AUTHOR=Wouters Anke , Nysten Céline , Thijs Vincent , Lemmens Robin 

TITLE=Prediction of Outcome in Patients With Acute Ischemic Stroke Based on Initial Severity and Improvement in the First 24 h

JOURNAL=Frontiers in Neurology

VOLUME=Volume 9 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2018.00308

DOI=10.3389/fneur.2018.00308

ISSN=1664-2295

ABSTRACT=Introduction: Stroke severity measured by the baseline National Institutes of Health Stroke Scale (NIHSS) is a strong predictor of stroke outcome. Early change of baseline severity may be a better predictor of outcome. Here we hypothesized that the change in NIHSS in the first 24h after stroke improved stroke outcome prediction. 
Material and methods: Patients from the Leuven Stroke Genetics Study were included when the baseline NIHSS (B-NIHSS) was determined on admission in the hospital and NIHSS after 24 hours could be obtained from patient files. The delta NIHSS, relative reduction (RR NIHSS) and major neurological improvement (MNI; NIHSS of 0-1 or≥8 point improvement at 24h) were calculated. Good functional outcome (GFO) at 90 days was defined as a mRS of 0-2. Independent predictors of outcome were identified by multivariate logistic regression. We performed a secondary analysis after excluding patients presenting with a minor stroke (NIHSS 0-5) since the assessment of change in NIHSS might be more reliable in patients presenting with a moderate to severe deficit.
Results: We analyzed the outcome in 369 patients. B-NIHSS was associated with GFO (OR: 0.82; 95%CI 0.77-0.86). In a multivariate model with B-NIHSS and age as predictors the accuracy (AUC: 0.82) improved by including the delta NIHSS (AUC: 0.86; p<0.01). In 131 patients with moderate to severe stroke, the predictive multivariate model was more accurate when including the RR NIHSS (AUC: 0.83) to the model which included B-NIHSS, age and ischemic heart disease (AUC: 0.77; p=0.03). 
Conclusion: B-NIHSS is a predictor of stroke outcome. In this cohort the prediction of GFO was improved by adding change in stroke severity after 24h to the model.