OBJECTIVES: In February 2007 new software, Prenatal Risk Calculation (PRC), for calculating the risk of fetal aneuploidy was introduced in Germany. Our aim was to investigate its test performance and compare it with that of the PIA Fetal Database (PIA) software developed and used by The Fetal Medicine Foundation. METHODS: Between 31 August 1999 and 30 June 2004 at the Women's Hospital of the Medical University of Hanover in Germany, 3120 singleton pregnancies underwent combined first-trimester screening at 11 + 0 to 13 + 6 weeks of gestation. Calculation of risk for fetal aneuploidy was computed prospectively using the PIA software. In a subsequent retrospective analysis, we recalculated risks for the 2653 of these datasets with known fetal outcome using the PRC software and compared the results. RESULTS: Of the 2653 datasets analyzed, 17 were cases of aneuploidy. At a cut-off of 1 : 230, for the detection of fetal aneuploidy, the respective sensitivity, false-positive rate and positive predictive value were 70.6%, 4.1% and 9.9% for PRC and 76.5%, 2.9% and 14.6% for PIA. At a cut-off of 1 : 300, the equivalent values were 70.6%, 5.6% and 7.5% for PRC and 76.5%, 4.0% and 11.0% for PIA. The differences in test performance between the two types of software were highly significant (P < 0.0001). DISCUSSION: The test performance of PRC was inferior to that of PIA, the sensitivity for detection of fetal aneuploidy being lower and the false-positive rate higher. Had PRC been employed prospectively in our study, 40% more women examined would have been offered unnecessarily an invasive procedure for fetal karyotyping.
OBJECTIVES: In February 2007 new software, Prenatal Risk Calculation (PRC), for calculating the risk of fetal aneuploidy was introduced in Germany. Our aim was to investigate its test performance and compare it with that of the PIA Fetal Database (PIA) software developed and used by The Fetal Medicine Foundation. METHODS: Between 31 August 1999 and 30 June 2004 at the Women's Hospital of the Medical University of Hanover in Germany, 3120 singleton pregnancies underwent combined first-trimester screening at 11 + 0 to 13 + 6 weeks of gestation. Calculation of risk for fetal aneuploidy was computed prospectively using the PIA software. In a subsequent retrospective analysis, we recalculated risks for the 2653 of these datasets with known fetal outcome using the PRC software and compared the results. RESULTS: Of the 2653 datasets analyzed, 17 were cases of aneuploidy. At a cut-off of 1 : 230, for the detection of fetal aneuploidy, the respective sensitivity, false-positive rate and positive predictive value were 70.6%, 4.1% and 9.9% for PRC and 76.5%, 2.9% and 14.6% for PIA. At a cut-off of 1 : 300, the equivalent values were 70.6%, 5.6% and 7.5% for PRC and 76.5%, 4.0% and 11.0% for PIA. The differences in test performance between the two types of software were highly significant (P < 0.0001). DISCUSSION: The test performance of PRC was inferior to that of PIA, the sensitivity for detection of fetal aneuploidy being lower and the false-positive rate higher. Had PRC been employed prospectively in our study, 40% more women examined would have been offered unnecessarily an invasive procedure for fetal karyotyping.
Authors: S Kate Alldred; Yemisi Takwoingi; Boliang Guo; Mary Pennant; Jonathan J Deeks; James P Neilson; Zarko Alfirevic Journal: Cochrane Database Syst Rev Date: 2017-03-15