Abstract
Background
Stroke volume (SV) using Doppler echocardiography is a significant prognosticator in patients with atrial fibrillation (AF). However, tracing multiple left ventricular outflow tract (LVOT) Doppler flow velocity envelope is tedious. Although ASE recommends to average 5 -10 beats of SV in AF, this has not been validated. Recently developed fully automated Doppler flow velocity analytical software can analyze Doppler parameters in multiple consecutive beats on the screen within a few seconds.
Purpose
We aimed to determine the usefulness of the software, and to validate the minimum number of beats required to approximate SV in patients with AF.
Methods
We selected 21 AF patients who had undergone 2D echocardiography using GE ultrasound machine (E95, GE healthcare). LVOT area was calculated by 3.14×(LVOT diameter/2)². LVOT pulse-wave Doppler velocity was recorded from an apical approach with a quiet breathing. To maximize the number of flow envelope on the screen, sweep speed was set at 12.5 mm/s. LVOT velocity time integral was measured by both manual tracing method and fully automated method (Cardiac Auto Doppler, GE healthcare). The grand truth of mean SV (reference SV) in each patient was defined as the averaged values of SV from all consecutive beats. We also calculated the mean SV value with the successive addition of sequential beats started from the 1st beat. Each value was compared with the reference SV and % variability was calculated. We determined the minimum number of beats showing %variability becoming <5%.
Results
Mean age was 77 years. Mean heart rate and reference SV index (SVI) were 80 ± 12 bpm and 35 ± 10 mL/m2. A total number of beats for recording was ranged from 16 to 25 in each patient. The fully automated software could analyze Doppler envelope in 395 out of 412 beats (Feasibility: 96%). Although there was a good correlation of SV in individual beats between the manual and automatic method (r = 0.92), the automatic method significantly overestimated SV (mean bias: 2.6 mL, p < 0.001) compared with the manual method. The median values of minimum number of beats showing % variability < 5% were 4 (interquartile range: 2 -7) for manual tracing method. The corresponding values were 5 (2 -6) for automatic method. If we used mean values of SV during consecutive 10 beats, 92% of patients using manual method and 96% of patients using automatic method showed % variability < 5%. There were excellent correlation between reference SV and averaged SV from the 1st beat to 10th beat (manual: r = 0.98, automatic: r = 0.99). If we defined low flow status as < 35 mL/m2, averaged SVI during consecutive 10 beats using the automatic method had a correct diagnosis in 20 out of 21 patients.
Conclusions
We concluded that minimum number of required beats for averaging was 10 in most AF patients. Rapid and reliable SV analysis with a novel fully automated Doppler software has a potential for its adoption in busy echocardiography laboratories.
P768 Potential utility of fully automated Doppler flow velocity tracing software in patients with atrial fibrillation