Median values and maternal weight specific to the ethnic group should be used

EDITOR,—Lucy Gilbert and colleagues report a 3.5-fold increase in the proportion of women found to be positive on screening for Down's syndrome when rates among Indian Asian women were compared with those among white women.1 This prompted me to examine the data from Oldchurch Hospital's screening programme. Unlike Gilbert and colleagues' study, this screening programme uses (alpha) fetoprotein and free β human chorionic gonadotrophin as biochemical markers, a combination shown to give a detection rate of 75% at a false positive rate of 5%.2 Screening with this test combination has been carried out in my department since 1991, and over the past four years about 50000 women have been screened. Women are assigned to one of four ethnic groups: white, Afro-Caribbean, Asian, and oriental/others/unknown. Table 1 shows the breakdown of the screened population by ethnic group and and the proportion found to be positive.

The findings in our much larger study population do not confirm Gilbert and colleagues' finding of an increased probability of a positive result of screening in Asian women. I agree that there is considerable evidence for ethnic differences in the median concentrations of the various analytes used, particularly (alpha) fetoprotein and intact human chorionic gonadotrophin in Afro-Caribbeans3 and free ß human chorionic gonadotrophin in oriental women.4 Provided due consideration is given to using medians specific to the ethnic group that have been derived from the test population and provided due regard is taken of the maternal weight in these groups (which may vary appreciably from that in a white population), then general screening principles can be applied. As has been shown in a large study of oriental women, provided these factors are taken into account detection rates of 84% at an amniocentesis rate of 50% are readily achievable—certainly when measurement of (alpha) fetoprotein and free β human chorionic gonadotrophin is used.