Abstract
BACKGROUND AND PURPOSE: Diagnosing spontaneous intracranial hypotension (SIH) remains a clinical challenge. The Bern score offered a structured scoring system for estimating the probability of spinal extradural cerebrospinal fluid leaks. We aimed to validate the Bern score in an East Asian cohort from a headache clinic. METHODS: Patients with SIH were recruited from 2022 to 2023, and age- and sex-matched controls with primary headache and normal brain MRI were included. All SIH patients underwent brain MRI and spine MR myelography before epidural blood patch. The Bern score was assessed and validated by two neurologists and one neuroradiologist. We evaluated its diagnostic performance with a receiver operating characteristic (ROC) curve and visualized the temporal profile of Bern scores using a LOESS (locally weighted scatterplot smoothing) graph. RESULTS: A total of 45 patients with SIH and 45 age-sex-matched controls were included. The ROC curve showed moderate diagnostic accuracy (area under the curve 0.775, 95% confidence interval 0.675-0.874; sensitivity 42.2%, specificity 95.6%) using the original cutoff of 3. Using a revised cutoff was ≥2, the Youden index was highest (sensitivity 66.7%, specificity 84.4%, accuracy 75.6%). Still, its performance remains suboptimal. The Bern score declined after 50 days from symptom onset. CONCLUSIONS: The Bern score showed high specificity but limited sensitivity in our cohort, indicating it should not be used alone as a screening tool. A lower cutoff may improve its diagnostic performance, and timing from symptom onset should be considered when interpreting the score.