Introduction Cardiovascular MRI (CMR) is the gold standard for calculation of left ventricular ejection fraction (LVEF). Echocardiography (echo) has been shown to consistently underestimate LVEF when compared to CMR. However, this has not been demonstrated for different categories of LVEF.

Methods This was an observational study of patients with both CMR and echo studies, between May and October 2017, identified via local UK National Heart Failure Audit data or CMR records. Patients were excluded if the imaging tests were performed >60 days apart or if they had a diagnosis that would predictably change between tests. LVEF categories were defined according to ESC guidelines.

Results 103 cases with both imaging modalities were included. Overall, echo underestimated LVEF compared to CMR (CMR 46.5%±1.8%, echo 37.33%±1.4%, p<0.0001). There was no significant variation in discrepancy over time (r² 0.04). Subgroup analysis according to echo technique (Simpson’s vs 3D) and underlying diagnosis (where at least n=10 available) did not alter the overall trend. Importantly, echo underestimated LVEF compared to CMR for all categories of LVEF (<40% CMR 34.0±1.7% vs echo 26.1%±1.0%, p<0.0001; 40%–50% CMR 60.1±2.8% vs 40%–50% echo 45.7%±0.7%, p<0.0001;>50% CMR 63.7±2.1% vs echo 56.4%±0.9%, p=0.0019). Using echo as the default measurement, CMR LVEF assessment would result in 36 patients (35.0%) being reclassified.

Conclusion Echo underestimates LVEF compared to CMR across all LVEF subgroups and diagnoses, regardless of technique used. This is of interest in HFmrEF trials where, if echo is used for LVEF assessment, many included patients may have normal systolic function.