OBJECTIVE: To compare Down syndrome screening efficiency of the standard serum triple analyte screen to that of a four-component screen consisting of ultrasound biometry and serum markers in the second trimester. METHODS: The Down syndrome screening efficiency of the triple screen, i.e. alpha-fetoprotein (AFP), unconjugated estriol (E3), hCG and maternal age, was compared with the four-marker algorithm, i.e. humerus length, nuchal thickness, AFP and hCG plus maternal age. A quadrivariate Gaussian algorithm was used to calculate individual Down syndrome odds. Receiver operating characteristic (ROC) curves plotting sensitivity against false-positive rate were constructed for each algorithm and the areas under the curves were compared to determine which was superior. Sensitivity and false-positive rates at different Down syndrome risk thresholds were also compared. RESULTS: There were 46 cases of Down syndrome (1.9%) with 2391 normal singleton pregnancies in a referral population in which triple screen, fetal biometry and karyotype had been done. The gestational age range for the study was 14-24 completed weeks. The median maternal age for the study group was 35.0 years (14.0-46.0 years). The areas (SE) under the ROC curves were 0.75(0.04) and 0.93(0.02) for the standard triple and the four-marker screen, respectively (P < 0.001). At a 10% false-positive rate, detection was 45.7% for the triple and 80.4% for the four-marker screen. CONCLUSIONS: A new algorithm combining humerus length and nuchal thickness measurement with serum AFP, hCG and maternal age substantially improved Down syndrome screening efficiency compared with the traditional triple screen. The model appears promising and should be evaluated in an independent data set.
OBJECTIVE: To compare Down syndrome screening efficiency of the standard serum triple analyte screen to that of a four-component screen consisting of ultrasound biometry and serum markers in the second trimester. METHODS: The Down syndrome screening efficiency of the triple screen, i.e. alpha-fetoprotein (AFP), unconjugated estriol (E3), hCG and maternal age, was compared with the four-marker algorithm, i.e. humerus length, nuchal thickness, AFP and hCG plus maternal age. A quadrivariate Gaussian algorithm was used to calculate individual Down syndrome odds. Receiver operating characteristic (ROC) curves plotting sensitivity against false-positive rate were constructed for each algorithm and the areas under the curves were compared to determine which was superior. Sensitivity and false-positive rates at different Down syndrome risk thresholds were also compared. RESULTS: There were 46 cases of Down syndrome (1.9%) with 2391 normal singleton pregnancies in a referral population in which triple screen, fetal biometry and karyotype had been done. The gestational age range for the study was 14-24 completed weeks. The median maternal age for the study group was 35.0 years (14.0-46.0 years). The areas (SE) under the ROC curves were 0.75(0.04) and 0.93(0.02) for the standard triple and the four-marker screen, respectively (P < 0.001). At a 10% false-positive rate, detection was 45.7% for the triple and 80.4% for the four-marker screen. CONCLUSIONS: A new algorithm combining humerus length and nuchal thickness measurement with serum AFP, hCG and maternal age substantially improved Down syndrome screening efficiency compared with the traditional triple screen. The model appears promising and should be evaluated in an independent data set.