What to make of Medical Dogs

For thousands of years we have guided the evolution of dogs to fulfill our needs for work and companionship.  Service dogs are pretty remarkable.  I love to watch herd dogs mimicking the dance of predator and prey.  When you see a guide dog help someone navigate a building or street, you can’t help but to be impressed by the dogs “devotion” and “skill”.

It seems there is a new canine skill in the news every day.  Now, in addition to the traditional roles guiding the blind and deaf, and helping the physically disabled, dogs are claimed to be able to calm autistic children, detect blood pressure changes and seizures, and find cancers. Dogs have been used in the bed bug epidemic to find the critters (with little scientific evidence of success).

Humans and dogs have co-evolved successfully to create strong owner-dog attachments (to the point of pit bull owners defending their dogs rather than acknowledging a dog’s danger to humans).  It seems intuitive, and is quite plausible, that dogs can calm us, can help lead us in ways analogous to their roles in nature (if “natural” can even be applied to dogs). It’s easy to see how herding behavior can be adapted into guide dog behavior, or hunting behavior into chemical detection.

What’s less clear is whether any of these roles are based on fact rather than intuition.

What raised my interest this week was a dog featured on ABC News. The show followed a young woman with a “rare heart condition” that goes unnamed.  It shows her dog alerting her to an impending fainting spell. The film itself is useless as data—it simply shows the dog nudging the patient and the patient responding by lying down. But even assuming that the family has seen “successful” alerts of fainting, there is no evidence in the literature to suggest dogs can actually do this in any context.  It is likely that the dog improves the woman’s sense of security and well-being (something the literature supports in human-service dog relationships) but there is no way to know if the dog is “sensing” changes in blood pressure as claimed.

One of the most interesting medical claims for dogs is an ability to detect cancers in humans, perhaps before conventional testing might. While anecdotes abound, there is scant literature to support this ability. One unimpressive pilot study looked at dogs’ potential ability to detect bladder cancers from urine samples. The idea behind cancer dogs is that there may be volatile compounds produced in cancer patients that dogs can detect by scent.  In these studies, the compounds are not identified, not tested for, not named. There are many confounders, for example, in the few samples used, there may be other differences being detected by the dogs.

In the other study (I found very few) dogs were “trained” to detect lung and breast cancers in humans. The methodology of breath sampling is not validated as far as I can see, and once again, the putative compounds in breath are not identified. Statistically, the efficacy is marginal at best.

Another controversial use medical use of dogs is for seizure detection. So-called “seizure alert dogs” are purported to warn their masters of impending seizures, and to stay with them during the event.  It doesn’t take a bucket of skepticism to wonder what this actually means.  Many seizures are preceded by marked changes in behavior as the patient experiences an aura, something anyone/anydog should be able to notice. Staying by a bonded human isn’t too unusual either. I do wonder how the humans know what the dog does, as many true seizures are accompanied by a period of confusion and memory loss. An objective witness would be needed to note the behavior.

And some case studies have done just that. There aren’t a lot of data, but one patient monitored in a seizure unit experienced nine seizures, one of which was “sensed” by her dog.  In another, the dog sensed a “pseudo-seizure”, that is a fit that is not, neurologically-speaking, a seizure but a non-seizure set of behaviors, often precipitated by stress.

That dogs can be trained to help humans is pretty clear—we’ve bred them for this.  But these abilities must necessarily have limits, limits set by biology. If a behavior seems implausible, the evidence for it must (from a Bayesian standpoint) be pretty damned solid.

I don’t doubt the social and emotional value of dogs as companions, and as active helpers in many circumstances. But beyond this, the evidence is wanting.


Allen K, Shykoff BE, & Izzo JL Jr (2001). Pet ownership, but not ace inhibitor therapy, blunts home blood pressure responses to mental stress. Hypertension, 38 (4), 815-20 PMID: 11641292

Lane, D., McNicholas, J., & Collis, G. (1998). Dogs for the disabled: benefits to recipients and welfare of the dog Applied Animal Behaviour Science, 59 (1-3), 49-60 DOI: 10.1016/S0168-1591(98)00120-8

Willis, C. (2004). Olfactory detection of human bladder cancer by dogs: proof of principle study BMJ, 329 (7468) DOI: 10.1136/bmj.329.7468.712

McCulloch, M. (2006). Diagnostic Accuracy of Canine Scent Detection in Early- and Late-Stage Lung and Breast Cancers Integrative Cancer Therapies, 5 (1), 30-39 DOI: 10.1177/1534735405285096

Dalziel DJ, Uthman BM, Mcgorray SP, & Reep RL (2003). Seizure-alert dogs: a review and preliminary study. Seizure : the journal of the British Epilepsy Association, 12 (2), 115-20 PMID: 12566236

Strong V, Brown S, Huyton M, & Coyle H (2002). Effect of trained Seizure Alert Dogs on frequency of tonic-clonic seizures. Seizure : the journal of the British Epilepsy Association, 11 (6), 402-5 PMID: 12160671

Doherty, M., & Haltiner, A. (2007). Wag the dog: Skepticism on seizure alert canines Neurology, 68 (4), 309-309 DOI: 10.1212/01.wnl.0000252369.82956.a3

Posted in: Science and Medicine

Leave a Comment (25) ↓