Variability among strain variables derived from two-dimensional speckle tracking echocardiography in dogs by use of various software

OBJECTIVE: To determine variability of global longitudinal strain (GLS) and strain rate (SR) measurements in dogs with and without cardiac disease derived from 2-D speckle tracking echocardiography (STE) by use of various software. ANIMALS: 2 cohorts comprising 44 dogs (23 cardiovascularly healthy and 21 with cardiac disease) and 40 dogs (18 cardiovascularly healthy and 22 with cardiac disease). PROCEDURES: Transthoracic echocardiographic images in each cohort were analyzed with vendor-independent software and vendor-specific 2-D STE software for each of 2 vendors. Values for GLS and SR obtained from the same left parasternal apical views with various software were compared. Intraobserver and interobserver variability was determined, and agreement among results for the various software was assessed. RESULTS: Strain analysis was not feasible with vendor-independent software for 20% of images obtained with the ultrasonography system of vendor 1. Intraobserver and interobserver coefficient of variation was < 10% for GLS values, whereas SR measurements had higher variance. There was a significant difference in GLS and SR obtained for each cohort with different software. Evaluation of Bland-Altman plots revealed wide limits of agreement, with variance for GLS of up to 6.3 units in a single dog. CONCLUSIONS AND CLINICAL RELEVANCE: Results of longitudinal strain analysis were not uniform among software, and GLS was the most reproducible measurement. Significant variability in results among software warrants caution when referring to reference ranges or comparing serial measurements in the same patient because changes of < 6.5% in GLS might be within measurement error for different postprocessing software.