Effect of measurement variability on Rossavik growth model specification and prediction of growth outcome at birth.

Overview

We have studied how variability in second-trimester ultrasound measurements affects the process of fetal growth evaluation based on individual fetal growth curve standards specified by Rossavik growth models. The head and abdominal short axes of two second-trimester scans of a patient with normal fetal growth were increased or decreased by increments ranging from 0.1 cm to values equal to the two standard deviations of the interobserver variability for these measurements--i.e., 0.3 cm for head short axis and 0.4 cm for abdominal short axis. The largest increments affected the Growth Potential Realization Index for weight at birth by 1% to 13% when applied to head short axis, and by 2% when applied to abdominal short axis. In contrast, the same increments had little effect (1%) on the Growth Potential Realization Indices for head and abdominal circumferences. Whereas 0.1-cm increments had no effect on any of the individual growth models or predicted birth characteristics, some combinations of these small errors involving both head and abdominal short axes changed the Growth Potential Realization Index for weight by as much as 15%, and those for head and abdominal circumferences by 3% and 8%, respectively. Under some scenarios, fetal growth status could be falsely classified as normal or abnormal. These results show that the reliability of individualized growth assessment depends to a great extent on excellent ultrasound technique, at least for certain parameters.