Abstract
BACKGROUND
All available echocardiographic methods to assess single, systemic right ventricle (sRV) systolic function have limitations. Subjective grading is prone to bias and varies among readers. Quantitative methods that require significant manual input, such as fractional area change (FAC), are often not reproducible. The aim of this study is to determine whether global longitudinal strain (GLS) is more reproducible than FAC and subjective grading in sRV patients among individual readers and across different levels of experience.
METHODS
Clinically indicated echocardiograms for 40 patients with functional systemic right ventricles were assessed by 5 readers with varying reading experience: one sonographer, one cardiology fellow, and three attending cardiologists at different career stages. All readers were blinded to patient data and other reader responses. Each reader reviewed the same images for subjective grade (scale 1-8, normal to severely depressed), RV end-diastolic and end-systolic area measurements, and longitudinal strain analysis. A repeat analysis was performed under identical conditions after at least 2 weeks on all 40 patients. Inter- and intra-reader reproducibility was assessed with intraclass correlation coefficient (ICC). Correlations between responses were assessed with Spearman's correlation coefficient.
RESULTS
The subjective method had fair to good reproducibility (ICC 0.7, interquartile range (IQR) 0.60,0.72) while the FAC method was poor (ICC 0.46, IQR 0.39,0.51) between readers. Reproducibility for GLS was excellent (ICC 0.88, IQR 0.88,0.89). Intra-reader reproducibility was excellent by subjective grading (ICC 0.85, IQR 0.73,0.88), poor by FAC (ICC = 0.63, IQR 0.35,0.66) and excellent by GLS (ICC 0.93, IQR 0.88,0.96). Attending-level readers were more consistent with their subjective grading, while all readers were excellent with GLS.
CONCLUSION
GLS is more reproducible than conventional methods at assessing sRV systolic function between readers with different levels of experience. For most readers it was more consistent than their own subjective grade of RV function. Laboratories staffed by multiple readers are likely to be more consistent in grading systemic RV systolic function using GLS.
Collapse