The SAH Score: A Comprehensive Communication Tool

Abstract

The Hunt and Hess grade and World Federation of Neurological Surgeons (WFNS) scale are commonly used to predict mortality after aneurysmal subarachnoid hemorrhage (aSAH). Our objective was to improve the accuracy of mortality prediction compared with the aforementioned scales by creating the "SAH score." The aSAH database at our institution was analyzed for factors affecting in-hospital mortality using multiple logistic regression analysis. Scores were weighted based on relative risk of mortality after stratification of each of these variables. Glasgow Coma Scale (GCS) was subdivided into groups of 3-4 (score=1), 5-8 (score=2), 9-13 (score=3), and 14-15 (score=4). Age was categorized into 4 subgroups: 18-49 (score=1), 50-69 (score=2), 70-79 (score=3), and 80 years or more (score=4). Medical comorbidities were subdivided into none (score=1), 1 (score=2), or 2 or more (score=3). In total, 1134 patients were included; all-cause SAH hospital mortality was 18.3%. Admission GCS, age, and medical comorbidities significantly affected mortality after multivariate analysis (P<.05). Summated scores ranged from 0 to 8 with escalating mortality at higher scores (0=2%, 1=6%, 2=8%, 3=15%, 4=30%, 5=58%, 6=79%, 7=87%, and 8=100%). Positive predictive value (PPV) for scores in the range 7-8 was 88.5%, whereas 6-8 was 83%. Negative predictive value (NPV) was 94% for range 0-2 and 92% for 0-3. The area under the curve (AUC) for the SAH score was .821 (good accuracy), compared with the WFNS scale (AUC .777, fair accuracy) and the Huntand Hess grade (AUC .771, fair accuracy). The SAH score was found to be more accurate in predicting aSAH mortality compared with the Hunt and Hess grade and WFNS scale.