JOURfoetal infection, will not by itself prevent completely the occurrence of congenital syphilis. Other factors which will lead to the achievement of this goal will be discussed in a further communication. It will be noted that the Liverpool controls are slightly lower in blood group A than those of the rest of S.W. Lancashire, though the difference is not significant. The numbers in the control series are fairly equal, whereas many more patients are drawn from the Liverpool postal district. Hence, addition of the two control series, which seems desirable in the interests of simplicity and numbers, has the effect of slightly reducing the difference between diabetics and controls which is described in this paper.
ResultsThe basic data, with subdivisions by area, sex, and age at onset of diabetes mellitus, are shown in Table II. If men and women of all ages are compared with the controls of Table I, the findings may be summarized as shown in Table III. The control figures are weighted according to the numbers of diabetic men and women drawn from each area. The one striking difference is that compared with the controls the men show a higher proportion of patients of group A and a lower proportion of patients of group 0. On the other hand, there is no such difference in the women.The relative differences in groups 0 and A, for which considerable numbers are available, were examined first. The method recommended by Aird et al. (1954) for examining the differences in blood-group proportions and the homogeneity of the areas was used, and the comparison is shown in Table IV. The relative increase in blood group A in men is shown to be fairly highly significant. A direct comparison of the men and women patients also yields a significant difference. The women are closely similar to the controls. The three areas are perfectly homogeneous.The men patients are lower in blood group B than are the controls, but the numbers are small and the difference is not significant. The women patients show a rather high proportion of group AB. This has been examined in two ways. First, a direct comparison has been made with the controls in regard to the percentage of group AB. With the method of Aird et al. (1954), x2 for the difference is 3.25 and so does not attain the 5% level of significance. x2 for the heterogeneity of the three areas is 2.88, which is not significant, but does to some extent reflect the fact that the excess of those of group AB is confined to two of the areas. Secondly, if the figures are added together, irrespective of area, Fisher's test for unreasonableness of group proportions (Dobson and Ikin, 1946) gives an expected number of those of group AB of 25.804 against the 38 observed, x2 for one degree of freedom being 3.99, which just exceeds the 5% level of significance. It may be concluded that, while the women are rather high