Using Radiological Data to Estimate Ischemic Stroke Severity

Area of Science:

• Neurology
• Radiology
• Health Services Research

Background:

• Assessing stroke care quality often relies on risk-adjusted mortality, but stroke severity measures like the NIHSS are not always in administrative data.
• Radiology reports, particularly head CT scans, are frequently available in electronic health records.

Purpose of the Study:

• To determine if admission head computed tomography (CT) data can be utilized to estimate stroke severity.
• To develop and validate a radiographic measure of stroke severity (BIS) using head CT findings.

Main Methods:

• A cohort of 1348 acute ischemic stroke patients (1998-2003) was analyzed.
• A radiographic score (BIS) was developed and validated using chart review data, comparing it against the dichotomized National Institutes of Health Stroke Scale (NIHSS) and in-hospital mortality.
• The BIS was assessed for its ability to model severe stroke and in-hospital mortality.

Main Results:

• 86.5% of patients had abnormal findings on initial head CT scans.
• The BIS demonstrated moderate predictive ability for severe stroke (c-statistics: development .581, validation .579).
• The BIS showed moderate performance in predicting in-hospital mortality (c-statistics: development .623, validation .678).

Conclusions:

• The developed 2-variable radiographic score (BIS) provided significant risk stratification information for stroke severity and mortality.
• While c-statistics were moderate, radiographic data from head CT can offer valuable information for risk adjustment in stroke care quality assessment when NIHSS is unavailable.