Related Experiment Video
Updated: Dec 28, 2025
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
16.1K
Deriving a Passive Surveillance Stroke Severity Indicator From Routinely Collected Administrative Data: The PaSSV
Amy Y X Yu1,2, Peter C Austin2, Mohammed Rashid2
1Department of Medicine (Neurology), University of Toronto, Sunnybrook Health Sciences Centre, ON, Canada (A.Y.X.Y.).
Circulation. Cardiovascular Quality and Outcomes
|February 19, 2020
Summary
Researchers developed an estimated stroke severity measure using administrative data, crucial for stroke outcomes research when direct severity information is unavailable. This new indicator aids in risk adjustment and health system performance assessments.
Area of Science:
- Neurology
- Health Services Research
- Biostatistics
Background:
- Stroke severity is vital for outcomes research but often missing in administrative data.
- Existing administrative data lacks a reliable indicator for baseline stroke severity.
- Accurate stroke severity assessment is needed for robust research and performance evaluation.
Purpose of the Study:
- To derive an indicator of baseline stroke severity using administrative healthcare data.
- To evaluate the utility of this estimated stroke severity measure in predicting mortality.
- To enable risk adjustment in population-based stroke outcomes research.
Main Methods:
- A linear regression model was used to estimate the Canadian Neurological Scale (CNS) from administrative variables.
- Cox-proportional hazards models assessed the association between stroke severity (observed and estimated CNS) and 30-day mortality.
- Model discrimination was evaluated using C statistics, with validation in external cohorts and using the National Institute of Health Stroke Scale.
Main Results:
- An estimated stroke severity measure was successfully derived from administrative data for over 41,000 patients.
- The association between stroke severity and mortality was comparable between observed and estimated CNS.
- Cox models showed improved mortality prediction with estimated CNS (C statistic: 0.76) compared to no severity data (C statistic: 0.69).
Conclusions:
- An estimated stroke severity measure can be reliably derived from administrative data.
- This derived measure is valuable for risk adjustment in stroke outcomes research.
- The findings support the use of administrative data for assessing health system performance in stroke care.
