172 Determining systematic bias in the assessment of left ventricular function between transthoracic echo and cardiac MRI using new generation imaging systems. (4th June 2021)

• Record Type: Journal Article
• Authors: Chapman, Sally; Fent, Graham; O'Toole, Laurence; Watson, Oliver; Mackay, Jane; Garg, Pankaj
• Is Part Of: Heart. Volume 107(2021)Supplement 1
• Page Start: A133
• Page End: A134
• Publication Date: 2021-06-04
• Subjects: Function -- Bias -- Imaging; Heart -- Diseases -- Treatment -- Periodicals; Cardiology -- Periodicals; 616.12
• DOI: 10.1136/heartjnl-2021-BCS.169
• Languages: English
• ISSNs: 1355-6037

Abstract

Abstract : Introduction: Estimation of left ventricular systolic function (LVEF) commonly underpins clinical decision making in Cardiology. TTE has previously been demonstrated to underestimate LVEF compared with CMR. Sheffield Teaching Hospitals Trust (STH) recently upgraded to the latest generation of TTE scanners (GE E95) and CMR scanner (Siemens Magnetom Aera 1.5T). This service evaluation investigated correlation and bias between TTE and CMR LVEF reporting at our institution. Methods: Consecutive TTE studies at STH from April-October 2018 were cross-referenced, identifying patients who had also undergone CMR during this period. All patients – in and outpatients - with both TTE and CMR within 30 days were included. Patients with missing data, reversible aetiology or poor endocardial definition were excluded. The determination of LV function by echo was made using Simpson's biplane where endocardial definition was sufficient. If not, it was determined by visual assessment and a 5% range quoted. A second opinion was sought if LV function was assessed as <40%. LV function by cardiac MRI was determined by standard semi-automated endocardial contouring at end diastole and end systole using Siemens Syngovia software. The cut off for severe LV dysfunction was <35% and >55% for normal. Results: 12301 TTE and 398 CMR scans were performed in the study period. 120 patients met inclusion criteria with TTE and CMR within 30 days. Reasons for exclusion included missing data (n=2), reversible cause for LV dysfunction (n=1) and poor endocardial definition (n=4) leaving a study population of 113 patients. LVEF by TTE and CMR ( figure 1 ) were strongly correlated (R=0.85, p <0.001). Bland Altman analysis demonstrated a signal of systematic bias with LVEF by CMR 4.4% higher than TTE (mean bias 4.4%, 95% CI -0.3 to 9.0, p = 0.06). Severe LV function on both TTE and CMR were in accordance in 91% of cases. Normal LV function by TTE and CMR was in accordance in 85% of cases. Conclusion: These results show good correlation of LVEF by TTE and CMR with the latest generation of scanners with a signal of systematic bias where CMR LVEF reads 4.4% higher than TTE. These results are used locally to guide clinical decision making and form part of on-going quality assurance processes. Further work to improve agreement between modalities is required. Conflict of Interest: None