Introduction
Stress echocardiography (SE) is a noninvasive, functional method of assessing for presence of flow-limiting coronary artery disease (CAD) without employing radiation and is a relatively low cost alternative to nuclear and magnetic resonance imaging. Dobutamine stress echocardiography (DSE) has a reported sensitivity and specificity of 81-86% and 84-86% respectively \cite{Sicari_2017}\cite{Picano_2008}; the sensitivity and specificity of exercise SE is similar (83% and 84% respectively) \cite{Sicari_2017}. However, it requires skill to both perform and interpret this modality and the American Society of Echocardiography recommends that it only be performed by physicians with specific expertise in the technique in centers of adequate volume and supplemented by continuing medical education \cite{Pellikka_2007}. The differences in both image acquisition and assessment criteria have led to a significant variation in interpretation of stress echo among cardiologists and among institutions \cite{Hoffmann_1996}.
There is a prevailing consensus that core laboratory interpretations are more accurate than clinical site results. We thus sought to determine concordance and discordance between site and core laboratory interpretation of stress echocardiograms in this substudy of the PROMISE trial \cite{2015} and identify factors that may influence disagreement
Methods
Study protocol
We analyzed data from the stress echocardiogram subgroup who were prospectively recruited as part of the PROMISE trial between July 27, 2010, and September 19, 2013. The PROMISE trial enrolled over 10 000 patients from 193 sites in North America who had symptoms suspicious for CAD requiring elective non-invasive testing and randomized to the use of coronary computed tomographic angiography (CTA) or to functional testing. Functional testing was performed with either exercise electrocardiography, nuclear stress testing, or stress echocardiography. The study sites represented both community and academic centers with all study sites certified by a central Coordinating Center to perform and interpret all tests. All tests were performed and interpreted by local physicians and clinical decisions were made based on local interpretation of test results. The design of the PROMISE trial has been previously described in detail \cite{2015} and the study protocol was approved by each local institutional review board at each enrolling site. Patients gave informed consent for participation.
Clinical data
Study sites entered patient demographic information as well as information on cardiovascular risk factors, presenting symptoms, physician characterization of chest pain, medication use, and electrocardiographic findings. Accreditation status was obtained from study site reporting and from the Intersocietal Accreditation Commission echocardiography directory. Canadian study sites which did not fall within Intersocietal Accreditation Commission jurisdiction were considered accredited due to their high volumes. Clinical study outcomes were referral for a second non-invasive test or coronary angiography within 90 days of randomization, revascularization within 90 days, and composite outcomes of cardiovascular death and myocardial infarction at one year and all-cause mortality, myocardial infarction, and hospitalization for unstable angina at one year.
Stress echocardiography
The method of stress (exercise versus dobutamine) was at the discretion of the site including the addition of atropine or handgrip to achieve target heart rate. Blood pressure and electrocardiographic data as well as patient symptoms during exercise were gathered by study sites. Stress echocardiograms were interpreted locally at study sites and a random subset as well as site-interpreted abnormal studies were reviewed at a central core laboratory. Only images were sent to the core laboratory for interpretation and no ancillary clinical information was available. Wall motion was divided according to the three main vascular territories and scored according to a 17-segment model. A stress echocardiogram was considered positive if at least two segments developed a wall motion abnormality at peak stress. Uninterpretable studies, and those with two or more segments missing at rest or stress were excluded from analysis. Study quality was adjudicated by study sites and use of contrast for left ventricular opacification was at the discretion of the performing physician.