Mathematical Models to Predict Long Bone Lengths by Transvaginal Scan at 11–16 Weeks’ Gestation

Abstract

Objective: To propose a new mathematical model to estimate the length of fetal long bones in early pregnancy that can be used in the routine clinical setting. Methods: In a group of 400 singleton normal fetuses, referred for transvaginal ultrasound examination between 11 and 16 weeks’ gestation prior to genetic amniocentesis, a regression analysis was performed to evaluate the relationship between femur length (FLl) (mm) and biparietal diameter (BPD) (mm) and gestational age (GA) (days), as well as between humerus length (HL) (mm) and BPD (mm) and GA (days). The confidence intervals (CIs) of the predicted values for different values of BPD and for different gestational periods and CIs for the regression coefficients are stated as the mean ± SD of standardized residuals. The accuracy of our best models obtained were evaluated at each gestational week between 11 and 16 with a 10% error cut-off limit. Results: The best relationships between FL and HL versus BPD and GA are: expected FL = –16.92108 + 0.4569402 BPD +0.171617 GA (R² = 0.86) and expected HL = –16.28531 + 0.4283019 BPD + 0.1696017 GA (R² = 0.88), respectively. When a cut-off limit of 10% in estimating fetal long bones was utilized, the mathematical models revealed a good accuracy particularly at 13–14 weeks’ gestation, a period when transvaginal biometric and morphologic examination is advisable and the highest percentage of scans are performed. Discussion: Our proposed two linear models had a very good ability to estimate FL and HL and, due to the simplicity of the calculations, would be particularly useful in the routine clinical setting.