Breast Cancer and the Politics of Abortion.

Epidemiology, like any branch of medical science, functions within a social and historical context. That context influences what questions are asked, how they are investigated, and how their conclusions are interpreted, both by researchers and by the public. The international debate over whether abortion increases breast cancer risk, which has been the subject of many studies and much heated controversy in recent decades, became so intensely politicized in the United States that it serves as a particularly stark illustration of how elusive the quest for scientific certainty can be. Although a growing interest in reproductive factors and breast cancer risk developed after the Second World War, it was not until the early 1980s, after induced abortion had been legalized in many countries, that studies began to focus on this specific factor.

In the US these were the years following Roe v Wade, when anti-abortionists mounted their counterattack and pro-choice forces were on the defensive. As a result, epidemiologists found themselves at the centre of a debate which had come to symbolize a deepening divide in American culture. This paper traces the history of the scientific investigation of the alleged abortion-breast cancer link, against the backdrop of what was increasingly termed an “epidemic” of breast cancer in the US. That history, in turn, is closely intertwined with the anti-abortion movement's efforts, following the violence of the early 1990s, to regain respectability through changing its tactics and rhetoric, which included the adoption of the “ABC link” as part of its new “women-centred” strategy.

Background

Of all the cancers which afflict the Western world, breast cancer has been the focus of the longest historical controversy over what its causes may be and why some women are more vulnerable to the disease than others. Common explanations have ranged from blows to the breast to the effect of unhappy emotions, but have also focused on the influence of events in women's reproductive lives. Opinion was long divided over whether childless women were at greater or lesser risk, and the same controversy surrounded the significance of breast-feeding. More definitive answers awaited the improvement of data collection and analysis as the field of epidemiology developed in the twentieth century. The first study to include both a large group of breast cancer patients and a control group of healthy women was conducted in Britain in the 1920s by Janet Lane-Claypon, and she found consistent differences between the two groups in relation to age at marriage, fertility, and lactation. She also posited that lifetime exposure to oestrogen was a significant factor, and, by the 1940s, endocrinologists were paying closer attention to the role of hormones in breast cancer.

The study of reproductive factors in relation to breast cancer continued after the Second World War, as researchers expanded the number of variables taken into consideration, including the use of hormone replacement therapy and, by the mid-1960s, the use of oral contraceptives. Even though attention was not focused on abortion until the early 1980s, it also began to receive brief mention as a possible risk factor. The earliest study cited in the epidemiological literature dates back to the 1950s, when researchers in the Department of Public Health at Tohoku University Medical School published a wide-ranging study of all common cancers in Japan. They found slightly higher rates of spontaneous abortion and significantly higher rates of induced abortion among cancer patients, but were hesitant to draw any firm conclusions because of methodological weaknesses in their study. Although their cautionary comments were usually ignored by later researchers when citing their report, the Tohoku team were explicit about their concern that the women in the control group, who were not facing a life-threatening disease, were less likely to report a past history of induced abortion, thus making it impossible to draw any conclusions about a cancer link. In later years, this phenomenon would be referred to as “recall bias”.

Intensive research into the role abortion might play in breast cancer risk would not begin for another quarter century but, during the intervening years, the legal status of the procedure changed in many countries, making it possible for the subject to be discussed more openly, and for more accurate statistical records to be kept. By far the greatest controversy surrounding this area of research would develop in the US, where the debate over abortion was passionate and deeply divisive.

Prior to 1973, abortion was illegal in thirty US states, with the exception of cases where the life of the mother was in danger, while in at least a dozen others abortion was permitted only in cases of rape or incest, or if the fetus suffered severe abnormalities. Abortion law was thus very inconsistent from one jurisdiction to another; in some states, even providing information on how to obtain an abortion elsewhere could lead to arrest. Challenges to abortion laws across the US eventually led to the sweeping decision by the Supreme Court on 22 January 1973. While Roe v Wade did not recognize a woman's “absolute right” to an abortion under the Constitution, it allowed (but did not require) states to regulate abortion only after the fetus was viable (i.e. in the third, or in some cases the second, trimester). By 1980, over one and half million abortions were being reported each year. The majority of them were performed in freestanding clinics, which mushroomed even in the weeks that followed Roe v Wade.

The ruling brought an immediate reaction from opponents of abortion. As James Risen and Judy Thomas explain, even some pro-choice advocates, such as SC Justice Ruth Bader Ginsburg, believed that Roe v Wade “went too far, too fast, taking a giant leap the country was not prepared to make”. The battle had only begun, and would now become closely tied to a major transformation in American religious and political life during the last quarter of the century, for the anti-abortion movement grew very much in tandem with the rise of evangelical Christianity in the United States. Before abortion was legalized, the Roman Catholic hierarchy had been the force behind most of the lobbying but, following Roe v Wade, they were joined by increasingly militant, and increasingly numerous, Protestant fundamentalists dedicated to the anti-abortion cause. Although the drift towards direct action strategies had its origins in the tactics of Catholic leftists of the 1960s, Christian fundamentalism would soon transform “civil disobedience into a conservative tool” and become associated with right-wing politics and opinion. Of even greater significance, however, was the grassroots, anti-establishment nature of that conservatism. Anti-abortionists discovered that conservatives in high places could not necessarily be counted upon, and their leaders, like their preachers, were drawn from the ranks of ordinary Americans, so that the increasingly apocalyptic rhetoric of the movement was infused with a strongly populist impulse. Christian revivalism, fused with “pro-life” activism, strove for the awakening of all of America to “the horrors, the evil, and the truth of abortion”.

Abortion-Breast Cancer Research to 1994

When research into a possible link between abortion and breast cancer began in earnest in the early 1980s, there was no hint that, within a few short years, epidemiologists would be thrust into the wider debate over abortion in the US. The first published studies represented a growing interest in an aspect of women's reproductive lives which could now be reported and discussed more openly, and they were part of an expanding international dialogue in which American researchers took a very active part.

This section of the paper will examine the tentative and conflicting conclusions reached during the 1980s and early 1990s, but will first consider how some earlier studies were used to provide a foundation for later work. It should be noted that the common means of conveying results throughout this literature was to indicate the degree of “relative risk” (RR) posed by abortion. For example, a relative risk of 1.5 suggests that women who have had abortions have a 50 per cent higher risk than other women, when all other factors have been controlled. The meaningfulness of such figures depends on many elements, including the ability of researchers to separate abortion from confounds and from other risk factors, the appropriateness of the control group, the numbers of women in the study, the accuracy of reporting abortions, and the extent to which low levels of relative risk can be considered significant.

In epidemiological literature, when interest in a potential risk factor is developing, its possible significance sometimes gains credibility from early, tentative evidence which seems to provide some shaky support for the hypothesis being tested. In this manner, a historical process begins, common in scientific research, whereby these early studies are cited again and again, without due attention to the context in which they were conducted, the validity of their results, or even the precise nature of their conclusions. This was very much the case with the abortion-breast cancer debate. The early Japanese study was often cited in support of a link, and much weight was given to a series of international studies on breast cancer epidemiology conducted during the late 1960s by Brian MacMahon in collaboration with researchers in both eastern Europe and Asia. They investigated a wide range of reproductive factors, but abortion did receive brief mention in each study. The findings were inconsistent, and in one study fewer abortions than expected or predicted were reported by women who had become cancer patients. But when MacMahon and his co-authors summarized these results in a much cited article published by the American Journal of the National Cancer Institute in 1973, they inaccurately concluded that, even though abortion was not consistently linked to cancer, “where a relationship was observed, abortion was associated with increased, not decreased, risk”.

Proceeding from an understanding that abortion was an already established risk factor for human breast cancer, Jose and Irma Russo of the Michigan Cancer Foundation in Detroit set out in 1978 to discover the mechanism behind this link. They introduced carcinogens into rats in order to investigate the influence of pregnancy interruption, compared with full pregnancy and lactation, on tumour development. Their study supported the theory that structural changes in breast tissue are responsible for the lasting, protective effect of full-term pregnancy. They observed that abortion left the rats highly susceptible to developing cancer, but that the aborted rats “were at the same risk as virgin animals treated with the carcinogen”. Over the next two decades, however, their findings would be cited repeatedly as evidence that pregnancy begins a process of breast change which, when stopped by abortion, put female rats (and thus humans) at greater risk of cancer than those who had never been pregnant.

The Russo study was published in 1980, and an international dialogue among researchers focused on the abortion-breast cancer question commenced in the British Journal of Cancer shortly thereafter. It began with a study conducted by Malcolm Pike and colleagues at the University of Southern California. Their objective was to determine the influence of both oral contraceptive use and early abortion, and, because oral contraceptives had only been in common use for just over a decade, they looked at younger women only—all 163 of their cases had been diagnosed with breast cancer by the time they were 32 years of age. Despite the methodological problems inherent in this study, it gained long-lasting attention because of its finding that women who had a first trimester induced or spontaneous abortion before their first full-term pregnancy (often abbreviated as FFTP) had almost a two and a half times greater risk of breast cancer than women who had no abortions.

A later pregnancy reduced that level of risk, and abortions occurring after the first trimester or after a FFTP carried no increased risk at all. Pike and his colleagues saw their study as the first to report a “substantial increase in risk” specifically associated with early abortion, and, in light of the growing abortion rate among young women in many countries, they predicted their findings would be “of major importance”. The gauntlet had been thrown down, and a British research team, already engaged in a long-term study of reproductive factors in breast cancer patients treated at London and Oxford hospitals, responded immediately to Pike's “provocative and worrying results”. But in contrast to Pike, they found no increase in risk associated with either oral contraceptive use or abortion, and attributed this outcome to differences in methodologies and the “effect of chance”.

Conflicting results, and conflicting interpretations of results, would characterize the abortion-breast cancer debate as it intensified throughout the decade. An international array of cancer and epidemiological journals published studies which variously demonstrated an increased risk, no evidence of risk, and even decreased risk of cancer following abortion. The interpretation of statistical findings was inconsistent as well, for a certain level of risk was considered significant by some and not by others—an issue still hotly debated in recent years, and part of the reason why some of the same studies have been used as supporting evidence by opposing sides in the debate. Some reported that miscarriage was associated with a lesser risk (or even a negative risk) in comparison with induced abortion, while a much publicized study of Connecticut women which included only spontaneous abortion found a 3.5 fold increase in cancer incidence in women who miscarried before their first live birth.

Positive findings also varied with respect to whether abortion posed a danger only if it occurred before the first full-term pregnancy, or only among women who never carried a later pregnancy to term, in which case there was often ambiguity as to whether or not women in the cancer group were being compared only with other nulliparous women. Not unexpectedly, researchers usually drew upon preceding reports which they considered significant in relation to their own findings, with some giving undue weight to studies involving very few women reporting induced abortions or glossing over the inconsistencies which emerge from a closer examination of the data. The great majority of these studies were retrospective, case-control studies, meaning that women were reporting their abortion histories after being diagnosed for breast cancer (rather than before, as in prospective studies). During the 1980s, few researchers openly considered the question of inaccurate reporting of past abortions by participants.

The interest in younger women and cancer, initiated by Pike, persisted throughout this period for a variety of reasons. Records of abortion were more readily available for this age group, and researchers were very conscious of the fact that more and more young women were choosing abortion. There was also some evidence, inconsistent and yet alarming, that breast cancer rates among young women were on the rise; this trend was reported in the US and Sweden, although not in the United Kingdom, and was apparently more pronounced among young African-American women in the US. This last factor attracted the attention of the epidemiologist Janet Daling at the Fred Hutchinson Cancer Research Center in Seattle, Washington, who had already published widely on cancer epidemiology and whose earlier research had examined the implications of abortion for future childbearing.

In 1987, she and her colleagues published a study of breast cancer incidence, in relation to age and ethnicity, in the western portion of Washington state. They observed a 22 per cent increase in breast cancer in women aged 25 to 44 over an eight-year period, a change which could not, they felt, be accounted for by earlier detection or an overall increase in breast cancer rates, as they reported (inaccurately, as they later discovered) a declining incidence for women aged 45 to 54. They found the greatest increase was among women in low-income urban areas (53 per cent) and among African-American women (118 per cent), which contradicted the longstanding association between breast cancer and white, middle-class prosperity. As a possible explanation for the overall trend, they pointed to changes in reproductive patterns in the post-war period; as they noted, among American women “born between 1950 and 1954, 48% were nulliparous at age 25 years, compared with 29% of women born between 1935 and 1939”. In addition, they pointed out that many young women now delayed pregnancy through the use of oral contraceptives and abortion, both of which were under suspicion as risk factors, and that abortion rates were higher among black women than white women.

Daling and her colleagues were awaiting the results of a long-term study focused on abortion and cancer in young women which would cause a considerable stir once it was released. Meanwhile, the findings of other researchers contributed to the growing confusion. A Swedish study reported that, although the increase in abortion rates had paralleled a 40 per cent increase in breast cancer among women aged 20 to 44, they found no association between the two, while the study by Holly Howe and her colleagues in New York state reached the opposite conclusion. As for the rising incidence of breast cancer among young African-American women, an examination of this trend in the San Francisco Bay area found that breast cancer had been increasing in the African-American population throughout the post-war period, well before the legalization of abortion. A team in the eastern US concluded that there was some association between abortion and cancer in African-American patients diagnosed after the age of fifty, but that spontaneous abortion seemed to offer a small protective effect among the same group of women.

By the early 1990s, the debate over the possible link between abortion and breast cancer was about to move beyond the pages of medical journals and into the public eye. This was not, as yet, because of a growing concern among clinicians; a study of physicians' perceptions of breast cancer risk conducted in southern California during 1991–2 revealed that none of those interviewed mentioned abortion among the twenty-nine potential risk factors listed. Instead, public awareness of the controversy would come in the wake of Janet Daling's new study, outlined below.

Like others, this study had its flaws, and would not have received mass media coverage and aroused the interest of anti-abortion politicians if it were not for two factors. Firstly, the alarming claim that there was an epidemic of breast cancer in the United States, especially among younger women, had been widely reported in the early 1990s and had become an intensely political issue. Secondly, the anti-abortion campaign had reached a stage in its increasingly violent history when new strategies were needed, and activists recognized that an association between abortion and breast cancer could be very useful to their cause.