Critical appraisal of the use of nuchal fold thickness measurements for the prediction of Down syndrome

OBJECTIVE: Nuchal fold thickness is the best ultrasonographic predictor of fetal trisomy 21. However, the risk assigned on the basis of the commonly u sed threshold of nuchal fold thickness greater than or equal to 6 mm does n ot take into consideration the significant associations between nuchal fold thickness and gestational age and between maternal age and Down syndrome. We propose a new method of calculating Down syndrome probability that takes into account both gestational age at examination and previously assessed p robability of Down syndrome. STUDY DESIGN: Nuchal fold thickness was measured at ultrasonographic examin ation at 14 to 22 weeks' gestation without previous knowledge of the fetal karyotype. Nuchal cystic hygromas were excluded from analysis. Statistical analyses included correlation, logistic regression to control for other ult rasonographic predictors of trisomy 21 and for maternal age, receiver opera ting characteristic curve, and likelihood ratios (defined as the ratio of t he sensitivity to the false-positive rate). P <.05 was considered significa nt. RESULTS: Mean gestational age at ultrasonography was 16.9 weeks' gestation (range, 14-22 weeks' gestation). Mean (+/-SD) nuchal fold thickness in fetu ses with trisomy 21 (4.7 +/- 1.6 mm; n = 29) was greater than in euploid fe tuses (3.2 +/- 0.9; n = 780; P<.001). Logistic regression analysis establis hed that nuchal fold thickness was a significant predictor of trisomy 21 in dependent both of the other ultrasonographic markers and of maternal age (P <.001). Regression analysis showed that nuchal fold thickness was significa ntly correlated with gestational age among both fetuses with trisomy 21 and euploid fetuses and that the regression line of fetuses with trisomy 21 ha d a slope similar to that of euploid fetuses. The difference between observ ed and expected nuchal fold thicknesses on the basis of the biparietal diam eter (as a function of gestational age) was used to obviate the confounding effect of gestational age. Differences between observed and expected nucha l fold thicknesses were then used to calculate likelihood ratios. These lik elihood ratios could then be multiplied by the individual prior probability to obtain a patient-specific Down syndrome probability. CONCLUSION: Nuchal fold thickness is correlated with gestational age in bot h euploid fetuses and fetuses with Down syndrome. Use of the difference bet ween observed and expected nuchal fold thicknesses to determine likelihood ratios allows the calculation of individual posterior probabilities of Down syndrome that take into consideration both gestational age and maternal ag e.