A direct method for ultrasound prediction of day of delivery: a new, population-based approach


  • H. K. Gjessing,

    Corresponding author
    1. Division of Epidemiology, Norwegian Institute of Public Health, University of Oslo, Oslo, Norway
    2. Section for Epidemiology and Medical Statistics, Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
    • Division of Epidemiology, Norwegian Institute of Public Health, PO box 4404 Nydalen, N-0403 Oslo, Norway
    Search for more papers by this author
  • P. Grøttum,

    1. Section of Medical Informatics, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • S. H. Eik-Nes

    1. National Center for Fetal Medicine, Women and Children's Center, St Olavs University Hospital, Norwegian University of Science and Technology, Trondheim, Norway
    Search for more papers by this author



To introduce a direct population-based method for prediction of term based on ultrasound measurements of the biparietal diameter and femur length in the second trimester of pregnancy.


Our data consisted of 41 343 ultrasound scans from a non-selected population, prospectively collected during the years 1987–2004. Using measurements of biparietal diameter and femur length, we constructed prediction curves for term by computing median remaining time of pregnancy from the ultrasound measurement to birth. A local linear quantile regression method was used to smooth the median and quantile curves.


The quality of term prediction was stable over the prediction range for both biparietal diameter (25–60 mm) and femur length (11–42 mm). The femur-based predictions were nearly as good as those of the biparietal diameter. For the biparietal diameter, the median of the prediction residual was − 0.09 days; 87.2% of the births fell within ± 14 days of the predicted day of delivery, 3.5% births were classified as preterm and 4.3% as post-term. The corresponding figures for femur length were − 0.04 days, 86.7%, 3.6% and 4.5%. The covariates maternal age, parity, mother's smoking habits, sex of the fetus and examination year generally affected the predicted term by less than 1 day.


This direct ultrasound-based prediction of term using population-based data avoids selection biases possibly present in smaller prospective samples. The model obviates the dependence on last menstrual period found in standard methods for term prediction, and allows an immediate assessment of prediction quality in a population setting. The femur-based predictions had a quality similar to those based on the biparietal diameter. The model can be updated continuously as new data are collected. Copyright © 2007 ISUOG. Published by John Wiley & Sons, Ltd.