Objective This study assesses two methods used to define relatively short f
emur in screening for trisomy 21 and examines changes in performance of scr
eening with gestational age.
Design Retrospective analysis of data on menstrual age, femur length (FL) a
nd biparietal diameter (BPD) in 49 trisomy 21 pregnancies and 6069 normal c
ontrols. Reference ranges were derived for BPD/FL versus menstrual age and
for FL versus BPD. Two methods of defining short femur (BPD/FL) and observe
d-to-expected FL ratio) were examined for false-positive rates and detectio
n rates for trisomy 21 at different gestational ages.
Results In the control group the BPD/FL ratio and its standard deviation de
creased with menstrual age. Trisomy 21 was associated with a significantly
higher BPD/FL ratio (P < 0.001) and the deviation increased significantly w
ith menstrual age (P < 0.05). Eleven percent of 28 fetuses examined at 15-1
7 weeks had a BPD/FL above the 95th centile compared with 24% of 21 fetuses
examined at 18-20 weeks (P = 0.40). The median observed-to-expected FL rat
io in the control group was 1.0 throughout the gestational age range but th
e standard deviation decreased significantly with menstrual age (P <0.01).
Trisomy 21 was associated with a significantly reduced observed-to-expected
FL ratio (P < 0.001) and the deviation increased significantly with menstr
ual age (P <less than> 0.05). A fixed cut-off of 0.91 for observed-to-expec
ted FL ratio provided a false-positive rate of 12% at 15-17 weeks compared
with 6% at 18-20 weeks of gestation (P < 0.001) with detection rates of 29
and 38%, respectively (P = 0.73).
Conclusion Irrespective of the definition used to define the condition, rel
atively short femur is a poor marker for trisomy 21 particularly when the a
ssessment takes place before 18 weeks of gestation.