Left ventricular ejection fraction: real-world comparison between cardiac computed tomography and echocardiography in a large population.

PURPOSE: This study compared cardiac computed tomography (CT) and two-dimensional transthoracic echocardiography (ECC) for assessing left ventricular ejection fraction (LVEF) using real-world data from a large patient population. MATERIALS AND METHODS: We studied 450 patients (284 males; mean age 64±12 years; range 12-88) who underwent CT and ECC due to suspected coronary artery disease. For CT, we used multiphase short-axis reconstructions and evaluated them with a dedicated software tool that uses Simpson's rule to compute LV volumes. For ECC, computation was based on the biplane Simpson's method. Results in terms of EF were compared with the paired Student's t test, Pearson's correlation coefficient (r), and Bland-Altman analysis. RESULTS: EF was 52%±15% for CT and 55%±13% for ECC. Statistically significant differences, albeit with good correlation, were observed between the measurements (r=0.71; p<0.05). ECC showed a slight tendency to overestimate EF. When the population was divided into subgroups according to EF, this was underestimated by ECC in the subgroup with EF >50% and overestimated in those with EF 35%-50% and <35%, with consistently significant differences between ECC and CT (p<0.05) and progressively lower levels of agreement. CONCLUSIONS: In the real-world assessment of EF, ECC provides significantly different data from CT, with a bias that increases proportionally to LV systolic dysfunction.