- Email: [email protected]
Reliability of the classification and treatment of dislocations of the acromioclavicular joint
J Shoulder Elbow Surg (2014) 23, 665-670
www.elsevier.com/locate/ymse
INTERNATIONAL SHOULDER AND ELBOW ARTICLES
Reliability of the classification and treatment of dislocations of the acromioclavicular joint Chul-Hyun Cho, MD, PhDa,*, Ilseon Hwang, MD, PhDb, Jae-Sung Seo, MDc, Chang-Hyuk Choi, MDd, Sang-Hun Ko, MDe, Hyung Bin Park, MDf, Jinmyoung Dan, MDg a
Pain Research Center, Department of Orthopedic Surgery, Dongsan Medical Center, School of Medicine, Keimyung University, Daegu, South Korea b Department of Pathology, School of Medicine, Keimyung University, Daegu, South Korea c Department of Orthopedic Surgery, Yeungnam University Hospital, Daegu, South Korea d Department of Orthopedic Surgery, Daegu Catholic University Medical Center, Daegu, South Korea e Department of Orthopedic Surgery, Ulsan University Hospital, Ulsan, South Korea f Department of Orthopedic Surgery, Gyeongsang National University Hospital, Jinju, South Korea g Department of Orthopedic Surgery, Gumi CHA Hospital, CHA University, Gumi, South Korea Background: We evaluated interobserver and intraobserver reliability of the classification and treatment of acromioclavicular (AC) joint dislocations and assessed the impact of adding 3-dimensional computed tomography (3D CT) on the reliability of classification and treatment choice. Methods: Ten surgeons independently reviewed plain radiographs and 3D CT in 28 cases with AC joint dislocation. Images from each case were randomly presented to the observers, with plain radiographs alone being presented first, followed by plain radiographs plus 3D CT 2 weeks later. Four weeks later, they repeated the same survey to evaluate intraobserver reliability. Reliability was assessed on the basis of Fleiss k values. Results: On the basis of plain radiographs alone, interobserver and intraobserver reliability of the Rockwood classification were fair (k ¼ .214) and moderate (k ¼ .474), respectively. Interobserver and intraobserver reliability of treatment were both fair (k ¼ .213 and .399, respectively). On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability of the Rockwood classification were slight (k ¼ .177) and moderate (k ¼ .565), respectively. Interobserver and intraobserver reliability of treatment were fair (k ¼ .253) and moderate (k ¼ .554), respectively. There were no significant differences in reliability between the two groups in terms of any k values. Conclusion: This study suggests an overall lack of reliability of the Rockwood classification of AC joint dislocations and of decisions regarding their treatment. There is especially poor agreement between experienced shoulder surgeons. The addition of 3D CT did not improve reliability of classification and treatment of AC joint dislocations. Level of evidence: Level I, Diagnostic Study. Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees. Keywords: Acromioclavicular joint; dislocation; Rockwood classification; treatment; reliability; interobserver; intraobserver IRB approval: The authors obtained the approval of Keimyung University Dongsan Medical Center institutional review board (IRB number 1305030).
*Reprint requests: Chul-Hyun Cho, MD, PhD, Department of Orthopedic Surgery, Dongsan Medical Center, School of Medicine, Keimyung University, 56 Dalseong-ro, Jung-gu, Daegu 700-712, South Korea. E-mail address: [email protected] (C.-H. Cho).
1058-2746/$ - see front matter Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees. http://dx.doi.org/10.1016/j.jse.2014.02.005
666
Acromioclavicular (AC) joint dislocation is a common shoulder injury, representing 9% to 10% of all shoulder injuries.7,8 Despite its prevalence, however, the lack of consensus regarding its diagnosis and treatment makes it one of the most controversial shoulder injuries.3,15 Although diagnosis and treatment are based on the surgeon’s qualitative assessment combining medical history, physical examination, and radiographic evaluation, injury severity that is based on a classification system is the main determinant of whether management will be conservative or operative.5 In 1963, Tossy et al18 classified AC joint injuries into types I, II, and III. Then in 1984, Rockwood et al12 expanded the classification to include types IV, V, and VI. The expanded classification recognized a variety of complete AC joint dislocations. The Rockwood classification system is important for surgeons to accurately diagnose AC joint injuries and is used in the literature to guide nonoperative versus operative management.10 However, there is substantial variation in the classification and diagnosis of AC joint dislocations.14 Classification systems should be reliable and reproducible, yet there is little evidence proving the reliability of the Rockwood classification for AC joint dislocations. To the best of our knowledge, only one study of the system’s interobserver and intraobserver reliability has been reported. Kraeutler et al5 documented that individual surgeons were consistent in their grading with use of the Rockwood classification system but that observer agreement among surgeons was lower. The management of AC joint dislocations, in particular of Rockwood type III versus type IV and of type III versus type V, remains a difficult issue. Because there is a trend toward nonoperative treatment of type III dislocations but operative treatment of type IV and type V dislocations, distinction between these types is critical in the clinical setting.5,9 Although axial radiography of the shoulder is important for making that distinction, type IV injuries are sometimes difficult to diagnose and require an adequate axial view to determine posterior translation.5,7,11,16 Several studies have documented that plain radiographs may not be sufficiently accurate for determining type IV injuries.11,16 Additional radiographic tools, including functional axial radiography, computed tomography (CT), ultrasound, and magnetic resonance imaging, are needed for more exact assessment of the severity and direction of AC joint dislocations.1,4,8,11,16,20 The primary objective of this study was to evaluate interobserver and intraobserver reliability of the classification and treatment of AC joint dislocations. The secondary objective was to assess the impact of adding 3dimensional computed tomography (3D CT) on the reliability of classification and treatment choice. Our hypothesis was that there would be poor interobserver and intraobserver reliability regarding Rockwood classification
C.-H. Cho et al.
and treatment and that the addition of 3D CT would increase reliability over that of plain radiographs alone.
Materials and methods Patients and images Twenty-eight consecutive patients with acute AC joint dislocation were prospectively enrolled during a period of 3 years at our tertiary care hospital. Inclusion criteria were the presence of closure of the physeal plate, acute dislocation within 4 weeks after initial trauma, and availability of plain radiographs and 3D CT at the time of injury. Exclusion criteria were chronic dislocation, any other concomitant shoulder injury on the affected side, and history of shoulder surgery or trauma. The mean age of the patients was 42.2 15.1 years (range, 17-66 years); 25 (89.3%) were men, and 3 (10.7%) were women. Plain radiographs, including an anteroposterior view of the affected and contralateral AC joints and an axial view of the shoulder, and 3D CT were taken by the same technique with the patient in the supine position at the time of initial injury. Both shoulder anteroposterior and axial plain radiographs and both anterior and superior 3D CT reconstructed images were scanned (400 dpi) and uploaded with a web-based survey system (Google Docs survey tool, Google Inc, Mountain View, CA, USA).
Observers and survey Ten independent observers who had a mean of 11.2 years (range, 4-18 years) of experience as shoulder surgeons were recruited from 10 residency or fellowship training hospitals. We designed a web-based survey and presented 28 cases randomly to the observers in 2 formats: (1) plain radiographs alone or (2) a combination of plain radiographs and 3D CT. We did not provide observers with any information about the patients, including purpose of the study, medical history, or physical examination, except for the patients’ age and sex. For each case, the observers were asked 2 questions: (1) What is the Rockwood type of AC joint dislocation? and (2) Which treatment do you recommenddconservative or operative? With use of the Rockwood classification, each injury was classified into 1 of 6 types: type I, AC ligament sprain with the AC joint intact; type II, AC ligament tear with AC joint subluxation and coracoclavicular ligaments intact; type III, AC and coracoclavicular ligaments torn with 100% AC joint dislocation; type IV, complete dislocation with posterior displacement of the distal clavicle into or through the trapezius muscle; type V, exaggerated superior AC joint dislocation of 100% to 300%, including disruption of the deltotrapezius fascia; and type VI, complete dislocation with inferior displacement of the distal clavicle into a subacromial or subcoracoid position.12 Images from each case were randomly presented to observers, with plain radiographs alone being presented first, followed by plain radiographs plus 3D CT after 2 weeks. Four weeks later, the observers repeated the same survey with randomization to evaluate intraobserver reliability. There was no feedback for the previous survey.
Reliability of acromioclavicular dislocation
Statistical methods The SPSS statistical package (version 20.0; IBM, Armonk, NY, USA) was used for data analysis. Intraobserver reliability and interobserver reliability were assessed by calculation of the k correlation coefficient (with 1.0 representing total agreement and 0 representing no agreement) as described by Fleiss,19 and the interpretation of k coefficients was performed with the criteria of Landis and Koch.6 They defined a k value of > .8 as almost perfect agreement, between .6 and .8 as substantial agreement, between .4 and .6 as moderate agreement, between .2 and .4 as fair agreement, and < .2 as slight agreement. We used the paired Student t test to calculate statistical differences between the mean k values. P values of < .05 were considered significant.
Results The observers’ Rockwood classification and treatment choice for 28 cases with AC joint dislocation are listed in Table I.
Interobserver and intraobserver reliability of classification On the basis of plain radiographs alone, interobserver and intraobserver reliability of the Rockwood classification for AC joint dislocations were slight (k ¼ .214) and moderate (k ¼ .474), respectively. On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability for the classification were poor (k ¼ .177) and moderate (k ¼ .565), respectively. There were no significant differences in interobserver and intraobserver reliability between plain radiographs alone and a combination of plain radiographs and 3D CT (P > .05; Table II).
Interobserver and intraobserver reliability of treatment On the basis of plain radiographs alone, interobserver and intraobserver reliability of treatment were both slight (k ¼ .213 and .399, respectively). On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability for treatment were slight (k ¼ .253) and moderate (k ¼ .554), respectively. There were no significant differences in interobserver and intraobserver reliability between plain radiographs alone and a combination of plain radiographs and 3D CT (P > .05; Table III).
Case presentation Case 1 involved a 58-year-old man with AC joint dislocation. On the basis of plain radiographs alone, 7 observers classified the injury as type III, 1 observer as type IV, and 2 observers as type V. Seven observers decided that operation was necessary, and 3 decided that conservative treatment
667 Table I Distribution of observers’ Rockwood classification and treatment choice for 28 cases with acromioclavicular joint dislocation Case No. Rockwood classification
Treatment
Type Type Type Type Operative Conservative II III IV V 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
0 2 1 0 0 6 3 0 0 0 3 0 7 1 0 0 0 2 0 1 0 0 4 0 0 0 0 0
7 6 8 2 5 3 5 6 0 0 6 2 3 8 0 3 5 6 3 3 3 0 5 8 0 7 3 6
1 1 0 6 0 1 2 1 1 7 1 7 0 0 0 6 3 0 6 6 6 2 1 0 0 0 5 0
2 1 1 2 5 0 0 3 9 3 0 1 0 1 10 1 2 2 1 0 1 8 0 2 10 3 2 4
7 4 5 10 10 3 3 8 10 10 3 10 2 3 10 9 7 6 9 8 8 10 3 3 10 7 8 8
3 6 5 0 0 7 7 2 0 0 7 0 8 7 0 1 3 4 1 2 2 0 7 7 0 3 2 2
was sufficient. On the basis of a combination of plain radiographs and 3D CT, 3 observers classified the injury as type III, 5 observers as type IV, and 2 observers as type V. Nine observers decided that operation was necessary, and 1 observer decided that conservative treatment was sufficient (Fig. 1).
Discussion This study suggests an overall lack of reliability of the Rockwood classification for AC joint dislocations and of decisions regarding their treatment. There is especially poor agreement among experienced shoulder surgeons regarding the decision of injury type and treatment. The addition of 3D CT did not improve reliability of classification and treatment of AC joint dislocations. Acute AC joint dislocations still represent a diagnostic and therapeutic challenge for shoulder surgeons.15 The Rockwood classification is the most widely used system for
668
C.-H. Cho et al.
Table II Interobserver and intraobserver reliability of the Rockwood classification for AC joint dislocations
Table III Interobserver and intraobserver reliability of treatment of AC joint dislocations
Aspect
Plain radiographs alone
Plain radiographs and 3D CT
P value
Aspect
Plain radiographs alone
Plain radiographs and 3D CT
P value
Interobserver reliability First round Second round Intraobserver reliability
.214
.177
.667
.213
.253
.562
.292 .136 .474
.191 .163 .565
Interobserver reliability First round Second round Intraobserver reliability
.299 .127 .399
.290 .215 .554
.083
.066
AC, acromioclavicular; 3D CT, 3-dimensional computed tomography.
these dislocations, with grading based on radiologic measurements: amount of superior displacement of the distal clavicle above the superior border of the acromion, coracoclavicular distance, and posterior displacement of the distal clavicle. An ideal classification system should be reliable, reproducible, clinically useful, and an accurate guide to treatment and should be able to predict prognosis. Kraeutler et al5 studied the reliability of the Rockwood classification in patients with types III, IV, and V AC joint dislocations. They documented that individual surgeons were consistent in their grading of dislocations (k ¼ .694, substantial agreement) but that there was less observer agreement between surgeons (k ¼ .602, substantial agreement).5 Our results are inconsistent with their findings. We found only slight interobserver (k ¼ .214) and moderate intraobserver (k ¼ .474) reliability for the classification system, which translated to an overall lack of reliability, especially poor interobserver reliability. Regarding the management of AC joint dislocations, there is little debate about conservative treatment of Rockwood types I and II and operative treatment of types IV, V, and VI.2,5,10,13,14,17 However, type III dislocations are still being disputed.2,3,10,13,14,17 Furthermore, the surgeon’s decision for operation can have substantial variation because surgeons may have a different grading in the same patient with AC joint dislocation. In this study, interobserver and intraobserver reliability of treatment of AC joint dislocations were slight (k ¼ .213) and fair (k ¼ .399), respectively, and we found poor interobserver and intraobserver reliability for treatment. These results are different from those of Kraeutler et al.5 In the management algorithm for acute AC joint dislocations, complete radiologic evaluation represents the key to successful therapy.15 Although the management of acute type III injuries remains controversial, there is a trend toward conservative treatment, so distinction of type III, IV, and V injuries is critical in clinical practice.5 A clear axial view of the shoulder is important in differentiating a type III injury from a type IV injury. However, type IV injuries are difficult to diagnose because
AC, acromioclavicular; 3D CT, 3-dimensional computed tomography.
acute pain sometimes makes it impossible to obtain an adequate axial view to determine posterior translation.15 Several studies have documented that plain radiographs may not be sufficiently accurate for identifying type IV injuries.11,15,16 Newer diagnostic modalities, including axillary functional radiography, CT, ultrasound, and magnetic resonance imaging, allow assessment of horizontal instability and detection of underestimated or misdiagnosed AC joint dislocations.1,4,8,11,16,20 The secondary objective of this study was to assess the impact of adding 3D CT on the reliability of the Rockwood classification. The decision to use 3D CT was based on the belief that it would better delineate posterior displacement of the distal clavicle. This was thought to be due to the difficulty in identifying posterior displacement on plain radiographs because of the patient’s pain in the acute setting.15 With the addition of 3D CT, the interobserver and intraobserver reliability of the Rockwood classification changed from k ¼ .214 to k ¼ .177 and from k ¼ .474 to k ¼ .565, respectively, and interobserver and intraobserver reliability of treatment choice changed from k ¼ .213 to k ¼ .253 and from k ¼ .399 to k ¼ .554, respectively. Our results revealed that although the addition of 3D CT did improve reliability, the improvement did not reach statistical significance. Because CT involves increased radiation and cost, we think that the use of 3D CT should be carefully considered for diagnosis and treatment of AC joint dislocation. Inconsistent results after nonsurgical treatment of type II and type III injuries led us to believe that horizontally unstable injuries are missed during primary radiologic examination.16 However, the clavicle may appear subluxated posteriorly on standard axial radiographs of healthy shoulders.11 This subluxation might be viewed as an indication for surgical stabilization in AC joint dislocations. The standard axial radiograph has a high sensitivity but poor accuracy in identifying posterior translation in the AC joint.11 Heers and Hedtmann4 described the radiologic definition of type IV injury as 100% posterior dislocation of the anteroposterior width of the clavicle. However, there is
Reliability of acromioclavicular dislocation
Figure 1
669
Case presentation: acromioclavicular joint dislocation in a 58-year-old man.
still no consensus on the definition of this type of injury. Because of this lack of consensus, we think there is a need for an exact quantitative definition of posterior displacement in type IV injuries. The distinction between grades III and V makes it important to identify patients who need operative treatment.5 The lack of consensus among surgeons in this situation could be due to the lack of agreement regarding classification, particularly between type III and type V injuries.5 Perhaps the greatest reason for this disagreement is that although the designation of type V dislocation is based on coracoclavicular distance between the top of the coracoid and the bottom of the clavicle, usually at approximately 300% compared with that in the normal side, surgeons may still make a diagnosis of type V if the displacement exceeds 100%, even if it is not as large as 300%.5,20 For displacements close to 100%, it may be difficult to decide between a type III and a type V diagnosis.5 Our study had several limitations. First, we did not perform a power test. To assess the impact of additional radiologic tools for reliability, further well-designed studies with larger cohorts may be needed. Second, decisions regarding diagnosis and treatment for each patient in our study were based on radiologic evaluation and limited information: age and sex only. It may be difficult to grade an AC joint dislocation and to decide how best to treat it on the basis of such limited information. The patient’s age, sex, hand dominance, level of activity or sports participation, and medical status may often be more important than the injury itself in choosing the method of treatment. Third, recall bias is possible because of the repeated survey.
However, to minimize this effect, images of each case were randomly presented to observers after an interval of 2 weeks every time. In terms of strengths, ours is the first study to evaluate interobserver and intraobserver reliability of the classification and treatment of AC joint dislocations and to assess the impact of adding 3D CT on the reliability of classification and treatment choices by experienced shoulder specialists. Because level of clinical experience was expected to affect the reliability of classification assessment and treatment choice, we recruited 10 independent shoulder surgeons with an average of 11.2 years of experience in hospital residency or fellowship training.
Conclusion Our study demonstrated that the Rockwood classification system has limited interobserver and intraobserver reliability, even with the help of 3D CT. Inconsistent agreement with use of this system limits its role in clinical decision-making. We suggest a need for a better grading system and for radiographic tools to quantify injury severity and to improve interobserver agreement.
Acknowledgment We thank Gi Hyuk Moon, MD, In Hyeok Rhyou, MD, Dong-Ju Shin, MD, Jong Pil Yoon, MD, and Ho Min
670 Lee, MD, for participating in the survey. We also thank Sang-Hee Rho for support with data collection.
Disclaimer The authors, their immediate families, and any research foundations with which they are affiliated did not receive any financial payments or other benefits from any commercial entity related to the subject of this article.
References 1. Di Francesco A, Zoccali C, Colafarina O, Pizzoferrato R, Flamini S. The use of hook plate in type III and V acromio-clavicular Rockwood dislocations: clinical and radiological midterm results and MRI evaluation in 42 patients. Injury 2012;43:147-52. http://dx.doi.org/10. 1016/j.injury.2011.04.002 2. Farber AJ, Cascio BM, Wilckens JH. Type III acromioclavicular separation: rationale for anatomical reconstruction. Am J Orthop (Belle Mead NJ) 2008;37:349-55. 3. Gstettner C, Tauber M, Hitzl W, Resch H. Rockwood type III acromioclavicular dislocation: surgical versus conservative treatment. J Shoulder Elbow Surg 2008;17:220-5. http://dx.doi.org/10.1016/j.jse. 2007.07.017 4. Heers G, Hedtmann A. Correlation of ultrasonographic findings to Tossy’s and Rockwood’s classification of acromioclavicular joint injuries. Ultrasound Med Biol 2005;31:725-32. http://dx.doi.org/10. 1016/j.ultrasmedbio.2005.03.002 5. Kraeutler MJ, Williams GR Jr, Cohen SB, Ciccotti MG, Dines JS, Altchek DW, et al. Inter- and intraobserver reliability of the radiographic diagnosis and treatment of acromioclavicular joint separations. Orthopedics 2012;35:e1483-7. http://dx.doi.org/10.3928/ 01477447-20120919-16 6. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 7. Mazzocca AD, Arciero RA, Bicos J. Evaluation and treatment of acromioclavicular joint injuries. Am J Sports Med 2007;35:316-29. http://dx.doi.org/10.1177/0363546506298022
C.-H. Cho et al. 8. Nemec U, Oberleitner G, Nemec SF, Gruber M, Weber M, Czermy C, et al. MRI versus radiography of acromioclavicular joint dislocation. AJR Am J Roentgenol 2011;197:968-73. http://dx.doi.org/10.2214/ AJR.10.6378 9. Nissen CW, Chatterjee A. Type III acromioclavicular separation: results of a recent survey on its management. Am J Orthop (Belle Mead NJ) 2007;36:89-93. 10. Pallis M, Cameron KL, Svoboda SJ, Owens BD. Epidemiology of acromioclavicular joint injury in young athletes. Am J Sports Med 2012;40:2072-7. http://dx.doi.org/10.1177/0363546512450162 11. Rahm S, Wieser K, Spross C, Vich M, Gerber C, Meyer DC. Standard axillary radiographs of the shoulder may mimic posterior subluxation of the lateral end of the clavicle. J Orthop Trauma 2013;27:622-6. http://dx.doi.org/10.1097/BOT.0b013e31828f912c 12. Rockwood CJ, Williams G, Young D. Disorders of the acromioclavicular joint. In: Rockwood CA, Matsen FA III, Wirth MA, Lippitt SB, editors. The Shoulder. 3rd ed. Philadelphia: WB Saunders; 2004. p. 521-95 (ISBN: 978-0721601489). 13. Salem KH, Schmelz A. Treatment of Tossy III acromioclavicular joint injuries using hook plates and ligament suture. J Orthop Trauma 2009; 23:565-9. http://dx.doi.org/10.1097/BOT.0b013e3181971b38 14. Smith TO, Chester R, Pearse EO, Hing CB. Operative versus nonoperative management following Rockwood grade III acromioclavicular separation: a meta-analysis of the current evidence base. J Orthop Traumatol 2011;12:19-27. http://dx.doi.org/10.1007/s10195011-0127-1 15. Tauber M. Management of acute acromioclavicular joint dislocations: current concepts. Arch Orthop Trauma Surg 2013;133:985-95. http:// dx.doi.org/10.1007/s00402-013-1748-z 16. Tauber M, Koller H, Hitzl W, Resch H. Dynamic radiologic evaluation of horizontal instability in acute acromioclavicular joint dislocations. Am J Sports Med 2010;38:1188-95. http://dx.doi.org/10.1177/ 0363546510361951 17. Tischer T, Salzmann GM, El-Azab H, Vogt S, Imhoff AB. Incidence of associated injuries with acute acromioclavicular joint dislocations types III through V. Am J Sports Med 2009;37:136-9. http://dx.doi. org/10.1177/0363546508322891 18. Tossy JD, Mead NC, Sigmond HM. Acromioclavicular separations: useful and practical classification for treatment. Clin Orthop Relat Res 1963;28:111-9. 19. Venkataraman G, Ananthanarayanan V, Paner GP. Accessible calculation of multirater kappa statistics for pathologists. Virchows Arch 2006;449:272. http://dx.doi.org/10.1007/s00428-006-0209-2 20. von Heideken J, Bostr€om Windhamre H, Une-Larsson V, Ekelund A. Acute surgical treatment of acromioclavicular dislocation type V with a hook plate: superiority to late reconstruction. J Shoulder Elbow Surg 2013;22:9-17. http://dx.doi.org/10.1016/j.jse.2012.03.003
www.elsevier.com/locate/ymse
INTERNATIONAL SHOULDER AND ELBOW ARTICLES
Reliability of the classification and treatment of dislocations of the acromioclavicular joint Chul-Hyun Cho, MD, PhDa,*, Ilseon Hwang, MD, PhDb, Jae-Sung Seo, MDc, Chang-Hyuk Choi, MDd, Sang-Hun Ko, MDe, Hyung Bin Park, MDf, Jinmyoung Dan, MDg a
Pain Research Center, Department of Orthopedic Surgery, Dongsan Medical Center, School of Medicine, Keimyung University, Daegu, South Korea b Department of Pathology, School of Medicine, Keimyung University, Daegu, South Korea c Department of Orthopedic Surgery, Yeungnam University Hospital, Daegu, South Korea d Department of Orthopedic Surgery, Daegu Catholic University Medical Center, Daegu, South Korea e Department of Orthopedic Surgery, Ulsan University Hospital, Ulsan, South Korea f Department of Orthopedic Surgery, Gyeongsang National University Hospital, Jinju, South Korea g Department of Orthopedic Surgery, Gumi CHA Hospital, CHA University, Gumi, South Korea Background: We evaluated interobserver and intraobserver reliability of the classification and treatment of acromioclavicular (AC) joint dislocations and assessed the impact of adding 3-dimensional computed tomography (3D CT) on the reliability of classification and treatment choice. Methods: Ten surgeons independently reviewed plain radiographs and 3D CT in 28 cases with AC joint dislocation. Images from each case were randomly presented to the observers, with plain radiographs alone being presented first, followed by plain radiographs plus 3D CT 2 weeks later. Four weeks later, they repeated the same survey to evaluate intraobserver reliability. Reliability was assessed on the basis of Fleiss k values. Results: On the basis of plain radiographs alone, interobserver and intraobserver reliability of the Rockwood classification were fair (k ¼ .214) and moderate (k ¼ .474), respectively. Interobserver and intraobserver reliability of treatment were both fair (k ¼ .213 and .399, respectively). On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability of the Rockwood classification were slight (k ¼ .177) and moderate (k ¼ .565), respectively. Interobserver and intraobserver reliability of treatment were fair (k ¼ .253) and moderate (k ¼ .554), respectively. There were no significant differences in reliability between the two groups in terms of any k values. Conclusion: This study suggests an overall lack of reliability of the Rockwood classification of AC joint dislocations and of decisions regarding their treatment. There is especially poor agreement between experienced shoulder surgeons. The addition of 3D CT did not improve reliability of classification and treatment of AC joint dislocations. Level of evidence: Level I, Diagnostic Study. Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees. Keywords: Acromioclavicular joint; dislocation; Rockwood classification; treatment; reliability; interobserver; intraobserver IRB approval: The authors obtained the approval of Keimyung University Dongsan Medical Center institutional review board (IRB number 1305030).
*Reprint requests: Chul-Hyun Cho, MD, PhD, Department of Orthopedic Surgery, Dongsan Medical Center, School of Medicine, Keimyung University, 56 Dalseong-ro, Jung-gu, Daegu 700-712, South Korea. E-mail address: [email protected] (C.-H. Cho).
1058-2746/$ - see front matter Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees. http://dx.doi.org/10.1016/j.jse.2014.02.005
666
Acromioclavicular (AC) joint dislocation is a common shoulder injury, representing 9% to 10% of all shoulder injuries.7,8 Despite its prevalence, however, the lack of consensus regarding its diagnosis and treatment makes it one of the most controversial shoulder injuries.3,15 Although diagnosis and treatment are based on the surgeon’s qualitative assessment combining medical history, physical examination, and radiographic evaluation, injury severity that is based on a classification system is the main determinant of whether management will be conservative or operative.5 In 1963, Tossy et al18 classified AC joint injuries into types I, II, and III. Then in 1984, Rockwood et al12 expanded the classification to include types IV, V, and VI. The expanded classification recognized a variety of complete AC joint dislocations. The Rockwood classification system is important for surgeons to accurately diagnose AC joint injuries and is used in the literature to guide nonoperative versus operative management.10 However, there is substantial variation in the classification and diagnosis of AC joint dislocations.14 Classification systems should be reliable and reproducible, yet there is little evidence proving the reliability of the Rockwood classification for AC joint dislocations. To the best of our knowledge, only one study of the system’s interobserver and intraobserver reliability has been reported. Kraeutler et al5 documented that individual surgeons were consistent in their grading with use of the Rockwood classification system but that observer agreement among surgeons was lower. The management of AC joint dislocations, in particular of Rockwood type III versus type IV and of type III versus type V, remains a difficult issue. Because there is a trend toward nonoperative treatment of type III dislocations but operative treatment of type IV and type V dislocations, distinction between these types is critical in the clinical setting.5,9 Although axial radiography of the shoulder is important for making that distinction, type IV injuries are sometimes difficult to diagnose and require an adequate axial view to determine posterior translation.5,7,11,16 Several studies have documented that plain radiographs may not be sufficiently accurate for determining type IV injuries.11,16 Additional radiographic tools, including functional axial radiography, computed tomography (CT), ultrasound, and magnetic resonance imaging, are needed for more exact assessment of the severity and direction of AC joint dislocations.1,4,8,11,16,20 The primary objective of this study was to evaluate interobserver and intraobserver reliability of the classification and treatment of AC joint dislocations. The secondary objective was to assess the impact of adding 3dimensional computed tomography (3D CT) on the reliability of classification and treatment choice. Our hypothesis was that there would be poor interobserver and intraobserver reliability regarding Rockwood classification
C.-H. Cho et al.
and treatment and that the addition of 3D CT would increase reliability over that of plain radiographs alone.
Materials and methods Patients and images Twenty-eight consecutive patients with acute AC joint dislocation were prospectively enrolled during a period of 3 years at our tertiary care hospital. Inclusion criteria were the presence of closure of the physeal plate, acute dislocation within 4 weeks after initial trauma, and availability of plain radiographs and 3D CT at the time of injury. Exclusion criteria were chronic dislocation, any other concomitant shoulder injury on the affected side, and history of shoulder surgery or trauma. The mean age of the patients was 42.2 15.1 years (range, 17-66 years); 25 (89.3%) were men, and 3 (10.7%) were women. Plain radiographs, including an anteroposterior view of the affected and contralateral AC joints and an axial view of the shoulder, and 3D CT were taken by the same technique with the patient in the supine position at the time of initial injury. Both shoulder anteroposterior and axial plain radiographs and both anterior and superior 3D CT reconstructed images were scanned (400 dpi) and uploaded with a web-based survey system (Google Docs survey tool, Google Inc, Mountain View, CA, USA).
Observers and survey Ten independent observers who had a mean of 11.2 years (range, 4-18 years) of experience as shoulder surgeons were recruited from 10 residency or fellowship training hospitals. We designed a web-based survey and presented 28 cases randomly to the observers in 2 formats: (1) plain radiographs alone or (2) a combination of plain radiographs and 3D CT. We did not provide observers with any information about the patients, including purpose of the study, medical history, or physical examination, except for the patients’ age and sex. For each case, the observers were asked 2 questions: (1) What is the Rockwood type of AC joint dislocation? and (2) Which treatment do you recommenddconservative or operative? With use of the Rockwood classification, each injury was classified into 1 of 6 types: type I, AC ligament sprain with the AC joint intact; type II, AC ligament tear with AC joint subluxation and coracoclavicular ligaments intact; type III, AC and coracoclavicular ligaments torn with 100% AC joint dislocation; type IV, complete dislocation with posterior displacement of the distal clavicle into or through the trapezius muscle; type V, exaggerated superior AC joint dislocation of 100% to 300%, including disruption of the deltotrapezius fascia; and type VI, complete dislocation with inferior displacement of the distal clavicle into a subacromial or subcoracoid position.12 Images from each case were randomly presented to observers, with plain radiographs alone being presented first, followed by plain radiographs plus 3D CT after 2 weeks. Four weeks later, the observers repeated the same survey with randomization to evaluate intraobserver reliability. There was no feedback for the previous survey.
Reliability of acromioclavicular dislocation
Statistical methods The SPSS statistical package (version 20.0; IBM, Armonk, NY, USA) was used for data analysis. Intraobserver reliability and interobserver reliability were assessed by calculation of the k correlation coefficient (with 1.0 representing total agreement and 0 representing no agreement) as described by Fleiss,19 and the interpretation of k coefficients was performed with the criteria of Landis and Koch.6 They defined a k value of > .8 as almost perfect agreement, between .6 and .8 as substantial agreement, between .4 and .6 as moderate agreement, between .2 and .4 as fair agreement, and < .2 as slight agreement. We used the paired Student t test to calculate statistical differences between the mean k values. P values of < .05 were considered significant.
Results The observers’ Rockwood classification and treatment choice for 28 cases with AC joint dislocation are listed in Table I.
Interobserver and intraobserver reliability of classification On the basis of plain radiographs alone, interobserver and intraobserver reliability of the Rockwood classification for AC joint dislocations were slight (k ¼ .214) and moderate (k ¼ .474), respectively. On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability for the classification were poor (k ¼ .177) and moderate (k ¼ .565), respectively. There were no significant differences in interobserver and intraobserver reliability between plain radiographs alone and a combination of plain radiographs and 3D CT (P > .05; Table II).
Interobserver and intraobserver reliability of treatment On the basis of plain radiographs alone, interobserver and intraobserver reliability of treatment were both slight (k ¼ .213 and .399, respectively). On the basis of a combination of plain radiographs and 3D CT, interobserver and intraobserver reliability for treatment were slight (k ¼ .253) and moderate (k ¼ .554), respectively. There were no significant differences in interobserver and intraobserver reliability between plain radiographs alone and a combination of plain radiographs and 3D CT (P > .05; Table III).
Case presentation Case 1 involved a 58-year-old man with AC joint dislocation. On the basis of plain radiographs alone, 7 observers classified the injury as type III, 1 observer as type IV, and 2 observers as type V. Seven observers decided that operation was necessary, and 3 decided that conservative treatment
667 Table I Distribution of observers’ Rockwood classification and treatment choice for 28 cases with acromioclavicular joint dislocation Case No. Rockwood classification
Treatment
Type Type Type Type Operative Conservative II III IV V 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
0 2 1 0 0 6 3 0 0 0 3 0 7 1 0 0 0 2 0 1 0 0 4 0 0 0 0 0
7 6 8 2 5 3 5 6 0 0 6 2 3 8 0 3 5 6 3 3 3 0 5 8 0 7 3 6
1 1 0 6 0 1 2 1 1 7 1 7 0 0 0 6 3 0 6 6 6 2 1 0 0 0 5 0
2 1 1 2 5 0 0 3 9 3 0 1 0 1 10 1 2 2 1 0 1 8 0 2 10 3 2 4
7 4 5 10 10 3 3 8 10 10 3 10 2 3 10 9 7 6 9 8 8 10 3 3 10 7 8 8
3 6 5 0 0 7 7 2 0 0 7 0 8 7 0 1 3 4 1 2 2 0 7 7 0 3 2 2
was sufficient. On the basis of a combination of plain radiographs and 3D CT, 3 observers classified the injury as type III, 5 observers as type IV, and 2 observers as type V. Nine observers decided that operation was necessary, and 1 observer decided that conservative treatment was sufficient (Fig. 1).
Discussion This study suggests an overall lack of reliability of the Rockwood classification for AC joint dislocations and of decisions regarding their treatment. There is especially poor agreement among experienced shoulder surgeons regarding the decision of injury type and treatment. The addition of 3D CT did not improve reliability of classification and treatment of AC joint dislocations. Acute AC joint dislocations still represent a diagnostic and therapeutic challenge for shoulder surgeons.15 The Rockwood classification is the most widely used system for
668
C.-H. Cho et al.
Table II Interobserver and intraobserver reliability of the Rockwood classification for AC joint dislocations
Table III Interobserver and intraobserver reliability of treatment of AC joint dislocations
Aspect
Plain radiographs alone
Plain radiographs and 3D CT
P value
Aspect
Plain radiographs alone
Plain radiographs and 3D CT
P value
Interobserver reliability First round Second round Intraobserver reliability
.214
.177
.667
.213
.253
.562
.292 .136 .474
.191 .163 .565
Interobserver reliability First round Second round Intraobserver reliability
.299 .127 .399
.290 .215 .554
.083
.066
AC, acromioclavicular; 3D CT, 3-dimensional computed tomography.
these dislocations, with grading based on radiologic measurements: amount of superior displacement of the distal clavicle above the superior border of the acromion, coracoclavicular distance, and posterior displacement of the distal clavicle. An ideal classification system should be reliable, reproducible, clinically useful, and an accurate guide to treatment and should be able to predict prognosis. Kraeutler et al5 studied the reliability of the Rockwood classification in patients with types III, IV, and V AC joint dislocations. They documented that individual surgeons were consistent in their grading of dislocations (k ¼ .694, substantial agreement) but that there was less observer agreement between surgeons (k ¼ .602, substantial agreement).5 Our results are inconsistent with their findings. We found only slight interobserver (k ¼ .214) and moderate intraobserver (k ¼ .474) reliability for the classification system, which translated to an overall lack of reliability, especially poor interobserver reliability. Regarding the management of AC joint dislocations, there is little debate about conservative treatment of Rockwood types I and II and operative treatment of types IV, V, and VI.2,5,10,13,14,17 However, type III dislocations are still being disputed.2,3,10,13,14,17 Furthermore, the surgeon’s decision for operation can have substantial variation because surgeons may have a different grading in the same patient with AC joint dislocation. In this study, interobserver and intraobserver reliability of treatment of AC joint dislocations were slight (k ¼ .213) and fair (k ¼ .399), respectively, and we found poor interobserver and intraobserver reliability for treatment. These results are different from those of Kraeutler et al.5 In the management algorithm for acute AC joint dislocations, complete radiologic evaluation represents the key to successful therapy.15 Although the management of acute type III injuries remains controversial, there is a trend toward conservative treatment, so distinction of type III, IV, and V injuries is critical in clinical practice.5 A clear axial view of the shoulder is important in differentiating a type III injury from a type IV injury. However, type IV injuries are difficult to diagnose because
AC, acromioclavicular; 3D CT, 3-dimensional computed tomography.
acute pain sometimes makes it impossible to obtain an adequate axial view to determine posterior translation.15 Several studies have documented that plain radiographs may not be sufficiently accurate for identifying type IV injuries.11,15,16 Newer diagnostic modalities, including axillary functional radiography, CT, ultrasound, and magnetic resonance imaging, allow assessment of horizontal instability and detection of underestimated or misdiagnosed AC joint dislocations.1,4,8,11,16,20 The secondary objective of this study was to assess the impact of adding 3D CT on the reliability of the Rockwood classification. The decision to use 3D CT was based on the belief that it would better delineate posterior displacement of the distal clavicle. This was thought to be due to the difficulty in identifying posterior displacement on plain radiographs because of the patient’s pain in the acute setting.15 With the addition of 3D CT, the interobserver and intraobserver reliability of the Rockwood classification changed from k ¼ .214 to k ¼ .177 and from k ¼ .474 to k ¼ .565, respectively, and interobserver and intraobserver reliability of treatment choice changed from k ¼ .213 to k ¼ .253 and from k ¼ .399 to k ¼ .554, respectively. Our results revealed that although the addition of 3D CT did improve reliability, the improvement did not reach statistical significance. Because CT involves increased radiation and cost, we think that the use of 3D CT should be carefully considered for diagnosis and treatment of AC joint dislocation. Inconsistent results after nonsurgical treatment of type II and type III injuries led us to believe that horizontally unstable injuries are missed during primary radiologic examination.16 However, the clavicle may appear subluxated posteriorly on standard axial radiographs of healthy shoulders.11 This subluxation might be viewed as an indication for surgical stabilization in AC joint dislocations. The standard axial radiograph has a high sensitivity but poor accuracy in identifying posterior translation in the AC joint.11 Heers and Hedtmann4 described the radiologic definition of type IV injury as 100% posterior dislocation of the anteroposterior width of the clavicle. However, there is
Reliability of acromioclavicular dislocation
Figure 1
669
Case presentation: acromioclavicular joint dislocation in a 58-year-old man.
still no consensus on the definition of this type of injury. Because of this lack of consensus, we think there is a need for an exact quantitative definition of posterior displacement in type IV injuries. The distinction between grades III and V makes it important to identify patients who need operative treatment.5 The lack of consensus among surgeons in this situation could be due to the lack of agreement regarding classification, particularly between type III and type V injuries.5 Perhaps the greatest reason for this disagreement is that although the designation of type V dislocation is based on coracoclavicular distance between the top of the coracoid and the bottom of the clavicle, usually at approximately 300% compared with that in the normal side, surgeons may still make a diagnosis of type V if the displacement exceeds 100%, even if it is not as large as 300%.5,20 For displacements close to 100%, it may be difficult to decide between a type III and a type V diagnosis.5 Our study had several limitations. First, we did not perform a power test. To assess the impact of additional radiologic tools for reliability, further well-designed studies with larger cohorts may be needed. Second, decisions regarding diagnosis and treatment for each patient in our study were based on radiologic evaluation and limited information: age and sex only. It may be difficult to grade an AC joint dislocation and to decide how best to treat it on the basis of such limited information. The patient’s age, sex, hand dominance, level of activity or sports participation, and medical status may often be more important than the injury itself in choosing the method of treatment. Third, recall bias is possible because of the repeated survey.
However, to minimize this effect, images of each case were randomly presented to observers after an interval of 2 weeks every time. In terms of strengths, ours is the first study to evaluate interobserver and intraobserver reliability of the classification and treatment of AC joint dislocations and to assess the impact of adding 3D CT on the reliability of classification and treatment choices by experienced shoulder specialists. Because level of clinical experience was expected to affect the reliability of classification assessment and treatment choice, we recruited 10 independent shoulder surgeons with an average of 11.2 years of experience in hospital residency or fellowship training.
Conclusion Our study demonstrated that the Rockwood classification system has limited interobserver and intraobserver reliability, even with the help of 3D CT. Inconsistent agreement with use of this system limits its role in clinical decision-making. We suggest a need for a better grading system and for radiographic tools to quantify injury severity and to improve interobserver agreement.
Acknowledgment We thank Gi Hyuk Moon, MD, In Hyeok Rhyou, MD, Dong-Ju Shin, MD, Jong Pil Yoon, MD, and Ho Min
670 Lee, MD, for participating in the survey. We also thank Sang-Hee Rho for support with data collection.
Disclaimer The authors, their immediate families, and any research foundations with which they are affiliated did not receive any financial payments or other benefits from any commercial entity related to the subject of this article.
References 1. Di Francesco A, Zoccali C, Colafarina O, Pizzoferrato R, Flamini S. The use of hook plate in type III and V acromio-clavicular Rockwood dislocations: clinical and radiological midterm results and MRI evaluation in 42 patients. Injury 2012;43:147-52. http://dx.doi.org/10. 1016/j.injury.2011.04.002 2. Farber AJ, Cascio BM, Wilckens JH. Type III acromioclavicular separation: rationale for anatomical reconstruction. Am J Orthop (Belle Mead NJ) 2008;37:349-55. 3. Gstettner C, Tauber M, Hitzl W, Resch H. Rockwood type III acromioclavicular dislocation: surgical versus conservative treatment. J Shoulder Elbow Surg 2008;17:220-5. http://dx.doi.org/10.1016/j.jse. 2007.07.017 4. Heers G, Hedtmann A. Correlation of ultrasonographic findings to Tossy’s and Rockwood’s classification of acromioclavicular joint injuries. Ultrasound Med Biol 2005;31:725-32. http://dx.doi.org/10. 1016/j.ultrasmedbio.2005.03.002 5. Kraeutler MJ, Williams GR Jr, Cohen SB, Ciccotti MG, Dines JS, Altchek DW, et al. Inter- and intraobserver reliability of the radiographic diagnosis and treatment of acromioclavicular joint separations. Orthopedics 2012;35:e1483-7. http://dx.doi.org/10.3928/ 01477447-20120919-16 6. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 7. Mazzocca AD, Arciero RA, Bicos J. Evaluation and treatment of acromioclavicular joint injuries. Am J Sports Med 2007;35:316-29. http://dx.doi.org/10.1177/0363546506298022
C.-H. Cho et al. 8. Nemec U, Oberleitner G, Nemec SF, Gruber M, Weber M, Czermy C, et al. MRI versus radiography of acromioclavicular joint dislocation. AJR Am J Roentgenol 2011;197:968-73. http://dx.doi.org/10.2214/ AJR.10.6378 9. Nissen CW, Chatterjee A. Type III acromioclavicular separation: results of a recent survey on its management. Am J Orthop (Belle Mead NJ) 2007;36:89-93. 10. Pallis M, Cameron KL, Svoboda SJ, Owens BD. Epidemiology of acromioclavicular joint injury in young athletes. Am J Sports Med 2012;40:2072-7. http://dx.doi.org/10.1177/0363546512450162 11. Rahm S, Wieser K, Spross C, Vich M, Gerber C, Meyer DC. Standard axillary radiographs of the shoulder may mimic posterior subluxation of the lateral end of the clavicle. J Orthop Trauma 2013;27:622-6. http://dx.doi.org/10.1097/BOT.0b013e31828f912c 12. Rockwood CJ, Williams G, Young D. Disorders of the acromioclavicular joint. In: Rockwood CA, Matsen FA III, Wirth MA, Lippitt SB, editors. The Shoulder. 3rd ed. Philadelphia: WB Saunders; 2004. p. 521-95 (ISBN: 978-0721601489). 13. Salem KH, Schmelz A. Treatment of Tossy III acromioclavicular joint injuries using hook plates and ligament suture. J Orthop Trauma 2009; 23:565-9. http://dx.doi.org/10.1097/BOT.0b013e3181971b38 14. Smith TO, Chester R, Pearse EO, Hing CB. Operative versus nonoperative management following Rockwood grade III acromioclavicular separation: a meta-analysis of the current evidence base. J Orthop Traumatol 2011;12:19-27. http://dx.doi.org/10.1007/s10195011-0127-1 15. Tauber M. Management of acute acromioclavicular joint dislocations: current concepts. Arch Orthop Trauma Surg 2013;133:985-95. http:// dx.doi.org/10.1007/s00402-013-1748-z 16. Tauber M, Koller H, Hitzl W, Resch H. Dynamic radiologic evaluation of horizontal instability in acute acromioclavicular joint dislocations. Am J Sports Med 2010;38:1188-95. http://dx.doi.org/10.1177/ 0363546510361951 17. Tischer T, Salzmann GM, El-Azab H, Vogt S, Imhoff AB. Incidence of associated injuries with acute acromioclavicular joint dislocations types III through V. Am J Sports Med 2009;37:136-9. http://dx.doi. org/10.1177/0363546508322891 18. Tossy JD, Mead NC, Sigmond HM. Acromioclavicular separations: useful and practical classification for treatment. Clin Orthop Relat Res 1963;28:111-9. 19. Venkataraman G, Ananthanarayanan V, Paner GP. Accessible calculation of multirater kappa statistics for pathologists. Virchows Arch 2006;449:272. http://dx.doi.org/10.1007/s00428-006-0209-2 20. von Heideken J, Bostr€om Windhamre H, Une-Larsson V, Ekelund A. Acute surgical treatment of acromioclavicular dislocation type V with a hook plate: superiority to late reconstruction. J Shoulder Elbow Surg 2013;22:9-17. http://dx.doi.org/10.1016/j.jse.2012.03.003