Reliability of the classification and treatment of dislocations of the acromioclavicular joint

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.

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 κ values.

On the basis of plain radiographs alone, interobserver and intraobserver reliability of the Rockwood classification were fair (κ = .214) and moderate (κ = .474), respectively. Interobserver and intraobserver reliability of treatment were both fair (κ = .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 (κ = .177) and moderate (κ = .565), respectively. Interobserver and intraobserver reliability of treatment were fair (κ = .253) and moderate (κ = .554), respectively. There were no significant differences in reliability between the two groups in terms of any κ values.

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.