The use of head CT scan in adult trauma victims: an algorithm

INTRODUCTION: The purpose of this study was to identify, through recursive partitioning, clinically relevant criteria which predict the need for acute neurosurgical intervention in a group of patients with mild head injury. MATERIALS AND METHODS: A retrospective cohort of all adult patients, from April 2000 to March 2001, who sustained a blunt trauma and underwent head CT scan, was reviewed. The following inclusion criteria for mild head injury were used: initial Glasgow Coma Scale (GCS) ranging from 13 to 15; no loss of consciousness lasting more than one hour; no obvious skull fracture; a cranial CT scan performed. We collected demographic and trauma related data, interventions and outcome. Univariate and multivariate analyses were undertaken. In parallel, recursive partitioning was carried out using all variables to elaborate a decision algorithm. RESULTS: There were 405 patients in the sample. CT identified lesions in 12% of patients. Twelve patients (3%) required acute neurosurgical intervention. The recursive partitioning analysis identified three significant sequential nodes: deterioration of the GCS; an initial GCS of 13 vs 14 or 15; and the presence of associated injuries or comorbid conditions. CONCLUSIONS: A simple three step rule predicts the need for acute neurosurgical intervention based on clinical findings: a deteriorating GCS; an initial GCS of 13; and the presence of associated injuries or comorbid conditions.