dog nose Bigstock

Five dogs (Labs and Golden retrievers) were trained to detect contaminated urine in scent detection boxes. When they came across a contaminated urine sample, they sat in front of the box. That’s a pretty standard training approach.

After the training phase, they took urine samples with fairly high levels of bacterial contamination and samples with no bacteria, and saw how good the dogs were at identifying the contaminated samples. They then tested urine samples that were diluted to 1% and 0.1% and did the same thing.

The results were pretty impressive:

• For E. coli, the most common bug involved in urinary tract infections, the sensitivity of the “sniff test” was 99.6%. That means the dogs detected 99.6% of positive samples (and therefore had few false negatives, aka samples that were actually positive that they missed.) The specificity was 91.5%, so 8.5% of positives were false positives (aka negative samples that were mistakenly identified as positive). Overall, the results mean that if bacteria were present, the odds were very good that the dogs would detect it, with a relatively small percentage of additional positives being false alarms.
• Results were similar with other bacteria.
• Diluting samples to 1% or even 0.1% did not impact the sensitivity or specificity.

So, it seems that dogs can do a good job figuring out whether urine has bacteria in it.

Is having a dog that is able to detect bacteria in urine samples useful?

• Probably not, at least in practical terms. I doubt someone’s going to keep a trained dog around to pre-screen urine samples.
• Testing urine for bacteria only provides part of the needed information. The susceptibility of the bacterium to antibiotics is another key piece of information, something that even the best dog nose can’t sort out.

What if the dogs can detect bacteriuria directly in certain people, such as those with spinal cord injuries or other conditions that hide other signs of urinary tract infections?

• That might be more interesting. However, asymptomatic bacteriuria (presence of bacteria in the absence of disease) is common in these same groups of patients and treatment is rarely needed, so it might not be a useful screening tool anyway. In fact, there’s a lot of effort being taken to reduce unnecessary cultures and unnecessary treatment of people with asymptomatic bacteriuria, so finding more of those individuals might be counterproductive.

How well would dogs do trying to detect bacteriuria directly in people?

• That’s another good question.  I suspect there’d be a lot of other distracting smells when sniffing urine directly from a person or other animal.
• However, an interesting anecdote is reported in the paper. One month after the study, one of the dogs alerted to a person at the training centre. That person wasn’t feeling well but didn’t suspect a urinary tract infection. However, based on the dog’s response, a urine culture was done and bacteria were present.

Regardless, it’s an interesting study and certainly not the last report of using dogs to detect a human health problem that we’ll see in the near future.

It also makes me wonder about owning or handling one of these dogs. It might be a bit awkward explaining to the person you meet walking down the street why your dog is sitting in front of them and staring at them. A little too much information perhaps.

In terms of my own dog, Merlin (a yellow Lab), well, he has a great nose and he’s great at the “staring at someone’s crotch” component that a dog like this would need. Unfortunately, the connection to his brain (what there is of that) is pretty weak. He’ll remain blissfully unemployed.

(Reprinted with permission from Worms and Germs Blog)