Study aim To validate in a UK population a clinical prediction rule which identifies children who are at very low risk of intra-abdominal injury (IAI) undergoing acute interventions following blunt torso trauma and who can therefore safely avoid computed tomography and its associated risks.

Methods The study was a single centre retrospective case note analysis of children under 16 years of age who had sustained blunt torso trauma during a 3 year period at a Level 1 trauma centre.

Results Of 357 trauma calls, 120 (34%, 95% CI 29% to 34%) were assessed as sustaining blunt torso trauma. Of the 120 cases, 90 patients met inclusion criteria, with 86 cases having complete data for all variables investigated. 13 (15.1%; 95% CI 7.5% to 22.7%) had IAI. The main outcome of interest, IAI requiring acute intervention, was identified in 2 of the 13 patients. In assessing the performance of the clinical prediction rule, 19 (22%) were at very low risk of IAI requiring acute intervention, as assessed by the absence of any variables in the prediction rule. The rule held true for all 19 cases, none required acute intervention for IAI. However, 1 of the 19 cases did sustain an intra-abdominal injury which did not require acute intervention. The test characteristics for the clinical prediction rule’s performance in identifying IAI (with or without acute intervention) are: sensitivity 92%, specificity 25%, negative predictive value 95%, and positive predictive value 18%. 8 of 19 (42%) patients identified by the prediction rule as being very low risk for intra-abdominal injury had an abdominal CT. This represents 14% of all abdominal CT scans in the study sample.

Conclusion Our sample size was insufficient to robustly test the prediction rule for identifying those at low risk of IAI requiring acute intervention. A further assessment of the rule in identifying those at low risk with or without acute intervention was performed. Test characteristics of the rule in both instances were very similar to those reported in the original study and in our study sample the prediction rule was valid. A prospective multi-centre study is required to further assess the performance of the prediction rule.