Objective To determine the sensitivity of antenatal screening methods for neural tube defects in population-based screening in South Australia in 1986-1991, and whether ultrasound can replace serum alpha-fetoprotein screening in terms of achieving an equivalent level of sensitivity. Design and setting Ascertainment of all births and terminations of pregnancy with neural tube defects from multiple sources for 1986-1991 in South Australia. Serum and amniotic fluid alpha-fetoprotein results were obtained from the only laboratory performing the tests as a Statewide antenatal screening programme, and information on ultrasound screening from case notifications, hospital case records and medical practitioners who cared for the women. Subjects All 243 births and terminations of pregnancy with neural tube defects in South Australia in 1986-1991. Main outcome measures The sensitivity of individual screening methods and of all methods used, particularly for spina bifida. Results For pregnancies with neural tube defects screened by any method (serum alpha-fetoprotein, ultrasound or amniocentesis), 86% sensitivity was achieved. Ultrasound screening for anencephaly achieved 100% sensitivity even in low risk pregnancies, compared with 92% for serum alpha-fetoprotein. For spina bifida, the sensitivity of ultrasound screening increased with the level of risk in pregnancy: it was 60% in low risk pregnancies, which was equivalent to that of serum alpha-fetoprotein screening (64%); 89% in high risk pregnancies and 100% for women referred for confirmation of a suspected spina bifida by another ultrasonographer (chi(1)(2) for trend = 23.49, P < 0.0001). Ultrasound screening in high risk pregnancies for spina bifida achieved higher sensitivity in teaching hospitals compared with other ultrasound services in the State (97% vs 65%), but sensitivity was equivalent for low risk pregnancies. It is estimated that, had the serum screening programme not been in place, the level of sensitivity achieved for spina bifida by ultrasound and amniocentesis would have been 62% compared with the actual situation of 76% with the programme in existence, a difference of nearly 15% (95% CI 2.5 to 26.7) (chi(1)(2) = 5.45, P = 0.02). Conclusions Antenatal screening for neural tube defects in South Australia achieved a higher lever of sensitivity with the maternal serum alpha-fetoprotein programme in place. We conclude that the serum screening programme should continue in South Australia pending a significant improvement in the sensitivity of routine ultrasound screening for spina bifida.
