Audit on nuchal translucency thickness measurements in Flanders, Belgium: a plea for methodological standardization

To audit nuchal translucency thickness (NT) measurements for fetal aneuploidy screening in Flanders, and to estimate the impact of small variations in NT measurement on the screening result of two first-trimester screening algorithms: maternal age + NT (Algorithm A), and maternal age + NT + pregnancy associated plasma protein-A + free beta-human chorionic gonadotropin (Algorithm B).

We used the database of first-trimester combined screening, as collected by the General Medical Laboratory AML in Antwerp, Belgium, between 1 January 2001 and 1 April 2004. Audit was performed by establishing a delta-NT distribution curve for one trainee of The Fetal Medicine Foundation (FMF) and for a group of 263 other sonographers, in comparison with the FMF reference values. Risks for fetal aneuploidy were calculated at a cut-off value of 1 : 300 for Algorithm A and 1 : 150 for Algorithm B. These risks were recalculated in both algorithms after a modeled increase of all NT values by 0.1 or 0.2 mm.

In a total of 592 measurements performed by the FMF trainee, the 5th, 50th and 95th percentiles of delta-NT measurements were at -0.41, +0.03 and +0.68 mm, respectively. These values were close to the FMF reference values. The screen-positive rate for this set of data was 4.4% (26/592) in both algorithms. For the 12 555 measurements of the 263 other sonographers, the 5th, 50th and 95th percentiles of delta-NT were at -0.81, -0.14 and +0.73 mm, respectively, which clearly indicates underestimation of NT in the lower range. In this set of data the screen-positive rate was 3.5% for both algorithms (439/12 555 for Algorithm A and 436/12 555 for Algorithm B). Also in this group, 5% (59/1186) of negative screening results at maternal age > or = 35 years in Algorithm A became positive after a modeled 0.1-mm increase in NT, whereas this was only in 1.2% (134/11 369) of tests at maternal age < 35 years (P < 0.0001). The overall increase of screen-positive rate in Algorithm A after an NT modification of +0.1 mm was 1.2% (152/12 555), significantly more than in Algorithm B (86/12 555; 0.7%) (P < 0.0001).

In Flanders, there is a systematic underestimation of NT in comparison with the FMF reference range. Attempts to change these measurements according to the FMF criteria are crucial. This will mainly influence the screening results of women at advanced maternal age and of NT-based algorithms without the use of other parameters.