For easing newcomers along , please consult the glossary of terms that I’ve written to make the terminology very understandable. Also, consult the post that explains the essential background
Title: Induced and Spontaneous Abortion and Incidence of Breast Cancer Among Young Women
Authors: Karen B. Michels, Sc.D, Ph.D., Fei Xue, M.D., Graham A. Colditz, M.D., Dr.P.H., Walter C. Willett, M.D., Dr. P.H.
Author affiliations are with Harvard Medical School and Harvard’s School of Public Health.
Journal: Archives of Internal Medicine, 2007. Vol. 167:814-820
This study included 105, 716 women in a prospective cohort of young women between the ages of 29 46 years old at the start of follow-up in 1993. The participants were asked about their history of spontaneous and induced abortion at the outset, and every two years thereafter. This method was employed to avoid the issue of recall bias, as articulated by the opponents of the retrospective studies.
Breast Cancer status was also part of the questionnaire and follow-up surveys. Hazard Ratios (Comparable to relative risks) were established with 95% confidence intervals.
Using healthy subjects as the controls, The Hazard Ratios for women with breast cancer were:
Age at first abortion:
30 HR=1.12 (12% increased risk)
When parous (women who had children) and nulliparous (women who did not have children) women were analyzed separately (what is called:stratified) the results are as follows:
Nulliparous:
No Induced Abortion HR=1.00 (Reference value)
Induced Abortion HR=1.26
1 Abortion HR=1.28
> 2 Abortions HR=1.18
Parous:
No Induced Abortion HR=1.00 (Reference value)
Induced Abortion HR=0.96
1 Abortion HR=0.98
> 2 Abortions HR=0.87
The authors analyzed the data after ten years from the point of commencing the study with its follow-ups. So what do we see.
The authors conclude an overall 6% increased risk of abortion, which is a bit disingenuous statistically, given the trends reported and the design flaws.
In Parous women v. nulliparous women, we see the obvious risks of abortion and the loss of protective effect of a full term pregnancy in the nulliparous women, whereas we see the protective effect of full term pregnancy in the parous women.
Several design flaws combine to depress the increased risks already evident in this study’s data. First, breast cancer takes at least ten years to develop after the exposure event (abortion). This study stopped at ten years, not accounting for many cancers that will develop after the premature analysis. Also, the most credible analysis would have been a study of the women with induced abortion at the outset in 1993. Two year follow-up surveys were done since then. So the study includes abortions done as close to the analysis as 8 years, 6 years, 4 years and 2 years. Including those abortions in the analysis, while cancer risk is still in its accepted latency period is simply reckless and irresponsible at best.
In truth, it is part of the lie.
For those who love the scientific details, a more thorough scientific analysis by Dr. Joel Brind is available:
In short, this analysis shows all that we have been saying is evident in the retrospective analyses. But this paper was published after the 2003 NCI Fraudulent Workshop and funded with an NCI grant CA50385, and its authors are loathe to bite the hand that feeds them.
More great prospective studies and their shoddy design and reporting all this week.
[…] We’ll consider the paper and its flaws in Part II. […]
Hope you’re feeling better.
I wonder if there will be any studies tracking rates of BC in Japan after the pill was approved here (which was quite recently)?
Japanese women have a very low BC rate — so low, in fact, that mammograms are not standard in public health prevention.