Abortions don't cause cancer any more than parties do
Thursday, Aug 28, 2014, 04:57 AM | Source: The Conversation
By Louise Keogh
Abortions don't cause cancer any more than parties doLouise Keogh, University of Melbourne
The purported link between abortion and breast cancer is based on research that's no longer accepted as valid because its methods are so flawed. But that hasn't stopped politicians such as Fred Nile and groups such as the World Congress of Families, a global conservative Christian group that claims to be a defender of the so-called "natural family", from repeating it.
The association stems from research from the 1950s and has been used to publicise and bolster the conservative, anti-choice position on abortion. Numerous articles published since the 1950s that fail to confirm any such association between abortion and breast cancer seem not to have dented the enthusiasm of those repeating it. Nor have they reduced the newsworthiness of this early work.
Historian Patricia Jasen suggests that, of all cancers, breast cancer has been the focus of the most controversy regarding its cause. At one stage, a "blow to the breast" was thought to be the cause, at another it was "unhappy emotions".
The current consensus among experts is that the strongest risk factors for breast cancer are age, a family history of the cancer, and breast conditions, such as previous disease or high breast tissue density. Reproductive factors also play a role, as do certain lifestyle factors, but these are minor.
The 2009 Cancer Australia report summarising the data explicitly states that abortion has been shown to not impact the risk for breast cancer.
But despite abortion not appearing among the many risk factors cited by trusted cancer organisations, there are some studies that propose this potential link. And it is to these studies that conservative politicians and anti-choice activists keep harking back.
Parties cause stomach cancer?
The original paper reporting an association between breast cancer and abortion is an epidemiological study on cancer in Japan, from 1957. It's often cited by people promoting the purported link.
Let's look at it closely to highlight the problems of such work.
In this research, Segi and colleagues gathered information on all cancer patients in 33 hospitals across Japan between 1948 and 1952. They also collected information on a similar number of people attending health centres in Japan, and who did not have a diagnosis of cancer. This latter group were known as the controls in the study.
They compared the two groups based on their responses to a large number of questions. And analysed each risk factor independently of the others.
They found men who had "much opportunity to participate in parties" were more likely to have stomach cancer than controls, while people living in rural areas "using large quantities of soy sauce" were less likely to have it. For lung cancer, they found no difference between cancer cases and controls who smoked tobacco.
The authors were at pains to explain their findings were only observations, and that further research was needed to determine whether any of the associations they found were real.
For their findings on breast cancer, they considered a large range of variables, including responses to "is the size of the right and left breast equal or not? If not, which size is the larger?"
They found women with breast cancer were later in marrying and having children, that they had fewer children, were more likely to have had mastitis, and had menstruated earlier. They also found that "spontaneous interruption of pregnancy" and "artificial interruption of pregnancy" was higher for women with breast cancer.
With regard to this later findings, the authors claimed to be:
rather hesitant … in inducing some definite conclusions from them, because we know that our approach is not necessarily satisfactory.
They also said they didn't trust the accuracy of abortion reports by controls. This proviso reflects the findings of one of the largest studies (83,000 women from 53 studies) to address this association.
Published in 2004, that research found studies in which participants self-report previous abortions feature a strong recall bias. And it found women with cancer were more likely to report a previous abortion than controls.
The 1957 study was no doubt a bold endeavour for the time, but the approach it used is no longer considered appropriate. Many of its findings are known to be anomalies resulting from flawed methodology and the poor quality of some of their data.
It's not the only case-control epidemiological study to produce findings that were later overturned. Epidemiologists are now more aware of the flaws in these methods, including recall and selection bias, and analysis methods that can produce obscure findings.
These less than-perfect studies have led to improvements in how epidemiology is conducted because science is not static. New research about breast cancer is published every day and research methods are constantly adjusted and improved to take account of the latest findings.
The public needs to be able trust the scientific process, but when politicians and groups like the World Congress of Families selectively choose research that supports their position and promote these at the expense of the best and most current evidence, they damage science, people's trust in politicians and in this case, women's access to correct information about their bodies and risks to their health.
The question for us now is not whether there's a link between breast cancer and abortion but, when will women's health cease to be a political football, and their right to correct information be prioritised ahead of the conservative agenda.