Comparison of Clinician Suspicion Versus a Clinical Prediction Rule in Identifying Children at Risk for Intra-abdominal Injuries After Blunt Torso Trauma

Objectives: Emergency department (ED) identification and radiographic evaluation of children with intra-abdominal injuries who need acute intervention can be challenging. To date, it is unclear if a clinical prediction rule is superior to unstructured clinician judgment in identifying these children. The objective of this study was to compare the test characteristics of clinician suspicion with a derived clinical prediction rule to identify children at risk of intra-abdominal injuries undergoing acute intervention following blunt torso trauma.

Methods: This was a planned subanalysis of a prospective, multicenter observational study of children (<18 years old) with blunt torso trauma conducted in 20 EDs in the Pediatric Emergency Care Applied Research Network (PECARN). Clinicians documented their suspicion for the presence of intra-abdominal injuries needing acute intervention as <1, 1 to 5, 6 to 10, 11 to 50, or >50% prior to knowledge of abdominal computed tomography (CT) scanning (if performed). Intra-abdominal injuries undergoing acute intervention were defined by a therapeutic laparotomy, angiographic embolization, blood transfusion for abdominal hemorrhage, or intravenous fluid administration for 2 or more days in those with pancreatic or gastrointestinal injuries. Patients were considered to be positive for clinician suspicion if suspicion was documented as ≥1%. Suspicion ≥ 1% was compared to the presence of any variable in the prediction rule for identifying children with intra-abdominal injuries undergoing acute intervention.

Results: Clinicians recorded their suspicion in 11,919 (99%) of 12,044 patients enrolled in the parent study. Intra-abdominal injuries undergoing acute intervention were diagnosed in 203 (2%) patients. Abdominal CT scans were obtained in the ED in 2,302 of the 2,667 (86%, 95% confidence interval [CI] = 85% to 88%) enrolled patients with clinician suspicion ≥1% and in 3,016 of the 9,252 (33%, 95% CI = 32% to 34%) patients with clinician suspicion < 1%. Sensitivity of the prediction rule for intra-abdominal injuries undergoing acute intervention (197 of 203; 97.0%, 95% CI = 93.7% to 98.9%) was higher than that of clinician suspicion ≥1% (168 of 203; 82.8%, 95% CI = 76.9% to 87.7%; difference = 14.2%, 95% CI = 8.6% to 20.0%). Specificity of the prediction rule (4,979 of the 11,716; 42.5%, 95% CI = 41.6% to 43.4%), however, was lower than that of clinician suspicion (9,217 of the 11,716, 78.7%, 95% CI = 77.9% to 79.4%; difference = -36.2%, 95% CI = -37.3% to -35.0%). Thirty-five (0.4%, 95% CI = 0.3% to 0.5%) patients with clinician suspicion < 1% had intra-abdominal injuries that underwent acute intervention.

Conclusions: The derived clinical prediction rule had a significantly higher sensitivity, but lower specificity, than clinician suspicion for identifying children with intra-abdominal injuries undergoing acute intervention. The higher specificity of clinician suspicion, however, did not translate into clinical practice, as clinicians frequently obtained abdominal CT scans in patients they considered very low risk. If validated, this prediction rule can assist in clinical decision-making around abdominal CT use in children with blunt torso trauma.