When Angel the yellow Labrador starts barking in Sarah Specht’s house, the countdown begins. She must find her 7-year-old son Hunter and then run a small magnet over the vagal nerve stimulator implanted in his chest. Time is of the essence—she has mere minutes to block one of Hunter’s incoming seizures.
Hunter was 3 years old when he was diagnosed with epilepsy in 2013. He had episodes 95% of the day—and constantly in his sleep—with a mixed bag of seizures that momentarily gave him blank stares, arm jerks, and muscles that either stiffened so much that he would fall to the ground like a plank or loosened to the point where Specht says he “goes into a noodle.”
Because of the vagal nerve stimulator, the boy now has most of his seizures automatically blocked by electrical pulses from the device. It won’t catch everything, however, with ten seizures on average a day slipping through the cracks for Hunter. But seizures that do occur can be stopped in their tracks or at least made less severe with manual intervention—a magnet swiped over the implant—and that’s where Angel’s usefulness comes in for Specht.
Many epilepsy patients swear by dogs like Angel. Special trained behaviors such as barking and licking are an apparent warning against risky activities like taking baths, climbing stairs, or driving a car. And cases like Specht’s, they may offer advance warning that could help stave off a seizure altogether.
But despite such anecdotal accounts, there is no proof that dogs can be trained to detect seizures, let alone predict their onset far enough in advance to tell humans about it. What’s more, even the dog trainers themselves don’t guarantee that their dogs can detect every seizure, and there is no regulation to ensure that the dogs were trained properly in the first place.
“Any good detector has positive value,” says Dr. Gary Mathern, a pediatric neurosurgeon at Mattel Children’s Hospital UCLA. “But as a physician, a key unanswered question is whether the dogs really work.”
Demand for the dogs is strong, even with the unanswered questions regarding their effectiveness. Over the course of a year, 4 Paws for Ability, the organization that trained Angel, will place 100 dogs with children, about 40% of which will be seizure alert dogs.
For many, the idea that dogs can recognize seizures is based on the idea that they’ll be able to detect seizures using their sense of smell. Some organizations believe that dogs can be trained to pick up a signal, others think they must be born with the skill. One, Canine Partners For Life, thinks only certain dogs are fit for duty. Susann Guy, COO of organization, says that dogs’ talent seems to be innate—whether the ability to detect a seizure signal is utterly untrainable, she says.
Canine Partners for Life receives more than 2,000 applications per year for service dogs. Located in rural Cochranville, in southern Pennsylvania, the facility overlooks acres of vast open space. Her dogs enjoy a good life here before they’re put to work. Two canines share each indoor pen, which contain dog beds and a play area. To acclimate them to a home environment, volunteers play with the dogs in the “cuddle room,” complete with a sofa, books, a rug, and a TV. Next to the vet’s office is the “zen room,” a dimly lit space where the canines can regularly relax to classical music.
When trainers spot a puppy that seems to have a particularly good nose, they give it a low-tech test: A local man with epilepsy takes the dog and observes it over a weekend, Guy says. In the comfort of his own home, he’s able to see if the dog changes its behavior in the minutes leading up to one of his seizures. If it does, it’s a candidate for special training after its initial fostering period.
Before the dog graduates from seizure alert school, it is matched to a potential new owner. In Specht’s case, she mailed in a shirt that Hunter wore during a seizure to another organization, 4 Paws For Ability, where Angel was trained. Its founder, Karen Shirk, says that her dogs are trained to detect seizures by a scent. Shirk says she matched Angel with Hunter based on his personality, a video of his daily life, and his needs, which were documented in a questionnaire that his mom filled out.
Preparing a service dog is time consuming and labor intensive, and Shirk says that each of her canines costs $26,000 to train, feed, and house over the course of a year.
Depending on the trainer and the patient’s eligibility for financial aid, seizure alert dogs can range in price from free to upwards of $20,000. The IRS considers service dogs to be “medical durable equipment” and therefore deductible medical expenses, though they aren’t covered by health insurance and Medicare. Patients often resort to private fundraising with the help of friends and family.
Canine Partners for Life charges between $1,000 and $3,000 for a seizure alert dog that has been raised for one year in a foster home and has passed one year of specialized training. Because the organization has cultivated a strong donor base—65% of its revenue comes from individual donors across the country—Guy’s organization is able to sell the dogs for a relatively low price.
And even if patients can raise funds, Canine Partners for Life will not sell its service dogs to anyone—namely, children under the age of 12, much less a kid Hunter’s age. The reasoning, Guy says, is to make sure that each dog recipient is ready for a working bond between human and service dog. Young children may treat their dogs as pets, not as partners, which can degrade a dog’s skills and motivation.
In those cases, some parents turn to other organizations. Specht tried and was rejected by many organizations like Canine Partners For Life before coming upon one that would give her a canine helper for Hunter. Two months after her son was diagnosed with epilepsy, Specht learned that 4 Paws For Ability would provide Hunter with a seizure alert dog. Shirk’s team agreed to start training a dog as soon as Specht delivered the $15,000 fee.
Founded in 1998, the organization now owns hundreds of dogs that are fostered across local homes and colleges. In addition to seizure alert dogs, the company also provides dogs that are specialized for diabetes and autism. “We don’t [just] look at what their diagnosis is,” says Shirk, who explains that dogs are trained to the specific problem and temperament of each child. Once families provide the $15,000 fee, there’s no waiting list: they are matched with a dog that is expected to graduate training 18 months later, she says.
Those dogs’ training regimens aren’t necessarily backed by scientific studies. The few studies done on alert dogs in the 1990s and 2000s did not find conclusive evidence for the dogs’ seizure detecting abilities. “My experience is that they are not very accurate and they have not been tested in a proper randomized trial,” Dr. Mathern says.
One study was published in 2004. In it, two patients’ brains were continuous monitored via electroencephalogram, also known as EEG, while their trained dogs sat nearby. Neither of their dogs could predict seizures accurately. Still, the study’s very small sample size makes it hard to apply its conclusions universally. Other studies claim that dogs can alert their owners up to an hour before a seizure, but those have relied on patient-reported surveys, which are not always reliable, or were written by dog-training organizations, which have an interest in the matter.
Researchers have been stumped by how, precisely, some dogs might be able to detect seizures minutes or even hours in advance. “There’s almost no research on it. It’s actually close to zero research so we have no idea,” says Nathaniel Hall, director of the Canine Olfaction Research and Education Laboratory at Texas Tech University.
Some say that the dogs might be picking up on electrical signals, such as irregular brain waves, or subtle movements that are unnoticeable by humans. The most common explanation given by canine trainers, though, is that their dogs can smell trace chemical signals released by people who are about to experience a seizure. “The olfactory hypothesis would make sense if biochemical changes were producing an odor that a dog could detect,” Hall says. “As far as I know, that’s no more than a guess.”
“If we knew what the odors were, that would be ground-breaking,” he says. Such information could give researchers something that they could measure in a scientific study. Hall has searched for funding to launch studies on seizure-related olfactory cues, but he hasn’t had much luck. Those studies never got off the ground because of the expense—training the dogs and controlling for myriad variables proved too costly. In the end, he says, “what we think of as the dog’s amazing sense of smell is backed by scant evidence. It’s an exciting area, since there’s so little to go off of.”
Michael Sapp, CEO of Michigan-based service dog organization Paws With A Cause, is also wary of claims that dogs can alert people to seizures. He estimates that there are fewer than 100 dogs that truly have these developed capabilities in the entire U.S.
“Many times people with epilepsy cannot be left alone because their seizures could be life threatening. To give someone a false hope that they can now rely on this dog to tell him or her when a seizure is coming is at best unethical,” Sapp says, whose organization doesn’t claim to provide seizure alert dogs.
They may even cause harm: Researchers reported one case in which a patient’s dog supposedly alerted him seven minutes before his seizures, but those turned out to be psychosomatic, non-epileptic episodes that were reinforced by the alerts.
Lack of Regulation
Despite the lack of evidence and potential for harm, Dr. Mathern says he has witnessed an increase in interest in seizure alert dogs by patients and their families.
The service dog industry is lightly regulated. There is an accrediting agency, though—Assistance Dog International, a nonprofit—that sets voluntary standards for service dog training and processes and approves increasingly more service dog centers every year.
To become accredited by ADI, an organization like Canine Partners for Life undergoes a campus assessment every five years and complies with certain standards in ethics and training. But ADI’s standards for dog training are “minimal,” Sapp says, adding that it is by no means a regulatory body.
Even then, there’s no rule that service dog providers need to be ADI-approved. In fact, U.S. laws don’t even require that service animals be professionally trained. People with disabilities are free to train their dogs themselves or use an outside trainer, as long as the result is to their liking. Desperate patients or parents may opt for non-ADI certified organizations like 4 Paws For Ability. These organizations may not have gone through the ADI approval process, but they still claim to produce high-quality service dogs.
Once the dog is delivered, it is ultimately up to patients to determine how a service dog will fit into their home and daily routine. In contrast with traditional service pairings, Angel doesn’t necessarily stay by Hunter’s side at every moment. Detecting many seizures over a short period of time makes Angel upset, Specht says, so she takes the dog to another room to unwind.
Though this can leave Hunter unmonitored for periods of time, Specht says that she is willing to separate the pair for Angel’s benefit on occasion. “Dogs can develop PTSD. Rather than have her keep alerting, she’s given the free command,” she says. She doesn’t want Angel’s anxiety to be forced her to stop working, “so I give her breaks when I see that she’s nervous.”