Reliability of the Traditional Classification Systems for Acromioclavicular Joint Injuries by Radiography

Abstract

Background The present study aimed to examine the reliability of the radiographic classification systems for acromioclavicular (AC) joint injuries. Methods We initially polled 47 orthopaedic surgeons regarding what common technique they used for applying the Rockwood and the Tossy and Allman classification systems. All used a single standard AC joint view (Zanca view). We then presented 24 Zanca view radiographs of patients who had sustained AC joint injuries to 19 specialist shoulder surgeons and asked each of them to classify the injuries using the Rockwood and the Tossy and Allman classification systems. We then altered the order of radiographs and repeated the survey with the same group of surgeons 1 month later. Results The mean inter-observer agreement and the corresponding weighted kappa for the Rockwood and the Tossy and Allman classification system were 64.6% and 0.258; and 68.1% and 0.309, respectively. The mean intra-observer agreement and the corresponding weighted kappa for the systems were 59.4% and 0.150; and 67.4% and 0.113, respectively. Conclusions We conclude that the classification of AC joint injuries using a radiograph alone has limited reliability and consistency in clinical practice.