An external validation of clinical-based score to predict traumatic intracranial hemorrhage on computed tomography scan and surgical intervention in mild traumatic brain injury patients

Background: Prediction rules for computed tomography (CT) use in mild traumatic brain injury (MTBI) often have limited generalizability due to restrictive inclusion criteria and development in high-income countries. A clinical-based prediction score was recently developed for a broader MTBI population in a low- to middle-income country. This study aims to externally validate this score in a new patient dataset. Methods: We prospectively collected data from adult patients with MTBI (Glasgow Coma Scale score 13–15) presenting to a hospital in Thailand between 2014 and 2016. The performance of the 13-predictor clinical-based score was evaluated for two outcomes: traumatic intracranial findings on CT and need for neurosurgical intervention. We assessed discrimination using the area under the receiver operating characteristic curve (AuROC) and diagnostic accuracy at predetermined cut-points. Results: The study included 565 patients. Seventy-two patients (12.7%) had a positive CT scan, and 8 (1.4%) required neurosurgical intervention. The AuROC was 0.70 for predicting a positive CT scan and 0.78 for predicting neurosurgical intervention. At a cut-point score of ≥2, the sensitivity for a positive CT was 97.2% with a specificity of 12.6%. At a cut-point score of ≥3, the sensitivity for neurosurgical intervention was 100% with a specificity of 21.0%. Conclusion: The clinical prediction score demonstrated acceptable and safe diagnostic performance in an external validation cohort, maintaining high sensitivity for critical outcomes. The score may be a useful tool to help guide management policy for MTBI patients, particularly in low- to middle-income countries.