For ease in online reading, when you click on a reference link, a pop up window will appear with the reference. If you are using a pop-up stopper, you will have to disable the pop-up stopper while you read this document.

The paper in the Lancet entitled "a collaborative reanalysis of data from 53 epidemiological studies, including 83,000 women with breast cancer from 16 countries," by Valerie Beral et al.(1) , is an apparently comprehensive meta-analysis of worldwide published and unpublished data. It has been widely touted as definitive proof that there is no link between induced abortion and the risk of subsequent breast cancer (ABC link). In Beral's own words (as reported by the Associated Press): "The totality of the worldwide epidemiological evidence indicates that pregnancies ended by induced abortion do not have adverse effects on women's subsequent risk of developing breast cancer"; and, (as reported in the Atlanta Journal-Constitution) "Scientifically, this really is a full analysis of the current data".

In fact, the Beral study reanalyzed the data only after a highly biased selection process which had many studies showing valid evidence of the ABC link inappropriately excluded, invalid studies whose flaws had been documented in the scientific literature inappropriately included, and valid studies whose data had been published simply not mentioned at all.

• Eleven valid studies (7-17) were excluded for unscientific and inappropriate reasons, including:

It is also very important to note that the relation of induced abortion to breast cancer is measured epidemiologically in a very artificial way. This is clear just from the title of the key figure which shows the data on induced abortion in the Beral et al. reanalysis: "Relative risk of breast cancer, comparing the effects of having a pregnancy that ended as an induced abortion versus effects of never having had that pregnancy." Obviously, a woman considering abortion is already pregnant, and does not have the option of "never having had that pregnancy". In other words, the Beral et al. study, only measures the independent, additive effect of the abortion to a woman's breast cancer risk, ignoring the fact that abortion definitely leaves a woman at a higher risk of breast cancer than would apply had she chosen to carry the pregnancy to term. As Beral et al. put it right in the opening line of the paper's introduction: "Pregnancies that result in a birth are known to reduce a woman's long-term risk of developing breast cancer,". It is therefore quite misleading to state that abortion has no effect on future breast cancer risk, even if it could be shown not to increase risk beyond the "never having had that pregnancy" level.

It is particularly telling that, for example, another risk factor which is widely acknowledged, is measured by a different standard. The case in point is combination hormone replacement therapy (HRT) for menopausal women. Menopause is a lot like full-term pregnancy, in terms of its effect on future breast cancer risk. Just as a full-term pregnancy lowers a woman's risk of breast cancer (and the younger a woman is when she has her first child, the more her future risk is lowered), the younger a woman is when she goes through menopause, the lower her risk of breast cancer. This is attributed to lower estrogen exposure due to cessation of the ovaries' production of the hormone.

When induced abortion is studied as a risk factor, women with abortion are compared (as noted above) to women who did not have a pregnancy then, rather than to women who carried the pregnancy to term. The latter comparison would show elevated risk with abortion, since post-abortive women would not have the risk-lowering effect of a full-term pregnancy.

In stark contrast, when HRT is studied as a risk factor, women taking HRT are not compared to women of the same age who did not go into menopause. Were this the case, HRT would not show up as a risk factor either, for the premenopausal women in the comparison group would not have gotten the protective effect of menopause. It would simply be concluded that HRT leaves a woman at the same risk as if she had not yet gone into menopause, and that it is not a risk factor for breast cancer.

Instead, when HRT is studied as a risk factor, premenopausal women are excluded from the analysis, (as is made explicitly clear in a major study by Beral et al. published just last year) and so postmenopausal women taking HRT are compared to postmenopausal women not taking HRT. The women in the comparison group, therefore, have gotten the protective effect of menopause. This protective effect is blocked by HRT, and so HRT shows up as a risk factor, as well it should. But abortion is judged by a different standard; one that makes it appear "safe" for women.

Finally, it is noteworthy that the authorship of the Beral et al. study is presented in a misleading way. The by-line of the paper simply says "Collaborative Group on Hormonal Factors in Breast Cancer*". This implies that the authors of all the studies included in the reanalysis are responsible, as co-authors, for the content of the paper. However, as is indicated only in a footnote at the end of the text, the "Analysis and Writing Committee" consists of Valerie Beral and four co-authors, who "analysed data and wrote the paper, taking into account comments on earlier drafts by collaborators". By current internationally accepted standards of authorship, only these five people are responsible for the paper's content, and therefore qualify as authors of this paper. It also hardly seems coincidental that this group, based at the Radcliffe Infirmary of Oxford University, UK, represents a continuum of authorship dating back to 1982. This 2004 paper represents the third paper by a group with at least one common author, which papers (1,5,45) also have in common the use of inappropriate databases to draw the conclusion that induced abortion does not increase the risk of breast cancer. Unfortunately, however, it does.