Predicción de la isquemia cerebral tardía en la hemorragia subaracnoidea aneurismática mediante escalas radiológicas

Abstract

BACKGROUND: Delayed cerebral ischemia (DCI) is the leading treatable cause of death and disability in around 20% of patients with aneurysmal subarachnoid haemorrhage (SAH). Several studies have proposed markers to predict this complication, with the main associated factor being the amount and distribution of blood in the initial unenhanced computed tomography (CT) scan. While numerous radiological risk scales have been developed, their predictive capabilities have not been widely studied and compared. METHOD: This observational retrospective and single-centre four-year study included patients with aneurysmal SAH consecutively attended to in the Hospital Clínico Universitario Virgen de la Arrixaca (Spain). The amount and distribution of ventricular and cisternal blood were measured by applying five radiological scales to CT scans performed on admission: Fisher, modified Fisher, Eagles, Vasograde and Barrow scales. Its predictive ability, progressiveness and between-scale agreement were analysed and compared. Patients were also comparatively analysed in terms of DCI presentation to identify which other radiological and clinical variables were possible independent risk factors for DCI. The predictive capacity of a modified version of the Hijdra scale was also analyzed. RESULTS: The study population was 147 patients, of whom 21.8% developed DCI. The Barrow scale presented the highest predictive capacity and category III was outlined as the one that best predicted the occurrence of ICT, with a sensitivity of 75% and a specificity of 65.2%. The degree of radiological severity according to the Modified Fisher scale, the degree of clinical severity according to the WFNS scale, the amount of blood in the ventricular system according to the Modified Graeb scale, the presence of space occupying cisternal clots or with a thickness greater than 15 millimeters and the location of the aneurysm showed a statistically significant association for ICT in the bivariate analysis. After adjusting these factors in a multivariate model, only the degree of radiological severity according to the Modified Fisher scale was still predictive for ICT. A value of 11.5 on the Modified Hijdra scale was the most appropriate score to detect ICT, with a sensitivity of 75% and a specificity of 50.4%. CONCLUSIONS: Among the scales analyzed, Barrow's is the one that presents the greatest predictive capacity with moderate sensitivity and specificity. The clinical and radiological factors present at the time of admission are scarcely useful for improving the predictive capacity of risk scales. The Modified Hijdra scale has a low predictive capacity.