Validation of Diagnoses of Transient Ischemic Attack in the Swedish Stroke Register (Riksstroke) TIA-Module

BACKGROUND: In 2010, the Swedish Stroke Register (Riksstroke; RS) established a module for transient ischemic attacks (RS-TIA). We report a diagnostic validation study of patients included in RS-TIA. METHODS: During the first year, 7,825 patients were registered at 59 out of 74 Swedish hospitals. A time-based TIA definition was applied. A sample of 180 patients (30 patients each from 6 hospitals), with a similar distribution of age and sex as in RS-TIA, was prepared. Two independent observers assessed medical records for quality of documentation and assigned a diagnosis of likely, possible, unlikely TIA or ischemic stroke, according to prespecified criteria. RESULTS: The 2 observers agreed in 77% of cases that the event was a likely or possible TIA, in 3% that the event was an ischemic stroke, and in 2% that the event was an unlikely TIA. The observers disagreed in 8% of patients on TIA vs. ischemic stroke, and in 11% on a vascular vs. non-vascular cause. Quality of documentation was fair. CONCLUSIONS: There was interobserver agreement on diagnosis of TIA in the majority of patients included in RS-TIA. Diagnostic accuracy may be further improved by more systematic documentation of symptoms and signs.