A game-changer on abortion & breast cancer? You make the call

The long-standing controversy over the link between therapeutic abortion and breast cancer found its way onto unexpected territory—the Globe and Mail website—on Friday. The pro-life movement has long been quarrying the epidemiological literature for the smoking gun of what it calls “ABC“. This is what pro-lifers ask Santa for Christmas, or wish for when they see a falling star: that abortion will turn out to carry previously unsuspected harms which might become the pretext for outlawing it completely, for imposing severe restrictions on it, or, at the very least, for stigmatizing it like tobacco and allowing clients to receive a scary mandatory lecture on cancer risk in the name of informed consent.

Thus far, science hasn’t been much help to them. ABC is a tricky topic because there are confounders in the picture: in general, spending less time pregnant (and more time menstruating) gives women a slightly greater lifetime risk of breast cancer. Abortion probably does increase breast cancer risk insofar as it eliminates one pregnancy—just as being able to drive increases one’s risk of ending up with shards of windshield glass under one’s eyelids.

Whether abortion imposes a distinct burden of cancer risk is another question, one much harder to answer. Occasionally a study will turn up that suggests it might. And that’s what has happened now. Gloria Galloway writes:

Three years ago, [Saskatchewan MP Maurice Vellacott] helped to bring an American doctor and activist to Parliament Hill to tell Canadian women that abortion increases the risk of breast cancer. It turned out that the doctor, Angela Lanfranchi, was speaking from a defined religious point of view that had little apparent basis in science.

And, at the time, the link between the procedure and the disease had been discounted by the National Cancer Institute in the United States, the Society of Obstetricians and Gynaecologists of Canada (and their U.S. counterparts), as well as the Canadian Cancer Society and the Canadian Breast Cancer Network.

But a study released last fall (available here but only for a fee) by the respected Fred Hutchinson Cancer Research Institute in Seattle by a number of distinguished cancer experts including Louise Brinton, the chief of the Hormonal and Reproductive Epidemiology Branch of the National Cancer Institute, lists induced abortion as being “associated with an increased risk for breast cancer.” Background documents further suggest that it increases the risk of the disease by 40 per cent.

An e-mail to Dr. Brinton on Friday was returned by an Institute spokesman named Michael Miller who said: “NCI has no comment on this study. Our statement and other information on this issue can be found at http://www.cancer.gov/cancertopics/ere.” …Requests for an explanation of the apparent discrepancy between that position and the information contained in the study released last spring went unanswered by NCI.

I visited the library to double-check whether Galloway had characterized the study’s findings correctly. The data behind the study come from a breast-cancer surveillance project in the Seattle area that included interviews with 897 women who had suffered invasive breast cancers before the age of 45. Here’s the part that’s the cause of all the excitement—a line in a table of odds ratios for “known and suspected risk factors among women 45 years of age and younger”:

Odds-ratio table from "Risk Factors for Triple-Negative Breast Cancer in Women Under the Age of 45 Years"The odds ratios were derived by adjusting for age, family history of breast cancer, lactation history, and duration of oral contraceptive use: the double dagger indicates that only women who had been pregnant at least once were included in the “never” row under the “Abortion” heading, so the statistically significant 40% apparent increase in background risk actually leaves never-pregnant women out of the background completely. This is notable, especially given that the study is population-based (the authors boast that it is the “largest of its kind”; their goal was not just to measure breast-cancer risk but to differentiate between etiologic subtypes of breast cancer).

On the other hand, it’s not that notable. If you look at the raw numbers, you’ll see that the randomized control group of 1,569 Seattle-area women with no history of breast cancer broke down between “Never [had an abortion] and “Ever” pretty much the same way that the breast-cancer victims did. Most of the “40%” extra risk, in other words, is the product of statistical adjustments, and may, in part, be attributable to confounding variables that weren’t controlled for. Income wasn’t controlled for, and as you can see in the table itself, it might make a difference; neither was obesity. And 40% is not a big number in epidemiology. In general researchers don’t get worked up about an odds ratio until it is at least 2.0, and it is seen over and over again in multiple studies.

Galloway is, frankly, not being careful enough when she describes the study as implying that abortion “increases the risk of the disease by 40 per cent.” This study is strictly about breast cancer in women under 45—a small fraction of all breast-cancer cases (though, to be sure, it is a fraction that is of special concern). In no way can it provide justification for any statement about overall lifetime breast-cancer risk.

Moreover, there is really no “discrepancy” between the NCI’s stated position on ABC and this particular study. Here’s what the NCI says officially:

The relationship between induced and spontaneous abortion and breast cancer risk has been the subject of extensive research beginning in the late 1950s. Until the mid-1990s, the evidence was inconsistent. Findings from some studies suggested there was no increase in risk of breast cancer among women who had had an abortion, while findings from other studies suggested there was an increased risk. Most of these studies, however, were flawed in a number of ways that can lead to unreliable results. Only a small number of women were included in many of these studies, and for most, the data were collected only after breast cancer had been diagnosed, and women’s histories of miscarriage and abortion were based on their “self-report” rather than on their medical records. Since then, better-designed studies have been conducted. These newer studies examined large numbers of women, collected data before breast cancer was found, and gathered medical history information from medical records rather than simply from self-reports, thereby generating more reliable findings.

Although the new Seattle study is large and features randomized controls, it too is a retrospective, questionnaire-based study, reliant on self-reporting; it does not meet the gold standard for epidemiological evidence. The NCI has no reason I can see to change, or apologize for, its position.