The findings of a study examining 39 years of breast cancer death rates have hit the headlines, with The Guardian reporting that, 'Breast cancer screening not shown to reduce deaths.' The value of breast cancer…
The findings of a study examining 39 years of breast cancer death rates have hit the headlines, with The Guardian reporting that, 'Breast cancer screening not shown to reduce deaths.' The value of breast cancer screening has been the subject of debate for many years. Every time it seems the question has been settled – as some presumed was the case after the publication of a 2012 review into screening – new evidence emerges that reignites the debate.
The latest study by researchers at Oxford University found that declines in mortality rates over time were highest in women under the age of 40, who are not normally invited for screening. The researchers also found significant downward changes in trend in women aged between 50 and 64 years old, the age group screening is targeted at.
But this change occurred in 1979 in Oxford and in 1990 throughout England, with the researchers concluding that better treatment for breast cancer is behind the trend, not screening programmes. This is because the downward trend began either before screening was introduced or too soon after the introduction of screening for it to have an effect.
Breast cancer screening is an extremely complex topic and it is difficult to assess the value of screening programmes. It may be possible the benefits of screening have been obscured by other risk factors and improvements in treatment. It is hoped the picture will become clearer as more evidence becomes available.
Where did the story come from?
The study was carried out by researchers from the University of Oxford and was funded by the National Institute for Health Research in the UK.
It was published in the peer-reviewed Journal of the Royal Society of Medicine.
The results of the study were reported well by the media.
What kind of research was this?
This study was a time-trend analysis of mortality (death) data in England to see whether breast cancer screening using mammography reduced deaths from breast cancer.
As this is an observational study, it is possible the benefits of screening programmes may be obscured by changes in both treatment and risk factors that have occurred over time.
Ideally a randomised controlled trial would be performed to assess the benefits of a screening programme. However, it is unlikely that any new randomised controlled trials of breast cancer screening in the UK will be performed.
To perform a randomised controlled trial, women would have to be prepared to be randomised to screening or to no screening. As there is currently a national screening programme in place, it is unlikely that enough women would be prepared to possibly forgo screening.
What did the research involve?
Researchers analysed the number of women who died from breast cancer in the Oxford region between 1979 and 2009. They focused on data from this region because all causes of death are mentioned on death certificates there, not just the underlying cause of death.
The researchers wanted to try to exclude the possibility that ambiguity about the underlying cause of death or changes in reporting practices distort the true picture. A total of 20,987 death certificates where female breast cancer was noted were included.
The researchers also analysed the rate of death from breast cancer between 1971 and 2009 for the whole of England, where only the underlying cause of death is reported on the death certificate.
Researchers compared trends in the rate of death from breast cancer before and after the English National Breast Cancer Screening Programme was introduced in 1988. Three groups of women were included for the same time period:
- women who had been screened once
- who had been screened several times
- unscreened women
The researchers used a statistical technique called joinpoint analysis to estimate the years in which trends changed. Joinpoint analysis makes use of specialist statistical software to track trends over time. Each joinpoint corresponds to the estimated location of a change in a trend – in this case, mortality.
What were the basic results?
In the Oxford region, of the women with breast cancer mentioned on their death certificate, breast cancer was the underlying cause of death in 96% of women aged under 65 at death, 88% of women aged between 65 and 74, 78% aged between 75 and 84, and 66% of women aged 85 or older.
Trends for breast cancer-related deaths were very similar for whether breast cancer was listed as the underlying cause or whether it was mentioned on the death certificate. This suggests that it is unlikely that changes in death certification practices or changes in the rules for selecting the underlying cause of death affect the change in deaths due to breast cancer over time.
For all ages combined, death rates peaked in 1985 (both when breast cancer was the underlying cause and when breast cancer was mentioned) and then started to decline. This occurred prior to the introduction of the screening programme in 1988.
Between 1979 and 2009, for deaths due to breast cancer as the underlying cause, rates declined uniformly (without a detected change in trend over time):
- for unscreened women aged 40-49 there was a decline of -2.1% per year, and
- for screened women aged 50-64 there was a similar decline of -2.1% per year
There was also a significant change in trend downward in deaths caused by breast cancer in 1987 in women aged 65-74, and in those aged 75 years or older in 1989. These changes occurred before the screening programme was introduced, or before it was likely to have had an effect.
Between 1979 and 2009, rates of breast cancer mentioned on the death certificate also declined uniformly in women aged 40-49 (unscreened) and women aged 50-64 (screened). There was a significant downward change in trend in breast cancer deaths among women aged 65-74 in 1990, and among women aged 75 years or older in 1996.
In England, the first estimated changes in trend occurred prior to the introduction of screening, or before screening was likely to have had an effect (between 1982 and 1989). A second downward change in trend occurred in 2001 in women under the age of 40 (who are not routinely screened) and in 1990 in women aged between 50 and 64.
Most significantly, there was no evidence that declines in mortality rates were consistently greater in women in age groups and cohorts that had been screened, or screened several times, compared with other unscreened women in the same time periods.
How did the researchers interpret the results?
The researchers say that, "mortality statistics do not show an effect of mammographic screening on population-based breast cancer mortality in England."
This study of rates of death caused by breast cancer over a 39-year period has found no evidence of the benefits of breast cancer screening. Age-specific mortality rates for women aged between 40 and 49, 50 and 64, and 64 and 74 years peaked prior to the introduction of breast cancer screening in 1988. Declines in mortality were greatest in women under the age of 40 and smallest among women aged 75 years or older.
The researchers found that there were significant changes in downward trend in women aged between 50 and 64 years old – the age group screening is targeted at – but that these occurred in 1979 in Oxford and in 1990 in England. Both of the changes occurred before the introduction of screening, or too soon after the introduction of screening for it to be likely that screening caused the change.
In addition, significant declines in mortality rates per year were seen in women aged under 40, who would not normally be invited for screening.
As an observational study of population level data, several points are worth noting:
- Direct comparisons of individuals who were screened with those who weren't is not possible with this type of study design. Researchers were only able to compare mortality for women in age groups that were likely to have been screened with those who were unlikely to have been screened.
- The results do not rule out a benefit at the level of individual women, but the effect is not large enough to be detected at the population level.
- "Secular" effects – that is, effects that occur over time independently of screening – can obscure the screening effects. For example, the effect of better drug treatments or changes in risk factors such as childbearing patterns over time might have outweighed smaller improvements thanks to screening.
Overdiagnosis is when people receive unnecessary cancer treatment for a cancer that would not otherwise have been diagnosed. For every death prevented by screening, there are estimated to be three cases of overdiagnosis.
This study provides additional valuable population data to inform the breast cancer screening debate. There is a great deal of information on both the pros and cons of screening. The 2012 review into breast cancer screening estimated that for every 10,000 women invited for screening from the age of 50 for 20 years:
- 43 deaths from breast cancer will be prevented
- 681 breast cancers will be diagnosed
- 129 of these diagnoses will be "overdiagnosed"
Analysis by Bazian. Edited by NHS Choices. Follow Behind the Headlines on Twitter.