External Validation and Modification of the EDEMA Score for Predicting Malignant Brain Edema After Acute Ischemic...

Accurate prediction of malignant brain edema (MBE) after stroke is paramount to facilitate close monitoring and timely surgical intervention. The Enhanced Detection of Edema in Malignant Anterior Circulation Stroke (EDEMA) score was useful to predict potentially lethal malignant edema in Western populations. We aimed to validate and modify it to achieve a better predictive value for MBE in Chinese patients.

Of ischemic stroke patients consecutively admitted in the Department of Neurology, West China Hospital between January 2010 and December 2017, we included patients with anterior circulation stroke, early signs of brain edema on computed tomography within 24 h of onset, and admission National Institutes of Health Stroke Scale (NIHSS) score ≥ 8. MBE was defined as the development of signs of herniation (including decrease in consciousness and/or anisocoria), accompanied by midline shift ≥ 5 mm on follow-up imaging. The EDEMA score consisted of five parameters: glucose, stroke history, reperfusion therapy, midline shift, and cistern effacement. We created a modified score by adding admission NIHSS score to the original EDEMA score. The discrimination of the score was assessed by the area under the receiver operating characteristics curve (AUC). Calibration was assessed by Hosmer-Lemeshow test and calibration plot. We compared the discrimination of the original and modified score by AUC, net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Clinical usefulness of the two scores was compared by plotting net benefits at different threshold probabilities in the decision curve analysis.

Of the 478 eligible patients (mean age 67.3 years; median NIHSS score 16), 93 (19%) developed MBE. The EDEMA score showed moderate discrimination (AUC 0.72, 95% confidence interval [CI] 0.67-0.76) and good calibration (Hosmer-Lemeshow test, P = 0.77). The modified score showed an improved discriminative ability (AUC 0.80, 95% CI 0.76-0.84, P < 0.001; NRI 0.67, 95% CI 0.55-0.78, P < 0.001; IDI 0.07, 95% CI 0.06-0.09, P < 0.001). Decision curves showed that the modified score had a higher net benefit than the original score in a range of threshold probabilities lower than 60%.

The original EDEMA score showed an acceptable predictive value for MBE in Chinese patients. By adding the admission NIHSS score, the modified score allowed for a more accurate prediction and clinical usefulness. Further validation in large cohorts of different ethnicities is needed to confirm our findings.