Title: New first trimester crown-rump length's equations optimized by structured data collection from a French general population

Abstract: --- Objectives --- Prior to foetal karyotyping, the likelihood of Down's syndrome is often determined combining maternal age, serum free beta-HCG, PAPP-A levels and embryonic measurements of crown-rump length and nuchal translucency for gestational ages between 11 and 13 weeks. It appeared important to get a precise knowledge of these scan parameters' normal values during the first trimester. This paper focused on crown-rump length. --- METHODS --- 402 pregnancies from in-vitro fertilization allowing a precise estimation of foetal ages (FA) were used to determine the best model that describes crown-rump length (CRL) as a function of FA. Scan measures by a single operator from 3846 spontaneous pregnancies representative of the general population from Northern France were used to build a mathematical model linking FA and CRL in a context as close as possible to normal scan screening used in Down's syndrome likelihood determination. We modeled both CRL as a function of FA and FA as a function of CRL. For this, we used a clear methodology and performed regressions with heteroskedastic corrections and robust regressions. The results were compared by cross-validation to retain the equations with the best predictive power. We also studied the errors between observed and predicted values. --- Results --- Data from 513 spontaneous pregnancies allowed to model CRL as a function of age of foetal age. The best model was a polynomial of degree 2. Datation with our equation that models spontaneous pregnancies from a general population was in quite agreement with objective datations obtained from 402 IVF pregnancies and thus support the validity of our model. The most precise measure of CRL was when the SD was minimal (1.83mm), for a CRL of 23.6 mm where our model predicted a 49.4 days of foetal age. Our study allowed to model the SD from 30 to 90 days of foetal age and offers the opportunity of using Zscores in the future to detect growth abnormalities. --- Conclusion --- With powerful statistical tools we report a good modeling of the first trimester embryonic growth in the general population allowing a better knowledge of the date of fertilization useful in the ultrasound screening of Down's syndrome. The optimal period to measure CRL and predict foetal age was 49.4 days (9 weeks of gestational age). Our results open the way to the detection of foetal growth abnormalities using CRL Zscores throughout the first trimester.