Abstract 28: Defining the Diagnosis of TIA: An Inter-rater Agreement Analysis of a New Precise Diagnostic Score (PREDISC)

Abstract

Background: the “gold standard” for the diagnosis of transient ischemic attack (TIA) is typically the final diagnosis of a neurologist. However, inter-rater agreement between physicians for the diagnosis of TIA is low, even among vascular neurologist. We developed a diagnostic scoring scale (PREDISC), which features well defined clinical and imaging parameters and investigated inter-rater agreement in patients with suspected TIA. Methods:fellowship-trained vascular neurologists, blinded to patient’s final diagnosis or outcomes, independently reviewed the history, clinical examination and MRI (including DWI and perfusion imaging) findings of patients with suspected TIA who were referred to our institution's TIA program. Each patient’s clinical manifestation and MRI findings were rated using the PREDISC score, an 8-point scale composed of a clinical score (0-4 points) and an Imaging score (0-4 points). The imaging score includes specific criteria for DWI positivity (4 points) as well as criteria for assessment of clinically relevant perfusion lesions (maximum 3 points for patients who are DWI negative). The final PREDISC score for confirmation that the event was caused by brain ischemia is determined as: 0-1 = unlikely, 2= possible, 3= probable, 4-8 = very likely. We assessed global agreement between the raters. Results: Three raters reviewed data from 114 patients. Using PREDISC all 3 raters scored a similar percentage of the clinical events as being “probable” or “very likely” caused by brain ischemia: 57%, 55% and 58%. Agreement was high for both total PREDISC Score (Intraclass correlation coefficient [ICC] of 0.94) and for the 4-point scale of TIA likelihood of diagnosis (AC1 agreement coefficient of 0.84). Agreement was excellent both on the clinical and imaging sub scores (ICC 0.88 and 0.95). Conclusion:Based on comparison with prior studies, inter-rater agreement for the diagnosis of TIA appears to be substantially improved with the PREDISC structured scoring system. This scale is the first to include specific criteria to assess the clinical relevance of DWI and perfusion lesions and supports the added value of MRI for assessing patients with suspected TIA. The diagnostic value of PREDISC should be confirmed in a prospective multicenter trial.