Stress echocardiography, stress sing le-photon-emission computed tomography and electron beam computed tomography for the assessment of coronary artery disease: A meta-analysis of diagnostic performance

Background Many studies have been published on the diagnostic performance of noninvasive tests for the assessment of coronary artery disease. The objective of the present study was to compare the published literature on the diagnostic performance of stress echocardiography, stress sing le-photon-emission computed tomography (SPECT), and electron beam computed tomography (EBCT). Methods Meta-analytic studies on the diagnostic performance of imaging tests for coronary artery disease were searched in the Cochrane Library, PubMed, and bibliographies of selected articles. Sensitivities, specificities, and diagnostic odds ratios of the source studies were calculated per modality. Taking into account differences between studies, a random effects summary receiver operating characteristic analysis was performed.. Results We analyzed the data of 351 patient series, which were reported in I I meta-analyses. The sensitivity of EBCT was significantly higher than that of stress SPECT, which had a significantly higher sensitivity than stress echocardiography (respectively, 93.1% [95% confidence interval, 90.7-95.6], 88.1 [95% confidence interval, 86.6-89.6], and 79.1% [95% confidence interval, 77.6-80.51). The specificity of stress echocardiography was significantly higher than that of stress SPECT, which had a significantly higher specificity than EBCT (respectively, 87.1% [95% confidence interval, 85.7-88.5], 73.0% [95% confidence interval, 69.1-76.9], and 54.5% [95% confidence interval, 45.3-63.81). The diagnostic odds ratios did not differ significantly between the 3 modalities, which resulted in one underlying summary receiver operating characteristic curve. Conclusions This study suggests that there are no significant differences in the overall diagnostic performance between stress echocardiography, stress SPECT, and EBCT for the diagnosis of coronary artery disease. However, differences exist in sensitivity and specificity estimates, which may make each modality useful in different settings.