Iatrogenic Radial Nerve Palsy After Humeral Shaft Nonunion Repair: More Common Than You Think

Transcription

1 ORIGINAL ARTICLE Iatrogenic Radial Nerve Palsy After Humeral Shaft Nonunion Repair: More Common Than You Think Rafael Kakazu, MD, Steven K. Dailey, MD, Amanda J. Schroeder, MD, John D. Wyrick, MD, and Michael T. Archdeacon, MD, MSE Objectives: To determine the rate of iatrogenic radial nerve palsy (RNP) after surgical repair of established humeral shaft nonunion (HSNU). Design: Retrospective chart review. Setting: Level I trauma center. Patients/Participants: Fifty-four patients with HSNU, 10 (18.5%) of whom developed an iatrogenic RNP after nonunion repair. Intervention: HSNU repair with compression plate stabilization with or without autogenous bone graft. Main Outcome Measurements: Postoperative iatrogenic RNP. Results: Ten (18.5%) patients developed iatrogenic radial nerve palsies: 8 experienced complete resolution (mean, 2.5 months) and 2 experienced partial resolution. There were no statistically significant differences between patients who developed nerve palsy and those who did not in regard to age, gender, tobacco use, diabetic status, previous RNP, initial management (operative vs. nonoperative), surgical approach, presence of infected nonunion, number of previous surgeries, or operative time (P. 0.05). Conclusions: The occurrence of iatrogenic RNP for patients undergoing surgical repair of an HSNU was 18.5%. According to historical data, this rate is nearly 3 times higher than for those undergoing open reduction and internal fixation of either acute humeral shaft fractures or HSNUs. Key Words: radial nerve palsy, humeral shaft nonunion, complications, iatrogenic Accepted for publication December 22, From the Department of Orthopaedic Surgery, University of Cincinnati, Cincinnati, OH. Presented in part at the Annual Meeting of the American Academy of Orthopaedic Surgeons, March 27, 2015, Las Vegas, NV, and the Mid- America Orthopaedic Association, April 23, 2015, Hilton Head Island, SC. Presented as a poster at the Annual Meeting of the Orthopaedic Trauma Association, October 7 10, 2015, San Diego, CA. M. T. Archdeacon is a paid consultant for Stryker, lectures for Stryker and AO North America, receives royalties from Stryker and SLACK Incorporated, and receives research grants from the OTA. J. D. Wyrick is a paid consultant for Stryker. The remaining authors report no conflict of interest. Institutional review board approval for this study was obtained. Reprints: Amanda J. Schroeder, MD, Department of Orthopaedic Surgery, University of Cincinnati, 231 Albert Sabin Way, ML 0212, Cincinnati, OH ( Amanda.Schroeder@uc.edu). Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved. Level of Evidence: Therapeutic Level IV. See Instructions for Authors for a complete description of Levels of Evidence. (J Orthop Trauma 2016;30: ) INTRODUCTION Fractures of the humeral shaft account for 1% 3% of all skeletal injuries 1 and exhibit a bimodal age distribution. 2,3 Approximately 95% of these fractures heal with nonoperative management Operative treatment has a nonunion rate of approximately 5% 10%. 2,11 17 Radial nerve palsy (RNP) is an associated injury encountered with acute fractures of the humeral shaft. 18 The incidence of primary RNP associated with humeral shaft fracture has been well documented in the literature, with an occurrence rate of approximately 12%. 19,20 Iatrogenic RNP after open treatment of acute humeral shaft fractures with plates or nails occurs in approximately 6.5% of cases, with a reported high of 23.8%. 7,13,21 24 However, iatrogenic RNP have been shown to spontaneously recover at rates comparable with that of primary RNP after acute fracture, reportedly between 70% and 90%. 19,20 To our knowledge, there is very little data specifically evaluating the occurrence of RNP after repair of humeral shaft nonunion (HSNU). We conducted a retrospective review of 54 patients undergoing repair of an HSNU with plate fixation with or without autogenous bone grafts and determined the rate of iatrogenic RNP. We also evaluated possible risk factors for iatrogenic RNP after HSNU repair. PATIENTS AND METHODS Inclusion criteria consisted of older than 18 years, a nonpathologic humerus fracture resulting in clinically and radiographically confirmed nonunion, documented normal radial nerve function before definitive treatment, and definitive operative treatment of the HSNU by 1 of 2 fellowshiptrained orthopaedic trauma surgeons. Initial treatment of fracture, whether by operative or nonoperative means, was noted. For the operative group, the index procedure consisted of plate fixation or intramedullary (IM) nail stabilization. For the group initially treated nonoperatively, nonunion (Figs. 1A, B) was defined by the operative surgeons as no evidence of progressive healing, both clinically and radiographically, at a minimum of 6 weeks after injury. For the patients who underwent operative treatment, nonunion diagnosis was based on continued pain at the fracture site with no evidence of J Orthop Trauma Volume 30, Number 5, May 2016

2 J Orthop Trauma Volume 30, Number 5, May 2016 Iatrogenic Radial Nerve Palsy FIGURE 1. A and B, AP and lateral radiographs of an HSNU. fracture healing after a minimum of 12 weeks or failure of fixation at any time point. After institutional review board approval, our fracture database was queried over a 13-year period for the inclusion criteria and yielded 61 patients. Medical records were retrospectively reviewed for demographic data, including age and gender. Injury and treatment data that were abstracted included body mass index (BMI), smoking status, diabetic status, fracture type (open vs. closed), operating room time, estimated blood loss, history of previous RNP, index treatment (nonoperative vs. operative), index fixation (open reduction and internal fixation with plates vs. IM nails), operative approach, autogenous bone grafting, number of previous surgeries, and history of infection. Preoperative notes were reviewed for radial nerve function as well as clinical and radiographic diagnosis of HSNU. Postoperative notes were reviewed for both diagnosis and resolution of RNP. Diagnostic criteria for iatrogenic RNP included explicit documentation by the operative surgeon of RNP and clear notation of the associated radial nerve deficits. Seven patients were excluded from the initial database query. Three patients lacked adequate follow-up, 2 patients had RNP at the time of definitive nonunion repair, and 2 patients were misdiagnosed in the database (1 acute humerus fracture and 1 acute periprosthetic fracture). Our treatment protocol for repair of the nonunion was fairly standard. Fractures occurred in the shaft and at the proximal and distal metadiaphyseal regions. All fracture variants were encountered in the series. Articular fractures were not included. In general, for most mid-third to distal third nonunions and those that had a previous posterior surgery, we used a triceps-splitting approach, identifying the radial nerve proximally and tracking it carefully into the zone of injury or nonunion. For proximal third to mid-third fractures, we typically used a standard brachialis-splitting anterior-lateral approach, where the nerve was only specifically identified distally and laterally but not proximally or posteriorly. Positioning of the patients depended on the approach used; those undergoing the anterior-lateral approach were positioned supine and those undergoing the posterior approach were positioned prone. No tourniquet was used for any of the patients. Hardware was removed in typical fashion, and broken screws were removed using trephines and extractors in cases of infection. Debridement was performed both sharply and bluntly, with care taken to avoid undue stress on the softtissue envelope. For reduction and plating, many techniques were used including carefully placed clamps, compression plate reductions, and manual manipulation. All patients were managed with decortication and/or excision of any necrotic bone. In addition, patients presenting with infected nonunion were managed with antibiotic bead placement and irrigation and debridement until the infection subsided. For reconstruction, compression plate stabilization was performed and supplemented with autogenous bone graft at the discretion of the operative surgeon (Figs. 2A, B). Forty-eight (89%) of the patients had an adjunctive bone graft. The bone graft techniques included 41 autogenous grafts (5 local calluses, 13 iliac crests, 20 proximal tibias, 2 IM femurs by means of reaming irrigation aspirator, and 1 segmental fibula), 5 cancellous allografts, and 2 patients who were managed with bone morphogenetic proteins. Six patients had a combination of autograft and allograft. Postoperatively, radiographs and clinical examinations were performed at days after surgery and at 6-week intervals until healing was confirmed (Figs. 3A, B). Physical therapy to regain strength and range of motion was based on patient needs and was not standardized. Patients who developed an RNP following surgical treatment of HSNU were followed clinically. If the palsy did not show signs of resolution by 6 12 weeks postoperatively, electromyography or nerve conduction study was obtained. Statistical analysis included the x 2 test to compare proportions. Fisher exact test was used in place of x 2 when,5 observations were available for a particular variable. Student t test was used for normative data. For all statistical tests, P, 0.05 was considered to indicate a significant result. Data were Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved

3 Kakazu et al J Orthop Trauma Volume 30, Number 5, May 2016 FIGURE 2. A and B, anteroposterior and lateral radiographs after operative fixation with a single plate by means of an anterior-lateral approach. managed and statistics analyzed in Microsoft Excel (Microsoft Corporation, Redwood, WA). RESULTS After applying inclusion and exclusion criteria, we identified 54 patients who made up the cohort. Demographic, injury, and treatment data are summarized in Table 1. The patient population comprised 24 men and 30 women with a mean age of 54 years (range 21 93). Twenty-six patients were initially treated nonoperatively and 28 were managed surgically with plates (25) or nails (3). Eight patients presented with an open fracture. In the operative group, 3 patients had staged initial treatment with excisional debridement and/or external fixation before fracture repair. Twelve of the nonunion reconstructions underwent staged treatment including removal of implants, antibiotic bead or spacer placement, and excisional debridement. Eleven of these had a culturepositive nonunion. Of the patients who were treated surgically at the time of the initial injury, the mean number of procedures before definitive nonunion reconstruction was 2.21 (range 1 10). Of the 54 patients, 10 (18.5%) developed RNP after definitive repair of the nonunion. Eight had complete resolution of the palsy at an average of 2.5 months, whereas 2 had partial recovery. Both declined further treatment. One patient, who did not develop an RNP, had a recalcitrant septic nonunion; however, this patient also declined further treatment. The remaining 53 patients achieved complete union with no signs of acute infection or residual infection at the time of union. No other complications were identified. Comparing the group of patients who developed RNP with those who did not, there were no significant differences observed with regard to age, gender, history of previous RNP, initial treatment, surgical approach, open injury, number of previous surgeries, infection, or operative time (P. 0.05) (Table 2). Factors associated with healing BMI, tobacco use, and diabetic status also showed no difference between FIGURE 3. A and B, AP and lateral films demonstrating a united humerus approximately 1 year after operative intervention Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved.

4 J Orthop Trauma Volume 30, Number 5, May 2016 Iatrogenic Radial Nerve Palsy TABLE 1. Characteristics of the Study Population Total number 54 Age (median, range) 54.5 (21 93) Gender (male:female) 24:30 Previous radial nerve palsy 7 Initial treatment Nonoperative 26 Plates 25 Nails 3 Open fracture 8 Infection 11 Approach (anterolateral:posterior) 34:20 groups (P = 0.45, 0.80, and 0.37, respectively). One factor identified as statistically different was the utilization of bone grafts. Interestingly, 93% of the patients who did not develop an iatrogenic RNP had an adjunctive bone graft, whereas only 70% of the group that were diagnosed with a palsy had an adjunctive bone graft (P = 0.04). In a subset analysis, 7 patients (7/54; 13%) had an initial RNP with the acute injury that completely resolved before the definitive reconstruction of the HSNU. Five of these patients (5/7; 71%) did not develop a second RNP after definitive nonunion surgery, whereas 2 (2/7; 29%) did develop a second RNP (P = 0.89). To clarify the effect of the initial fracture, patients who developed an RNP that resolved before definitive treatment of HSNU were compared with those who did not develop a previous RNP. No significant differences were observed with regard to postoperative RNP, age, gender, laterality, nonoperative versus operative TABLE 2. Comparison of RNP to Non-RNP Group RNP Non-RNP P Number 10 (18.5%) 44 (81.5%) Age (mean, range) 51.4 (23 72) (21 93) 0.30 Gender (male/female) 4/6 20/ Previous RNP 2 (20.0%) 5 (11.4%) 0.46 Initial treatment Nonoperative 5 (50.0%) 21 (47.7%) 0.90 Plates 5 (50.0%) 20 (45.5%) 0.79 Nails 0 (0.0%) 3 (6.82%) 0.40 Open 1 (10.0%) 7 (15.9%) 0.63 No. surgeries (mean, range) 0.8 (0 4) 1.27 (0 10) 0.33 Infected nonunion 2 (20.0%) 9 (20.5%) 0.97 Surgical approach Anterolateral 6 (60.0%) 28 (63.6%) 0.83 Posterior 4 (40.0%) 16 (36.4%) 0.83 Bone graft 7 (70.0%) 41 (93.2%) 0.04 Tobacco use 3 (30.0%) 15 (34.1%) 0.80 Diabetes 1 (10.0%) 10 (22.7%) 0.37 BMI (mean, SD) OR time (mean, SD) EBL (mean, SD) EBL, estimated blood loss. treatment, open fracture, infection, surgical approach, autogenous bone grafts, diabetes, BMI, operative time, or blood loss (P. 0.05). The only statistically significant factor for the development of a second RNP before definitive HSNU fixation was tobacco use (P = 0.03). Patients initially treated nonoperatively for the acute fracture were compared with those treated surgically. Of the 26 patients initially treated conservatively, 5 (19%) developed RNP. Similarly, of the 28 patients managed surgically for the acute fracture, 5 (17.8%; P = 0.90) developed RNP. To clarify the effect of multiple procedures, we further analyzed those patients initially treated surgically for the acute fracture in terms of the number of procedures required to repair the nonunion. Of that group, 14 underwent a single surgical procedure before the definitive surgery for the nonunion and 14 had multiple procedures before the definitive repair of the nonunion. In the multiple surgery group, 1 patient developed RNP (7.0%), whereas 4 (29%) developed RNP in the single surgery group (P = 0.14). DISCUSSION The aim of this study was to determine the rate of iatrogenic RNP after surgical repair of HSNU. Many reports document this rate after open treatment of acute fractures; it is generally found to be approximately 6% In our series, an RNP occurred in 18.5% of HSNU treated with surgical repair, approximately 3 times than that of acute humeral shaft fractures. Eight of 10 (80%) fully recovered, and 2 (20%) had residual palsy with partial recovery. Additionally, 7 patients with normal nerve function at the time of definitive nonunion repair originally had RNP after the acute injury. Of those 7, 2 (28.5%) developed a second palsy and 5 (71.5%) did not (P = 0.89). Thus, it does not seem that a previous RNP is a significant risk factor for palsy after nonunion repair. Only 1 factor was found to be statistically significant when comparing patients who developed RNP with those who did not: the use of autogenous bone grafts. In our study, autogenous bone grafts seemed to have a protective effect, as bone grafts were used in 93.2% of patients (41/44) who did not develop a postoperative RNP and in 70.0% of patients (7/10) who did develop an RNP (P = 0.04). Intuitively, we would have assumed that bone grafts would have been used in more patients who developed palsies as likely more dissection was performed; however, that was not the case. How this finding can be clinically explained is not clear. Several series in the literature discuss RNP after repair of HSNU with an approximate range of occurrence from 4% to 8%. A recently published series evaluating the use of locking plate fixation for HSNU initially treated nonoperatively identified 2 of 24 patients with a postoperative RNP (8%). 25 Abalo et al 26 treated 46 nonunions with open reduction and internal fixation and reported 4 with RNP, a rate of 8.7%. Another study by Bernard de Dompsure et al 27 examined 21 patients treated for uninfected HSNU and reported 1 patient with a transient radial paresis (4.7%). Martinez et al 28 used a 2-plate fixation for HSNU and reported that 1 of 22 patients developed a postoperative, transient RNP (4.5%). In another study that evaluated the use of compression plate and Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved

5 Kakazu et al J Orthop Trauma Volume 30, Number 5, May 2016 cancellous bone graft for HSNU, 4 of the 105 patients with HSNU had temporary RNP (3.8%). 29 Although these series report rates of occurrence substantially lower than in our reported series, we believe that the focus of our report was RNP; thus, we only included patients with documented neurovascular examinations. Therefore, we are of the opinion that our critical evaluation demonstrates a fairly accurate representation of the risk of iatrogenic RNP after the treatment of HSNU. If anything, we possibly have underdiagnosed palsy as we excluded patients without confirmatory physical examination. There are several studies that investigate the use of external fixators for treatment of HSNU. The rates of RNP found in these series are more varied. One series identified 1 RNP of 28 atrophic HSNU that were treated with circular frame fixation, for a rate of 3.6%. 30 In 2008, Micic et al 31 reviewed 31 patients treated using either plates or external fixators for HSNU; 2 of these patients developed transient RNP, giving a rate of 6.5%. A final study comparing plates and unilateral and circular external fixators for HSNU treatment found that 10 of their 80 patients developed transient RNP (12.5%), 32 more similar to our rate of 18.5%. Our data specifically reflect the incidence of RNP after plate fixation of HSNU and thus cannot be extrapolated to RNP after external fixator placement. The etiology of the RNP after repair of HSNU is likely multifactorial. Callus formation at the nonunion site may encase the radial nerve or place it under tension so that reduction or dissection may result in a stretch palsy. In fact, the radial nerve is tethered proximally as a terminal branch of the brachial plexus and distally as it exits through the lateral intermuscular septum. 33 This puts the radial nerve at substantial risk as it sensitive to excessive force and stretch. In our series of 10 RNP cases, a posterior approach was used in 4 cases and it was documented that the radial nerve was clearly identified, found to be intact, and carefully protected. Despite these precautions, the patients still developed an RNP postoperatively. This would suggest that the nerve was susceptible to stretch injury during the process of preparing the fracture site for fixation clearing the callus and developing the fracture ends. This study is not without limitations that must be taken into consideration. First, the retrospective nature of the study makes it possible that some important data elements were not collected accurately, and more importantly, some elements that may be influential were never collected including clear documentation of the status of the nerve during dissection for all nonunion repairs. Second, owing to the fact that RNP after repair of HSNU is relatively uncommon, our series is small. However, our series represents 2 very mature referral-based practices over a 13-year period, so that larger series of similar treatment may be difficult to collect. Furthermore, we defined nonunion in the nonoperative group as no signs of healing 6 weeks after injury, and as pain with no signs of healing 12 weeks after surgery in the operative group. This definition may not meet the criteria for a strict, confirmed nonunion suggested by all individuals, but it was standardized throughout this study and would likely not influence the occurrence of RNP. Also, although RNP was the primary outcome for this investigation, another limitation of the study is the fact that exact motor and sensory deficits and detailed electromyography reports were not available. Moreover, the results might have been different if we excluded patients with infections, multiple operations, or other characteristics. However, inclusion of a heterogenous group of patients could also be considered a strength as the results thus encompass all patients with an HSNU. Additionally, several patients were excluded from analysis because immediate or follow-up postoperative examination of neurologic status was not clearly documented. Thus, it is possible that our study actually underestimates the potential occurrence of iatrogenic RNP after surgical repair of HSNU. Another consideration involves nerve recovery. Wang et al 21 suggest that the timing and pattern of recovery from iatrogenic RNP is similar to the course in primary RNP. However, it is unclear whether the same can be applied to RNP after repair of HSNU. Thus, the recovery rate quoted here may be inconclusive as the palsy group was small with only 10 patients. Finally, we examined numerous variables to determine risk factors for developing RNP and were unable to show significance for many factors; however, the possibility of a type II error exists as our overall sample size was small with only 54 patients. Nevertheless, most surgeons are well aware that operative treatment of an acute humeral shaft fracture has a risk for RNP. However, surgeons must be cognizant of the fact that RNP after repair of HSNU may be substantially higher, approximately 3 times higher in our study when compared with repair of an acute fracture. Patients should be informed of the higher risk for RNP during these procedures; however, many factors seem to be outside the control of the surgeon based on the results of our study. Further studies should focus on determining risk factors for development of RNP after repair of HSNU and examining the natural history of this variant of RNP. REFERENCES 1. Ekholm R, Adami J, Tidermark J, et al. Fractures of the shaft of the humerus. An epidemiological study of 401 fractures. J Bone Joint Surg Br. 2006;88: Mast JW, Spiegel PG, Harvey JP, et al. Fractures of the humeral shaft: a retrospective study of 240 adult fractures. Clin Orthop Relat Res. 1975; Tytherleigh-Strong G, Walls N, McQueen MM. The epidemiology of humeral shaft fractures. J Bone Joint Surg Br. 1998;80: Sarmiento A, Zagorski JB, Zych GA, et al. Functional bracing for the treatment of fractures of the humeral diaphysis. J Bone Joint Surg Am. 2000;82: Rutgers M, Ring D. Treatment of diaphyseal fractures of the humerus using a functional brace. J Orthop Trauma. 2006;20: Rosen H. The treatment of nonunions and pseudarthroses of the humeral shaft. Orthop Clin North Am. 1990;21: Ring D, Chin K, Taghinia AH, et al. Nonunion after functional brace treatment of diaphyseal humerus fractures. J Trauma. 2007;62: Koch PP, Gross DFL, Gerber C. The results of functional (Sarmiento) bracing of humeral shaft fractures. J Shoulder Elbow Surg. 2002;11: Ekholm R, Tidermark J, Törnkvist H, et al. Outcome after closed functional treatment of humeral shaft fractures. J Orthop Trauma. 2006;20: Papasoulis E, Drosos GI, Ververidis AN, et al. Functional bracing of humeral shaft fractures. A review of clinical studies. Injury. 2010;41: e Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved.

6 J Orthop Trauma Volume 30, Number 5, May 2016 Iatrogenic Radial Nerve Palsy 11. McCormack R, Brien D, Buckley RE, et al. Fixation of fractures of the shaft of the humerus by dynamic compression plate or intramedullary nail. J Bone Joint Surg Br. 2000;82: Bhandari M, Devereaux PJ, McKee MD, et al. Compression plating versus intramedullary nailing of humeral shaft fractures a meta-analysis. Acta Orthop. 2006;77: Bell MJ, Beauchamp CG, Kellam JK, et al. The results of plating humeral shaft fractures in patients with multiple injuries. The Sunnybrook experience. J Bone Joint Surg Br. 1985;67: Foster RJ, Dixon GL, Bach AW, et al. Internal fixation of fractures and non-unions of the humeral shaft. Indications and results in a multi-center study. J Bone Joint Surg Am. 1985;67: McKee MD, Seiler JG, Jupiter JB. The application of the limited contact dynamic compression plate in the upper extremity: an analysis of 114 consecutive cases. Injury. 1995;26: Vander Griend R, Tomasin J, Ward EF. Open reduction and internal fixation of humeral shaft fractures. Results using AO plating techniques. J Bone Joint Surg Am. 1986;68: Tingstad EM, Wolinsky PR, Shyr Y, et al. Effect of immediate weightbearing on plated fractures of the humeral shaft. J Trauma. 2000;49: Liu G, Zhang C, Wu H. Comparison of initial nonoperative and operative management of radial nerve palsy associated with acute humeral shaft fractures. Orthopedics. 2012;35: Shao YC, Harwood P, Grotz MRW, et al. Radial nerve palsy associated with fractures of the shaft of the humerus: a systematic review. J Bone Joint Surg Br. 2005;87: Pollock FH, Drake D, Bovill EG, et al. Treatment of radial neuropathy associated with fractures of the humerus. J Bone Joint Surg Am. 1981;63: Wang JP, Shen WJ, Chen WM, et al. Iatrogenic radial nerve palsy after operative management of humeral shaft fractures. J Trauma. 2009;66: Wang X, Zhang P, Zhou Y, et al. Secondary radial nerve palsy after internal fixation of humeral shaft fractures. Eur J Orthop Surg Traumatol. 2014;24: Hee HT, Low BY, See HF. Surgical results of open reduction and plating of humeral shaft fractures. Ann Acad Med Singapore. 1998;27: An Z, He X, Zeng B. A comparative study on open reduction and plating osteosynthesis and minimal invasive plating osteosynthesis in treating mid-distal humeral shaft fractures [in Chinese]. Zhongguo Xiu Fu Chong Jian Wai Ke Za Zhi. 2009;23: Kumar MN, Ravindranath VP, Ravishankar M. Outcome of locking compression plates in humeral shaft nonunions. Indian J Orthop. 2013; 47: Abalo A, Dosseh ED, Adabra K, et al. Open reduction and internal fixation of humeral non-unions: radiological and functional results. Acta Orthop Belg. 2011;77: Bernard de Dompsure R, Peter R, Hoffmeyer P. Uninfected nonunion of the humeral diaphyses: review of 21 patients treated with shingling, compression plate, and autologous bone graft. Orthop Traumatol Surg Res. 2010;96: Martinez AA, Cuenca J, Herrera A. Two-plate fixation for humeral shaft non-unions. J Orthop Surg. 2009;17: Hsu TL, Chiu FY, Chen CM, et al. Treatment of nonunion of humeral shaft fracture with dynamic compression plate and cancellous bone graft. J Chin Med Assoc. 2005;68: Tomic S, Bumbasirevic M, Lesic A, et al. Ilizarov frame fixation without bone graft for atrophic humeral shaft nonunion: 28 patients with a minimum 2-year follow-up. J Orthop Trauma. 2007;21: Micic ID, Mitkovic MB, Mladenovic DS, et al. Treatment of the humeral shaft aseptic nonunion using plate or unilateral external fixator. J Trauma. 2008;64: Atalar AC, Kocaoglu M, Demirhan M, et al. Comparison of three different treatment modalities in the management of humeral shaft nonunions (plates, unilateral, and circular external fixators). J Orthop Trauma. 2008;22: McKee MD. Fractures of the shaft of the humerus. In: Bucholz RW, Heckman JD, Court-Brown CM, eds. Rockwood and Green s Fractures in Adults. Philadelphia, PA: Lippincott Williams & Wilkins; 2006: Copyright 2016 Wolters Kluwer Health, Inc. All rights reserved