The CLUE examined 4 targets using 4 previously-validated, nonquantitative ultrasound “signs” [2
]. In brief, the parasternal long axis view was assessed for LV systolic dysfunction (LVSD) and left atrial enlargement (LAE). LVSD was present if the anterior leaflet of the mitral valve during diastole did not appear to encroach upon the LV outflow tract and approach the septum to within approximately 1
cm by nonquantitative estimation. LAE was present if the anteroposterior diameter of the LA appeared larger than that of the overlying aorta throughout the cardiac cycle. Longitudinal views at the mid-infraclavicular intercostal space of each lung apex were assessed for extravascular lung water by ultrasound lung comet-tail artifacts (ULC+) [7
], considered present when 3 vertical hyperechoic lines per image were seen to emanate from the pleural line in the near field and reach the far field, in either lung apical view. Each lung apex was a separate CLUE view and a positive ULC sign was defined when either lung apex or both lung apices were ULC+. The subcostal longitudinal view of the proximal intrahepatic inferior vena cava as it entered the right atrium, within approximately 1-2
cm of the diaphragm, was assessed for elevated central venous pressures (eCVP), defined when the IVC appeared plethoric by visual interpretation with parallel vessel walls and a luminal diameter reduction of <50% with respiratory motion of the diaphragm, without forced “sniffing” (see ).
Components of the CLUE Protocol.
Three trainees (medical student, intern, and pharmacy resident) with less than 1-hour device orientation and no prior ultrasound experience or training performed the CLUE using a PUD (either the Vscan, GE Healthcare, or the P10, Siemens Healthcare) on 27 subjects (22 outpatients and 5 normal volunteers) at a medical office while being directed by an off-site cardiologist through an iPhone 4 or iPod (Apple Inc.) attached to the PUD as follows. The iPhone or iPod had been affixed to the base of the PUD using a small commercially available dashboard phone mount so that the iPhone/iPod's front facing 0.3 megapixel VGA camera was directed at the opposing PUD display screen (see Figures , , and ). Using the application Apple FaceTime (WPA2 Enterprise, 128-bit AES Encryption; HIPAA Compatible; VGA resolution; 30
fps), data was transmitted wirelessly through currently available off-the-shelf Wi-Fi networks (2.4
g, wireless router) to the cardiologist's receiving iPod. The cardiologist provided remote audiovisual guidance with real time feedback on operation of the PUD, image acquisition, and interpretation for CLUE signs to the trainee. The Vscan employs a 1.7–3.8
MHz phased array probe and the Siemens P10 uses a 2–4
MHz phased array probe. Only one device was used throughout a CLUE exam. As the study tested real-time wireless guidance, no images were stored or reviewed off-line. All subjects were asymptomatic. Trainees, the sonographer, and echocardiography expert were blinded to their clinical histories.
Screening accuracy and technical quality of transmitted CLUE images were compared to gold-standard echocardiograms that included lung apical imaging performed on a standard platform (Acuson Cypress, Siemens Healthcare). The gold-standard echocardiograms were obtained by a registered sonographer blinded to the CLUE results according to standard guidelines [8
] and was interpreted by an expert echocardiographer in a blinded and randomized fashion. LVSD was defined as an interpreted ejection fraction <40% using the interpreter's discretion of all available methods including nonquantitative expert estimation, the biplane method of discs (modified Simpson's rule), or fractional shortening. LAE was defined by an anteroposterior LA diameter >4.0
cm or LA volume index >28
measured using the area-length method [8
]. Interpretation of the eCVP and ULC+ findings on the echocardiogram used the same method as the PUD. All images were assigned a quality score: 0 (no image), 1 (only motion detected; off-axis), 2 (“suboptimal,” poor delineation of structures), 3 (“adequate” for diagnosis of particular sign), or 4 (“optimal,” good delineation of all structures equivalent to idealized standard echocardiographic view). Views with scores >2 were considered technically adequate.
The diagnostic sensitivity, specificity, positive and negative predictive values, and accuracy were derived for the CLUE diagnostic criteria for LVSD, LAE, ULC+, and eCVP by comparing the interpretation of technically adequate CLUE views with the results of LVSD, LAE, ULC+, and eCVP from the reference standard echocardiogram. The Scripps Institutional Review Board approved the study.