Testing the assumptions of customized intrauterine growth charts using national birth studies

Customized intrauterine growth charts are widely used for growth monitoring and research. They are based on three assumptions: (1) estimated fetal weight (EFW) has a normal distribution with a constant coefficient of variation at all gestational ages; (2) Hadlock's growth curve accurately describes the relation between EFW and gestational ages; (3) associations between EFW and the fetal and maternal characteristics included in the customization model (fetal sex, pre-pregnancy weight, height, parity) are proportional throughout pregnancy. The aim of this study was to test whether these underlying assumptions are verified.

Data came from (1) the French Longitudinal Study of Children (ELFE) cohort, which recruited births after 32 weeks' gestation in 349 maternity hospitals in France in 2011, and (2) the National Perinatal Survey, which included births from all French maternity hospitals in 2016. The study population included, respectively, 6 920 and 8 969 singleton non-malformed term live births with data on customization characteristics and EFW. We computed the coefficient of variation by gestational age and then modeled the association of gestational age, maternal and fetal characteristics with EFW at the second and third trimester ultrasound and with birthweight using linear regression. To assess the proportionality of the impact of maternal and fetal characteristics, we computed the percent change in weight associated with these characteristics at these three time points.

The coefficient of variation was close to 12% at each gestational age, but EFW was not normally distributed, leading to small but systematic underestimation of fetuses under the 10th percentile. Weights representing the 50th and 10th percentiles based on Hadlock's growth trajectory were lower than observed or predicted weights. Most characteristics more strongly impacted weight at birth than during pregnancy. In the French Longitudinal study of Children (ELFE) cohort, boys were 1.8% (95% confidence interval [CI] 1.3-2.4) heavier than girls in the third trimester, whereas this percentage was 4.6% (95% CI 4.0-5.2) at birth. In the National Perinatal Survey, these percentages were 2.3% (95% CI 1.8-2.8) and 4.3% (95% CI 3.8-4.8).

These results from two independent sources revealed discrepancies between routine clinical EFW data used for growth monitoring and the customized growth model's assumptions.