The effect of screening – what are we actually measuring?

Women who regularly have dental check-ups have a twenty per cent lower risk of dying from breast cancer – regardless of whether they are screened for breast cancer or not. Another curious finding is that women who regularly show up for breast cancer screening reduce their risk of being beaten to death, dying in an accident or killing themselves, by fifty per cent.

This is the conclusion of a new study in which researchers from Aarhus University have examined the value of the comparison groups that are included when measuring the effect of a national screening programme – in this case mammography screening for breast cancer in Danish women over the age of fifty.

The research results have just been published in the International Journal of Epidemiology, and according to one of the authors, Associate Professor Henrik Støvring from the Department of Public Health, they should sound a warning to other researches about over-optimistic conclusions about effect.

Does the comparison group stand up to comparison?

“We ought to be almost paranoid about our conclusions when it is not clear that the comparison group is a good match to compare to. It might sound like common sense, and all professionals are aware of the problem, but it’s nonetheless common within the field of screening to use ‘non-participants’ as a comparison group - of course, because there often aren’t other possibilities,” says Henrik Støvring.

He points out that studies from abroad showing the strongest effect of breast cancer screening have in common that they make a comparison with ‘non-participants’.

"Seen in this light, the study also provides an opportunity to ask the key question about whether the observed effects occur because screening is working, or if something else is at play. In any case, we’ve published a study which confirms that this ‘something else’ plays an indisputable role," he says.

The starting point for the research group’s study were 37,000 Danish women aged 50-52 who were collectively cleared of having breast cancer at their first possible breast cancer screening in the Copenhagen area or on Funen in the period 1991-2001. The research group took the 37,000 women who were cleared and started a ‘timer’ before dividing the women into two groups that they then followed until 2014: One group was those women who participated in the subsequent rounds of breast cancer screening, while the other was the women who failed to turn up to the screenings they were offered for various reasons.

The groups do not look after themselves as well

The study documents that those who failed to turn up have a general mortality level which is twice as high when compared with the screening participants, as well as a two-and-a-half times greater risk of dying in accidents, self-harm or as the result of being attacked. In other words, it is not the ‘same type of women’ being compared to one another, as Henrik Støvring points out.

“The comparison group is generally more exposed and its behaviour appears to be unhealthier. Maybe some of them are struggling with forms of abuse or mental disorders, which would be plausible grounds for not reacting to an appointment for a preventative mammography. On the other hand, it appears that women who do attend the breast cancer screening also look after themselves a little better in other areas and may therefore also have a lower underlying risk of dying from breast cancer,” he says.

The latter is supported by the part of the research result which shows that women who regularly have dental check-ups have a twenty per cent lower risk of dying from breast cancer. This is true regardless of whether these women are screened or not.

According to Henrik Støvring, the research results are not only interesting in the field of screening, but also for other observational studies in which researchers can easily be fooled if the comparison group differs in unknown ways from the group who make use of the healthcare service being studied.

This has been demonstrated by the American researcher Alan Brookhart, who has been a source of inspiration to Henrik Støvring. A few years ago, Brookhart documented that patients who were careless with their statins – a medicine for elevated cholesterol levels – die more frequently in traffic accidents than patients who have taken their medicine regularly over a number of years.

The research results - more information

• The study is a registry-based study which utilises an epidemiological analysis tool known as negative controls.
• Henrik Støvring is the senior researcher behind the project, while Postdoc Mette Lise Lousdal from the Department of Public Health is the lead author (she is presently on maternity leave).
• The research project is funded by the Danish Cancer Research Foundation, Fru Astrid Thaysens Scholarship for Medical Basic Research and Aarhus University.
• The scientific article Negative controls to detect uncontrolled confounding in observational studies of mammographic screening comparing participants and non-participants is published in the International Journal of Epidemiology.

Contact

Associate Professor,  DMSc, PhD Henrik Støvring
Department of Public Health at Aarhus University
Email: stovring@ph.au.dk
Mobile: (+45) 2233 5586