Reproducibility of transvaginal Doppler velocimetry measurements in the uterine arteries of postmenopausal women.

Intraobserver and interobserver reproducibility of transvaginal Doppler velocimetry measurements in uterine arteries were assessed by two observers in 20 postmenopausal women. In addition, the agreement between the observers regarding the detection of an ovary as well as the presence or absence of intraovarian arterial blood flow was documented. Pulsatility index (PI), peak systolic velocity (PSV) and time-averaged maximum velocity (TAMV) measurements in uterine arteries were made twice by the same investigator (IJ). Thereafter, the same measurements were made by the second investigator (AT). The agreement in categorical data was studied by using 2 x 2 tables and Cohen's kappa-coefficient. Reproducibility of the Doppler measurements was analyzed by using coefficient of variation, repeatability coefficient, intraclass correlation coefficient, mean differences and limits of agreement. In addition, the 95% confidence interval was calculated as appropriate. There was agreement of 95-100% between the observers with respect to the detection of an ovary in transvaginal scanning. Agreement regarding the presence or absence of intraovarian arterial blood flow, however, was only 75-81%. Intraobserver repeatability was very good, the intraclass correlation coefficient being 0.98-0.99 and the coefficient of variation 6% in PI measurements. The PSV measurements showed an intraclass correlation coefficient of 0.94-0.96 and a coefficient of variation of 10-12%. The intraclass correlation coefficient for TAMV measurements was 0.94 and the coefficient of variation varied from 12 to 16%. Interobserver agreement was good. There was no bias between the observers' measurements. The intraclass correlation coefficient for PI measurements was between 0.93 and 0.95, and the coefficient of variation was 11%. For PSV measurements, the intraclass correlation coefficient was 0.79-0.80 and the coefficient of variation was 18-29%. TAMV measurements showed an intraclass correlation coefficient of 0.83-0.84 and a coefficient of variation of 19-29%. When the limits of agreement for Doppler velocimetry measurements are considered, we expect the two observers to give PI measurements that differ by less than 0.7, with any discrepancy being equally likely in either direction. In conclusion, based on the high intraclass correlation coefficients, the intraobserver repeatability of all Doppler parameters was very good, the most reliable measurement being the PI. Variation in PSV and TAMV measurements increased considerably, however, when Doppler velocimetry was carried out by two investigators instead of one.