Femur length in the US prediction of trisomy 21 and other chromosomal abnormalities.

After review of 461 consecutive amniocentesis examinations, 428 fetuses with normal karyotype and 13 with chromosomal abnormalities were analyzed. Six of the abnormal fetuses had trisomy 21, and seven had other chromosomal abnormalities. The ratio of measured-to-expected femur length (MFL/EFL) was determined for each fetus. The MFL/EFL values were 1.0 +/- 0.7 (mean +/- standard deviation) for the control group, 0.94 +/- 0.06 for trisomy 21 fetuses, 0.94 +/- 0.10 for non-trisomy 21 fetuses, and 0.94 +/- 0.08 for the entire group of fetuses with chromosomal abnormalities. The mean MFL/EFL value for the abnormal fetuses differed significantly from that for the control group (P less than .003). An MFL/EFL of 0.90 or less resulted in a sensitivity of 46.2% and a specificity of 94.1% for detection of all chromosomal abnormalities. For detecting trisomy 21, the sensitivity and specificity were 50% and 93.5%, respectively. Assuming a pretest probability for trisomy 21 of one in 250, an MFL/EFL of 0.90 or less resulted in a positive predictive value of 3%. It is concluded that an MFL/EFL of less than 0.90 modifies the relative risk for chromosomal abnormalities and may significantly influence the decision for amniocentesis.