Abstract
BACKGROUND: Diagnosing cerebral venous thrombosis (CVT) is challenging due to varied symptoms. Computed tomography venography (CTV) or magnetic resonance venography (MRV) are the gold standard, but resource-demanding with potential side effects. We evaluated the diagnostic properties of a previously proposed CVT probability score combined with D-dimer levels in detecting CVT. METHODS: Medical records of consecutive adults who underwent CTV or MRV due to suspicion of CVT at a Norwegian primary hospital were retrospectively analyzed. The CVT probability score was calculated based on previous medical history and clinical findings/symptoms at presentation to stratify patients into low, moderate or high CVT probability groups. Performance characteristics were calculated for the CVT probability score with and without D-dimer levels. RESULTS: Among 578 patients evaluated from 2016 to 2019, CVT was confirmed in 6.6%. The CVT probability score was available for 444 patients and distributed one-third of CVT patients across each CVT probability group, with a 0.33 sensitivity for each group. A negative predictive value (NPV) of 0.96 was noted for the high CVT probability group. The NPV was 0.99 for the combined low/moderate CVT probability group including D-dimer > 500 µg/L. The positive predictive values were all fairly low. CONCLUSION: Our results indicate that using a CVT probability score and D-dimer levels can help exclude CVT; however, false positive cases present a clinical challenge and carry the risk of excessive advanced imaging.