Taking account of vaginal bleeding in screening for Down's syndrome

OBJECTIVE: To derive a method for revising the risk of Down's syndrome in maternal serum marker screening when there is vaginal bleeding. The effect on screening performance of routinely allowing for the presence or absence of bleeding in all women is also assessed. DESIGN: Overview of published studies on the rate of reported vaginal bleeding in pregnancies with Down's syndrome, on the rate according to maternal age and on the association of bleeding with alpha-fetoprotein (AFP) level. The publications are supplemented with data on unconjugated oestriol (uE3), human chorionic gonadotrophin (hCG) and AFP levels in a consecutive series of screened women. SETTING: Routine Down's syndrome screening tests carried out on women having antenatal care at the St James's University Hospital, Leeds. SUBJECTS: Eight hundred and nine screened women. RESULTS: In five studies the rate of vaginal bleeding in Down's syndrome pregnancies was 1.7 times that in unaffected pregnancies on average. In three studies, the vaginal bleeding rate increased proportionally by 2.2% on average for each year of maternal age. Three studies and our own data were consistent with a 10% increase in the mean AFP level associated with vaginal bleeding, but it did not appear to materially alter uE3 and hCG levels or the standard deviations and correlation coefficients for any of the three analytes. An individual woman's risk was calculated by multiplying her age-specific odds of Down's syndrome by two likelihood ratios, one relating to the vaginal bleeding itself and one from the marker levels. Routine allowance for the presence or absence of vaginal bleeding was estimated to increase the detection rate by less than 1%. CONCLUSION: Our method is of clinical value in revising the risk when there is concern that vaginal bleeding might be responsible for a negative maternal serum Down's syndrome screening result. A policy of routinely incorporating information on vaginal bleeding in risk estimation for all women would have too small an effect on overall screening performance to recommend it.