ASPECTS Interobserver Agreement of 100 Investigators from the TENSION Study

PURPOSE: Evaluating the extent of cerebral ischemic infarction is essential for treatment decisions and assessment of possible complications in patients with acute ischemic stroke. Patients are often triaged according to image-based early signs of infarction, defined by Alberta Stroke Program Early CT Score (ASPECTS). Our aim was to evaluate interrater reliability in a large group of readers. METHODS: We retrospectively analyzed 100 investigators who independently evaluated 20 non-contrast computed tomography (NCCT) scans as part of their qualification program for the TENSION study. Test cases were chosen by four neuroradiologists who had previously scored NCCT scans with ASPECTS between 0 and 8 and high interrater agreement. Percent and interrater agreements were calculated for total ASPECTS, as well as for each ASPECTS region. RESULTS: Percent agreements for ASPECTS ratings was 28%, with interrater agreement of 0.13 (95% confidence interval, CI 0.09-0.16), at zero tolerance allowance and 66%, with interrater agreement of 0.32 (95% CI: 0.21-0.44), at tolerance allowance set by TENSION inclusion criteria. ASPECTS region with highest level of agreement was the insular cortex (percent agreement = 96%, interrater agreement = 0.96 (95% CI: 0.94-0.97)) and with lowest level of agreement the M3 region (percent agreement = 68%, interrater agreement = 0.39 [95% CI: 0.17-0.61]). CONCLUSION: Interrater agreement reliability for total ASPECTS and study enrollment was relatively low but seems sufficient for practical application. Individual region analysis suggests that some are particularly difficult to evaluate, with varying levels of reliability. Potential impairment of the supraganglionic region must be examined carefully, particularly with respect to the decision whether or not to perform mechanical thrombectomy.