Amar C. Gupta BA, Presenter: Nothing to Disclose

Zeshan Chaudhry MD, Abstract Co-Author: Nothing to Disclose

Thabele Mbuso Leslie-Mazwi MD, Abstract Co-Author: Nothing to Disclose

Ronil Vikesh Chandra MBBS, Abstract Co-Author: Nothing to Disclose

Ramon Gilberto Gonzalez MD, PhD, Abstract Co-Author: Nothing to Disclose

Joshua A. Hirsch MD, Abstract Co-Author: Consultant, CareFusion Corporation Consultant, Johnson & Johnson Consultant, Koninklijke Philips Electronics NV Shareholder, IntraTech Medical Ltd

Pamela Whitney Schaefer MD, Abstract Co-Author: Nothing to Disclose

Albert J Yoo MD, Abstract Co-Author: Nothing to Disclose

The Alberta Stroke Program Early CT Score (ASPECTS) is a semi-quantitative system developed to assess early ischemic changes in ten regions of the middle cerebral artery territory on NCCT. ASPECTS score >7 has been shown to predict good outcomes following intra-arterial therapy (IAT). To date, there has been controversy regarding the interobserver reliability of ASPECTS.  We sought to evaluate its reliability and its affect on treatment decision making in the setting of proximal artery occlusions undergoing IAT. We also examined the interobserver reliability of ASPECTS by topographic region.

This retrospective cohort study evaluated 169 consecutive major ischemic stroke patients who underwent emergent IAT. NCCT (5mm axial slices) imaging was scored using ASPECTS by two experienced neuroradiologists. The readers independently evaluated each scan, blinded to all clinical information except stroke side. 14 patients were excluded due to chronic infarcts or lack of appropriate imaging. Statistical analysis of agreement included Cohen’s kappa, concordance correlation coefficient, Spearman’s rank correlation coefficient (ρ), and prevalence adjusted bias adjusted kappa (PABAK) scores for ASPECTS dichotomized at >7 versus ≤7 and overall scores.     

For ASPECTS dichotomized at >7 versus ≤7, moderate overall agreement was achieved (κ = 0.53, PABAK = 0.54) with a 76.8% observed rate of agreement (bias index = 0.04, prevalence index = 0.11). There was also moderate agreement for overall ASPECTS (weighted κ = .58) with a Spearman’s rank correlation ρ = 0.77 (95%CI:0.69, 0.83; p < 0.0001), and concordance correlation coefficient = 0.77 (0.70, 0.82). Agreement for each individual ASPECTS region is provided in the Table. Adjusting for bias and prevalence, the worst agreement was seen with the lentiform nucleus (PABAK=0.535). There was moderate to excellent agreement for the remainder of the regions.

Among highly experienced readers, there is moderate interobserver reliability for dichotomized ASPECTS > 7 versus ≤7, despite good agreement for individual regions. In approximately 25% of patients, treatment decisions for IAT would be affected by the inter-rater reliability. Our results warrant further investigation into improving the diagnostic performance of ASPECTS for clinical decision making.  

ASPECTS shows moderate interobserver reliability, despite good agreement for individual ASPECTS regions.  

Gupta, A, Chaudhry, Z, Leslie-Mazwi, T, Chandra, R, Gonzalez, R, Hirsch, J, Schaefer, P, Yoo, A, Interobserver Reliability of ASPECTS for Endovascular Stroke Therapy.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11009881.html