Empirical validation of risk screening for Down's syndrome.

Abstract

To validate individual risk estimates in antenatal serum screening for Down's syndrome. Women screened for Down's syndrome using maternal serum alpha fetoprotein (AFP), unconjugated oestriol (uE3), and human chorionic gonadotrophin (hCG) with maternal age (the triple test) or AFP, uE3, free beta subunit and free alpha subunit of hCG with maternal age (the quadruple test) were grouped according to their predicted risk of having an affected pregnancy. The mean predicted risk in each category was then compared with the observed prevalence based on the number of affected and unaffected pregnancies in each category. About 100,000 pregnant women screened for Down's syndrome from 1989 to 1995. There was close agreement between the predicted term risk and the prevalence at birth for both the triple test and the quadruple test. For example, with the quadruple test the predicted risk in the highest risk group was 1 in 3.3 and the prevalence was 1 in 2.6; in the lowest risk group these were 1 in 3000 and 1 in 2300 respectively. Risk estimates based on multiple marker screening for Down's syndrome are accurate. The technique used to demonstrate this is simple and offers a useful empirical check on screening performance.