Strain derived from speckle-tracking echocardiography is emerging as a useful tool in the assessment of single ventricle function. The purpose of this study is to compare layer-specific longitudinal strain values in children with single, systemic right ventricles (sRV) using two commercially available software platforms (GE EchoPAC (EP) and TomTec (TT)). Two readers analyzed two-dimensional longitudinal strain on EP (v 202) and TT (v 2.21.25) in 40 pediatric sRV patients. Intravendor reproducibility and intervendor agreement between layer-specific measurements were assessed by intraclass correlation coefficient and Bland-Altman analysis. Absolute difference (AbΔ) and relative mean errors (RME) were calculated. Subgroup comparisons (stratified by age, heart rate (HR), and frames per second (FPS): HR ratio) were made. Median age was 4.4years. 32 (80%) patients had hypoplastic left heart syndrome; 19 (48%) were post-Fontan. Intravendor reproducibility was excellent with high ICC (0.86-0.97). AbΔ between readers was small (1.2%-1.5%) with interobserver RME slightly higher for TT (11%-12% vs 8%-9% for EP). Layer-specific intervendor agreement was poor (ICC 0.45-0.62). Default layer comparisons (EP mid vs TT endo) showed good agreement (ICC 0.72-0.77) and less variability (AbΔ 2%, RME 15%) than layer-to-layer. There were no differences in ICC for groups dichotomized by age, HR, or FPS:HR ratio. sRV strain values are more negative when using EP. Intravendor reproducibility for sRV peak longitudinal strain in children is excellent with acceptable variability between experienced users. Intervendor, layer-specific strain agreement is poor. Vendor default layer strain values show better agreement but are not interchangeable.
Read full abstract