We investigated the measurement repeatability of four pulmonary venous
flow indices. The indices were measured on 45 anonymised, transthorac
ic Doppler recordings of adequate technical quality. Measurements were
taken by two independent observers, and repeated after 10 days. Plus/
minus the repeatability coefficient, which was used to quantify repeat
ability, gives the 95% probability limits for random variation between
repeated measurements. The index D-diff, which is the difference in d
uration of the pulmonary venous flow reversal during atrial systole an
d the transmitral A-wave, had repeatability coefficients of 50 and 57
ms intra- and inter- observer. For the fraction of antegrade pulmonary
venous now during ventricular systole, the coefficients were 12 and 1
3 percentage points, but improved to 6 and 7 among the high-quality re
cordings. The retrograde pulmonary venous flow during atrial systole a
s a fraction of the antegrade flow, had coefficients of 5 percentage p
oints both intra- and inter-observer. The coefficient for the peak vel
ocity of retrograde pulmonary venous flow was 0.05 m/s intra- and inte
r-observer. Thus, the systolic fraction was the only index that showed
a satisfactory repeatability. We suggest that if the other indices ar
e used, measurements should be taken by a blinded observer to avoid ob
server bias.