A comparison of three methods of assessing inter-observer variation applied to measurement of the symphysis-fundal height.

This study assesses the inter-observer variation of symphysis-fundal height measurements by three methods--the coefficient of variation, the correlation coefficient and the limits of agreement. The coefficient of variation was 4% and the correlation coefficient 0.959, yet the limits of agreement were very wide. Applying these limits to centile charts of symphysis-fundal height shows that the fundal height cannot be measured by different observers with sufficient agreement to separate small fundal heights from those which are not small, and this severely limits the usefulness of measurement of the symphysis-fundal height as a screening test for intrauterine growth retardation. We conclude that inter-observer variation should be assessed by the method of limits of agreement, and not by calculating the coefficient of variation or the correlation coefficient.