As an indicator of fetal limb growth, change in the femur diaphysis length (FDL) between 15 and 38 weeks, menstrual age, has been evaluated in a longitudinal study of 20 normal fetuses, as determined by prenatal biparietal diameter, head circumference, and abdominal circumference growth patterns as well as postnatal pediatric assessment. Individual FDL growth curves were uniformly parabolic and followed very similar trajectories. The projected start points [mean: 9.7 (+/- 1.5 SD) weeks] for these curves were in complete agreement with embryologic data, and the curves themselves were well characterized by the Rossavik model [R2 99.5 (+/- 0.5 SD)%]. No differences between males and females were detected, and good agreement was obtained between the average longitudinal growth curve and the cross-sectional growth curve derived from a previously studied data set. Individual growth curve standards, determined from Rossavik models based on data obtained before 26.1 weeks, menstrual age, were within 7% of actual measurements in 94% of the 97 time points studied after 26 weeks. These results indicate that the Rossavik model and its associated individual growth curve standards can be used to evaluate the growth of the femur and thus provide an improved means for detecting skeletal dysplasias and fetal growth retardation.

To characterize the growth of the thigh circumference (ThC) in individual fetuses, longitudinal studies of ThC growth were carried out in 20 fetuses between 19 (+/- 1.8) and 38 (+/- 1.5) weeks, menstrual age. Because of measurement uncertainties, analysis of growth patterns was limited to the data collected after 22 weeks. The Rossavik growth model fit the data well [R2: 95.8(+/- 2.9 SD)%], but considerable coefficient variability was seen, particularly for the coefficient k. Use of the coefficient k-value (1.138) derived from a cross-sectional data set reduced the variability of the coefficient c by 95% and made the results of the longitudinal and cross-sectional studies consistent with each other. This k-value was taken as the appropriate one for ThC growth curves. Studies of ThC growth in individual fetuses revealed considerable variability in growth curve shape, although the majority of curves could be approximated by a straight line. No differences between growth curves for males and females could be detected. The average longitudinal ThC growth curve was found to be very similar to the growth curve obtained in a cross-sectional study of ThC growth. ThC growth after 26.1 weeks could be predicted with an accuracy of +/- 15% from growth models derived from the data obtained before 26.1 weeks. These results indicate that ThC growth after 22 weeks (as with other parameters) can be followed in individual fetuses. However, because of the greater inherent variability in growth patterns, using each fetus as its own control may be more important.

To explore the use of three-dimensional parameters in characterizing fetal growth with ultrasound, the accuracy of volume measurements and the ability to make such measurements during pregnancy have been investigated. Total fetal volumes (TOTV) determined with ultrasound in utero were compared to volume measurements obtained by hydrostatic weighing following induced abortion at 19.4 (+/- 1.8 SD) weeks, menstrual age (MA), and to those calculated from birth weights and density values for term fetuses [39.3 (+/- 0.8 SD) weeks, MA]. This study indicated that ultrasound measurements underestimate the fetal volume [-9.8 (+/- 9.5 SD)% at 19 weeks and -19.0 (+/- 9.3 SD)% at 39 weeks]. To assess the possibility of obtaining growth curves for volume parameters, 20 normal fetuses [based on crown-rump length (CRL), biparietal diameter (BPD), head circumference (HC) and abdominal circumference (AC) growth curves, birth weight (WT), crown-heel length (CHL) and postnatal examination] with known dates of conception were studied. Head volume (HV), thoracic volume (TV), abdominal volume (AV), limb volume (LV) and TOTV were measured at 2- to 3-week intervals during the third trimester. Mathematical modeling indicated that the linear-cubic (LC) model was optimal for HV, TV, AV, and TOTV (R2: 97.3%, 97.2%, 98.1% and 99.4%) while the linear-quadratic (L-Q) model was optimal for LV (R2: 98.4%). Significant individual variation was suggested by the high coefficients of variation (CV) obtained for the optimal model coefficients. Inspection of individual growth curves revealed the presence of three types, "linear," "concave," and "convex" with the "concave" type predominating. The variability was less than that seen in the groups as a whole for most subgroups but was still greater than that observed with linear parameters. These results indicated that although volume growth curves can be obtained, their individual variability is significant and thus more fetuses must be studied before standard curves can be defined.