|
NRL News
Abortion
and Breast Cancer: Once Again AGI Misses the Link In reading the Alan Guttmacher Institute's (AGI) latest tome, "Abortion in Women's Lives," it would, of course, be shocking to discover any retreat from AGI's mantra that abortion is not only safe but wonderful for women. Likewise, it is no surprise that this "special research affiliate" of Planned Parenthood would tell the reader that an "exhaustive literature review and analysis" confirms that there is no link between having an induced abortion and having a higher risk of breast cancer years later (known as the abortion-breast cancer link or ABC link). But when read carefully, the new AGI volume, written by Heather Boonstra and three colleagues, also reveals some important nuanced changes that suggest AGI is straining under the burden of declining credibility. The authors begin a chapter entitled "Abortion is not associated with an increased risk of cancer" with an accurate account of the conclusions of a 1996 comprehensive review and meta-analysis of the ABC link compiled by myself and several associates. AGI's characterization was another matter. "Abortion opponents," we're told, "seized upon" the analysis because it "combined the results of multiple studies and reported that women who had had an abortion had a significantly elevated risk of breast cancer." Boonstra and her colleagues then write, "Other researchers and medical groups, however, found this study to be flawed, largely because the data were collected only after breast cancer had been diagnosed." It gets harsher. Our study was "further flawed," they say, since it relied on medical histories supplied by the women themselves, rather than from medical records. All epidemiological studies have their strengths and weaknesses. But when it is said that a study is "flawed," that means that there is some serious error in the way the study was conducted and/or the results compiled or interpreted. Also notice the implication that our study was only endorsed by "abortion opponents," while "other researchers and medical groups" (presumably more legitimate) rejected our "flawed" research. The anti-ABC link rhetoric has hardened quite a bit over the last decade. When our meta-analysis came out, leading "pro-choice" researcher Janet Daling called it "very objective and statistically beyond reproach." The 2000 clinical guide on abortion practice published by the United Kingdom's Royal College of Obstetricians and Gynaecologists (no friend of pro-lifers) said our meta-analysis "had no major methodological shortcomings and could not be disregarded." Yet in spite of this, to believe AGI and their pro-abortion colleagues, our study has now sunk to the new level of being doubly "flawed." The "flaw" at the center of the criticism is the tiresome canard known as "reporting bias" (a.k.a. "response bias" or "recall bias"). This type of bias is actually a potential shortcoming of any epidemiological study that is based on the self-reporting of medical history retrospectively, that is, after the cancer diagnosis has been made. But it is noteworthy that most epidemiological research uses this type of methodology, so it makes little sense to view the standard, accepted methodology as flawed. Hypothetically, it is possible that those women who have been diagnosed with cancer will be more forthcoming about such a sensitive topic as abortion, compared to women who are cancer-free (known as the "control" group). If fewer of the cancer-free women reveal having had an abortion on the study questionnaire, then it would appear as if, in comparison, that more of the cancer patients had abortions, even if they had abortions at the same rate as the controls. This bias would translate to an apparent--but artificial--increased breast cancer risk among women who'd had an abortion. The issue is not whether this is theoretically possible but whether the studies bear out that this reporting bias exists. They don't! But that has not prevented critics of the ABC link from arguing that reporting bias explains the 30–50% increased risk of breast cancer associated with abortion. To support this claim, reference is frequently made to a 1991 Swedish study which claimed to show evidence of reporting bias. Was the study valid? In a word, no. A rigorous mathematical disproof of the Swedish study was published in 1998. (This AGI report aside, the discredited Swedish study is no longer being cited by critics.) Many other studies have conclusively proven that reporting bias simply does not exist in the published literature on the ABC link. I have written about them in many issues of NRL News. That did not stop the AGI authors from twice invoking the reporting bias canard--in the main text and again in a side box entitled "Methodological issues in abortion research." Noteworthy is that the only reference they cite as justification is a 2001 American Cancer Society "fact sheet." In other words, there is no longer any study in the medical literature that even the AGI is willing to cite as credible evidence of reporting bias. I spend all this time on the reporting bias argument because it is central to the denial of the ABC link. If accepted on its face, it allows the wholesale dismissal of almost four decades' worth of studies published between 1957 and 1996 that show an increased breast cancer risk. By dismissing what women say, it sets up the argument that only research that is based on medical records--rather than retrospective self-reporting by women--can yield reliable results. That paved the way for a body of deeply flawed studies published over the last decade--studies which are based on medical records--which insist there is no ABC link. Boonstra et al. cite the most notorious of these studies, the 1997 paper by Melbye et al. on women in Denmark. The Melbye study is the largest, based on all 1.5 million women born in Denmark between 1935 and 1978, encompassing over 300,000 abortions and over 10,000 cases of breast cancer. The study was based entirely on medical records. The authors found no overall increase in risk. However, the Melbye study also embodies the most egregious departures from proper methodology. For example, it included all records of breast cancer records beginning in 1968, but abortion records were only examined beginning in1973--as if the effect could precede the cause! Worse yet, 60,000 women who had undergone abortions were misclassified as having never having had an abortion! Predictably, the Boonstra study also relies on a 2003 U.S. National Cancer Institute (NCI) "workshop." Contrary to NCI's claims, not only was there no review of the data, a presentation of the evidence for the ABC link was never allowed. (My "minority report" can be found at our web site: www.bcpinstitute.org.) Just as predictably, Boonstra et al. tout the 2004 "exhaustive literature review and analysis" by Valerie Beral et al. in the UK. This review has the advantage of neatly selecting out almost every study showing a significant ABC link (the 15 excluded studies revealed an overall 80% increase in risk) on such dubious, non-scientific grounds as that "principal investigators ... could not be traced," or "original data could not be retrieved," or "researchers declined to take part in the collaboration." In addition to excluding relevant studies, Beral et al. included 28 studies which had not been previously published to buttress her conclusion there was no link. Finally, it is not at all surprising that the authors never mention my own review of the last decade's parade of supposedly definitive studies said to disprove the ABC link. This is not only intellectually dishonest, it also means that a well-connected think-tank is not telling women that abortion raises a woman's future risk of breast cancer. Joel Brind, Ph.D., is a professor of human biology and endocrinology at Baruch College of the City University of New York. He is founder and president of the Breast Cancer Prevention Institute and a frequent contributor to NRL News. |