Author(s): Stephen W. Brown, Laura H. Goldstein
Affiliations: King’s College London, Peninsula College of Medicine & Dentistry
Publication: Epilepsy Research, Vol. 97, pp. 236-242
Date: November 2011
Full Text: available from EFNC-YALC
Key: Â Â Â
           SAD – Seizure Alert Dog
           SRD – Seizure Response Dog
           PNES – Psychogenic Non-Epileptic Seizures
Introduction:
Both Seizure Alert Dogs (SADs) and Seizure Response Dogs (SRDs) have been associated with positive outcomes for epilepsy patients (summary below), but the research surrounding SAD behavior and mechanisms has been controversial for several reasons. First, the mechanism of seizure detection is unknown, although several plausible explanations exist. Second, performing experimental research on the subject with untrained controls has been deemed unethical due to preexisting records of both pet and human injury when untrained animals are involved. Third, several studies have found that SADs will alert to non-epileptic seizures (PNES), which has created skepticism around the existing mechanism hypotheses.
Summary of Previous Research on SRDs/SADs:
The first training groups from 1999-2002 developed the use of reward-based operant conditioning to train dogs to recognize and anticipate tonic-clonic seizures in paired human subjects. From these studies, UK researchers found: (1) all dogs involved in the studies had the ability to function as SADs with good reliability (about 15-45min lead time), and (2) unexpectedly, participation in a SAD training program was associated with a significant drop in seizure frequency for the human subjects, for unknown reasons. These studies had relatively small sample sizes, but the data were remarkably consistent.
Some later studies and surveys of families with pets (2003-2008) found that many pet dogs (particularly females) will spontaneously acquire the ability to function as both SADs and SRDs, although the length of warning time varies greatly (from 10s – 20+min). Additionally, it was found that dogs will alert to both PNES and true epileptic seizures.
Current Programs & Research:
The Support Dogs of Sheffield UK currently employ a similar method used in the 1999-2002 studies, with an emphasis on socializing and training the individual dog with the person with epilepsy from the beginning. Video footage for about two months after training is used to collect data on the dogs’ behavior.
Alert Time: From these data, the mean alert time for tonic-clonic seizures is about 30min (as found in earlier studies), while the alert time for complex partial seizures is about 15min.
Alert Mechanism(s): Some possible mechanisms for the alert ability have been posited. The current research has somewhat debunked the idea that the alert happens as a result of a premonition, which was originally founded on observations of dogs seeming to alert from another room or from sleep. Observations from the current research have found that the dogs will regularly check on the human subjects, and the alert will happen during one of those “check-insâ€.
Some probable explanations include: (1) sensitivity to respiratory and/or heart rate, and (2) sensitivity to a change in body odor or other autonomic characteristics.
 