By Joel Brind, Ph.D.
Joel Brind, Ph.D.
After researching and publishing on the ABC link for over 20 years, I’ve managed to get the hang of the formulaic manner in which studies come out periodically to “reassure” the public about the “safety” of abortion. Such studies often come out in the most prominent medical journals, coincidentally on the heels of studies that have reconfirmed the reality of the ABC link.
Why? To keep up the mainstream medical journals’ wall of denial.
But one needn’t be an epidemiologist to appreciate the way standard epidemiological (epi) methods are used inappropriately, so as to make the facts of the ABC link disappear. The techniques are remarkably few. With a few real life examples from the ABC literature, the lay person should be able to learn to spot these tricks rather easily.
First, it is important to understand the two basic types of epi study. The most common is the case-control study. In a case-control study, a group of patients with the disease in question (e.g., breast cancer) is recruited, along with a group of women without cancer, but who are otherwise similar in age, ethnicity, and country (or smaller location) of residence. The latter group is the “control” group.
The researchers then query the women in the two groups, typically by having a trained nurse interviewer go over medical and reproductive history with them. Data is entered into a questionnaire that can be processed by a computer.
One of the reproductive history data is typically “number of induced abortions.” These data are often reduced to the dichotomous form–yes or no, a woman has had or hasn’t had an induced abortion.
(For comparison purposes, the baseline number for “relative risk” is “1″. If the number is over 1, there is an increased risk. For example, if the relative risk is 1.4, there is a 40% increased risk. If the relative risk is 0.7, there is a protective effect; a 30% reduced risk. In case-control studies, the actual statistic used to provide an estimate of relative risk is called the odds ratio—OR.)
When the data are tallied, if there are more women among the breast cancer patients who reported having had an induced abortion than there are among the cancer-free women, this “association” can be interpreted as increased risk, represented by OR>1 [more than 1).
Conversely, if more abortions are reported among the healthy women, an OR<1 would result [less than 1], and indicate that abortion lowers the risk. Of course, such associations do not prove cause-and-effect, but that gets established by the biology.
Most epi studies are case-control studies, and since most of them indicate an increased breast cancer risk for women who have had an abortion (i.e., that there is an ABC link), what are pro-abortion researchers to do?
Usually they first seek to discredit the methodology in such studies. They argue that such studies suffer from a flaw called “reporting bias” or “response bias.” How does that supposedly work?
They declare that women with breast cancer will be more honest about their abortion history. Thus if the same number of women without breast cancer as breast cancer patients have had an abortion—but the breast cancer patients are more “honest” and report more of their abortions–it will falsely appear that there is a “positive association,” i.e., an increased risk for breasts cancer associated with having an induced abortion.
The trouble with this response bias argument is that it has been amply disproven many times. One good example was a 1989 case-control study in upstate New York that was not based on after the fact interviews with patients v. healthy “control” women. Instead it was based on medical records from the time of the abortion.
That meant the record of who had and did not have an abortion was established long before anyone in the study got breast cancer. That study resulted in an OR of 1.9, i.e., a 90 increased risk associated with abortion.
Though published in a prominent epi journal, mainstream researchers often simply deny that such a study even exists!
But what about the relative minority of case-control studies that do not show an ABC link? Good examples of these are a large study on women from Shanghai, China, published in 2001 and a Serbian study published just last year.
In the Shanghai study population, no significant association with abortion was found (OR = 0.9). Note 66% of the women in the study had had at least one abortion.
In the Serbian study, there was actually reported a negative association (i.e., a protective effect, or reduced risk of breast cancer) with abortion (OR = 0.46). Note the prevalence of abortion among the study population was a whopping 89%.
Why does the high incidence of abortion in the population make a large difference in determining whether an ABC link exists?
As we explained in a letter published in the British Journal of Cancer in 2004, the high prevalence of abortion explained the effect. Essentially, epidemiology can identify exposures that are associated with a given disease, as long as those with the disease can be compared with typical members of the population who do not have the exposure in question.
However, when the majority of the population has had the exposure (in this case having had an abortion), the comparison group (those who did not have an abortion) is no longer typical. In the case of breast cancer, where most of the women have had an abortion, the women who have not had an abortion are generally those who’ve also had fewer children and who started having children at a later age.
But starting to have children when one is older and having fewer of them are both risk factors for breast cancer. The comparison group is really a high-risk group. Hence, the effect of abortion disappears when the majority of women are post-abortive.
And if the majority is overwhelming (like the 89% in the Serbian population), those who’ve had abortions actually show up with lower risk–that is, compared to the very small minority of sub-fertile, high-risk women who have not had any abortions.
This explanation was reiterated and expanded in the recent meta-analysis of 36 studies from China, published last year. The Chinese authors of that study also showed the clear trend of decreasing ABC link as the prevalence of abortion increased.
The bottom line is simple. If you are determined not to find a ABC link, just make sure you choose a study population in which most of the women are post-abortive. It also helps to restrict the study to women who’ve had at least one livebirth—as both the Shanghai and Serbian studies did. Why? Because childless post-abortive women have a higher breast cancer risk than women who’ve never been pregnant.
But if your main argument against the ABC link is that case-control (i.e., retrospective) studies are unreliable because of response bias (whose repeated disproof has not dissuaded most mainstream pro-abortion researchers from its invocation), then the study you conduct needs to be of a type routinely championed to be methodologically superior. This second major type of epi study is called a cohort study (i.e., prospective).
In a cohort study, you start with a population wherein the fact of abortion is recorded at the time of the abortion, and all women are then followed in time to see who gets breast cancer and who does not. Because it’s not even possible for such studies to suffer from response bias, cohort studies are routinely touted by mainstream pro-abortion researchers to be superior to case-control studies.
They trotted out a number of them in very prominent journals between 1997 and 2008, to “prove” to the world that there was no real ABC link. (The reader might remember that in 1996, we published a worldwide meta-analysis that found a significant ABC link among all studies. Our meta-analysis garnered major mainstream attention and credibility to the ABC link–and that could not be allowed to stand.)
The first and the worst (and the largest) of these cohort studies was published in 1997 in the “New England Journal of Medicine,” arguably the most influential medical journal in the world. The study was based on the medical records of all 1.5 million women born in Denmark between 1935 and 1978, including over 400,000 abortions and over 10,000 cases of breast cancer.
The Danish study was widely touted as disproving the ABC link. It found a relative risk of 1.0; there was no possibility of response bias; and the statistical sample was enormous.
As I have written before, the flaws of the Danish study are many and monstrous. However, the most important flaw concerns the question of follow-up time. (Bear in mind breast cancer takes years to develop.)
In a case-control study, one typically starts with older women–since they are cancer patients–and so they are mostly diagnosed many years after the abortion has taken place. This makes sense, because it takes about 8-10 years for cancer to develop after an exposure like abortion.
But suppose you design your cohort study such that you log both abortions and breast cancer diagnoses during the same time period. That’s what the Danish study and several influential cohort studies did. So, for example, in the Danish study, both abortions and breast cancer diagnoses were collected through 1992.
Remember that the study included all women born through 1978—i.e, including women as young as 14. So a teenager in Denmark who had an abortion in 1992 is included in the study as a woman who had an abortion but did not develop breast cancer.
Well of course not—breast cancer takes years to develop! So in fact, the Danish study included a total of 358,000 women under the age of 25—fully one fourth of the study population, who collectively had over 100,000 abortions—yet represented a total of only 8 cases of breast cancer!
So that’s how one stacks the deck of a cohort study with women who have abortions, but no breast cancer: allow for as little as zero follow-up time after the abortion. And to make sure the ABC link does not show up, it’s also important to eliminate the type of breast cancer that shows up most quickly following an abortion, so called “in situ” carcinoma. Typically, the cohort studies that do not show an ABC link simply eliminate in situ carcinoma diagnoses from the study.
Now you know the basic tricks for conducting an ABC link study—whether retrospective (case-control) or prospective (cohort) in design—that does not show the link. Watch for them.
Editor’s note. Joel Brind, Ph.D. is a Professor of Human Biology and Endocrinology at Baruch College, City University of New York; Co-founder of the Breast Cancer Prevention Institute, Somerville, NJ; and a frequent contributor to NRL News and NRL NewsToday.