Real-time three-dimensional echocardiography for left atrial volume assessment in Thoroughbred racehorses: Observer variability and comparison with two-dimensional echocardiography.

Background: Left atrial size predicts cardiac morbidity and mortality in humans and dogs. Real-time three-dimensional echocardiography (3DE) may be reliable for assessing left atrial volume (LAV) in horses.

Objectives: To determine intra- and interobserver variability estimates of 3DE-LAV and compare it to that of 2DE-LAV estimates.

Study Design: Method comparison.

Methods: 3DE datasets were obtained from 40 horses, then graded for quality, creating a final study population of 22 horses. The 3DE and 2DE maximum LAV (LAV ) and minimum LAV (LAV ) were measured, and left atrial emptying volume (LA EV) and left atrial ejection fraction (LA EF) were calculated, from the same 3D dataset on four occasions using (a) a semi-automatic surface recognition algorithm and (b) a modified Simpson's method of discs. 3DE LAV measurements were repeated by a second observer.

Results: For 3DE, median LAV was 596cm for observer one, and 852 cm for observer two, LAV was 373 cm for observer one and 533 cm for observer two. Low intraobserver measurement variation was observed for LAV and LAV , with horse-level intraclass correlation coefficients (ICC ) for both observers between 76% and 85% (horse added as random effect). The interobserver ICC was 58% for LAV and 50% for LAV on averaged measurements (with observer added as random effect), indicating consistent differences between observers. While intraobserver variation was similar for 2DE LAV measurements, it was greater for LAV (ICC  = 67%). The intermethod ICC for 3DE vs 2DE was low at 14% for LAV and ~0% for LAV , indicating less-consistent differences with method.

Main Limitations: Small study population, low observer number, use of different imaging modalities (fundamental frequency and octave harmonics).

Conclusions: 3DE assessment of LAV was reliable, suggesting suitability for longitudinal evaluation of clinical cases. Clinicians should be aware of differences in LAV measurements between observers. More defined measurement guidelines may improve repeatability.