Wednesday, June 17, 2015

Incidentalomas, ultrasound and diagnostic errors

Ultrasound, like any other form of diagnosis, is not perfect. Here is a sonographic image of a perirenal pseudocyst in a cat. The kidney is the bleb in the middle and the black stuff around it is fluid.

Recently a friend of mine has been through the emotional wringer over an “incidentaloma” – i.e. a benign lesion that is found, incidentally, when one is performing medical imaging (radiography, ultrasound, computed tomography etc.) for another reason. And no, I did not make up this word. Its even on Wiki. In my friend's case this lesion is located in an organ, cancers of which have been mentioned in the media as having a terrible, terrible prognosis. 

Incidentalomas, like their nasty malignant counterparts, have a way of popping up just about anywhere. But what is seen cannot be unseen, and the clinician must determine whether the lesion is indeed benign or clinically significant. 

Which can involve a pretty hefty work-up because, to use a phrase I think vet students should learn by rote, cancer is a tissue diagnosis. At least in veterinary medicine, you need at very least cells, and often a biopsy, to determine that a lesion is cancerous.

This is a pregnancy ultrasound. The patient is a carpet python and you can see the outline of eggs. Unfortunately this patient was egg-bound and her eggs were non-viable by the time of this examination. They were removed surgically and she made a full recovery.
The danger of an incidentaloma is that a clinician may jump to a conclusion – this is a cognitive error known as over-diagnosis. And when it comes to medical tests, errors can be made.

One example is abdominal ultrasound. Performing an abdominal ultrasound requires a lot of skill (according to one veterinary sonographer I spoke to, you need to have performed 1000 abdominal ultrasounds to approach competency). 

But even specialist sonographers don’t get it right every time.

In a study published in 2012, sonographic diagnoses were compared with the findings in exploratory surgery to determine if any diagnostic errors had been made. The ultrasounds were performed by experienced sonographers. They made an error 16.2 per cent of the time.

Not all errors are equal, in terms of either gravity or consequence. The authors of the paper classified errors according to whether they were

  • Perceptual – a matter of not seeing a lesion or abnormality in the first place;
  • Cognitive – either over-interpretation or misinterpretation of the findings, or flawed data collection (not getting a complete history, not performing a complete examination – e.g. “search satisfaction” or finding one lesion then assuming that is the only problem, lack of understanding of the patient’s underlying condition or conditions, or failure to consider all possible differential diagnoses). These could occur due to a rushed examination, poor patient preparation (e.g. not fasting or clipping the patient properly – alas, to perform a proper abdominal ultrasound one must permit one’s companion to wear the “ultrasound clip” – unless one is willing to mask it with the complete crew-cut);
  • Equipment related – equipment not working, not having the right probe, poor training, and poor supervision;
  • Inevitable or unavoidable errors – where the imaging findings of an abnormality were hidden, masked or absent so no one could be expected to know. Examples of unavoidable errors would be failure of ultrasound to diagnose adhesions, or failure to identify metastases. Often the only way to find these is to go to surgery. Similarly, gastrointestinal perforation is very challenging to diagnose on ultrasound and to try is perhaps to ask ultrasound too much. It’s more useful to go to endoscopy;
  • Multifactorial errors – a combination of some or all of the above.

Before you lose faith in ultrasound, this modality lead to a correct recommendation to take the patient to surgery in 83.8 per cent of cases.

Interestingly, most errors in this study fell into the cognitive category, suggesting that operators themselves need to be more thoughtful when interpreting the findings. For example, one dog diagnosed with a uterine infection – because of visualisation of a big fluid filled structure right next to the bladder – actually had a bladder diverticulum (at surgery it looked like an extra bladder stuck onto the back of a normal bladder). The sonographer in question didn’t consider a bladder diverticulum in their differential diagnoses in the first place, so they would never have come to that diagnosis. What are the odds of having a bonus bladder, you ask? Slim, but within the realms of possibility. That's why specialists often produced longer lists of differentials than new graduates. They know more, they've seen more of the odd cases, and they've been bitten before by leaving something a bit obscure off the list.

In another case, the ultrasound diagnosis was a mass in the tail of the spleen. But at surgery this was in the body of the spleen – the sonographer had failed to do a complete diagnosis and look at the whole spleen. The outcome for the dog was the same – it went to surgery and the mass (with the spleen) was removed.

The moral of the story is that we need to appreciate and accept the limitations of medical testing, and indeed of the operators and interpreters of such imaging.

If you want to read more about diagnostic errors, check out this post on Avoiding Errors in Medicine, or this one on How Doctors Think.

The tutor who shared this very thought provoking paper with me is Dr Cathy Beck, who co-teaches the sonology distance education course with Dr Karon Hoffman through the Centre for Veterinary Education.


Garcia DAA & Froes TR (2012) Errors in abdominal ultrasonography in dogs and cats. Journal of Small Animal Practice. 53:514-519.