Predictors of Clinically Significant Radiographic Shoulder Pathology in the Emergency Department

Abstract Background Although there are no clinical decision rules for radiograph use among persons with shoulder pain, they are ordered for most patients. Previously published reviews have demonstrated that radiography is overutilized in evaluating emergency department (ED) patients with shoulder pain, and clinical factors might define patients in whom plain film radiography need not be performed. Objectives The objectives of this study were to identify predictors of clinically significant shoulder pain and develop a clinical decision radiograph-ordering rule for adult ED patients with shoulder pain. Methods Records from adult ED visits resulting in shoulder radiographs were reviewed. Potential predictors of clinically significant shoulder pain were then identified. Univariate screening was performed to find variables associated with injury and were subsequently included in a multivariable prediction model. Results Five of the predetermined factors were found to be associated with the likelihood of injury: history of trauma, range of motion, deformity, age, and duration of pain. Receiver operating characteristics revealed an area under the curve of 80%. Conclusions Despite accounting for multiple variables, the area under the curve was 80%. Based on these results it is not practical to develop clinical decision radiograph ordering rules for ED patients with shoulder pain.