Monday October 8 2007
Breast cancer cells dividing
The “rising rate of abortion will lead to a sharp increase in breast cancer cases,” reported the Daily Mail. There are predictions of “a dramatic rise in breast cancer among those who have abortions before giving birth to their first child”, it said. The newspaper calculates a doubling in breast cancer rates from around 39,000 per year in 2004 to more than 65,000 per year in 2025, “with rising abortion rates a major reason”.
The story implies that the study looked at individual data to come up with this association. However, this is not the case. The story is based on a study that used a mathematical model that was developed to forecast the number of breast cancer cases in the future. The model was used to assess how fertility factors (including abortion rates) were linked to breast cancer rates; it was not set up to study if these factors cause breast cancer. This is an important distinction. To date, there is no evidence for a causal relationship between breast cancer and abortion in individuals.
Where did the story come from?
Patrick S. Carroll developed this model and published this research. The research was funded by LIFE and The Medical Education Trust. It was published in the peer-reviewed medical journal Journal of American Physicians and Surgeons.
What kind of scientific study was this?
Nationwide data from cancer registries and other sources in eight European countries were used to develop a mathematical model that examined the link between breast cancer rates and certain fertility factors (including the abortion rate). The author uses the model to predict the number of breast cancer cases in all eight countries, including the number of breast cancer cases that would be seen in 2025 in England and Wales.
The researcher assessed whether there was a link between breast cancer rates (in women aged 50 to 54 years) and abortion rates in England and Wales (using data collected since 1926), fertility rate, average age at first birth, and childlessness. There was no data from England and Wales available for the other factors in the Europe-wide study that were thought may have an association with breast cancer; breastfeeding, hormonal contraceptive use, hormone replacement therapy, so these were excluded from the model.
In order to test the accuracy of the model, it “predicted” known information: the number of breast cancer cases seen in 2003 and 2004. The model predicted slightly more of the cases in 2003 (100.5%) and slightly less of the cases seen in 2004 (97.5%), suggesting that it is an accurate way to predict breast cancer cases.
What interpretations did the researchers draw from these results?
The researcher concluded that of the potential risk factors related to fertility for breast cancer in women, the rate of induced abortions and the fertility rate in a population are useful predictors of the rates of breast cancer. The model based on these factors is used to forecast the number of breast cancer cases in England and Wales in 2025, forecasting an increase to 65,252 cases.
What does the NHS Knowledge Service make of this study?
- You may be led to believe from the newspaper that this study offers conclusive evidence that abortion causes breast cancer. This is not the case. Other real-time studies (i.e. those that have examined individual women and their risk factors) have not found a link.
- This modelling study assessed which of the fertility risk factors is most strongly linked to the number of breast cancer cases seen in England and Wales. When developing such models, researchers start with some idea of the potential factors that could be affecting breast cancer rates across a population, collect data on them and then see whether there is a link between that factor and risk of breast cancer. For England and Wales, data was only available on four of the seven fertility factors which are believed to be linked to breast cancer.
- As the researcher says, there is a complex relationship between socioeconomic status, risk of breast cancer and risk of abortion. Socioeconomic status is a confounder variable. Women in England and Wales from higher socioeconomic classes have more breast cancer incidences and are also believed to have a greater “preference for abortion when pregnant”. They are also more likely to have their first children at a later age (another potential risk factor for breast cancer). The model did not take into account socioeconomic status and so could not examine this relationship. The researcher states that “if abortions had been examined in the studies of this social gradient, the role of this factor could have been made clear”.
- Although this modelling study has highlighted an association between abortion rates and breast cancer rates, the researcher does not imply that this is a causal relationship. There are other complex factors including family history, height and weight, alcohol intake, exercise, diet and smoking which are definite breast cancer risks.
- The model here appears to be accurate in predicting breast cancer rates and it can be tested against actual rates each year. Of course, the model cannot take into account the potential changes in other risk factors (mentioned above) that may change over time.
It seems unlikely that this data will change an individual’s decision about abortion but it may help public health researchers to untangle the complex relationships between risk factors and socioeconomic markers.
Sir Muir Gray adds...
Models can be used to highlight issues that need further study and may be used to represent reality, but real life data usually needs to be studied to work out whether A causes B.