Using dogs to detect C. difficile

Researchers from the Netherlands have been testing the diagnostic powers of dogs. At least, they’ve been testing whether a two-year old beagle named Cliff can detect which hospital patients are infected with Clostridium difficile. Not surprisingly, it turns out that he can.

C. difficile often arises as a secondary infection in patients whose internal flora has been decimated by antimicrobials. Unfortunately, it’s also extremely contagious, requiring strict infection control. For this reason, the sooner patients with this illness are identified the better. Patients with a C. difficile infection often have diarrhea with a distinctive smell that even humans can detect well over half the time. The researchers suspected that dogs could do much better than that. To that end, they had professional detection dog instructor Hotsche Luik teach Cliff to alert to the scent of C. difficile in stool samples. This process took two months, and can be seen in the video at the bottom of the post.



Cliff, preparing for rounds.
DOI: 10.1136/bmj.e7396

For the testing phase, Cliff was presented with 100 stool samples, half of which had come from C. difficile patients. The dog successfully identified all 50 C. difficile samples. He gave an inconclusive response to three of the negative samples.

Next, Cliff was walked through thirty ten-bed wards, each of which contained one C. difficile patient and nine controls without that illness. Cliff’s trainer, who accompanied him, did not know which person out of the ten had C. difficile. The dog correctly identified 25 out of the 30 C. difficile patients and 261 out of the 270 controls. In many of the cases in which Cliff gave an inconclusive or incorrect response, he’d been distracted. For example, even though the dog wasn’t supposed to interact with or touch the patients, some of them offered him food or beckoned to him.

This leads me to a large caveat about using detection dogs: their eagerness to please their humans makes them susceptible to fudging the results (see my prior post about sniffer dogs). It’s not hard to envision that in a real hospital setting, the nursing staff would have strong suspicions about which patients have C. difficile, and this knowledge could very well influence the dogs into providing the answers they think their humans want. That said, it only took Cliff about ten minutes per ward. That’s a pretty quick and completely noninvasive way to screen patients even if dogs do have a small error rate.

You can see Cliff in action below.





Bomers, M., van Agtmael, M., Luik, H., van Veen, M., Vandenbroucke-Grauls, C., & Smulders, Y. (2012). Using a dog's superior olfactory sensitivity to identify Clostridium difficile in stools and patients: proof of principle study BMJ, 345 (dec13 8) DOI: 10.1136/bmj.e7396