Objective: To describe a method for deciding whether an individual's first-trimester Down syndrome screening test result justifies further testing in the second trimester.
Methods: Statistical modelling was used to estimate the distribution of second-trimester marker profiles for a given first-trimester profile and hence the probability of a final positive result, using a 1 in 250 term cut-off. A multi-variate log Gaussian model was used with published parameters. Markers were maternal serum pregnancy-associated plasma protein-A and free beta-human chorionic gonadotrophin (hCG) at 10 weeks, nuchal translucency at 11 weeks, and second-trimester maternal serum alpha-fetoprotein, total hCG, unconjugated estriol and inhibin-A. To illustrate the method, the model was applied to a published series of 24 Down syndrome and 367 unaffected pregnancies.
Results: Modelling predicts that for 63% Down syndrome and 0.4% unaffected pregnancies having first-trimester tests, there is a 50% or more probability of a final positive result. A step-wise sequential screening policy based on immediate prenatal diagnosis for those with high probability and second-trimester testing for the remainder would have a 90% detection rate and 1.7% false-positive rate. Modelling also predicts 8.0% Down syndrome and 89% unaffected pregnancies with probabilities below 3%. A contingent screening policy restricting second-trimester testing to those with 3-49% probabilities would have an 88% detection rate and 1.4% false-positive rate.
Conclusion: Predicting the probability of a positive final result from the first-trimester marker profile has potential utility, either as a decision aide for individual women or as a formal part of screening policy in selecting a subset of women for second-trimester testing.
Copyright 2005 John Wiley & Sons, Ltd