The use of radiographs to evaluate shoulder pain in the ED

This study was conducted to describe the utilization of plain radiography in the initial evaluation of shoulder pain in the emergency department (ED) and assess the feasibility of developing guidelines for the selective use of shoulder radiographs by determining whether clinical variables are able to discriminate between subjects having therapeutically informative versus uninformative X-rays. The study was a chart review of all adult ED patients presenting to the Boston University Medical Center Hospital (Boston, MA) between January 1994 and January 1996 with a chief complaint that included shoulder pain. Subjects with X-rays were classified into two groups, those with therapeutically informative X-rays (ie, identified conditions requiring specific therapy) and those with therapeutically uninformative X-rays (ie, did not result in specific therapy). Recursive partitioning techniques were then used to identify clinical variables that best distinguished between therapeutically informative and uninformative radiographs. Three hundred twelve patients were included in the analysis, of whom 185 (59%) had shoulder X-rays performed. Thirty-seven (20%) radiographs were therapeutically informative, including 13 glenohumeral dislocations, 3 acromioclavicular joint separations, and 21 fractures. Deformity present on shoulder examination was the strongest discriminating variable and correctly classified 21 of 23 subjects as having informative X-rays. Of the remaining patients (n = 162), only those older than 43.5 years with a history of a precipitating fall (n = 40) had a high likelihood of having therapeutically informative X-rays. No patients without a deformity or precipitating fall (n = 90) had an informative X-ray. These data show that X-rays for the initial evaluation of shoulder pain in the ED are overutilized. This preliminary model suggests that clinical variables are able to distinguish between patients with informative versus uninformative X-rays. Prospective studies are needed to derive valid decision rules for selective use of shoulder radiographs in the ED.