Analysis of interinstitutional observer agreement in interpretation of dobutamine stress echocardiograms

Objectives. This study sought to determine the degree of interinstitutional agreement in the interpretation of dobutamine stress echocardiograms. Background. Dobutamine stress echocardiography involves subjective interpretation. Consistent methods for acquisition and interpretation are of critical importance for obtaining high interobserver agreement and for facilitating communication of test results. Methods. Five experienced centers were each asked to submit 30 dobutamine stress echocardiograms (dobutamine up to 40 mu g/kg body weight per min and atropine up to 1 mg) obtained in patients undergoing coronary angiography. Thus, a total of 150 dobutamine stress echocardiograms were interpreted by each center without knowledge of any other patient data. Left ventricular wall motion was assessed using a 16-segment model but was otherwise not standardized. No patient was excluded because of poor image quality or inadequate stress level. Echocardiographic image quality was assessed using a five point scale. Results. Angiographically significant coronary artery disease (greater than or equal to 50% diameter stenosis) was present in 95 patients (63%). By a majority decision (three or more centers), the sensitivity, specificity and accuracy of dobutamine echocardiography were 76%, 87% and 80%, respectively, Abnormal or normal results of stress echocardiography were agreed on by four or all five of the centers in 73% of patients (mean kappa value 0.37, fair agreement only), Agreement on the left anterior descending artery territory (78%) was similar to that for the combined right coronary artery/left circumflex artery territory (74%), and for specific segments the agreement ranged from 84% to 97% and was highest for the basal anterior segment and lowest for the basal inferior segment, Agreement was higher in patients with no (82%) or three-vessel coronary artery disease (100%) and lower in patients with one- or two vessel disease (61% and 68%, respectively). Agreement on positivity or negativity of stress test results was 100% for patients with the highest image quality but only 43% for those with the lowest image quality (p = 0.003). Conclusions. The current heterogeneity in data acquisition and assessment criteria among different centers results in low inter institutional agreement in interpretation of stress echocardiograms. Agreement is higher in patients with no or advanced coronary artery disease and substantially lower in those with limited echocardiographic image quality. To increase interinstitutional agreement, better standardization of image acquisition and reading criteria of stress echocardiography is recommended.