Reproducibility of the amniotic fluid index: its effect on clinical practice.

Abstract

Two methods of measuring the amniotic fluid index (AFI) were subjected to tests of intra- and interobserver reproducibility. In the first method, amniotic fluid in each quadrant was measured by using the deepest pool perpendicular to the floor; in the second method, amniotic fluid in each quadrant was measured by using the deepest pool perpendicular to the uterine contour. Intraobserver and interobserver variability were assessed by using one-way analysis of variance and limits of agreement, respectively. Intraobserver standard deviations were all < 9 mm. When the AFI was measured using the first method, there was an inverse relationship between the interobserver difference and mean values of AFI. Plots of the differences between observers against their means showed that the lower the AFI, the greater the interobserver variability. No such correlation was noted when the AFI was measured by the second method. The limits of agreement (within which 95% of interobserver differences lie) were comparable for the two methods [(-42 mm to 29.4 mm) and (-46.4 mm to 41.6 mm), respectively]. The results suggest that the amniotic fluid index measured by the second method is more reproducible, especially in the presence of reduced amniotic fluid.