Despite the anecdotally reported benefits of dogs for people with epilepsy, many questions remain.  While there is little disagreement that dogs can provide emotional support, the actual role they play with regard to predicting seizures is controversial.

There are two types of dogs that assist patients with epilepsy.  Those that recognize and warn of an impending or ongoing seizure are called seizure alert dogs; those that remain with the person to assist with the aftermath of seizure activity are seizure response dogs.

Two recent reports in the medical literature question the effectiveness of seizure alert dogs in detecting epileptic seizures. 

In “Wag the dog: Skepticism on seizure alert canines” (Neurology 2007;68:262), Michael J. Doherty, MD, and Alan M. Haltiner, PhD describe a dog that warns of impending psychogenic non-epileptic seizures (PNES) and present a critical evaluation of the literature.  They said that “a dog’s companionship can be reassuring and relaxing and may cut down on seizure frequency. The ability for a dog to obtain help during or after a seizure could prove lifesaving.”  But they also caution that, “Without objectively clarifying if patients have epilepsy or PNES, the current literature fails to support that canines can warn of impending seizures.”

Gregory L. Krauss et al in “Pseudoseizure dogs” (Neurology 2007;68:308-309) report six patients with seizure response dogs. Four of the six experienced PNES while the other two patients had epilepsy. After the article went to press, they saw three additional patients with seizure response dogs – of whom two had PNES and one had epilepsy. They further cited one study in which a patient was “alerted by his dog 7 minutes prior to having psychogenic seizures.”  They noted that the authors of that study found that while the dog’s predictions were accurate, “they concluded that the dog’s behavior was misleading since it only reinforced the patient’s psychogenic events.”

Krauss et al found that the reported cases “demonstrate the importance of establishing an accurate diagnosis of epilepsy before patients obtain epileptic seizure response dogs.

Service animals should match the patient’s condition and should be provided by skilled professional organizations.”

Brian Litt, MD, and Abba Kriger, PhD in an accompanying editorial concluded: “Both studies suggest a potential therapeutic effect in owning such dogs, but that benefit is more likely to be psychological than neurologic.” Neurology 2007; 68:250-251)

Given the controversy, asked two experts, Gregory Krauss, MD, Department of Neurology, Johns Hopkins Hospital, and Stephen Brown, Cornwall North Heath Services, United Kingdom to share their views.


Gregory Krauss is an associate professor of neurology at the Johns Hopkins School of Medicine. He received his medical degree at the Oregon Health Sciences University and completed residency training in neurology and an epilepsy/electrophysiology fellowship at Johns Hopkins.

Dr. Krauss has studied and published extensively on issues of concern for patients with epilepsy, including driving crash risks, safety issues with antiepilepsy drugs and new medical and surgical therapies. He recently published a new interactive EEG training guide to help improve the accuracy of neurologists’ identification of epileptogenic activity.  

Professor Stephen Brown is an epileptologist and neuropsychiatrist working in Cornwall, in the far southwest of the United Kingdom. A graduate of Cambridge University, he trained in psychiatry and neurology at the Maudsley and King's College Hospitals in London, and was for 12 years medical director of the UK's largest epilepsy center, the David Lewis Centre, near Manchester.

His epilepsy-related academic interests have included cognitive developmental aspects of seizure disorders, sudden unexpected death in epilepsy, behavioral medicine approaches to treating epilepsy, and seizure-alert dogs. He is a former chair of the British Epilepsy Association, and has worked on commissions of the ILAE and the IBE. He was given the international award of Ambassador for Epilepsy by the ILAE & IBE in 1999.

What is the value of seizure dogs in your opinion?

Gregory Krauss: Seizure assist dogs provide companionship and emotional support for some patients with epilepsy.  However, there is, as of yet, no objective evidence that dogs can reliably predict seizures and be used to help patients by anticipating seizures.  In a recent Neurology article, we pointed out that many of our patients seeking seizure assist dogs turned out to have non-epileptic psychogenic seizures.  The main point in our series is that patients need an accurate diagnosis of epilepsy before obtaining specially trained seizure response dogs.

Dogs trained to assist patients with seizures have been termed seizure “assist” or seizure “response” dogs.  It has been reported that some dogs are able to alert patients when they were about to have a seizure — these have been called seizure “alerting” dogs. There is no objective evidence, however, that dogs can anticipate seizures and this term should be used only if this ability is confirmed.

Adam Kirton in Calgary did a study with families of children with epilepsy who had seizure response dogs. Forty percent of the families reported their dogs developed an ability to respond to seizures.  The families reported that about 20% of their dogs showed alerting behavior prior to the children’s seizures. Families were asked to keep diaries of their dogs’ responses to seizures in the study and it is unclear whether this alerting behavior was in response to subtle changes in behavior at the onset of the children’s seizures.  No dog in Dr. Kirton’s study, for example, incorrectly anticipated seizures in the study, which is pretty unlikely.

Stephen Brown: Our dogs are trained to have a personal relationship with the person.  Our particular approach is to train the dog with the person. There is an American model, in which dogs are trained independently, and we think it may cause all sorts of problems.  So we introduce the dog to the person and they are trained together so that the conditioning occurs together.  It is time consuming, but there seem to be benefits.

It was Val Strong, an animal trainer for people with disabilities, who began working with me on this.  She was approached by someone who needed help after immobilization resulting from a seizure.  She trained a seizure response dog for the person.  Some people can be given a warning from the dog.  When the person has a seizure while the dog is being trained, the dog is given a reward.  The trainer does this, and the dog learns that behavior can be shaped. You reward and shape.  Dogs like to live in a predictable universe and to know what is happening next.

The first six patients with epilepsy who had a dog trained with them saw that the dog had learned to anticipate the seizures and gave the people a warning.  This was the original idea for our work.  When we spoke further with the six patients, they said their seizures were less frequent.  We could not understand this.  They said that because of the dogs they were more likely to go out and do things.  If you stay at home and do nothing, you might have more seizures.  But you are less likely to have seizures if your social activity increases. So we have seen a positive benefit.  Val Strong, my colleague, currently works with Personal Assistance Dogs UK .

Should seizure response dogs be made available only to patients with epileptic seizures?

Gregory Krauss: Seizure response dogs are trained to provide companionship and to stay with patients during seizures. The dogs can be very helpful for patients who have recurring or disabling seizures, but are expensive to train and should not be used to support patients with psychogenic seizures.

As we pointed out in our article, most of our patients with seizure response dogs at our referral center turned out to have psychogenic seizures.  This shows that patients with abnormal illness behaviors may seek out support animals as part of the validation of their illness. Patients with non-epileptic psychogenic seizures need psychiatric evaluation and then psychotherapy or cognitive behavior therapy to help with their behaviors — not animals trained to respond to epileptic seizures.

Stephen Brown: We did very careful reviews of the patients by soliciting reviews from their neurologists.  Although some people came to us who did, in fact, have nonepileptic seizures, we offered them a dog. We could not ethically say to these people, “You can have a dog as a pet, but we are not giving you a trained one.”  

Therefore, although we gave them a trained dog and worked with them, we did not include them in our study.  So we have no data about them.  But we did try to distinguish between those with actual epileptic seizures and those with nonepileptic psychogenic seizures.

What, if any, is the downside to seizure response dogs and seizure alert dogs?


Gregory Krauss: For some people, seizure response dogs are very helpful as companions and in providing support during seizures. For the majority of patients who do not have frequent or serious seizures, however, having a seizure response dog may separate them from others and may be inconvenient.  While a support dog might be helpful and appropriate at home, it might not be in other situations, such as school. This often depends on the severity of the epilepsy and the benefits of having a dog stay with the patient. It may be appropriate, for example, to have a seizure response dog at home for a child, or while walking around the neighborhood, but not have the dog go to school with the child.

Additionally, my caution is that the majority of our patients who obtained seizure response dogs had non-epileptic psychogenic seizures.  These patients were misdiagnosed with epilepsy by neurologists.  This shows that dog assist programs need to be cautious and not match seizure response dogs with patients with pseudoseizures.  A major source of this misdiagnosis is that neurologists often misclassify minor rhythmic changes in the EEG called wicket patterns as epileptogenic patterns.  This often leads to patients with atypical seizures that turn out to be pseudoseizures being misdiagnosed with epilepsy.

Stephen Brown: In our study there was no downside.  There is a responsibility that people have towards the dog.  They need to take care of the animal.  Would a dog make their condition worse?  I haven’t seen any evidence of this.  And we were very meticulous in our studies.  However, here in England we have to be sensitive to the issue of animal rights and it’s important to demonstrate that by working with dogs in this way we are not harming or distressing them. In fact we think that dogs love it when properly trained.  But this research is difficult to fund with the usual funding streams since it does not involve drugs or molecular biology.

For dogs not trained to recognize a seizure on the other hand, it can be very distressing for them. To draw attention to this, we reported a series of cases in 2000 (Strong V, and Brown SW. “Should people with epilepsy have untrained dogs as pets?” Seizure 2000; 9:427-430). I am aware that such distress is not a universal phenomenon, but the fact that it can occur raises issues not only about animal training, but also quite specifically about the use of untrained dogs in a control group in any randomized trial.

What type of study might best demonstrate the effectiveness of the dogs?

Gregory Krauss: My colleague recently saw a patient with a dog in our clinic; the dog began pacing just prior to the patient’s seizure.  The dog was possibly responding to subtle changes in behavior during the patient’s aura.  We need better evidence, however, of whether dogs can detect impending seizures.  Dr. Kirton’s study showed that families with seizure response dogs reported higher quality of life compared to families without dogs.  The families also reported that their seizure response dogs could detect children’s seizures. This needs to be evaluated more objectively in studies using time-stamped diaries and video EEG.

Stephen Brown: We feel we need a randomized trial, as most of our previous work has been descriptive and what might be called “dose finding”. As mentioned above there are problems with appropriate controls however.   A delayed entry study where subjects are randomized to acquire a seizure alert dog straight away or after a period of time might get round this, and we have devoted some time to trying to devise such a study, but to date we have lacked funding. Ambulatory EEG recording and continuous video monitoring might help us to see what the dog was reacting to.

People have suggested that there might be something olfactory.  We have not ruled this out.  But we think it is behavior.  We have been carrying out some further work, but we did not get the funding, to do EEG monitoring during the training period.  However, if you watch a dog, they peruse and check the person on a regular basis.  We think they are looking for some subtle change, and it might be different for each patient.

Two things which we heard seemed a bit crazy – dogs that alert in the night and dogs that alert supposedly from another room.  We did have a case of Sudden Unexplained Death in Epilepsy where during the training period the dog did alert in the night by barking and going to a relative for help, but it wasn’t acted on.

With regard to barking and alerting from another room – how can they do that?  The videos suggest that dogs often go back and forth between rooms.  And when they sleep, they wake up every ten minutes or so. While we don’t necessarily notice this, in fact a dog is constantly checking to see if the patient is all right.

Under what circumstances might you recommend a seizure dog to your patients?

Gregory Krauss: I have a patient with disabling seizures who recently asked about obtaining a seizure response dog.  I referred him to the seizure response dog program supported by UCB Pharma. He has had a continuous seizure (epilepsia partialis continua) for 10 years with leg jerking and also has nocturnal convulsions.  He is very disabled by his condition and I think would be very supported by having a specially-trained dog.

I was in Santa Fe recently and saw an article about parents who felt relieved in obtaining a seizure assist dog that could help monitor their child for seizures at night.  The family had fundraisers to help raise $10,000 for the seizure response dog and another $2,500 for training.  This cost is probably worthwhile, but it is important to have a correct diagnosis of epilepsy before going to this expense and effort to train a dog.  

Stephen Brown: The actual answer is that people who do not have a useful warning may benefit from a seizure alert dog.  And two groups may benefit from the response dogs, those who are not able to anticipate a seizure and those who cannot use the information from a seizure alert dog because they become incapacitated by their seizures; for example, they might get cold, and the dog might get blankets for them. For all those with epilepsy, dogs might help to improve their quality of life by affording them a degree of predictability with regard to their seizures.