Observer bias in the interpretation of dobutamine stress echocardiography

Background and hypothesis : Blinded image analysis is typically utilized in published studies evaluating the accuracy of dobutamine stress echocardiography (DSE). However, in clinical settings, practical considerations may limit the use of blinded interpretations and thus the potential for observer bias arises. This study evaluated the relationships between clinical and blinded interpretations of DSE. Methods : Wall motion analysis from clinical and blinded DSE interpretations were compared and factors associated with their concordance were investigated in 115 consecutive patients with known or suspected coronary artery disease. Results : Clinical and blinded interpretations agreed on the presence or absence of inducible ischemia in 102 of 115 cases (88.7% k = 0.76, p<0.00001). In studies in which the clinical and blinded interpretations were in agreement, there was greater ST‐segment depression (STD) in echocardiographically positive compared with negative studies (mean STD 0.73 ± 0.65 vs. 0.42 ± 0.67 mm; p = 0.008). In contrast, studies in which there was disagreement had significantly less ST‐segment changes (mean STD 0.19 ± 0.56 mm; p = 0.012) despite comparable results on blinded wall motion analysis. Multiple logistic regression for factors related to the results of clinical and blinded wall motion analysis disclosed that angina pectoris and ST‐segment changes were related to clinical interpretations, whereas only angina pectoris was related to the findings on blinded analysis. Conclusions : Clinical interpretations of echocardiographic images during DSE overall demonstrate good agreement with the results of blinded analysis. Ancillary testing data may influence the analysis of wall motion abnormalities, and thus the potential for observer bias exists unless these interpretations are performed blinded to other clinical data.