Wednesday, June 19, 2013

New Research Reveals More about Seizure, Diabetic, and Migraine Alert Dogs: But Are They Always Service Dogs? If There's Doubt, Will a Doctor's Letter Help?

Recent studies of dogs alerting in advance to migraines and radical shifts in blood sugar levels indicate that these are functions that are likely to become much more prevalent for service animals in the coming years. These skills, and the already well-known ability of some dogs to alert to the onset of epileptic and other seizures, present legal issues that not all of the federal regulatory agencies have fully considered. It is time to do some speculating on that, but I will begin by summarizing three recent studies.

Alerting Reported by Dogs of Migraine Sufferers in Survey

An online survey of migraine sufferers concerning their dogs’ behavior before and during migraine episodes revealed that slightly over half of those responding said they recognized changes in the behavior of their dogs prior to getting a migraine.

The researchers who reported on the survey results were Dr. Dawn Marcus of the Department of Anesthesiology at the University of Pittsburgh (whose therapy dog research was discussed here recently) and Amrita Bhowmick of Health Union in Philadelphia. They note that migraines are episodic disorders “with disabling attacks often preceded by subtle changes during the hours before the painful part of the attack, called the migraine predrome.” This predrome “includes changes in mood; food cravings; digestive symptoms, yawning, frequent urination; neck pain; and neurological symptoms (e.g., difficulty concentrating, dizziness, blurred vision, and sensitivity to noise or lights).”

In a prior (2012) study, Dr. Marcus had catalogued behaviors of pet dogs whose owners had migraines, some reporting changes up to 24 hours before the onset of migraine symptoms. The current survey was designed to obtain a broader sampling. Most participants (94.9%) were women, who typically had migraines less than 8 days a month, though some had chronic migraines occurring more than 15 days a month.

The most common behavior reported was “the dog refusing to leave the migraineur.” The following table describes the different types of alerting behaviors reported, and the time before initial migraine symptoms:

Alerting Behavior
Number (%)
Staring at migraineur
126 (27%)
Barking at migraineur
15 (3.2%)
Sitting on migraineur
103 (22.1%)
Refusing to leave migraineur’s side
364 (78.1%)
56 (12%)
Pawing at migraineur
101 (21.7%)
Other (e.g., licking, restricting usual activity level, herding migraineur to couch or bed)
130 (27.9%)
Duration between alerting behavior and initial migraine symptoms

0-15 minutes
52 (11.2%)
16-30 minutes
70 (15%)
31-60 minutes
65 (13.9%)
1 to 2 hours
60 (12.9%)
More than 2 hours before migraine
20 (4.3%)
Total identifying alerting behavior before symptoms of migraine attack began
267 (57.3%)
Migraine symptoms usually begin before noticing dog’s behavior
199 (42.7%)

Some participants only reported one behavior in the dog, but others reported more, some more than four behaviors. Over a third of participants who noticed changes in a dog’s behavior regularly began treatment for the migraine as a result of the dog’s alert. This argues strongly that dogs may be a valuable therapeutic addition to the regimen of a migraine sufferer. For additional tasks that a migraine-alert dog can be taught to perform, see the 2015 comment by "Fuzzbutt" at the end of this blog.

UK Training Program for Seizure-Alert Dogs

Stephen Brown of the Peninsula College of Medicine and Dentistry in Cornwall, England, was one of the first medical researchers to look at dogs whose behavior was recognized as changing prior to the onset of a seizure of someone close to them, usually the master. He and a colleague, Laura Goldstein of the Institute of Psychiatry at King’s College London, describe the training regimen of dogs that are being paired with persons with epilepsy. This is the program being implemented by Support Dogs of Sheffield. The patient and the dog are admitted to an assessment center for three weeks, during which:

“The dog is trained to make an extreme focus on the person’s face, referred to as 'look at me' training. When a seizure occurs, the dog is rewarded. The reward, administered by the trainer, is usually a food treat that varies with dog, and this particular reward is only used for seizures. Depending on how many seizures occur, the dog may or may not be acquiring a seizure alert pattern after three weeks.”

The patient and the dog then go home, but video cameras record daily activities in the home. The dog should be alerting to seizures within two months of living with the patient. Dogs vary in what they do to alert, but Strong and Goldstein state that “in a typical premonitory alerting behaviour the dog will seek out the person and engage their attention by getting in their field of vision, staring at them and sometimes barking.”

The trainers of Support Dogs of Sheffield “regard the mean anticipation period for tonic-clonic seizures to be about 30 min, but shorter times, in the order of 15 min are observed with complex partial seizures. The action taken by the human after the dog alerts varies, and may for example include lying down and placing a cushion under the head, or going to lie on a bed.”

There is nothing psychic about what is going on. The researchers report that those “with knowledge of the field have been occasionally approached by journalists for opinions about psychic abilities of dogs,” some reporters suggesting that psychic ability must explain dogs that alert from other rooms than where the patient is. The researchers respond that “a dog situated in another room in the house typically enters the room and appears to check the human every 15 min or so,” and this happens whether the patient is awake or asleep. This attentiveness, rather than psychic ability, explains what is happening. Also, some dogs will go to other rooms to try to get help for the person suffering from a seizure.

As to what dogs sense when they alert, these researchers state that the “consensus from those who spend time reviewing the video evidence is that dogs probably alert to specific and subtle human behavior.” They note it is possible that dogs may detect changes in respiratory rate or even heart rate, or may detect a change in the master’s smell. Cats might conceivably detect oncoming seizures, but “the typical response of a cat would be to run away rather than stay with the person and this makes them less suitable for the role of an alert animal.”

The program costs £10,000 (about $15,000).

Large-Scale Diabetic-Alert Dog Study to Begin

Another online survey was conducted by scientists at the Behavioral Medicine Center at the University of Virginia and concerned diabetic-alert dogs. The survey gathered information from 36 diabetic-alert dog owners, 23 parents of children and 13 adults with type 1 diabetes. Over 90% of respondents reported dogs alerting to hypoglycemic episodes. Many reported significant decreases in the frequency of severe and moderate hypoglycemia since obtaining such a dog. Most also reported decreased worry about hypoglycemia and greater participation in physical activities after getting a dog. Dr. Linda Gonder-Frederick, the leader of this research team, has advised me that she will be undertaking a large-scale study of the effectiveness of seizure-alert dogs.

Legal Issues with Medical Alert Dogs

In the nature of a law school hypothetical, consider the following fact pattern:

Sarah is a diabetic and has several times gone into diabetic comas. She keeps insulin on hand at all times. She begins to notice that her pet Sheltie tries to push her to sit or lie down from ten to thirty minutes before her blood sugar gets dangerously high or low, with the result that she usually tests herself right away. She credits the dog with alerting her enough in advance that she has had a significant reduction in the number of episodes. She has never trained the dog.

Is this a service animal under federal regulations? It might depend on what agency you ask.

Taking a Medical Alert Dog into a Restaurant

The Department of Justice has issued regulations regarding the rights of individuals with disabilities to bring service animals into places of public accommodation, such as restaurants, theaters, motels, etc. DOJ defines a service animal as “any dog that is individually trained to do work or perform tasks for the benefit of an individual with a disability, including a physical, sensory, psychiatric, intellectual, or other mental disability.” DOJ provides examples of services that service animals can perform, including “assisting an individual during a seizure, alerting individuals to the presence of allergens….” (28 CFR 36.104, emphasis added)

In the 2008 release accompanying this regulation, DOJ noted that a service animal can provide “minimal protection” by doing such things as “alerting and protecting a person who is having a seizure.” The release also refers to “protecting the handler from injury resulting from seizures or unconsciousness.” (75 Fed. Reg. 56266, September 15, 2010) The latter phrase might arguably cover a situation where a dog alerts in advance of a seizure beginning, but it is certainly not explicit.

So, is Sarah’s untrained seizure-alert dog a service animal under DOJ regulations? Arguably not, since it is not “individually trained” to perform the task of alerting, even though that task is for the benefit of a person with a disability. An argument could be made, however, that Sarah’s dog should qualify for admission to a place of public accommodation since it will alert her to a threat in time to take action that will reduce the risks associated with the likely episode.

Bringing an Alert Dog into the Cabin of a Commercial Flight

The Department of Transportation includes diabetes in a list of “physical or mental impairments” (73 Fed. Reg. 27667, May 13, 2008). The same release states:

“Generally, a service animal is individually trained to perform functions to assist the passenger who is a qualified individual with a disability. In a few extremely limited situations, an animal such as a seizure alert animal may be capable of performing functions to assist a qualified person with a disability without individualized training. Also, an animal used for emotional support need not have specific training for that function. Similar to an animal that has been individually trained, the definition of a service animal includes: An animal that has been shown to have the innate ability to assist a person with a disability; or an emotional support animal.”

In guidance issued for airlines, DOT defines a service animal as “[a]ny animal that is individually trained or able to provide assistance to a qualified person with a disability or any animal shown by documentation to be necessary to support a passenger with an emotional or mental disability.” (77 Fed. Reg. 39804, July 5, 2012).

Therefore, Sarah’s dog satisfies DOT’s understanding of “service animal” for purposes of going into a cabin of a commercial airplane. Should Sarah get a letter from a medical professional? Letters can be required for psychiatric and emotional support animals under 14 CFR 382.117(e), but that regulation provides that the letter is to state that the “passenger has a mental or emotional disability recognized in the Diagnostic and Statistical Manual of Mental Disorders—Fourth Edition (DSM IV).” Since this is not the case with many alerting functions of dogs, the letter is not required. Nevertheless, I think it is a good idea if Sarah’s physician is willing to write it. First, it will add convincing evidence that the dog is really a service animal, not a pet that Sarah is trying to fly with in the cabin. Second, if the dog is not as well behaved as it should be, the airline staff will be less likely to invoke their right to exclude it because of disruptive behavior.

Migraines can be associated with mental illness or not, but because DOT has not referred to migraine-alert dogs in its Federal Register releases, I would recommend getting a letter for such a dog in any case.

Living with an Alert Dog

Fran Breitkopf and I analyzed the complex area of assistance animals in housing law in an article in the American Bar Association’s general practice journal, GP Solo, some years ago, which we have supplemented here on occasion. Although there are different issues at play depending on the type of housing involved, generally speaking a reasonably well-behaved assistance dog, after whom the owner picks up and keeps from disturbing the neighbors, will be entitled to a reasonable accommodation. If an apartment owner or co-op board is particularly adamant about a no-pets policy, a letter from a medical professional could be important.

For what medical professionals should include in a letter for a patient with a medical alert animal in any of the contexts discussed above, see the article that Dr. J. Lawrence Thomas and I recently wrote for the Journal of Forensic Psychology Practice, Writing Letters to Help Patients with Service and Support Animals. All of the eight issues that we list for a psychologist or psychiatrist to consider before writing such a letter are as important for a medical doctor or other professional who is not treating a patient for a psychological condition. See So What Can Psychologists Say, and What Should They Say? beginning on page 111 in the journal.

Possibility for Fraud

I was recently asked what kind of dog I would say Chloe is if I wanted to fake service dog status and, say, get her into a restaurant. After thinking for a minute, I said that I would probably call her an alert dog, perhaps a migraine alert dog. I have no obvious physical disabilities and Chloe, a highly obedient therapy dog, would not be expected to demonstrate any skill on the spot. If traveling, such a dog would make it easier to get around not having a letter from a psychologist. If asked what she does, I would explain that she paws my knee upon detecting that I may be about to have an episode. There would be no real way to check this.

This is what concerns many trainers in the service dog industry. Not all people who think like this—and unfortunately some do—will have dogs as obedient and socialized as Chloe. Disruptions by poorly trained or bogus service dogs make it harder for well-trained service dogs to be accepted in public environments. Even if an untrained animal might qualify for service dog status under the law, owners should respect the fact that other service dog users have highly trained animals that have worked long and hard to belong in public environments. (News reports recently described a dog trainer whose business ran afoul of various state authorities in New York. The trainer had been claiming to train dogs to alert to oncoming seizures.)


I was recently contacted by a family with a dog that recognizes the onset of precipitous and dangerous drops in blood pressure suffered by a member of the family. I wanted to know if there was any literature on this and sent out emails to some of the authors of the studies above and others. There appears to be no research on this phenomenon at present, but I have no reason to doubt that it is actually happening. Anecdotal accounts on some of the websites associated with hypoglycemia alert dogs indicate that other people are noticing the same thing when their blood pressure drops. If this continues to be observed, I suspect that it will in time be studied and perhaps a new branch of the service dog industry will begin.

I strongly recommend that people who have animals that begin to spontaneously alert to medical conditions have the dogs trained. Early studies found that dogs became so distressed at the onsets of severe seizures that they sometimes injured themselves or attacked people nearby. A few dogs even died. It is a good idea, in my opinion, to reward a dog for specific behaviors in advance of a seizure so that the alerting pattern becomes consistent. Training is also important for teaching an owner to understand how to work with a dog that is providing such an important service. More important for some, training will likely insulate the dog against claims that it is not really a service animal.

Thanks to Dailyah Rudek, Bart Sherwood, Fran Breitkopf, and Ronald Keats for comments and corrections.

For a summary of some subsequent papers on autism research and dogs, see my May 19, 2014, blog.  


Brown, Stephen W., and Goldstein, L.H. (2011). Can Seizure-Alert Dogs Predict Seizures? Epilepsy Research, 97, 236-242.

Ensminger, J., and Breitkopf, F. (2009). Service and Support Animals in Housing Law. GP Solo, 26(5).

Gonder-Frederick, Linda, Rice, Pam, Warren, Dan, Vajda, Karen, and Shepard, J. (April 2013). Diabetic Alert Dogs: A Preliminary Survey of Current Users. Diabetes Care, 36(4), e47,

Marcus, Dawn A. (February 2012). Canine Responses to Impending Migraines. Journal of Alternative and Complimentary Medicine, 18(2), 106-108.

Marcus, Dawn A., and Bhowmick, Amrita (2012). Survey of Migraine Sufferers With Dogs to Evaluate for Canine Migraine-Alerting Behaviors. Journal of Alternative and Complementary Medicine (in press).

Studies prior to 2010 are summarized in Service and Therapy Dogs in American Society, particularly at pp. 64-72.


  1. Hi John,

    My thought is that you are absolutely onto something here concerning the future of canine assistance work. Assistance dogs who alert to medical conditions, or those slated to work with people on the autism spectrum, need to be better understood and supported by our developing laws. The autism assistance dog field has been hurt by people extrapolating policies from traditional service work and stopping there, where cut and dried training is more important to assistance dog function. Due to the lack of money available to support autism assistance dogs as well as assistance dogs that alert, as well as due to the money to be made by unscrupulous people, rescues or haphazardly bred and/or trained dogs are being used to partner with vulnerable children and adults. Sometimes a rather dangerous practice known as "tethering" is used in the autism assistance dog field, where the dog and child are tethered together, sometimes when the child is actually having a tantrum or trying to get away. Sometimes the dog used in this way is not bred or socialized for this role, merely "trained" to be obedient to certain cues. To me, this superficial understanding of the modern assistance dog field is creating incredibly dangerous situations.

    I have long known that most of the benefits I see coming from autism assistance dogs happens naturally when the correct (ie, specifically bred, partnered and socialized for a specific child with a social emotional challenge such as autism) dog is used...naturally, it is up to the adults around to reward and shape naturally occurring behavior, as well as supervise the interactions for a long time to come between child and dog, but a dog that is simply obedient and has not been properly bred/socialized has no place with a vulnerable child who is apt to tax a ill suited dog to a dangerous degree.

    Here, having an autism assistance dog be ineffective is the least of our worries, as when children are bitten, they are frequently bitten on the face, and the same phenomena that you refer to of alert dogs on occasion being driven to attack when emotionally provoked is very important to note. There was a recent story about a dog (a Mastiff) who was in the Missouri home that a five year old boy temporarily joined when his home was destroyed by the recent tornadoes in Oklahoma; this boy had a meltdown when he was told to put on his shoes, and the dog attacked and killed him. Not all dogs have the tolerance for children who throw tantrums or behave in unusual ways, and dogs who are not properly socialized or suited temperamentally to this work should be kept as far away from children who face social/emotional challenges as possible.

    One more thought: I was on a plane recently with a woman who had a "emotional support dog" with her, and this dog nearly bit a flight attendant when she went by. I think that for a dog to earn public access, we not only have to include the criteria of relevant tasks performed by alert dogs, emotional support dogs and/or autism assistance dogs, but these dogs should also pass the strict criteria of passing the rather stringent standards of the ADI for public access certification.  If pet dogs do alert to their owners' conditions, then programs should exist to provide the second necessary piece of being sure this dog is safe in public.

    Just my 2 cents!

    Kind regards,

    Patty Dobbs Gross
    Executive Director
    North Star Foundation
    We help children find their way.

  2. Migraine disease is not a mental illness; migraine is a neurological disorder which effects the entire body.

  3. John, Do you consider Migraine a disability? While yes, they can be dibilitating when they occur, do you believe they rise to the definition of disability requiring a Service Dog?

    1. This is a good question and I'll give my initial response now, though I may supplement this later. A disability under 42 U.S.C. 12102 is "a physical or mental impairment that substantially limits one or more major life activities...." Major life activities can including seeing, speaking, reading, concentration, thinking, working and other activities that may be impeded by a migraine. An impairment may be episodic and the "determination of whether an impairment substantially limits a major life activity [is] made without regard to the ameliorative effects of mitigating measures such as... medication," etc. I think that at least some migraine conditions meet the definition of disability, such that a dog's recognition and response abilities could qualify it as a service dog.

    2. Anonymous, not everyone who has migraine is disabled by it, but plenty of people are disabled by their migraine disease and some even get government disability benefits because of it. Obviously, a person who has episodic migraine and has migraine episodes only once or twice a year (if that) is not disabled by it, but most people who have chronic migraine (15 or more migraine days per month) are disabled by it. Migraine has four phases (any of which can be skipped): prodrome, aura, headache, and postdrome (a.k.a. migraine hangover). There are many types of migraine, so symptoms range from diarrhea, vertigo, severe headache pain, phantom smells, sensitivity to light and sound, nausea, vomiting, memory problems, slowed thought processes, Alice in Wonderland Syndrome (visual distortions like in the story of the same name), partial to total blindness, partial to total paralysis, seizures, inability to speak, and hundreds more.

      There are many tasks a service dog can do for someone who is disabled by migraine, including alert to the onset of an episode, mobility work for vertigo and/or partial paralysis/weakness, help with self-care for partial to total paralysis and weakness, retrieval work (including bringing a medicine bag and bottle of water), turning lights on/off, going for help, reminding to take preventive medication or do other things the handler forgets, guide work for visual symptoms and/or slowed thought process symptoms (such as keeping the handler safe when crossing the street), calling an emergency number, help with trigger avoidance (depending on the trigger), and more.

      Migraine is related to epilepsy, by the way.

    3. "Fuzzbutt": Thank you for listing the additional items that may be performed by a migraine-alert dog. I added a reference to your second paragraph inside the blog.

    4. Glad it was helpful! I also wanted to add that the World Health Organization (WHO) has declared that migraine can be as disabling as quadriplegia.