It is a truism in radiology that the more we image the more we will find. Some of it will help to advance the patient’s health, but much of it won’t

Two fictional friends, Mick and Mack, both in their mid-fifties, had MRI scans for back pain. Mick’s scan showed a little “wear and tear” and a small disc prolapse, but nothing warranting intervention at present if his symptoms could be managed in other ways.

Mack’s scan showed very similar findings, including a small disc prolapse, but he was also told that there was a lump in his left kidney. After further scans, an inconclusive biopsy and a lengthy discussion, he decided it was best to have the kidney removed. The lump in his kidney turned out to be an oncocytoma—a benign tumour often hard to distinguish from cancer.  The surgery was complicated by a chest infection resulting in a prolonged stay on ITU, following which he developed chronic lung damage. Recuperation was slow and in the following winters he suffered with a series of further severe chest infections, culminating in an episode of overwhelming sepsis from which he died, almost five years to the day from his MRI scan.

The two friends stayed in close touch and during one of his visits to Mack in hospital, Mick mentioned in passing that his back pain—which had improved for a while—seemed to have come back with a vengeance. The MRI scan had shown a prolapsed disc so he thought that was probably the explanation. Then one morning he found that he couldn’t move his legs. He was found to have a tumour in his left kidney which had spread to his spine, causing spinal cord compression. His cancer responded poorly to chemotherapy and after a few weeks of treatment he developed an episode of overwhelming sepsis, from which he died in the same week as his friend.  Looking back at his original MRI scan, the radiologist was able to identify the earliest signs of a tumour in his kidney at that stage. There was considerable discussion as to whether it could or should have been identified at the time.

Overdiagnosis and underdiagnosis are two sides of the same coin, an unavoidable trade-off. Perhaps an ethicist could decide which of the two men suffered the greater wrong—the one who died following a false positive diagnosis or the one who died after a false negative. As a radiologist, and having witnessed variants of both scenarios, I feel about equally distraught in either case.

The term “incidentaloma” was coined to describe the proliferation of benign adrenal nodules which improvements in computed tomography (CT) imaging made apparent. In common usage, this term has come to imply that incidental findings are tiresome and generally harmless, but this is not always true.  It is already the case that in some populations the commonest route to diagnosis of renal cancer is as an incidental finding on imaging carried out for another purpose.   

Radiologists in general have an ambivalent attitude to incidental findings, at least when not directly incentivized to detect and report them. We know that serious harm and even death can come to patients from the pursuit of radiological abnormalities which carry a very small likelihood of affecting the patient’s future health. On the other hand we pride ourselves on our ability to consider the whole image and not just the obvious findings. The ability to identify the incidental lung cancer on a radiograph showing a dislocated shoulder is one of the factors which marks us out from the common herd of healthcare professionals. Detecting the subtlest signs of pneumonia on the chest X-ray of a child with a fever is no big deal, but identifying the signs of aortic coarctation on the same image is the stuff of legend.

But these are muddy waters. For something to be an incidental finding, the primary purpose of the examination must be clear and this is often no longer the case. When the clinical question on a CT scan request is succinctly phrased as “?pathology”—as is now not uncommon—who’s to say which findings are incidental and which are not? Moreover, as we image more and more people, for some of our tests the value of the “incidental” findings may start to exceed the value of the primary purpose, even when that primary purpose is made clear.

Perhaps we have already reached that point for some tests—perhaps, as the indications for spinal surgery in patients with low back pain become fewer, an MRI scan of the lumbar spine should no longer be regarded as an investigation of back pain, but as a screening test for renal cancer? If so, perhaps we should change the way in which we scrutinise the images—and maybe we ought to tell the patients?  Framed like that, we will miss fewer renal tumours, but probably more lateral disc herniations. Yes, of course it depends on the prevalence of both conditions, but it’s not immediately clear (at least to me) where the greater benefit would lie.

It is a truism in radiology that the more we image the more we will find. Some of it will help to advance the patient’s health, but much of it won’t. At a time when the general approach often seems to be “Scan because you can,” perhaps I could suggest a small amendment: “Think before you scan because you can”.

Giles Maskell is a radiologist in Truro. He is past president of the Royal College of Radiologists.

Competing interests: None declared.