Emergency room decision-making for urgent cranial computed tomography: selection criteria for subsets of non-trauma patients

Abstract

Background Computed tomography (CT) of the brain is used extensively in the urgent work-up of patients with a suspicion of intracranial pathology, but is often normal. Previously proposed selection criteria aim at limiting the ordering of urgent cranial CT in the non-trauma population, while maintaining high sensitivity for diagnoses demanding immediate attention. Purpose To retrospectively evaluate these selection criteria in a general non-trauma population from a Swedish tertiary hospital, as well as in a nested subgroup that lacks guidelines at present, namely where the chief complaint was not headache, symptoms clearly indicating stroke, seizures, or vertigo. Material and Methods Medical records of 346 patients (114 in the nested group) who had undergone urgent cranial CT were reviewed. Selection criteria as proposed by Rothrock (patient age ≥60 years, presence of new onset focal neurologic deficit, headache with vomiting, or altered mental status) were used. Acute cerebral infarction, intracranial hemorrhage, malignancy, infection, cerebral edema, or hydrocephalus were considered significant findings. Results The prevalence of significant findings was 10.1%. The Rothrock criteria had a sensitivity of 97.1% (identifying 34 of 35 significant findings) among all 346 patients and 100% (10/10) among the 114 patients in the subgroup and resulted in a potential scan reduction rate of 22.8% and 11.4%, respectively. In the patient with significant pathology, that was not selected for CT, focal neurological symptoms were not described as newly onset. Conclusion Although 100% sensitivity was not achieved, our results may contribute to the evidence that in the absence of focal neurologic deficit, headache with vomiting or altered mental status in patients aged <60 years cranial tomography can be refrained from, in the general population as well as in the subgroup defined above. Further research might validate patient history as a parameter.