Incorporating Stroke Severity Dynamics to Improve Prognostic Modelling in Ischaemic Stroke Patients: Evidence from Serial NIHSS Assessments in Bergen NORSTROKE Study
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Stroke severity evolves rapidly in the acute phase, and yet severity is commonly measured only on admission and treated as a static, single-timepoint measure in stroke research. This study explores routinely collected, repeated National Institute of Health Stroke Scale (NIHSS) assessments to better understand the dynamic nature of stroke severity and evaluate its value for outcome prediction.
Methods
We analysed real-world data from 1,622 ischaemic stroke patients admitted to the Bergen NORSTROKE study, all of whom had at least two NIHSS assessments within 48 hours of symptom onset. Stroke severity trajectories were explored using summary statistics, spaghetti plots, linear mixed-effects models (LMM), and group-based trajectory modelling (GBTM). Seven logistic regression models were compared for their ability to predict favourable short-term functional outcomes (modified Rankin Scale), incorporating various representations of stroke severity dynamics. Model performance was evaluated using AIC, BIC, and area under the ROC curve (AUC).
Results
NIHSS scores varied considerably over time. Most patients had mild strokes (median NIHSS = 2) and showed improvement within 48 hours. GBTM identified three latent groups: Very Low-Stable (40.3%), Moderate Low-Stable (41%), and High-Mildly Improving (18.7%). Models incorporating stroke severity dynamics outperformed those using admission NIHSS alone (AUC 0.835 vs. 0.778). The best predictive performance was achieved using random intercepts and slopes from the LMM. While GBTM improved model fit (AIC = 1624), its added discriminatory power was limited (AUC = 0.792).
Conclusion
Stroke severity evolves dynamically in the acute phase. Incorporating these dynamics into prognostic models improves predictive accuracy and model fit. Advanced modelling approaches that account for individual symptom trajectories offer a more accurate and clinically relevant framework for stroke outcome prediction. This also underscores the importance of incrementally updating prediction models with new clinical information, as changing severity is a key but often overlooked element in stroke studies.
