A closer look at 'incidental findings'

Provisions needed for informing patients

As more sophisticated imaging technologies are used in research, investigators and IRBs must grapple with an unintended side effect – an increase in incidental findings (IF), or new health data unrelated to the study that is revealed about participants.

Common incidental findings turned up by X-rays, MRIs and CT scans can range from minor issues requiring no follow-up action to potentially life-threatening conditions such as a fast-growing cancer or aneurysm. But it's hard for researchers and IRBs to plan to manage such findings without knowing more about their frequency and severity.

In an effort to better understand what to expect from incidental findings in imaging studies, a team of physicians, radiologists, bioethicists and IRB specialists at the Mayo Clinic in Rochester, MN, tracked and analyzed data from three months worth of research imaging exams conducted at the hospital.

They looked at the IFs generated by more than 1,400 imaging exams — tracking follow-up studies, final diagnoses and interventions. The result: Nearly 40% of all exams turned up at least one IF, and nearly 20% had multiple IFs. Results were published recently in the Archives of Internal Medicine.1

But Nicholas Orme, MD, an internal medicine resident at Mayo and lead author of the study, says it's important to look beyond that initial, startling number.

"Incidental findings are common, but they're not commonly significant enough to cause clinical inquiry," he says. "So even though 40% of our research subjects had an incidental finding, only about 6.2% of those went on to have further clinical action."

In cases where there was clinical action, he says, the eventual benefits to the participants were often unclear, or worse, the procedure proved to be an overall burden. A mass might be investigated and found to be benign, causing extra expense, discomfort and stress to the participant in the process.

But in a handful of cases, the IF turned out to be the first clue of an extremely grave condition, such as a brain tumor, necessitating life-saving treatments.

"In three months, there were six cases where there was clear medical benefit — where something was picked up incidentally and then went on to have a significant positive impact," Orme says. "You can't just say, 'Oh, we can't worry about these things because they're so uncommon,' because when they do happen, they can be a big deal for the patient."

Breaking down the data

Orme says that part of the value of this study is that it breaks down IFs by the mode of imaging and area of the body, showing that certain types of imaging are far more likely to produce serious IFs that require follow-up.

As a result, researchers and IRBs can look at a proposed imaging study and estimate how likely it is to generate serious IFs. For example, 9.2% of all the CT scans done of the abdomen and pelvis during the three months of the study resulted in an IF that required further action.

"That's really quite high," Orme says. "If you're going to scan 300 people with this CT abdomen/pelvis (in a study), we may have 30 people who need further workup.

"So the IRB and the research protocol should determine what they are going to do. What's the plan? How quickly are we going to follow these scans up? How quickly are we going to refer them? Are we going to communicate with the primary care physician? There should be a plan in place, because that's going to come up in that study."

On the other hand, the Mayo study found fewer incidental findings from knee X-rays, with almost none requiring follow-up. So in protocols requiring knee X-rays, the IF planning wouldn't need to be as extensive, Orme says.

William Tremaine, MD, director of Mayo's Office of Human Research Protection, who was part of the panel that evaluated the incidental findings for this study, agrees that the results should aid IRBs in modulating their response to the potential for IFs.

"Certainly it was the abdominal and pelvic CT and to a lesser extent the chest CT and head scans that carried the biggest risk," Tremaine says. "I think IRBs would want to see if there are clinically validated studies used in part of the research. And if those studies carry a substantial risk for identifying important unexpected clinical findings, then provisions should be made for informing the participants and dealing with those."

Both he and Orme say that the informed consent for these types of imaging studies must lay out not just the possibility of incidental findings, but the consequences that could flow from them — extra tests, potential side effects and expenses to the participant.

"If there's a spot that's found on the lung and it will require repeat CT scans every several years, depending on how things are set up on this particular study, it may be on (the participant's) own dime, " Orme says. "That's something that they should probably be aware of."

When to review scans?

Since 2003, the policy at Mayo has been to have a trained staff radiologist examine all research images for potential IFs the same day the scans are administered. Orme says he knows that kind of commitment requires resources that not all institutions have.

"But I think this does bring up issues that people should be talking about," he says. "In some places, these scans are taken and then they're not read until a year later when the radiologist on the team has time, and that's a window of opportunity. If you do see a brain cancer that you could resect, a year later it's now metastatic brain cancer."

Tremaine says research scans should be reviewed within the same time frame as the clinical standard of care for the institution.

"For Mayo, that's the same day, but for some places, it might be the same week, or the same two weeks," he says. "If you're using a clinically validated study, it should be held to the same standards as that same test being done for clinical purposes."

Orme says one of the strengths of this study was the diverse team of experts assembled to analyze the results.

"We had a former head of our IRB, we had bioethicists, we had a lawyer, we had all these people who saw it from different angles," he says.

And Tremaine was happy to have the IRB heavily involved in the project.

"I was thrilled to see that the IRB was included in the research process as well as just the research oversight," he says. "I think there's a tremendous opportunity for that. Research deals with crunching data, and the IRB has lots and lots of data at every institution, so it's a huge opportunity for participation in research, not just oversight of research."


  1. Orme NM, Fletcher JG, Siddiki HA, et al. Incidental findings in imaging research: Evaluating incidents, benefits and burden. Arch Intern Med2010;170(17):1525-1532.