OBJECTIVE: To derive and internally validate a clinical decision rule that will rule out major thoracic injury in adult blunt trauma patients, reducing the unnecessary use of chest computed tomographic (CT) scans. METHODS: Data were retrospectively obtained from a chart review of all trauma patients presenting to a Canadian tertiary trauma care centre from 2005 to 2008, with those from April 2006 to March 2007 being used for the validation phase. Patients were included if they had an Injury Severity Score > 12 and chest CT at admission or a documented major thoracic injury noted in the trauma database. Patients with penetrating injury, a Glasgow Coma Scale (GCS) score ≤ 8, paralysis, or age < 16 years were excluded. RESULTS: There were 434 patients in the derivation group and 180 in the validation group who met the inclusion criteria. Using recursive partitioning, five clinical variables were found to be particularly predictive of injury. When these variables were normal, no patients had a major thoracic injury (sensitivity 100% [95% CI 98.4-100], specificity 46.9% [95% CI 44.2-46.9], and negative likelihood ratio 0.00 [95% CI 0.00-0.04]). The five variables were oxygen saturation (< 95% on room air or < 98% on any supplemental oxygen), chest radiograph, respiratory rate ≥ 25, chest auscultation, and thoracic palpation (SCRAP). In the validation group, the same five variables had a sensitivity of 100% (95% CI 96.2-100%), a specificity of 44.7% (95% CI 39.5-44.7%), and negative likelihood ratio of 0.00 (95% CI 0.00-0.10). CONCLUSIONS: In major blunt trauma with a GCS score > 8, the SCRAP variables have a 100% sensitivity for major thoracic injury in this retrospective study. These findings need to be prospectively validated prior to use in a clinical setting.
OBJECTIVE: To derive and internally validate a clinical decision rule that will rule out major thoracic injury in adult blunt traumapatients, reducing the unnecessary use of chest computed tomographic (CT) scans. METHODS: Data were retrospectively obtained from a chart review of all traumapatients presenting to a Canadian tertiary trauma care centre from 2005 to 2008, with those from April 2006 to March 2007 being used for the validation phase. Patients were included if they had an Injury Severity Score > 12 and chest CT at admission or a documented major thoracic injury noted in the trauma database. Patients with penetrating injury, a Glasgow Coma Scale (GCS) score ≤ 8, paralysis, or age < 16 years were excluded. RESULTS: There were 434 patients in the derivation group and 180 in the validation group who met the inclusion criteria. Using recursive partitioning, five clinical variables were found to be particularly predictive of injury. When these variables were normal, no patients had a major thoracic injury (sensitivity 100% [95% CI 98.4-100], specificity 46.9% [95% CI 44.2-46.9], and negative likelihood ratio 0.00 [95% CI 0.00-0.04]). The five variables were oxygen saturation (< 95% on room air or < 98% on any supplemental oxygen), chest radiograph, respiratory rate ≥ 25, chest auscultation, and thoracic palpation (SCRAP). In the validation group, the same five variables had a sensitivity of 100% (95% CI 96.2-100%), a specificity of 44.7% (95% CI 39.5-44.7%), and negative likelihood ratio of 0.00 (95% CI 0.00-0.10). CONCLUSIONS: In major blunt trauma with a GCS score > 8, the SCRAP variables have a 100% sensitivity for major thoracic injury in this retrospective study. These findings need to be prospectively validated prior to use in a clinical setting.