Longitudinal ultrasound data were collected for 18 structures in 37 pi
gtailed macaque (Macaca nemestrina) fetuses to 1) generate standards o
f normal fetal growth, 2) measure operator reliability, 3) assess the
accuracy of linear and nonlinear regression models to estimate gestati
onal age and dates of delivery, and 4) evaluate the portability of equ
ations and absolute values derived from data at one facility (Seattle)
to describe independently collected data on the same species at anoth
er facility (Medical Lake Breeding Colony). Femur length, biparietal d
iameter, and head area were found to be the best predictors of gestati
onal age, as judged by maximum ''explained'' variance (R(2)) With mini
mum error estimates. Days to delivery could not be adequately predicte
d by any single fetal growth parameter or by the best combination of p
arameters. Operator reliability was very good: error was less than or
equal to 5% over all parameters studied. Although the form of growth c
urves from each facility was generally the same, variability was much
greater for most parameters at Medical Lake, and both femur and humeru
s lengths were overestimated relative to the Seattle data. The same re
sult was obtained for femur length when Seattle data on pigtailed maca
ques were compared with published data on rhesus macaques. We believe
that these differences in facilities and populations may be due to the
practices of the ultrasound operators. We suggest that portability of
regression coefficients and absolute values for structures at a given
gestational age can be accomplished only if operators are trained to
use the same standard methods and receive periodic reliability checks.
(C) 1995 Wiley-Liss, Inc.