Reevaluating humeral length for the detection of fetal trisomy 21.

OBJECTIVE The purpose of this study was to analyze humeral length (HL) in a normal population and to compare that with HL in a population of fetuses with trisomy 21 to determine the most efficient discriminating parameters for diagnostic accuracy. METHODS A nested case-control study comparing HLs from a normal population and a population of fetuses with trisomy 21 was conducted. Humeral length was regressed against gestational age for a consecutive well-dated population of normal singleton gestations presenting to the Washington University School of Medicine prenatal diagnosis units over a 5-year period. A second population of well-dated pregnancies with trisomy 21 diagnosed either prenatally or postnatally was also selected on the basis of the same criteria, except that anomalous fetuses were included. Various discriminating thresholds for a short HL were compared for efficiency in the detection of trisomy 21. These included the following: observed/expected HL (<or= 0.89), biparietal diameter/HL greater than 1.3, 1.4, 1.5, 1.6, 1.7, 1.75, 1.8, and 1.85 SD above the mean for gestation, HL less than 0.8 or less than 0.9 multiple of the median, and HL less than the fifth percentile for gestation. RESULTS A total of 620 normal pregnancies and 32 with trisomy 21 were extracted from the database. A receiver operating characteristic curve revealed HL less than the fifth percentile as the optimal discriminator for trisomy 21 detection (area under the receiver operating characteristic curve = 0.80). The positive likelihood ratio (LR+) was greatest (25.0) with HL less than the fifth percentile. When HL was considered in isolation without other sonographic markers of trisomy 21, the LR+ was 6.3. CONCLUSIONS Humeral length less than the fifth percentile was the most effective discriminator among the many studied.