Variability in Ejection Fraction Measured By Echocardiography, Gated Single-Photon Emission Computed Tomography, and Cardiac Magnetic Resonance in Patients With Coronary Artery Disease and Left Ventricular Dysfunction

IMPORTANCE: Clinical decisions are frequently based on measurement of left ventricular ejection fraction (LVEF). Limited information exists regarding inconsistencies in LVEF measurements when determined by various imaging modalities and the potential impact of such variability. OBJECTIVE: To determine the intermodality variability of LVEF measured by echocardiography, gated single-photon emission computed tomography (SPECT), and cardiovascular magnetic resonance (CMR) in patients with left ventricular dysfunction. DESIGN, SETTING, AND PARTICIPANTS: International multicenter diagnostic study with LVEF imaging performed at 127 clinical sites in 26 countries from July 24, 2002, to May 5, 2007, and measured by core laboratories. Secondary study of clinical diagnostic measurements of LVEF in the Surgical Treatment for Ischemic Heart Failure (STICH), a randomized trial to identify the optimal treatment strategy for patients with LVEF of 35% or less and coronary artery disease. Data analysis was conducted from March 19, 2016, to May 29, 2018. MAIN OUTCOMES AND MEASURES: At baseline, most patients had an echocardiogram and subsets of patients underwent SPECT and/or CMR. Left ventricular ejection fraction was measured by a core laboratory for each modality independent of the results of other modalities, and measurements were compared among imaging methods using correlation, Bland-Altman plots, and coverage probability methods. Association of LVEF by each method and death was assessed. RESULTS: A total of 2032 patients (mean [SD] age, 60.9 [9.6] years; 1759 [86.6%] male) with baseline LVEF data were included. Correlation of LVEF between modalities was r = 0.601 (for biplane echocardiography and SPECT [n = 385]), r = 0.493 (for biplane echocardiography and CMR [n = 204]), and r = 0.660 (for CMR and SPECT [n = 134]). Bland-Altman plots showed only moderate agreement in LVEF measurements from all 3 core laboratories with no substantial overestimation or underestimation of LVEF by any modality. The percentage of observations that fell within a range of 5% ranged from 43% to 54% between different imaging modalities. CONCLUSIONS AND RELEVANCE: In this international multicenter study of patients with coronary artery disease and reduced LVEF, there was substantial variation between modalities in LVEF determination by core laboratories. This variability should be considered in clinical management and trial design. TRIAL REGISTRATION: Clinicaltrials.gov Identifier: NCT00023595