In one of the under-reported news stories of late 2013, a “meta-analysis of the association between induced abortion and breast cancer risk among Chinese females” was published in the journal Cancer Causes Control. Joel Brind, a professor of biochemistry, physiology, and immunology at Baruch College at the City University of New York, and president of the Breast Cancer Prevention Institute, has done extensive research studying the connection between abortion and breast cancer, a connection we should all — particularly groups wearing pink ribbons — want to know the truth about. Brind talks to National Review Online’s Kathryn Jean Lopez about what we know and what we should be asking.
Kathryn Jean Lopez: What’s new about the Chinese study?
Joel Brind: There are several things that are new about this study:
First of all, it is the first meta-analysis of just the Chinese data, and it summarizes the data of a hefty number of studies (36 in all). Many of them were published in Chinese, so a number of new studies’ worth of data have now been made available to us.
Second, the article includes a number of studies whose data did not distinguish between induced and spontaneous abortion, the latter well established as not increasing breast-cancer risk. Typically, I and others exclude such studies from our studies on induced abortion and breast-cancer risk. However, the Chinese study (Huang et al.) justified their inclusion by also including data on the rate of spontaneous abortion in China, i.e., as being only about 10 percent of total abortions, owing to the extremely high induced-abortion rate. Then, they tallied the data separately for the two types of studies, and found that the studies restricted to induced abortion averaged a 49 percent increase in risk, compared to a 41 percent average increase in risk for the studies that had combined induced and spontaneous. As expected, therefore, the latter number is slightly but not significantly lower than the “pure” induced-abortion number. They thus reported the combined figure as a 44 percent risk increase, and this is therefore a slight underestimate. This Chinese study confirms — specifically — the results I obtained in my 1996 meta-analysis I did in collaboration with colleagues at the Penn State Medical College. In fact, they report a slightly stronger link, i.e., a 44 percent overall risk increase for one or more abortions, compared with the 30 percent we reported.
It also confirms the theorem we (Vern Chinchilli of Penn State and I) articulated in a 2004 letter in The British Journal of Cancer. Not only is our letter cited and approving attribution given to our argument, but Huang et al. also performed a “meta-regression” analysis of all the Chinese studies which confirmed the theorem.
Huang et al. describe the theorem thus in their discussion: “As argued by Brind and Chinchilli [citing Brind and Chinchilli 2004], once the prevalence of a given exposure rises to a level of predominance in the control group, statistical adjustment cannot remove all the confounding caused by the adjustment terms. This was well exemplified by the meta-regression analysis in our study.” In layman’s terms, the value of epidemiology is to identify an exposure factor (in the present case, induced abortion) which may increase or decrease the risk of a given disease (in the present case, breast cancer), by comparing the relative minority of exposed individuals to the typical healthy person in the general population. But if the exposure is so prevalent as to become the rule rather than the exception, the unexposed population is no longer typical, but may represent a subgroup that is at elevated risk for other reasons. In the present case, the unexposed women in China are likely to be those women who did not get pregnant and have children or did not have children at as young an age as the exposed (post-abortive) women. Since nulliparity and late age at first childbirth are strong risk factors for breast cancer, the post-abortive are thus being compared to women at elevated risk, so the risk increase due to abortion tends to disappear, in direct proportion to the prevalence of abortion in the population. This is precisely what the Huang et al. meta-regression analysis shows.
Lopez: What’s familiar about the Chinese study?
Brind: If one looks at the so-called “Forest plots” — standard graphical representations of the component studies of a meta-analysis, they look oh-so familiar to me, in that the overall, significant elevation in risk across the multiple studies is obvious at a glance. What is — sad to say — unfamiliar is the fact that it is a very straightforward study with no statistical manipulations or verbal contortions to “fudge” or hide the risk increase.
Lopez: What about this study was unique to China?
Brind: Here again, Huang et al. restate what we had argued in our earlier publications: “First, different from the USA where abortion is used predominantly to postpone first childbirth, almost all induced abortions in China were performed to limit family size after the first child. Therefore, more induced abortions may imply an early age of childbirth. The protective effects of early childbirth will probably dilute the harmful effect of more induced abortions.”
In lay terms: Even though they observed a dose effect (i.e., more abortions mean greater risk of breast cancer than one abortion), the effect of abortion is likely underestimated because the post-abortive women also experience the protective effects of having earlier first childbirths. This is really the flip side of the problem I’ve mentioned regarding the high prevalence of abortion. In short, what’s unique to China — the one-child policy — tends to lower the observed association between induced abortion and breast cancer.