Determination of interobserver variability for identifying inducible left ventricular wall motion abnormalities during dobutamine stress magnetic resonance imaging.

Abstract

To determine the interobserver variability for identifying inducible left ventricular (LV) wall motion abnormalities during high-dose dobutamine/atropine stress cardiovascular magnetic resonance (DSMR). Four readers from various institutions were supplied with the image data from 150 consecutive DSMR examinations and asked to grade wall motion and image quality throughout graded doses of dobutamine infusion administered to achieve 85% of the maximum age-predicted heart rate. Inducible ischaemia was identified if more than one segment demonstrated a new or worsening LV wall motion abnormality, and significant stenosis was defined as > or =50% luminal diameter reduction by quantitative contrast coronary angiography. Seventy-seven patients (51%) had luminal narrowings > or =50%. Diagnostic performance (sensitivity, specificity, diagnostic accuracy) of all readers was 78.2, 87.0 and 82.5%. Disagreement between two readers occurred in every seventh examination. Agreement on the presence or absence of inducible wall motion abnormalities was moderate (mean kappa value 0.59, range 0.52-0.76). Diagnostic performance and disagreement were independent of the presence of luminal narrowings > or =50% or the number of diseased coronary vessels. Image quality was regarded excellent in 89.3% of standard views. In the setting of multiple observers from different institutions performing a diagnostic reading of DSMR examinations carried out at a single centre, the interobserver variability was low for identifying inducible LV wall motion abnormalities indicative of coronary arterial luminal narrowings > or =50%.