Risk assessment adjusted for gestational age in maternal serum screening for Down's syndrome.BMJ 1993; 306 doi: http://dx.doi.org/10.1136/bmj.306.6891.1509 (Published 05 June 1993) Cite this as: BMJ 1993;306:1509
OBJECTIVE--To investigate the relation between errors in calculation of gestational age and assessment of risk of Down's syndrome and to analyse the implications for screening programmes. DESIGN--Retrospective analysis of dating of gestational age by menstrual history v ultrasound scan. Computer program with maternal age and concentrations of alpha fetoprotein and free beta human chorionic gonadotrophin to calculate risk for a range of expected dates of delivery. Computer simulated prospective application of new screening programme. SETTING--Teaching hospitals in Nottingham. SUBJECTS--31,561 women with singleton pregnancies with gestational age based on routine ultrasound scan. Computer simulation of 20,000 women in three age ranges (up to 37; up to 40; all). MAIN OUTCOME MEASURES--Distribution of error between gestational age based on ultrasound scan v menstrual history. Proportion of women in the population who require precise dating of pregnancy; proportion of women who require amniocentesis. RESULTS--With gestational age derived from ultrasound scan as reference the 95% confidence interval for gestational age by menstrual history was -27 to +9 days. A screening programme for Down's syndrome for women up to age 40 would yield a low risk (< 1:250) for this range of days in 86.0% of cases. The 14.0% of women remaining would have one or more high risk values in their report and would thus require an ultrasound scan for precise dating of the pregnancy; 30% of these--that is, 3.7% of the screened population--would be identified as high risk and require consideration for amniocentesis. CONCLUSIONS--Screening programmes for Down's syndrome require the facility for precise dating of pregnancy to improve the accuracy of risk assessment. This can be achieved without introducing additional scans for early dating in the whole population but by selecting only those cases (about 14%) when an error in dates is likely to affect the risk of Down's syndrome.