Stop asking radiologists to rule out disease


When you get frustrated with my interpreting a chest x-ray as “atelectasis at the right lung base, pneumonia can’t be excluded,” trust me, I don’t enjoy it. But when you ask me to rule out pneumonia you leave me no choice but to tell you that pneumonia can’t be ruled out.

To rule out a disease a test must have a sensitivity of 100%, meaning there should be no false negatives. No imaging modality has a perfect sensitivity but the chest x-ray is nowhere near that perfection.

There are often blobs at the bottom of the lungs on a radiograph. In the vast majority these are areas of atelectasis, closure of parts of the lungs. Nearly all patients admitted in hospital have atelectasis. Here is the problem: it looks just like pneumonia. If I call one pneumonia I must call all pneumonia. This would mean that some patient somewhere is going to be put on Imipenem unnecessarily, develop pseudomembranous colitis all because of my interpretation. To reduce that possibility, I throw the ball back in your court by asking you to clinically correlate.

This is not good medicine. We can do better. You can tell me what you are actually thinking and I can tell you what I am actually seeing. Because when you tell me you really suspect pneumonia and I see that blob at the lung base, I will call it pneumonia, because I trust your clinical acumen.

When you don’t really think your patient has pneumonia, but just want to be extra sure because the patient’s temperature has marginally straddled beyond a threshold, and you want to feel you’ve done something by ordering a chest x-ray, be honest. Again, I trust your clinical judgment. I will call that patch atelectasis and won’t disclaim.

Better still, don’t order the test. Yes, you heard that right, don’t order a chest radiograph when you don’t really think the patient has pneumonia: fewer chances of a false positive. This would also mean that whenever you do order a chest x-ray or a CT scan a bulb will light in my frontal cortex, because I trust your clinical reasoning, and I know you are not the type to order tests frivolously.

But when you cry wolf, well you’ve heard the fable. But it won’t be you or I that will suffer, but the patient.

I am in the business of ruling in disease not ruling out disease. I am an adjunct to your clinical reasoning, not a substitute for it. I should mostly confirm your clinical suspicions, occasionally challenge them.

I am only as smart as the appropriateness of your imaging request. A diagnostic test is only as good as you make it. If you ask me to “rule out pulmonary embolism and aortic dissection, and whilst you are can you make sure he doesn’t have bowel ischemia and arterial clot,” my interpretation will read as if transcribed by a decerebrate pigeon. This is because I don’t know what you are thinking or not thinking. I have to assume the worst. My sensitivity rises, and specificity falls, and false positives abound.

Imaging findings are not binary: they are seldom all or none. They are a spectrum. There are shades of gray. Some of those shades are shared by both normal and diseased individuals. If I am forced to rule out disease I will either give lots of normal people disease or have to disclaim.

Help me help you by telling me truthfully your clinical reasoning. United we can be cleverer than Sherlock. By being divided and second guessing each other, we will lead to waste, over testing and poorer quality care.

Saurabh Jha is a radiologist.


View 18 Comments >

Most Popular

✓ Join 150,000+ subscribers
✓ Get KevinMD's most popular stories