Inter-observer and Inter-vendor Variability in Strain Measurements in Patients with Single Right Ventricular Anatomy.

ABSTRACT

Myocardial strain offers new insights into ventricular performance, There are software packages from several different companies used to ascertain this, and little data is available in patients with single right ventricle (sRV) physiology. We aimed to compare the analysis of two strain software applications using a cohort of patients with sRV for both inter-vendor and inter-observer variability. Echocardiograms from 85 patients with sRV (122 separate studies) were prospectively evaluated. All had Glenn and/or Fontan palliation. Longitudinal 4-chamber (4LS), inflow/outflow (IO), circumferential, and radial strain were assessed using Velocity Vector Imaging (VVI, Seimens, Munich) and Automated Functional Imaging (AFI, General Electric, Boston) software. In a subset of 45 patients (61 separate studies), strain measurements were obtained by two sonographers so a paired "inter-observer" analysis could be performed. A moderate correlation between measurements made by the two systems was observed. Circumferential strain assessment had the highest R value (0.77) with all others having R values < 0.6. Both software packages showed modest inter-observer reproducibility for longitudinal and circumferential strain. VVI intraclass correlation coefficients (ICC) for 4LS and average circumferential strain (ACS) were 0.6 and 0.58, compared to 0.68 and 0.59 for AFI. Other than radial strain and VVI IO inferior strain, mean strain differences between AFI and VVI were ≤ 1%. Inter-observer variability is modest, however, mean differences are minimal suggesting reasonable clinical reliability. Inter-vendor variability is greater and not as clinically reliable. In patients with sRV, serial assessments with strain should be performed using the same software.