A relook at the reliability of Rockwood classification for acromioclavicular joint injuries

Controversies for treatment of acromioclavicular joint injuries in particular type III injuries may be partially attributed to the lack of a standardized method of radiography and measurement technique. Previous studies looking at the Rockwood classification showed poor inter- and intraobserver reliability (Kappa value approximately 0.20-0.50). We hypothesized that the use of unilateral instead of bilateral acromioclavicular joint radiographs was the cause of this finding. In this article, we standardized the methodology to perform the radiograph and to measure the coracoclavicular distances. We designed the study to focus on the reliability of differentiating type III and type V injuries.

A standardized radiographic protocol for bilateral Zanca view was established in our institution. All patients who underwent this radiographic examination over a 3-year period were reviewed. Radiographs of 55 patients with significant (type III or V) injury met the inclusion criteria. For the interobserver reliability, a retrospective radiographic review was performed by 6 orthopedic surgeons and graded as either type III or V. For intraobserver reliability, a similar process was repeated by 3 observers after a period of 6 weeks.

Going by the majority agreement of the 6 reviewers, there were 34 type III injuries and 19 type V injuries. The Fleiss kappa for interobserver reliability was calculated to be 0.624. The Cohen kappa for intraobserver reliability was calculated to be 0.696.

The use of a standardized radiographic protocol-taking bilateral Zanca views on the same radiographic plate-would help eliminate a significant amount of variability and improve the reliability of classifying acromioclavicular joint injuries using the Rockwood classification, which uses a relative measure to the contralateral site as its definition criteria. Other possible sources of poor reliability may include the masking of injuries by muscle spasm, resulting in a misdiagnosis of a high-grade injury as a lower-grade one and the possible need to subclassify type III injuries.

Reliability of the Rockwood classification can be improved through the use of a standardized radiographic protocol to improve the detection of vertical instability. Similar to Rockwood dividing up Tossy grade 3 injuries when he noted the differential outcome and intervention, Rockwood type III injuries would likely require further subclassification as it remains an anomalous tool with high variability. Further studies are required to understand the pathologic basis of transition of type III into type V injury.