Dear Sir,
Thalib and Hall 1 presented data on 32,003 Swedish women diagnosed with breast cancer in 1958-97. They confined their attention to those women among this sample who had had exactly one child each. These authors failed to find an association between (i) the sex of that child and (ii) the progress of the disease and the survival rates of patients.
However they published interesting data on a related issue: the sex ratio (proportion male) of these children by mother's age at diagnosis. The sex ratio of the 2,134 offspring of the women diagnosed at or before the age of 45 was 0.537: the sex ratio of the 3,095 offspring of the women diagnosed after the age of 45 was 0.512. The value of 0.537 is significantly higher than an expected sex ratio of around 0.513 in the relevant years in Sweden (v 2 5 4.93, p < 0.03). The difference between the 2 sex ratios (0.537 and 0.512) is not formally significant (v 2 5 2.96, p < 0.1) but is suggestive.
So it seemed worth reviewing the literature to see whether any other data of this sort had been previously published. Using Medline (1950 onwards) I was able to locate only one set of data which gave, simultaneously, the sexes of offspring by the mother's age at diagnosis with breast cancer. These were the data of Olsson and Brandt 2 who dichotomised maternal age at diagnosis at age 55 (rather than 45). In these data, the sex ratio of (all) the 153 children of mothers diagnosed before age 55 was 0.59; and the sex ratio of (all) the 141 children born to women diagnosed after this age was 0.34.
Comments on the data of Olsson and Brandt
Before embarking on further analysis, it is worth emphasising the extraordinary nature of these data of Olsson and Brandt. 2 Their 2 3 2 partition of frequencies is reproduced here in Table I. The resultant v 2 value is 18.0, p 5 0.00002. Yet according to the Science Citation Index, the paper has only been cited 8 times since, and none of these subsequent papers reports direct tests of the hypothesis that will be advanced here later. The circumstances are that Olsson and Brandt's 2 interpretation of their data was that the sex of the child affects the progress of the mother's disease. Subsequent workers tested this hypothesis and found little support for it. However let us reconsider an alternative hypothesis to account for the data of Olsson and Brandt. 2 This is that a third agent (maternal hormones) is responsible for both the offspring sex ratio and the disease.