Saturday, February 16, 2013

Clinical Medicine Goes to the Dogs: Poop of Principle


Cliff, the Clostridium Sniffing Beagle

In WTNK, I was skeptical that trained dogs would be of much practical use in sniffing out disease. Since then, a steady trickle of studies has claimed that dogs can smell cancer of the lung, colon, ovary and prostate. This doesn’t strike me as far fetched. However, the elaborate training and housing of cancer dogs along with laborious preparation of test samples make it an expensive and cumbersome proposition. A recent review of the scientific literature found additional reasons why cancer dogs might not yet be ready for grand rounds.

Now, using a male beagle named Cliff, a Dutch research team has staked out new territory in canine diagnostics. They trained Cliff to recognize the smell of Clostridium difficile, a nasty intestinal bacterium that can cause life-threatening diarrhea, especially in hospital settings.

Cliff’s training took a relatively quick two months, after which he displayed near perfect performance judging a series of positive and negative stool samples. (Cliff responds to the C. difficile scent by sitting or lying down.) For the crucial clinical tests he was walked on a leash past an actual patient in a hospital bed—no stool sample, no prep, no contact, just an open-air sniff.

Cliff alerted to 25 out of 30 C. difficile patients, and correctly rejected 265 out of 270 control cases. That’s some pretty good sniffing. There are, however, some caveats. His trainer did not know the patients’ status but he knew there would be only one positive case out of the ten presented that day, i.e., enough information to possibly communicate an unconscious bias to Cliff. The authors also note Cliff had a harder time making the right call when they had him sniff residents in a nursing home lounge, i.e., away from their (smelly?) beds.

I’m willing to give Cliff the benefit of the doubt. But that still leaves me with a question I’ve asked before: why not run a straight-up sniff test with human odor judges? After all, C. difficile is said to smell like horse manure and it should not be difficult for most people to detect.

In fact, a 2002 study examined the various clinical features nurses use to recognize C. difficile infections in their patients. Along with fever and recent use of antibiotics, “characteristic odor” was a statistically significant predictor of the infection. In a 2007 study, nurses without any odor-training recognized the odor of C. difficile in 55% of cases, and correctly ruled it out in 83% of controls.

This suggests to me that with a few training sessions any nurse or doctor should be able to make a highly reliable olfactory diagnosis of C. difficile infection. They could begin treatment before the confirming lab results—a real benefit for the patient.

I’m all for bringing smell back to the practice of medicine. Let’s start with our own noses.

The study discussed here is “Using a dog’s superior olfactory sensitivity to identify Clostridium difficile in stools and patients: proof of principle study,” by Marije K. Bomers, Michiel A. van Agtmael, Hotsche Luik, Merk C. van Veen, Vandenbroucke-Christina M.J.E. Vandenbroucke-Grauls, and Yvo M. Smulders, published in BMJ 345:7396, 2012.

2 comments:

Perfumeshrine said...

Apparently the "difficile" part refers to the ability to constrain one's urge to run to the toilet rather than another's ability to smell you going!

Seriously, imagine having to run dogs inside hospitals for these kinds of tests, with all the additional problems that might entail (hygiene, asthma attacks on sensitive patients/personnel etc.) As you say, sometimes the practicality of having a human sort out the mess (a bit less effectively, I'll grant you) defeats the canine olfactory epiphany.

Avery Gilbert said...

Perfumeshrine:

I suppose the impracticalities of having a dog in an infectious disease unit could be overcome if the speed or accuracy of the canine diagnosis was significantly superior.

What mystifies me is that the authors of this and other such papers assume that dogs have an awesomely better sense of smell. That's no longer a defensible position. (Read the literature, people!) It's almost as if they want to believe in human inferiority.