Agreement between centers on the interpretation of exercise echocardiography

A low-to-moderate level of agreement on the interpretation of dobutamine echocardiography has been reported, but there are no similar findings on exercise echocardiography. The objectives of this study were to assess the level of agreement between centers on the use of exercise echocardiography and to evaluate the accuracy of the technique when used in a blinded manner. Six institutions with experience in exercise echocardiography each sent 25 study results to the other centers. Of these, 15 were positive or negative studies on consecutive patients undergoing coronary angiography, and 10 were on non-diabetic patients who had non-coronary chest pain or were asymptomatic and whose pretest probability of coronary artery disease was < 10%. Each institution evaluated 150 studies: 125 blinded and 25 of their own with knowledge of clinical data. For 116 patients (78%), four or more of the five centers blindly evaluating each study agreed with the positive or negative result. The average kappa coefficient was 0.48 (intercenter range 0.45-0.52). The percentage agreement was higher with three-vessel disease (93%, range 85%-95%), with left anterior descending coronary artery disease (83%, range 80%-86%), and when the referring institution reported baseline dyssynergy (86%, range 82%-90%), dyssynergy in left anterior descending coronary artery territory (81%, range 76%-84%), or a peak wall motion score index > 1.50 (88%, range 85%-90%). When the technique was used blinded to detect > or = 50% coronary narrowing in > or = 1 vessel, its sensitivity, specificity and accuracy were 68%, 66% and 67%, respectively, with wide variability between centers. There was moderate agreement between centers on the interpretation of exercise echocardiography. When used blinded, the technique's accuracy was lower than that reported when clinical data is known.