Developing predictive models of excellent and devastating outcome after stroke.

Area of Science:

• Neurology
• Clinical Prediction Modeling
• Stroke Medicine

Background:

• Predictive models for functional status post-stroke are crucial for clinical trials and outcome comparisons.
• This study aimed to develop models for predicting excellent (mRS 0-1) and devastating (mRS 5-6) stroke outcomes.

Purpose of the Study:

• To develop and validate predictive models for excellent and devastating functional outcomes six months after stroke.
• To identify key clinical and radiological predictors for these outcomes.

Main Methods:

• Enrolled 538 patients with ischemic or hemorrhagic stroke.
• Utilized 16 clinical variables and 6 radiological variables from acute CT scans.
• Developed and externally validated predictive models for outcomes at 6 months.

Main Results:

• Thirty percent of patients achieved excellent outcomes, and 30% had devastating outcomes.
• Models for excellent outcome (AUC 0.866-0.882) included age, pre-stroke status, stroke severity, arm/leg function, GCS, and leukoaraiosis.
• Models for devastating outcome (AUC 0.859-0.874) included additional factors: living alone and total anterior circulation stroke.

Conclusions:

• Externally validated models for predicting excellent and devastating stroke outcomes were developed.
• Leukoaraiosis was the sole independent radiological predictor for both outcomes.
• Living alone pre-stroke independently predicted a devastating outcome.