Intraobserver and interobserver agreement in the classification and treatment of midshaft clavicle fractures.

Background: With the recent emphasis on performing open reduction and internal fixation on midshaft clavicle fractures with complete displacement, comminution, and >2 cm of shortening, it is important to determine the reliability of orthopaedic surgeons to assess these variables on standard plain radiographs and to determine the agreement among orthopaedic surgeons in choosing the treatment.

Purpose: To determine the intra- and interobserver reliability in the classification of midshaft clavicle fractures via standard plain radiographs and to determine the intra- and interobserver agreement in the treatment of these fractures.

Study Design: Cohort study (diagnosis); Level of evidence, 3.

Methods: Charts of patients seen by the 2 senior authors from 2006 to 2011 were reviewed to identify patients treated for clavicle fractures (CPT codes 23500 and 23515). Anteroposterior and 30° cephalad radiographs were selected, representing midshaft clavicle fractures treated both operatively and nonoperatively. Thirty pairs of radiographs were included in the investigation. The radiographs were standardized for size to allow accurate measurements within a non-PACS (picture archiving and communications system) program, and a PDF document was created with all representative radiographs. Clinical scenarios were created for each set of radiographs, and the evaluators were asked to (1) measure the degree of shortening in millimeters, (2) determine the percentage displacement, (3) determine whether the fracture was comminuted, and (4) state whether they would treat the fracture operatively or nonoperatively. The radiographs, along with instructions on how to use the measuring tool with Adobe Reader, were distributed to 22 shoulder/sports medicine fellowship-trained orthopaedic surgeons, then reordered and redistributed approximately 3 months later. Sixteen surgeons completed 1 round of surveys, and 13 surgeons completed both rounds.

Results: Interrater agreement was moderate for displacement of 0%-49% (κ = 0.71, P < .001) and >100% (κ = 0.73, P < .001), with minimal agreement for displacement of 50%-100% (κ = 0.39, P < .001). There was moderate interrater agreement for the presence/absence of comminution (κ = 0.75, P < .001). Interrater agreement was weak for shortening of 0-5.0 mm (κ = 0.58, P < .001) and >30.0 mm (κ = 0.51, P < .001), with minimal agreement for shortening of 5.1-10.0 mm (κ = 0.22, P < .001) and no agreement for the other 4 categories. Interrater analysis showed weak agreement on whether surgical treatment was recommended (κ = 0.40, P < .001). Intrarater agreement was strong for comminution (κ = 0.80, P < .0001), moderate for both displacement (κ = 0.76, P < .001) and operative treatment (κ = 0.64, P < .001), and minimal for shortening (κ = 0.38, P < .001). The following variables statistically predicted whether surgery was recommended (P < .001): (1) the odds of surgery were 2.26 if comminution was noted, holding displacement and the interaction between displacement and shortening constant, and (2) the odds of surgery were 3.37 if there is displacement of >100% compared with displacement of 0%-49%, holding comminution and shortening constant.

Conclusion: Standard plain unilateral radiographs of the clavicle are insufficient to reliably determine the degree of shortening of clavicle fractures and the need for surgery among shoulder/sports medicine fellowship-trained orthopaedic surgeons. Consideration should be made to not use shortening as the sole determinant for whether to proceed with surgical intervention or to use other radiographic modalities to determine the amount of shortening.