Derivation of a decision instrument for selective chest radiography in blunt trauma

BACKGROUND:: To derive a decision instrument (DI) that identifies patients who have virtually no risk of significant intrathoracic injury (SITI) visible on chest radiography and, therefore, no need for chest imaging. METHODS:: This is a prospective observational study. At three Level 1 trauma centers, physicians caring for blunt trauma patients aged >14 years were asked to record the presence or absence of 12 clinical criteria before viewing chest imaging results. SITI was defined as pneumothorax, hemothorax, aortic/great vessel injury, two or more rib fractures, ruptured diaphragm, sternal fracture, and pulmonary contusion on official radiograph readings. The κ (interrater reliability) and screening performance of individual criteria were determined. By using recursive partitioning, the most highly sensitive combination of criteria for SITI was derived. RESULTS:: Of the 2,628 subjects enrolled, 271 (10.3%) were diagnosed with a total of 462 SITIs, with rib fractures (73%), pneumothorax (38%), and pulmonary contusion (29%) as the most common injuries. Chest pain and chest wall tenderness had the highest sensitivity for SITI (65%). The DI of chest pain, distracting injury, chest wall tenderness, intoxication, age >60 years, rapid deceleration, and altered alertness/mental status had the following screening performance: sensitivity 99.3% (95% confidence interval [CI], 97.4-99.8), specificity 14.0% (95% CI, 12.6-15.4), negative predictive value 99.4% (95% CI, 97.8-99.8), and positive predictive value 11.7% (95% CI, 10.5-13.1). All seven criteria in the DI met the predetermined cut off for acceptable κ (range, 0.51-0.81). CONCLUSIONS:: We derived a DI consisting of seven clinical criteria that can identify SITI in blunt trauma patients with extremely high sensitivity. If validated, this instrument will allow for safe, selective chest imaging with potential resource savings.