Likelihood ratio-based quality control for nuchal translucency measurements at 11-14 weeks of gestation.

ABSTRACT

OBJECTIVE: To develop a quantitative quality control process of nuchal translucency (NT) measurement at 11-14 weeks of gestation based on the mixture model, characterized by analysis of the corresponding distribution of the likelihood ratio (LR). METHODS: Based on the published mixture model for NT measurement in the first trimester, we simulated the expected distribution of NT and the corresponding LR. This approach was then tested for the quality control of 15 048 NT measurements performed by four operators trained and certified by The Fetal Medicine Foundation. The new quality control approach based on LR was compared with existing processes based on NT plots, multiples of the median (MoMs) and analysis of the 95(th) centile. RESULTS: Each operator contributed 2176-4730 examinations. Median, 5(th) and 95(th) centile of NT values ranged from 1.4 to 1.6, 1.0 to 1.0 and 2.2 to 2.6 mm, respectively. Median of NT-MoM values ranged from 0.83 to 0.95. Analysis of the distribution of NT measurements confirmed departure from the assumptions of the delta-NT and NT-MoM models. Analysis of LR distributions demonstrated a significant difference between observed and expected distributions for all operators (P < 10(-4) ). CONCLUSION: An LR-based quality control process is feasible at 11-14 weeks of gestation. Because it is more sensitive to measurement bias around the critical area of the 95(th) centile, its use should be encouraged.