity fractures (all in the primary arthroplasty group) were described as nondisplaced and stable. One of these fractures was treated with suture fixation, three were treated with bone-grafting, and the remaining four were treated with no specific fixation. Postoperative physiotherapy consisted of a standard protocol8,9 for twelve patients and limited goals therapy10,11 for five; the remaining two patients were managed with an abduction splint for three weeks, followed by initiation of the standard protocol.
Greater Tuberosity and Humeral Shaft Fractures
Three patients had fractures of the greater tuberosity that extended into the proximal part of the humeral shaft. One fracture, which occurred during removal of a well-fixed press-fit humeral component, was treated with revision to a long-stemmed component and suture fixation of the greater tuberosity. Another fracture, which occurred during removal of a well-fixed cemented humeral component, was treated with suture fixation of the tuberosity and bone-grafting of the fracture site. The final fracture occurred during impaction of a press-fit humeral component. Once this fracture had occurred, the humeral implant was unstable; therefore, it was revised to a cemented humeral component after cerclage fixation of the fracture (Figs. 1-A, 1-B, and 1-C). Postoperative physiotherapy consisted of a limited goals protocol10,11 for the first two patients and a standard protocol8,9 for the last patient.
Humeral Shaft Fractures
Sixteen patients sustained an intraoperative humeral shaft fracture. Six patients sustained the fracture during reaming of the humeral canal; five, during insertion of the true humeral component; and one each, during exposure of the glenoid, final joint reduction, removal of a cement restrictor, or removal of a well-fixed cemented humeral component adjacent to a cortical window. In the case of one patient, the cause was unknown as the spiral humeral fracture was identified on the immediate postoperative radiograph.
Six patients were managed with exposure of the fracture, radial nerve identification, insertion of a long-stemmed implant, and the placement of cerclage wires. These six patients were managed postoperatively with a standard physiotherapy protocol. Four of the sixteen patients underwent intraoperative radiographic assessment of the fracture, and the humeral implant was examined carefully for loosening. All four patients were believed to have a stable fracture, and all were managed with limited goals physiotherapy for six weeks. One of the sixteen patients underwent exposure of the fracture with identification of the radial nerve and insertion of a long-stemmed implant without cerclage wires; one underwent fluoroscopically assisted fracture reduction with insertion of a long-stemmed implant to bridge the fracture gap; one was managed with open reduction and internal fixation with plates and screws; one was managed with open reduction, allograft strut fixation with cerclage wires, and a long-stemmed implant; one was managed with open reduction and cerclage wire fixation; and one was managed with a shoulder spica cast for four weeks.
Metaphyseal Fractures
Six patients sustained an intraoperative fracture in the metaphyseal region of the proximal part of the humerus. Two fractures occurred during insertion of a humeral trial component, one occurred during insertion of the true humeral implant, one occurred during extraction of a well-fixed press-fit component during revision surgery, one occurred during retraction of the humerus posteriorly during glenoid preparation, and one occurred when an osteotome was used to remove osteophytes from the inferior part of the humeral neck.
Three of these fractures were treated with autologous bone-grafting of the inner part of the metaphysis, followed by insertion of a standard press-fit humeral component. One fracture each was treated with cerclage wire fixation, cerclage suture, and Steinmann pin fixation followed by insertion of a cemented humeral component. Postoperative physiotherapy consisted of a standard protocol8,9 for four patients and a limited goals protocol10,11 for two.
One patient sustained fractures of the greater and lesser tuberosities during insertion of a press-fit humeral component. The fractures were described as nondisplaced and were treated with cerclage suture fixation around a standard-length humeral component. The patient was managed postoperatively with a standard physical therapy protocol.
Patient Evaluation
The study was carried out under a protocol approved by our institutional review board. Patients were identified through our institution's joint replacement database. The protocol calls for all patients undergoing a shoulder arthroplasty to return in person to undergo an interview, an examination, and a radiographic evaluation. These evaluations are scheduled at two to three months after the arthroplasty and at one year, two years, five years, and at each subsequent five-year interval until revision or death. Patients who are unable to return for evaluation complete standardized questionnaires or undergo a telephone interview. In addition, patients are asked to forward all records and radiographs from subsequent orthopaedic or medical consultations to the institutional database office. These records are then reviewed and documented by an orthopaedic surgeon. The accuracy of the information and the completeness of the database has been shown previously to be on the order of 95%12.
The joint database allowed retrieval of demographic data, the date of surgery, complications, implant type, and diagnosis. A retrospective chart review was then conducted to determine the cause of fracture, the fracture treatment, previous and subsequent procedures, complications, and functional scores.
At the time of the present review, four patients from the original cohort of forty-five patients had died of causes that were unrelated to the shoulder arthroplasty at a mean of twelve years (range, nine to fourteen years) after the procedure. These four patients had complete medical records, outcomes scores, and radiographs at a mean of nine years (range, six to twelve years) of follow-up, and therefore, these results are included in the analysis. The remaining forty-one patients were contacted by mail and were asked to participate in the study. All consented to participate, with twenty-two patients undergoing clinical evaluation and nineteen patients responding to a validated mail-in questionnaire13 with a follow-up telephone interview.
After a mean duration of follow-up of 7.6 years (range, twenty-five months to sixteen years), patients completed the American Shoulder and Elbow Surgeons (ASES) evaluation form and the Simple Shoulder Test14. Results were also graded according to the criteria of Neer as modified by Cofield10. The result was graded as excellent if the patient had no or slight pain, had active elevation to at least 140°, had external rotation to at least 45°, and was satisfied with the result. The result was graded as satisfactory if the patient had no, slight, or moderate pain with unusual activity; had elevation to at least 90°; had external rotation to at least 20°; and was satisfied with the result. The result was considered to be unsatisfactory if any of the above criteria for a satisfactory result were not met or if revision surgery was required.
Patients were followed until fracture union and beyond according to the standard protocol at our institution. Fracture-healing was thought to have occurred when there was both clinical and radiographic evidence of fracture union. Clinical fracture-healing was defined as a pain-free fracture site at the time of physical examination. Radiographic union was defined as evidence of bridging bone across the fracture site on two orthogonal radiographic views without signs of hardware failure. In some circumstances, such as in cases of nondisplaced greater tuberosity fractures or metaphyseal fractures, fracture lines were difficult to visualize, and therefore a time to healing could not be calculated.
Radiographic Analysis
Radiographs of the shoulder that were made preoperatively, intraoperatively, postoperatively, at the time of fracture-healing, and at the time of the latest follow-up were analyzed for osteopenia, preoperative deformity, fracture characteristics, fracture-healing, implant positioning, and loosening. The mean duration of radiographic follow-up was five years (range, seven months to sixteen years). All radiographs were reviewed by all authors, and a consensus method was used to determine healing, loosening, subsidence, resorption, and osteolysis. Some nondisplaced or minimally displaced fractures of the metaphysis, greater tuberosity, or humeral shaft were difficult to visualize on postoperative radiographs, and therefore an exact time to healing could not be calculated. In those cases, the latest follow-up radiographs were critically examined for signs of bone resorption, fracture nonunion, and implant loosening or migration. Five metaphyseal fractures, eight greater tuberosity fractures, and one nondisplaced humeral shaft fracture could not be visualized on the postoperative or latest follow-up radiographs.
Osteopenia was classified according to the system of Campbell et al.3 and was based on the ratio between the combined width of the mid-diaphyseal cortices to the diameter of the diaphysis. The bone was graded as normal if the percentage was >50%, as having mild osteopenia if the percentage was between 25% and 50%, and as having severe osteopenia if the percentage was <25%. At the time of surgery, five patients were classified as having normal bone, twenty-eight were judged to have mild osteopenia, and twelve were classified as having severe osteopenia.
Preoperative humeral deformity was present in three patients. Two patients had a valgus malunion of the proximal part of the humerus, and one patient had deformity of the humeral shaft because of a previous shaft fracture. All deformities were posttraumatic. All three patients with preexisting deformities sustained a greater tuberosity fracture.
Descriptive statistics were calculated for all variables. Unpaired and paired Student t tests were used to compare quantitative measures. The relative risk was calculated for each subgroup as compared with the remainder of the cohort. The number needed to harm, which indicates the number of patients that need to be exposed to a risk factor to cause harm in one patient (intraoperative fracture), was calculated for correctible risk factors.
Source of Funding
No external funding was received in support of the present study.
The rate of intraoperative fractures of the humerus during shoulder arthroplasty at our institution over the twenty-two-year study period was 1.5%. The rate was 1.2% during primary shoulder arthroplasty and 3.3% during revision shoulder arthroplasty. Patients undergoing revision shoulder arthroplasty were at a significantly higher risk of sustaining an intraoperative humeral fracture than patients undergoing primary arthroplasty (relative risk, 2.8; p = 0.003). The fracture rate was 1.7% for women and 0.5% for men. Female patients undergoing primary shoulder arthroplasty had a significantly higher risk of intraoperative fracture than male patients did (relative risk, 3.3; p = 0.006).
The fracture rate based on the primary diagnosis was 1.2% (relative risk, 0.8; p = 0.58) for osteoarthritis, 1.1% (relative risk, 0.7; p = 0.67) for rheumatoid arthritis, 2.5% (relative risk, 1.9; p = 0.11) for posttraumatic arthritis, and 1.6% (relative risk, 1.1; p = 0.75) for osteonecrosis.
The rate of occurrence of intraoperative humeral fractures during primary shoulder arthroplasty was also examined by comparing press-fit humeral components with cemented humeral components. The fracture rate was 1.7% for patients with press-fit humeral components, compared with 0.6% for those with cemented humeral components. Patients who were managed with a primary press-fit humeral component had a significantly higher likelihood of sustaining an intraoperative fracture than did patients who were managed with a cemented component (relative risk, 2.9; p = 0.046). As the selection of implant design may be considered to be a correctible risk factor, the number needed to harm was calculated. Eighty-eight patients would have to undergo primary arthroplasty with a press-fit humeral component before one patient would be harmed from this complication.
All fractures healed at a mean of seventeen weeks (range, six to fifty-six weeks). The greater tuberosity fractures that were described as displaced and that required reduction and fixation healed at a mean of 13.5 weeks (range, six to forty-four weeks), and the nondisplaced greater tuberosity fractures healed at a mean of 6.5 weeks (range, six to seven weeks). The combined tuberosity and shaft fractures healed at a mean of 9.7 weeks (range, six to twelve weeks), and the humeral shaft fractures healed at a mean of 22.5 weeks (range, seven to fifty-six weeks). Of the six metaphyseal fractures, only one was visible on postoperative radiographs, and it healed at sixteen weeks. The one combined greater and lesser tuberosity fracture healed at nine weeks.
Complications
A total of eighteen complications occurred in sixteen patients (36%), including eight patients (26%) in the primary arthroplasty group and eight patients (57%) in the revision arthroplasty group. Nerve injuries occurred in six patients. Three patients sustained a partial brachial plexus palsy that completely resolved at six weeks, four months, and twelve months. Two patients had a radial nerve palsy that resolved at ten days and five weeks. One patient sustained an isolated ulnar nerve neurapraxia that completely resolved in five days.
Two displaced greater tuberosity fractures that were treated with open reduction and suture fixation redisplaced. Both fractures were treated nonoperatively. Both healed, but both were malunited. Other complications (noted in one patient each) included large hematoma formation; extensive fracture blisters; partial failure of fixation of a shaft fracture with partial loss of reduction, which healed with nonoperative treatment; extension of a shaft fracture at six weeks postoperatively during physiotherapy, which healed with nonoperative treatment; refracture of a shaft fracture after a fall, which required open reduction and internal fixation; and symptomatic glenoid arthritis.
Two patients underwent revision arthroplasty after the intraoperative fracture. One patient with a nondisplaced greater tuberosity fracture that was deemed stable underwent revision shoulder surgery, twenty-five months after the fracture, because of posterior glenohumeral joint instability in association with a large rotator cuff tear. The operative report indicated that the greater tuberosity fracture had healed. The other patient underwent revision arthroplasty, sixteen years after the intraoperative fracture, for the treatment of glenohumeral instability and polyethylene wear. Two patients had development of large symptomatic rotator cuff tears that were not sustained at the time of the fracture repair. There were no revisions for implant loosening or migration. There were no deep infections.
Motion, Pain, and Result Rating
Preoperatively, the mean range of motion was 82° (range, 30° to 160°) of forward elevation and 17° (range, -45° to 70°) of external rotation in the primary arthroplasty group and 79° (range, 10° to 150°) of forward elevation and 25° (range, -20° to 60°) of external rotation in the revision arthroplasty group. Twenty-nine patients rated the preoperative shoulder pain as severe, and sixteen rated it as moderate. After a mean duration of follow-up of 7.6 years (range, twenty-five months to sixteen years), the mean forward elevation for the entire cohort was 108° (range, 20° to 170°) and the mean external rotation was 39° (range, -40° to 90°).
In the primary arthroplasty group, the mean forward elevation at the time of the latest follow-up was 115° (range, 20° to 170°), the mean external rotation was 42° (range, -40° to 90°), and the mean shoulder pain score was 2 (range, 0 to 8) of 10. There were significant improvements in terms of range of motion (forward elevation and external rotation) and pain when compared with the preoperative values (p < 0.05). At the time of the latest follow-up, the mean Simple Shoulder Test score was 7 points (range, 0 to 12 points), the mean ASES score was 66 points (range, 25 to 98 points), and the mean score on the functional portion of the ASES form was 27 points (range, 3 to 48 points). At the time of the latest follow-up, seven patients were rated as having an excellent result according to the Neer criteria as modified by Cofield, seventeen were rated as having a satisfactory result, and seven were rated as having an unsatisfactory result.
In the revision arthroplasty group, the mean forward elevation at the time of the latest follow-up was 92° (range, 20° to 150°), the mean external rotation was 33° (range, 10° to 60°), and the mean pain score was 4 (range, 1 to 8) of 10. In this group, there was no significant difference in terms of forward elevation (p = 0.38) or external rotation (p = 0.16) at the time of follow-up when compared with the preoperative values. However, there was significant pain relief at the time of follow-up (p < 0.005). At the time of the latest follow-up, the Simple Shoulder Test score was 4 points (range, 0 to 10 points), the mean ASES score was 48 points (range, 15 to 78 points), and the mean score on the functional portion of the ASES form was 18 points (range, 0 to 35 points). At the time of follow-up, no patient in the revision arthroplasty group was rated as having an excellent result according to the Neer criteria, seven were rated as having a satisfactory result, and seven were rated as having an unsatisfactory result.
The outcomes associated with displaced greater tuberosity fractures requiring fixation were compared with those associated with nondisplaced greater tuberosity fractures that were deemed to be stable intraoperatively. There were no significant differences between these two groups with regard to mean forward elevation (105° compared with 125°; p = 0.13), mean external rotation (40° compared with 44°; p = 0.68), the mean Simple Shoulder Test score (7 compared with 8; p = 0.26), the mean ASES score (60 compared with 65; p = 0.47), or the mean functional portion of the ASES score (23 compared with 30; p = 0.15). At the time of the latest follow-up in the displaced greater tuberosity group, seven patients were rated as having an excellent or satisfactory result according to the Neer criteria as modified by Cofield and four were rated as having an unsatisfactory result. In the nondisplaced greater tuberosity group, all patients were rated as having an excellent or satisfactory result, except for one patient who was rated as having an unsatisfactory result.
Radiographic Results
Radiographs were analyzed at a mean of five years (range, seven months to sixteen years) of follow-up. Five fractures displaced between the time of the early postoperative radiograph and the radiograph made at the time of fracture-healing. One patient with a minimally displaced greater tuberosity fracture, which was treated with suture fixation, felt a popping sensation seven weeks after surgery when she was pulling herself up from the seated position. Radiographs demonstrated displacement of the greater tuberosity posteriorly. This fracture was treated nonoperatively and went on to heal. One greater tuberosity fracture, which was treated with suture fixation, displaced posterosuperiorly in the early postoperative period. This fracture healed without additional operative treatment (Figs. 2-A and 2-B). One patient with a displaced distal humeral shaft fracture that was treated with immediate open reduction and internal fixation with a 3.5-mm pelvic reconstruction plate experienced partial hardware failure with further fracture displacement. This fracture eventually healed without additional operative treatment. One patient with a displaced humeral shaft fracture that was treated with a long-stem component and cerclage cables heard a crack six weeks postoperatively while performing passive stretching exercises under the supervision of a physiotherapist. Radiographs demonstrated distal fracture extension. The fracture was treated with functional bracing and was healed sixteen weeks later. The final patient sustained a minimally displaced humeral shaft fracture that initially was treated with a shoulder spica cast for four weeks. The fracture was partially healed at four months postoperatively, when the patient fell. The fall resulted in a displaced shaft fracture that was treated with open reduction and internal fixation.
No component shifted or subsided. Five greater tuberosities (four in the displaced fracture group and one in the nondisplaced fracture group) had undergone partial resorption. The medial metaphyseal area had resorbed in one patient with a combined greater tuberosity and proximal shaft fracture.
Radiolucent lines were present in seventeen glenoids: five patients had a 1-mm incomplete radiolucent line, eight patients had a 1-mm complete radiolucent line, three patients had a 1.5-mm incomplete radiolucent line, and one patient had a 1.5-mm complete radiolucent line. Radiolucent lines were present around fifteen humeral components: nine patients had a 1-mm incomplete radiolucent line, three had a 1.5-mm incomplete radiolucent line, and three had a 1.5-mm complete radiolucent line.
Periprosthetic fractures of the humerus during or after shoulder arthroplasty have been reported as an uncommon complication1,5,6. The data from the present study indicate that intraoperative periprosthetic humeral fractures occur at an overall rate of 1.5% (1.2% during primary shoulder arthroplasty and 3.3% during revision shoulder arthroplasty). This rate is considerable as it is comparable with the rates of periprosthetic infection and nerve injury4,15. The occurrence of an intraoperative periprosthetic fracture is not without consequence. This complication has been reported to increase operative time, to increase blood loss, to alter postoperative physiotherapy, and to lead to poorer outcomes1-3,15.
Several factors have been identified in the literature to increase the risk of periprosthetic fractures, such as osteopenia1,3-7,16, proximal humeral deformity or malunion1,3, soft-tissue contracture requiring excessive external rotation force for dislocation3, rheumatoid arthritis1,3,4, overreaming of the humeral canal3,17, insertion of an oversized broach3, and aggressive or overzealous manipulation3,16. In our series, a relative risk analysis indicated that female patients were three times more likely to sustain fractures than male patients were, that patients undergoing revision surgery were three times more likely to sustain fractures than patients undergoing a primary arthroplasty were, that the use of a press-fit humeral component was three times more likely to lead to a fracture than the use of a cemented humeral component was, and that patients with the diagnosis of posttraumatic arthritis were two times more likely to sustain a fracture than patients with a diagnosis of rheumatoid arthritis, osteoarthritis, or osteonecrosis were.
Osteopenia was certainly prevalent in our cohort as forty patients (89%) had a radiographic diagnosis of osteopenia, with twenty-eight patients (62%) being rated as having mild osteopenia and twelve patients (27%) being rated as having severe osteopenia. Unfortunately, as no normative values exist for proximal humeral osteopenia and our database did not prospectively document osteopenia grade, a risk assessment could not be done.
Eighty percent of the fractures occurred by one of five predominant mechanisms. Fourteen intraoperative fractures (31%) occurred during insertion or impaction of a humeral trial component or the true humeral prosthesis. Eight fractures occurred during reaming or broaching of the humerus, and eight occurred during removal of a humeral component (cemented or press-fit) during revision surgery. Three fractures occurred during glenoid exposure and retractor placement, and three occurred during glenohumeral joint dislocation or reduction.
The treatment of periprosthetic humeral fractures begins with prevention. Preoperative templating allows appropriate sizing of the humeral head and canal and identifies osteopenia, humeral deformity, and canal stenosis, which may influence broaching, reaming, and insertion of the trial prosthesis. When templating demonstrates substantial deformity or canal stenosis, intraoperative fluoroscopy may assist with humeral preparation. As seven fractures occurred during canal reaming, we now gently hand-ream the humeral canal with use of the reamer as a sound, rather than torquing the reamer to obtain so-called cortical chatter. Appropriate capsular and soft-tissue releases should be performed to facilitate exposure and joint dislocation. Extensile approaches such as the anteromedial approach18 should be utilized when required, and reaming and broaching should be conducted with great care. As revision procedures are associated with a higher rate of intraoperative fractures, with most fractures occurring during removal of a well-fixed humeral component, the judicious use of cortical windows may decrease the risk of fracture. However, they may also function as stress-risers as one patient in our series sustained a spiral shaft fracture adjacent to such a window.
We recommend a systematic approach to addressing intraoperative periprosthetic fractures. Once a fracture is thought to have occurred, it should be identified and characterized. Periprosthetic fractures can be identified by means of direct exposure, such as by extending the deltopectoral approach distally into the lateral or anterolateral approach for humeral shaft fractures, or they may be identified by means of intraoperative fluoroscopy or portable radiographs. Intraoperative fractures should be classified and treated according to the system of Campbell et al.3, who divided these fractures into four anatomic regions. Region-1 fractures involve the greater and/or lesser tuberosities. These fractures are assessed for stability, and, if deemed stable, with the periosteum intact and without displacement, they may be treated with insertion of a standard implant without specific fixation. However, we err on the side of fixation; therefore, if any fracture motion exists or if there is any degree of displacement, suture fixation of the fractured tuberosity to the humeral implant and circumferentially around the proximal part of the humerus is recommended. Region-2 fractures involve the metaphysis of the proximal part of the humerus and are treated with a standard-length implant, cerclage fixation, and autologous bone-grafting. Region-3 fractures involve the proximal part of the humeral shaft, and region-4 fractures involve the middle and distal parts of the humeral shaft. Region-3 and 4 fractures are best treated with longer-stemmed implants with cerclage fixation and, in some cases, with supplementary allograft cortical struts.
It is not possible to correlate outcomes in our patients with the specific type of fracture or fixation construct because of the heterogeneous nature of this cohort and the limited number of patients. Interestingly, all fractures in the present study healed. We theorize that the rate of union was high because intraoperative periprosthetic fractures typically occur secondary to low-energy trauma; therefore, disruption of the periosteal and intramedullary blood supplies is less likely.
When the outcomes of displaced intraoperative greater tuberosity fractures requiring fixation were compared with those of nondisplaced greater tuberosity fractures, there was a trend toward lower functional ASES scores and less forward elevation in association with the displaced fractures. This difference was less than expected, considering the historically high rate of tuberosity-related complications seen when hemiarthroplasty is used to treat proximal humeral fractures19-22. When the overall outcomes in this series of shoulder arthroplasties that were complicated by an intraoperative fracture are compared with averages from historical series of shoulder arthroplasties in previously published reports from our institution23-28, there is virtually no difference. The follow-up forward elevation and external rotation after primary shoulder arthroplasty in the historical series averaged 116° and 44°, respectively, whereas in the present series these values averaged 115° and 42°, respectively. The mean shoulder pain score at the time of follow-up in the historical primary arthroplasty group (mean, 2; range, 0 to 10) was the same as that in the present series (mean, 2; range, 0 to 8). This finding was unexpected as we assumed that the outcomes would be poorer in the patients who had sustained an intraoperative periprosthetic fracture. These findings should be interpreted with caution, however, as they may be due to the small number of patients in the fracture group, or there may be other variables influencing outcomes in these populations. The outcomes in the revision arthroplasty group in the present study are also similar to the results of other shoulder revision series in published reports from our institution29-32, indicating that intraoperative fractures during revision surgery do not appear to drastically alter outcome.
There were no revisions for aseptic component loosening in this cohort, although intuitively it would seem to have been a likely complication after intraoperative fractures. However, the overall complication rate in our cohort was high (36%). Three patients sustained a partial brachial plexus palsy; all had a humeral shaft fracture that was treated with a long-stemmed component and cerclage fixation. We theorize that the nerve injury in those cases was due to prolonged traction during exposure and fixation. Two patients sustained a radial nerve injury, one with a greater tuberosity fracture and the other with a metaphyseal fracture; therefore, nerve injury was likely due to traction rather than direct injury as one would expect to occur during surgical exposure of a fracture of the humeral shaft.
One of the limitations of the present study is that although the data from the database were prospectively gathered, they were analyzed retrospectively. In addition, although several risk factors for intraoperative periprosthetic fractures were identified, several more may exist that may be more important. One such factor is osteopenia; although it was prevalent in our cohort, we were unable to demonstrate a significant relative risk as normative data for patients undergoing shoulder arthroplasty do not exist. Another limitation was the heterogeneous follow-up, as some patients underwent clinical examination and others only completed a validated questionnaire with a telephone interview. Radiographs also were not standardized, and computed tomography was not used to visualize difficult-to-see fractures to precisely identify the amount of displacement and time of healing.
Intraoperative periprosthetic humeral fractures are not uncommon, and significant risk factors for these fractures include female sex, revision surgery, and the use of a press-fit implant. Periprosthetic fractures should be approached in a systematic way that includes anatomic reduction, rigid fixation, and achieving implant stability that permits early range of motion. Satisfactory results can be expected, although the high complication rate is concerning. 
Note: The authors thank Joy MacDermid, BScPT, MSc, PhD, for assistance with the statistical analysis.
Boyd AD Jr, Thornhill TS, Barnes CL. Fractures adjacent to humeral prostheses. J Bone Joint Surg Am.1992;74:1498-504.741498Â
1992Â
[PubMed] Â
Cameron B, Iannotti JP. Periprosthetic fractures of the humerus and scapula: management and prevention. Orthop Clin North Am.1999;30:305-18.30305Â
1999Â
[CrossRef] Â
Campbell JT, Moore RS, Iannotti JP, Norris TR, Williams GR. Periprosthetic humeral fractures: mechanisms of fracture and treatment options. J Shoulder Elbow Surg.1998;7:406-13.7406Â
1998Â
[CrossRef] Â
Cuomo F, Checroun A. Avoiding pitfalls and complications in total shoulder arthroplasty. Orthop Clin North Am.1998;29:507-18.29507Â
1998Â
[CrossRef] Â
Kumar S, Sperling JW, Haidukewych GH, Cofield RH. Periprosthetic humeral fractures after shoulder arthroplasty. J Bone Joint Surg Am.2004;86:680-9.86680Â
2004Â
Â
Worland RL, Kim DY, Arredondo J. Periprosthetic humeral fractures: management and classification. J Shoulder Elbow Surg.1999;8:590-4.8590Â
1999Â
[CrossRef] Â
Wright TW, Cofield RH. Humeral fractures after shoulder arthroplasty. J Bone Joint Surg Am.1995;77:1340-6.771340Â
1995Â
Â
Boardman ND 3rd, Cofield RH, Bengtson KA, Little R, Jones MC, Rowland CM. Rehabilitation after total shoulder arthroplasty. J Arthroplasty.2001;16:483-6.16483Â
2001Â
[CrossRef] Â
Wilcox RB, Arslanian LE, Millett P. Rehabilitation following total shoulder arthroplasty. J Orthop Sports Phys Ther.2005;35:821-36.35821Â
2005Â
Â
Cofield RH. Total shoulder arthroplasty with the Neer prosthesis. J Bone Joint Surg Am.1984;66:899-906.66899Â
1984Â
Â
Neer CS 2nd, Watson KC, Stanton FJ. Recent experience in total shoulder replacement. J Bone Joint Surg Am.1982;64:319-37.64319Â
1982Â
Â
McGrory BJ, Morrey BF, Rand JA, Ilstrup DM. Correlation of patient questionnaire responses and physician history in grading clinical outcome following hip and knee arthroplasty. A prospective study of 201 joint arthroplasties. J Arthroplasty.1996;11:47-57.1147Â
1996Â
[CrossRef] Â
Smith AM, Barnes SA, Sperling JW, Farrell CM, Cummings JD, Cofield RH. Patient and physician-assessed shoulder function after arthroplasty. J Bone Joint Surg Am.2006;88:508-13.88508Â
2006Â
[CrossRef] Â
Matsen FA 3rd, Ziegler DW, DeBartolo SE. Patient self-assessment of health status and function in glenohumeral degenerative joint disease. J Shoulder Elbow Surg.1995;4:345-51.4345Â
1995Â
[CrossRef] Â
Wirth MA, Rockwood CA Jr. Complications of shoulder arthroplasty. Clin Orthop Relat Res.1994;307:47-69.30747Â
1994Â
Â
Bonutti PM, Hawkins RJ. Fracture of the humeral shaft associated with total replacement arthroplasty of the shoulder. A case report. J Bone Joint Surg Am.1992;74:617-8.74617Â
1992Â
Â
Lee M, Chebli C, Mounce D, Bertelsen A, Richardson M, Matsen F 3rd. Intramedullary reaming for press-fit fixation of a humeral component removes cortical bone asymmetrically. J Shoulder Elbow Surg.2008;17:150-5.17150Â
2008Â
[CrossRef] Â
Gill DR, Cofield RH, Rowland C. The anteromedial approach for shoulder arthroplasty: the importance of the anterior deltoid. J Shoulder Elbow Surg.2004;13:532-7.13532Â
2004Â
[CrossRef] Â
Boileau P, Krishnan SG, Tinsi L, Walch G, Coste JS, Molé D. Tuberosity malposition and migration: reasons for poor outcomes after hemiarthroplasty for displaced fractures of the proximal humerus. J Shoulder Elbow Surg.2002;11:401-12.11401Â
2002Â
[CrossRef] Â
Mighell MA, Kolm GP, Collinge CA, Frankle MA. Outcomes of hemiarthroplasty for fractures of the proximal humerus. J Shoulder Elbow Surg.2003;12:569-77.12569Â
2003Â
[CrossRef] Â
Sperling JW, Cuomo F, Hill JD, Hertel R, Chuinard C, Boileau P. The difficult proximal humerus fracture: tips and techniques to avoid complications and improve results. Instr Course Lect.2007;56:45-57.5645Â
2007Â
Â
Tanner MW, Cofield RH. Prosthetic arthroplasty for fractures and fracture-dislocations of the proximal humerus. Clin Orthop Relat Res.1983;179:116-28.179116Â
1983Â
Â
Adams JE, Sperling JW, Hoskin TL, Melton LJ 3rd, Cofield RH. Shoulder arthroplasty in Olmsted County, Minnesota, 1976-2000: a population-based study. J Shoulder Elbow Surg.2006;15:50-5.1550Â
2006Â
[CrossRef] Â
Sperling JW, Cofield RH, Rowland CM. Minimum fifteen-year follow-up of Neer hemiarthroplasty and total shoulder arthroplasty in patients aged fifty years or younger. J Shoulder Elbow Surg.2004;13:604-13.13604Â
2004Â
[CrossRef] Â
Torchia ME, Cofield RH, Settergren CR. Total shoulder arthroplasty with the Neer prosthesis: long-term results. J Shoulder Elbow Surg.1997;6:495-505.6495Â
1997Â
[CrossRef] Â
Adams JE, Sperling JW, Schleck CD, Harmsen WS, Cofield RH. Outcomes of shoulder arthroplasty in Olmsted County, Minnesota: a population-based study. Clin Orthop Relat Res.2007;455:176-82.455176Â
2007Â
[CrossRef] Â
Rispoli DM, Sperling JW, Athwal GS, Schleck CD, Cofield RH. Humeral head replacement for the treatment of osteoarthritis. J Bone Joint Surg Am.2006;88:2637-44.882637Â
2006Â
[CrossRef] Â
Sperling JW, Cofield RH, Schleck CD, Harmsen WS. Total shoulder arthroplasty versus hemiarthroplasty for rheumatoid arthritis of the shoulder: results of 303 consecutive cases. J Shoulder Elbow Surg.2007;16:683-90.16683Â
2007Â
[CrossRef] Â
Antuna SA, Sperling JW, Cofield RH, Rowland CM. Glenoid revision surgery after total shoulder arthroplasty. J Shoulder Elbow Surg.2001;10:217-24.10217Â
2001Â
[CrossRef] Â
Cheung EV, Sperling JW, Cofield RH. Polyethylene insert exchange for wear after total shoulder arthroplasty. J Shoulder Elbow Surg.2007;16:574-8.16574Â
2007Â
[CrossRef] Â
Cheung EV, Sperling JW, Cofield RH. Revision shoulder arthroplasty for glenoid component loosening. J Shoulder Elbow Surg.2008;17:371-5.17371Â
2008Â
[CrossRef] Â
Sanchez-Sotelo J, Sperling JW, Rowland CM, Cofield RH. Instability after shoulder arthroplasty: results of surgical treatment. J Bone Joint Surg Am.2003;85:622-31.85622Â
2003Â
[CrossRef] Â