Cancer Epidemiology and the Abortion Wars.

During the same years as research into a possible abortion-breast cancer link was intensifying so, too, was the militancy of the anti-abortion movement. Fundamentalist Christianity and the New Right had become closely allied, but, after Republican Ronald Reagan was elected president in 1980, there was disappointment and a sense of betrayal, for the new administration did not seek to reverse Roe v Wade. In 1983 the Senate defeated an amendment that would have returned the abortion issue to the state level, and there continued to be about one and a half million abortions performed in the US each year.

As the lobbying efforts of the non-violent majority in the movement seemed to have accomplished little, supporters of direct action rose to prominence, first employing tactics, such as sit-ins, inherited from the tradition of civil disobedience, but moving on by the mid-1980s to clinic break-ins and bombings. Carol Mason has analysed the rise of an apocalyptic narrative within the movement—the understanding that if abortion were not stopped, God would cease to protect America. The evangelical leaders Pat Robertson and Jerry Falwell gave their support to the violent strategies of such groups as Operation Rescue, and at the 1988 Democratic convention in Atlanta hundreds of demonstrators were arrested. But most politicians knew by that point that a majority of Americans had come to accept the right of adults to seek early abortion, and the newly elected Republican leader, the elder George Bush, would not commit his party—in his words—to a “litmus test” on the abortion question. The belief that they had once again been abandoned by the conservative establishment encouraged even more desperate measures on the part of anti-abortion activists, but the 1993 murder of the physician David Gunn irreparably harmed the movement.

In 1994, President Clinton signed the Freedom of Access to Clinic Entrances Act (known as FACE), and the Supreme Court ruled that lower courts could establish “protest-free buffer zones” around abortion clinics. Arsons, bombings, and murders of clinic workers continued during that year, leading to the collapse of violent anti-abortion activism in the US. Organizations such as National Right to Life reasserted their leadership in the movement and pursued more targeted strategies, including opposition to late-term abortion (termed partial-birth abortion) and to FDA approval of the abortion drug RU-486. They also began to use the alleged abortion-breast cancer link to discourage women from abortion and to demand new legislation.

The widespread belief that a breast cancer epidemic was under way in the US seemed to substantiate the movement's theory that the “abortion generation”—younger women who had become sexually active in the twenty years since Roe v Wade—were now reaping the consequences of their freedom to terminate pregnancies at will. The notion of an epidemic requires closer examination, for it arose largely from an inaccurate understanding of breast cancer history. It was true that there had been a moderate but steady increase in breast cancer rates since the 1940s, but the more recent and alarming rise in diagnoses was in good part due to early detection through mammography. This conclusion was supported by the fact that the increase was seen mainly in the early stages of the disease, that the greater increase occurred not in young women but in those over fifty (the age group most likely to be effectively screened), and that rates had levelled off by the late 1990s, as would be expected if the increase were due to early detection. Paula Lantz and Karen Booth have examined the role played by the media in raising fears of an epidemic in the US—how the disease was portrayed as being out of control, and how the discussion of reproductive factors, such as birth control and delayed childbearing, underscored the frequent suggestion that it was young, white, “liberated” women who were in the greatest danger.

Lantz and Booth found, for example, that 85 per cent of the case studies and anecdotes found in popular magazines described women who were under fifty, whereas only 20 per cent of those diagnosed were in that age group. The number of such articles increased dramatically during the 1990s, and, after 1993, induced abortion was added to the list of risk factors mentioned. Lantz and Booth propose that the portrayal of women as victims of their own behaviour may be seen in the context of a backlash against the power and autonomy that women had achieved through controlling their fertility, and it was certainly the case that anti-abortionist literature increasingly linked the fight against breast cancer with the struggle to preserve conservative “family values”.

The most vocal proponent of an abortion-breast cancer link in the early 1990s was Joel Brind, a professor of biochemistry at Baruch College in New York City. His area of academic research had been the study of blood levels of steroids in relation to disease. In 1985, he experienced (in his words) “a major course correction” when he converted from Judaism to Christianity, and thereafter he sought to reconcile his professional life with his newfound religious conviction. Brind became involved with National Right to Life, and, wishing to devote his scientific expertise to the cause, he began by providing information on the steroid abortion drug, RU-486. Another turning point came when he read an article in Science News which discussed the most recent research of Malcolm Pike (who had drawn attention to abortion and cancer risk in young women back in 1981). This article focused on the apparent protection that pregnancy offered against breast cancer, but omitted to say that the pregnancy had to be full-term or to mention Pike's earlier findings regarding abortion. Now convinced that the link was real, that knowledge of it was “being actively suppressed”, and that the Lord wished him to “bring this life-saving knowledge into public awareness”, Brind protested, but Science News failed to publish his letter. Further efforts did not bring the recognition he desired, and his belief in a conspiracy of silence—and his vision of his own messianic role—seem to have been born at this time. He began publishing regularly on this issue in the anti-abortion press, confident that his efforts would “spare many women the agony of breast cancer”.

It was in this atmosphere of heightened tension and growing militancy that an article published in the Journal of the National Cancer Institute, along with its accompanying editorial, came to occupy centre stage in this debate. In the autumn of 1994, Janet Daling and her colleagues reported the findings of their much anticipated study of breast cancer in the generation of women “born recently enough to have had some or most of their reproductive years after the legalization of induced abortion”. As was the case in most such studies, their methodology involved in-person interviews with cases and controls in order to collect detailed information on the women's reproductive histories. After attempting to take other risk factors into account, they found that, among women who had been pregnant at least once, those who had had an induced (but not a spontaneous) abortion had a 50 per cent higher risk of developing breast cancer before the age of 45 than those who did not, and that the highest risk was associated with abortion in the last month of the first trimester.

Contrary to some previous studies, including that of Pike and his colleagues, they reported no difference in risk associated with the number of abortions or in women with completed pregnancies. Much would be made by Brind and others of the findings which concerned women who had aborted before the age of 18. For this group, the relative risk was 9.0 if the abortion took place between 9 and 24 weeks of pregnancy, and all twelve of the women with a family history of breast cancer who had aborted before the age of 18 had later been diagnosed with breast cancer. But these categories represented less than 3 per cent of the total of 845 cancer cases, and the interpretation of such figures would also be complicated by the fact that cancer patients who had never had a completed pregnancy were being compared with a control group of parous women. Daling herself warned against reaching “a firm conclusion at the time”. In fact, Daling and her team published a study two years later which found that abortion was associated with a relative risk of only 1.2, that “there was no excess risk of breast cancer associated with induced abortion among parous women”, and that there was no sub-group “in whom the relative risk associated with induced abortion is unusually high”. That report would go largely unnoticed.

The positive correlations Daling reported in 1994 would give proponents of the abortion-breast cancer link their strongest support to date, but, for Brind, the editorial that accompanied Daling's report was almost as important as the study itself. Written by Lynn Rosenberg of the Slone Epidemiology Unit at the Boston University School of Medicine, the editorial reflected some of the wider debates under way in the discipline of epidemiology in the 1990s, including the problem of recall bias inherent in retrospective studies, and the questionable statistical significance of fairly low elevations in risk. It also drew attention to some specific shortcomings and inconsistencies in Daling's study, and to the lack of a “convincing biologic mechanism” to explain why induced abortion posed a danger while spontaneous abortion did not.

Acknowledging the “intensity of emotion” surrounding the issue, Rosenberg ventured that, while Daling's results provided leads for the scientific community, it was questionable how they would be “informative to the public” at that time. She concluded that “whatever future results show, the decision to continue or terminate an unplanned pregnancy will still need to be based on a balanced consideration of the entire range of relevant issues—personal ethical considerations, the desire for a child, the ability to care for it, and the total health implications of continued pregnancy versus induced abortion”. For Brind, Rosenberg's efforts to defuse the issue gave credence to his repeated accusations that the National Cancer Institute was determined to cover up or discredit research pointing to an abortion-breast cancer link.

The publication of Daling's report and Rosenberg's response unleashed a reaction that put pro-choice advocates on the defensive. Time magazine reported that, months before the results were officially released, anti-abortionists “laid plans to trumpet the seven-year study's findings”, while “in the opposition camp, pro-choice groups marshalled the statistics they needed” to defend their position. While the study was still in progress, Daling was pursued for days by a Virginia lawyer employed by a right-to-life group trying to recruit her as a spokesperson, and she recounted how she finally told him, “I don't think you care one bit about breast cancer and women's health”. Once the report appeared, newspapers, magazines, and television news shows publicized the highlights, many cautiously, but some in a partisan fashion, either praising or criticizing the study. Daling herself repeatedly told the media that politics and personal views should not be allowed to cloud the issue, but it was inevitable that breast cancer would become a new weapon in the abortion wars. For example, Christ's Bride Ministries rented space in rapid-transit stations in the eastern US to advertise that “Women who choose abortion suffer more and deadlier breast cancer!”, and the federal order to remove the posters in Philadelphia fuelled charges of a cover-up by Washington.

Meanwhile, anti-abortionists in Congress began a long campaign demanding hearings on the abortion-breast cancer question, and more post-Reagan, New Right Republicans were drawn to the issue. Doubting the effectiveness of sheer denial on the part of pro-choice advocates in such a climate, the left-leaning magazine Mother Jones took the view that, by attacking Daling's study and dismissing its findings outright, “most pro-choice groups have played right into Newt [Gingrich]'s hands”.

Meanwhile, researchers produced new studies and engaged in a growing international debate over methodological questions. Because retrospective studies still predominated, the question of recall bias remained highly relevant, especially in light of a Swedish study which suggested that it could result in a 50 per cent increase in reported risk. One method of investigating the influence of recall bias was to examine differences in responses among sub-groups of women involved in the studies.

At the University of Wisconsin Comprehensive Cancer Center, Madison, Polly Newcomb found a higher level of risk for women who had abortions before 1973, suggesting a higher rate of under-reporting among controls for the period when abortion was still illegal, while M A Rookus of the Netherlands Cancer Institute found a higher association between abortion and breast cancer in the mainly Roman Catholic south-eastern portion of the country, suggesting that Catholic women in the control group were more likely to under-report. Other researchers addressed the question of why so many studies, including Daling's, showed no increased risk following spontaneous abortion. The standard answer was that failed pregnancies were the result of hormonal deficiencies and did not affect breast tissue the same way. But a Greek research team pointed out that this was true only in some cases, and that a percentage of spontaneous abortions should have the same effect on cancer risk as induced abortions. If the data did not reflect that fact, then “subtle information bias” must be considered a possibility. Given the conflicting nature of the data thus far and the apparent problem of bias, the American Medical Association warned that legislative initiatives already under way in some states by 1995 must be considered premature.

In Joel Brind's view, such caution was mere cowardice or compliance with a pro-abortion medical culture, and he heightened his efforts to reach a wider audience. He wrote frequently about the problem of denial, as he saw it, in the anti-abortion press, while continuing the fight against the legalization of RU-486, using the abortion-breast cancer link as part of his argument. When testifying at the FDA's advisory committee hearing in 1996, he predicted that “thousands upon thousands” of women would get breast cancer as a result of having used this drug. That year, Brind also published a lengthy review and meta-analysis of twenty-three studies dealing with induced abortion and cancer and thus entered into a dialogue with the epidemiological community.

In his version of the history of the abortion-breast cancer debate, he described how a positive association had first been observed in Japan in 1957, and how, since then, such findings were consistently ignored or their validity questioned even by those who had conducted the research. He determined that, taken together, the studies proved that abortion raised the risk of cancer by at least 30 per cent. In his view, there was “overwhelming evidence” that recall bias was not a factor, and he used data from a variety of studies to argue that, while spontaneous abortion did not lead to higher cancer risk, induced abortion elevated risk irrespective of when it occurred in a woman's reproductive life. He concluded that the studies published thus far, whether prospective or retrospective, produced a very consistent, positive association between induced abortion and breast cancer independent from all other risk factors, including nulliparity.

More studies might be useful, said Brind and his co-authors, but “there exists the more present need for those in clinical practice to inform their patients about what is already known”. The fact that Brind had gained a hearing through a reputable medical journal made him more difficult to ignore and affected the tenor of the debate in the wider research community. At least some researchers felt the need to respond directly to his challenge, while at the same time using this opportunity to confront the wider problem of bias as it affected their discipline.

At the Harvard University School of Public Health, Karin Michels and Walter Willett conducted their own survey of the medical literature on abortion and breast cancer and concluded, unlike Brind, that the investigation to date did not permit a final assessment of the issue. When interviewed for the Harvard University Gazette, Michels pointed again to the problem of recall bias in retrospective studies, and questioned Brind's method of combining a series of reports and attempting to arrive at an overall estimate, “instead of trying to understand why the studies differ in their results”.

Responding both to Brind and Michels, an editorial in the Journal of the National Cancer Institute asked how two reviews with the same objective could produce such different results, and posed the question of whether the problem of bias was a sign that epidemiology had “reached its limit”. Recall bias was one issue, but another was what they termed “wish bias”, or “the extent to which a reviewer believes a priori that the hypothesis is true”. They were critical of Brind's meta-analysis and his “blurring of association with causation”, and used his work as a case study in how epidemiologists must learn to recognize when biases are being reproduced in a succession of studies, and why this occurs. “Indeed”, they concluded, “after this excursion into the issue of abortion, bias, and breast cancer, it seems our future has as much to do with human behavior as with human biology”.

In this atmosphere of uncertainty, a major study which was free from the problem of recall bias was bound to gain attention. In 1997, the New England Journal of Medicine published the results of research by Mads Melbye and colleagues in Denmark, which used the Danish national registries of induced abortions and breast cancer cases to study a cohort of 1.5 million women born between 1935 and 1978. It reported that induced abortion carried a relative risk of 1.0, indicating that there was no link between abortion and breast cancer. Criticizing Brind directly, Melbye pointed out that he had relied almost entirely on case-control studies and had based his results on “a crude analysis of published odds ratios and relative risks with no attempt to incorporate the original raw data into a more sophisticated statistical analysis”. The accompanying editorial by Patricia Hartge of the National Cancer Institute also attacked Brind, touted Melbye's study as definitive, and added fat to the fire by concluding that “a woman need not worry about the risk of breast cancer when facing the difficult decision of whether to terminate a pregnancy”. But flaws in Melbye's research left that position open to attack.

By his own admission, the fact that pre-1973 abortion information was not included could lead to an under-estimation of a link with breast cancer among older women. Other researchers noted, as well, that it was far too early to know what the breast cancer incidence would be for women in the study who were born as recently as the 1970s—an age group for which abortion rates would have been comparatively high. Brind and his followers would soon exploit these limitations to the full, but they were arguably less motivated by a quest for scientific credibility than by the imperatives of the anti-abortion cause. In Brind's words, following the publication of Melbye's report, his telephone “was ringing off the hook, mostly from pro-lifers who had signed onto the ABC link based on my witness and were now being barraged with tough questions about their new found anti-abortion argument that went bust”.