Intra-and inter-observer agreement in the classification and treatment of midshaft clavicle fractures

Objectives: The purpose of this study was to determine the intraobserver and interobserver reliability in the classification of midshaft clavicle fractures via standard plain radiographs and to determine the intraobserver and interobserver agreement in the treatment of these fractures. Methods: Charts of patients seen by the two senior authors from 2006 to 2011 were reviewed to identify patients treated for clavicle fractures (CPT 23500 and 23515). AP and 30 degree cephalad radiographs were selected, representing midshaft clavicle fractures treated both operatively and non-operatively. Thirty pairs of radiographs were included in the investigation. The radiographs were standardized for size to allow accurate measurements within a non-PACS program. A PDF file was created with all representative radiographs. Clinical scenarios were created for each set of radiographs, and the evaluators were asked to perform the following tasks: 1) measure the degree of shortening in millimeters; 2) determine the percent displacement; 3) determine whether the fracture was comminuted; and 4) state whether they would treat the fracture operatively or non-operatively. The radiographs and clinical scenario handout, along with instructions on how to use the measuring tool with Adobe Reader, were distributed to 16 shoulder/sports medicine fellowship-trained orthopaedic surgeons who completed the evaluations. The radiographs and scenarios were then reordered and redistributed approximately three months later. Results: Intra and interobserver results are summarized in table 1. The following variables statistically predicted whether surgery was recommended (p< 0.001): 1) odds of surgery are 2.26 if comminution was noted holding displacement and the interaction between displacement and shortening constant, and 2) the odds of surgery are 3.37 if there is displacement > 100% compared to displacement 0-49% holding comminution and shortening constant. Conclusion: Our study demonstrated that there was moderate to strong interobserver and intraobserver agreement for both displacement and comminution when utilizing standard plain radiographs. However, there was only weak to no interobserver agreement and minimal intraobserver agreement on the amount of shortening. Also, there was minimal interobserver and only moderate intraobserver agreement on whether operative treatment should be selected, most likely due to the poor reliability of plain radiographs in determining the degree of shortening. Therefore, other modalities including comparison radiographs of the contralateral clavicle should be considered to more reliably measure the degree of shortening and help determine the need for surgery.

Duke Scholars

Author Brian David Lewis Orthopaedic Surgery