Validation of a prediction rule for the identification of children with intra-abdominal injuries after blunt torso trauma.

STUDY OBJECTIVE We validate the accuracy of a previously derived clinical prediction rule for the identification of children with intra-abdominal injuries after blunt torso trauma. METHODS We conducted a prospective observational study of children with blunt torso trauma who were evaluated for intra-abdominal injury with abdominal computed tomography (CT), diagnostic laparoscopy, or laparotomy at a Level I trauma center during a 3-year period to validate a previously derived prediction rule. The emergency physician providing care documented history and physical examination findings on a standardized data collection form before knowledge of the results of diagnostic imaging. The clinical prediction rule being evaluated included 6 "high-risk" variables, the presence of any of which indicated that the child was not at low risk for intra-abdominal injury: low age-adjusted systolic blood pressure, abdominal tenderness, femur fracture, increased liver enzyme levels (serum aspartate aminotransferase concentration >200 U/L or serum alanine aminotransferase concentration >125 U/L), microscopic hematuria (urinalysis >5 RBCs/high powered field), or an initial hematocrit level less than 30%. RESULTS One thousand three hundred twenty-four children with blunt torso trauma were enrolled, and 1,119 (85%) patients had the variables in the decision rule documented by the emergency physician and therefore made up the study sample. The prediction rule had the following test characteristics: sensitivity=149 of 157, 94.9% (95% confidence interval [CI] 90.2% to 97.7%) and specificity=357 of 962, 37.1% (95% CI 34.0 to 40.3%). Three hundred sixty-five patients tested negative for the rule; thus, strict application would have resulted in a 33% reduction in abdominal CT scanning. Of the 8 patients with intra-abdominal injury not identified by the prediction rule, 1 underwent a laparotomy. This patient had a serosal tear and a mesenteric hematoma at laparotomy, neither of which required specific surgical intervention. CONCLUSION A clinical prediction rule consisting of 6 variables, easily available to clinicians in the ED, identifies most but not all children with intra-abdominal injury. Application of the prediction rule to this sample would have reduced the number of unnecessary abdominal CT scans performed but would have failed to identify 1 child undergoing (a nontherapeutic) laparotomy. Thus, further refinement of this prediction rule in a large, multicenter cohort is necessary before widespread implementation.