Derivation of a clinical decision rule to exclude cerebral venous sinus thrombosis in emergency department patients: A retrospective cohort study

Abstract

AbstractObjectiveTo derive a clinical decision rule to exclude cerebral venous sinus thrombosis (CVST) in the ED. A secondary aim was to derive a rule that incorporated clinical parameters and the non‐contrast CT brain.MethodsSingle‐centre, retrospective cohort study. Patients suspected of CVST were identified from the radiology database for CT/MR venograms. Clinical features included in the rule were determined by literature review. The presence of these features in participants was determined by chart review. Variables were tested for univariate association with CVST using logistic regression. Variable selection was accomplished using a forward‐stepwise process, calculating the sensitivity/specificity of a rule containing the variable of most significance, then repeating the process after adding the next most significant variable.ResultsForty‐five out of 912 participants had confirmed CVST. The primary clinical rule was answering ‘no’ to all the following: any prothrombotic risk factor, age ≥54 years, confusion: sensitivity 95.6% (95% confidence interval [CI] 84.9–99.5%), specificity 40.9% (95% CI 37.6–44.2%), negative predictive value 99.4% (95% CI 97.9–99.9%) and positive predictive value 7.7% (95% CI 7.1–8.3%). The rule classified 39.5% of participants as CVST ruled out. The rule incorporating the non‐contrast CT brain was answering ‘no’ to all the following: abnormal non‐contrast CT brain, any prothrombotic risk‐factor, age ≥54 years, confusion: sensitivity 100.0% (95% CI 91.6–100.0%), specificity 42.0% (95% CI 38.7–45.4%), negative predictive value 100.0% (95% CI not calculated) and positive predictive value 7.8% (95% CI 7.4–8.2%). The rule classified 40.0% of participants as CVST ruled out.ConclusionsA clinical decision rule was derived to rule out CVST. These results require validation before adoption into clinical practice.