Skip to content

Incidentally… avoiding the problem of incidental findings

A new study from the Mayo clinic in the United States points to a frequent problem in certain types of medical research. When healthy volunteers or patients with a given condition take part in research studies they may have brain scans, CAT scans, blood tests or genetic tests that they wouldn’t otherwise have had. These tests are not done for the benefit of the individual, they are designed to answer a research question. But sometimes, quite often according to the authors of this new study, researchers may spot something on the scan that shouldn’t be there, and that could indicate a previously undiagnosed health condition. These ‘incidental findings’ generate an ethical dilemma for researchers. Should they tell the research participant about the shadow seen on their scan? Do they have an obligation to reveal to a research participant that they have found them to carry a gene increasing their risk for breast cancer, or Alzheimer’s disease? There is much agonising by ethics committees, ethicists and researchers about the problem of incidental findings, but there is a simple way of avoiding the problem. Anonymise research databases and tests so that there is no possibility of determining which participant has the breast cancer gene, or the lump in their kidney.

There are two main problems for researchers from incidental findings. Firstly, the researchers may not be qualified to assess whether or not incidental findings are present, and whether they are serious or not. It may add extra expense to the costs of the study if researchers are going to have to employ a medical specialist to check all scans for incidental findings. Secondly it may risk harm to the research participant if researchers find an incidental finding. These findings may lead to unnecessary further tests or even surgery. It may cause significant anxiety in the research participant to inform them of the finding, even if it turns out to be benign.

The Mayo clinic study looked at almost 1500 ultrasounds, x-rays and scans performed for research reasons and reviewed the medical records of participants to see what action was taken in response to them. 40% of all research studies (567) had incidental findings. Of those studies with incidental findings 6% led to further tests, referrals, procedures or treatments. A quarter of these were judged by an expert panel to have led to clear benefit to the patient. However 2/3 led to no clear benefit to the patient, and 10% led to apparent harm. Overall, 1% of incidental findings were worth detecting from the patient’s point of view.

Is anonymisation the answer? This solution is sometimes suggested for genetic research. If we make it impossible for researchers to work out which participant is the one with the breast cancer gene, then it removes the need to worry about whether or not to inform the research subject. The ethical dilemma is effectively sidestepped. We could do something similar for brain scans or CT scans performed as part of research. 
There are two principal arguments in favour of anonymisation. Firstly, anonymising research tests does not harm research subjects, where the research would not otherwise take place. People who enter a research study have tests that they wouldn’t otherwise have done. If they didn’t enter into this research, or if they entered a different research study they would not have had the test and found out about this incidental finding (or at least perhaps not until some later point). Secondly, anonymising research tests avoids harm to research subjects from taking part in research. If the Mayo clinic studies had been performed anonymously 6 individuals would have missed out on having a problem detected that would be good to detect. However, 571 incidental findings of no benefit to the individual would not have been detected. 29 individuals would have avoided unnecessary or harmful tests, referrals and/or treatment.

Of course, one reason to not anonymise research tests is if research participants want to be told if there is something possibly abnormal on their scan or on their genetic test. Where participants are aware of and consent to the risks (of unnecessary tests, anxiety from untreatable conditions or non-pathological findings), and where it is possible to do so, there is nothing wrong with providing the results of scans or tests to participants and their doctors. It may be that the possible benefit (of detecting a hidden cancer or aneurysm) is one of reasons why participants volunteer to have a brain scan. However, there is a reason why MRIs or CTs or genetic profiles are not performed routinely in healthy individuals to screen for unsuspected health conditions, these are not useful or cost effective forms of health screening. We should not force researchers to perform ineffective or usually non-beneficial forms of health screening just because they can.

Medical Research Images Frequently Reveal Unrelated Abnormalities in Study Subjects: Scientific American 2010

Arch Intern Med — Incidental Findings in Imaging Research: Evaluating Incidence, Benefit, and Burden, September 27, 2010, Orme et al. 170 (17): 1525 

The Ethics of Scan and Tell: Scientific American June 2005

Incidental findings on brain magnetic resonance imaging: systematic review and meta-analysis — Morris et al. 339 —BMJ 2009

Share on

1 Comment on this post

  1. I can’t agree with your proposition completely on anonomysing the results. You touch upon my argument only very slightly in your last paragraph.

    I think if the researchers spend the money doing those tests and ‘incidentally’ find something then they must tell the patient. From the ethical point of view anonymising will mean sidestepping this problem from the researchers perspective but it doesn’t bode well from a utilitarian perspective.

Comments are closed.