Canine Responses to Hypoglycemia in Patients with Type 1 Diabetes

Abstract

Anecdotal evidence suggests that domestic dogs may be able to detect hypoglycemia in their human caregivers; scientific investigation of this phenomenon, however, is sorely lacking. This study thus aimed to investigate how pet dogs respond to the hypoglycemic episodes of their owners with type 1 diabetes.
Two hundred and twelve dog owners (64.2% female) with medically diagnosed type 1 diabetes participated in the study. All participants owned at least 1 dog. Each person completed a purpose-designed questionnaire developed to collect information on their dogs' responses (if any) to their hypoglycemic episodes.
One hundred and thirty-eight (65.1%) respondents indicated that their dog had shown a behavioral reaction to at least one of their hypoglycemic episodes, with 31.9% of animals reacting to 11 or more events. Canine sex, age, breed status, and length of pet ownership were unrelated to hypoglycemia-response likelihood. Thirty-six percent of the sample believed that their dogs reacted most of the times they went "low"; 33.6% indicated that their pets reacted before they themselves were aware they were hypoglycemic. Dogs' behavioral responses to their owners' hypoglycemic episodes varied. Most animals behaved in a manner suggestive of attracting their owners' attention, for example, vocalizing (61.5%), licking them (49.2%), nuzzling them (40.6%), jumping on top of them (30.4%), and/or staring intently at their faces (41.3%). A smaller proportion showed behavioral responses suggestive of fear, including trembling (7.2%), running away from the owner (5.1%), and/or hyperventilating (2.2%).
The findings suggest that behavioral reactions to hypoglycemic episodes in pet owners with type 1 diabetes commonly occur in untrained dogs. Further research is now needed to elucidate the mechanism(s) that dogs use to perform this feat.

Full text

THE JOURNAL OF ALTERNATIVE AND COMPLEMENTARY MEDICINE
Volume 14, Number 10, 2008, pp. 1235–1241
© Mary Ann Liebert, Inc.
DOI: 10.1089/acm.2008.0288
Canine Responses to Hypoglycemia in Patients
with Type 1 Diabetes
Deborah L. Wells, Ph.D.,
1
Shaun W. Lawson, Ph.D.,
2
and A. Niroshan Siriwardena, Ph.D.
3
Abstract
Objective: Anecdotal evidence suggests that domestic dogs may be able to detect hypoglycemia in their hu-
man caregivers; scientific investigation of this phenomenon, however, is sorely lacking. This study thus aimed
to investigate how pet dogs respond to the hypoglycemic episodes of their owners with type 1 diabetes.
Methods: Two hundred and twelve dog owners (64.2% female) with medically diagnosed type 1 diabetes par-
ticipated in the study. All participants owned at least 1 dog. Each person completed a purpose-designed ques-
tionnaire developed to collect information on their dogs’ responses (if any) to their hypoglycemic episodes.
Results: One hundred and thirty-eight (65.1%) respondents indicated that their dog had shown a behavioral
reaction to at least one of their hypoglycemic episodes, with 31.9% of animals reacting to 11 or more events.
Canine sex, age, breed status, and length of pet ownership were unrelated to hypoglycemia-response likeli-
hood. Thirty-six percent of the sample believed that their dogs reacted most of the times they went “low”; 33.6%
indicated that their pets reacted before they themselves were aware they were hypoglycemic. Dogs’ behavioral
responses to their owners’ hypoglycemic episodes varied. Most animals behaved in a manner suggestive of at-
tracting their owners’ attention, for example, vocalizing (61.5%), licking them (49.2%), nuzzling them (40.6%),
jumping on top of them (30.4%), and/or staring intently at their faces (41.3%). A smaller proportion showed
behavioral responses suggestive of fear, including trembling (7.2%), running away from the owner (5.1%),
and/or hyperventilating (2.2%).
Conclusions: The findings suggest that behavioral reactions to hypoglycemic episodes in pet owners with type
1 diabetes commonly occur in untrained dogs. Further research is now needed to elucidate the mechanism(s)
that dogs use to perform this feat.
1235
Introduction
Pet-keeping is a widespread and well-accepted phenome-
non in today’s society. As a nation of self-confessed “ani-
mal lovers,” the British public now share their homes with
over 9 million cats and 6 million dogs.
1
These animals can play
an enormous role in their owners’ lives. As well as providing
a source of companionship, emotional support, and enter-
tainment, there is now substantial evidence to suggest that do-
mestic pets may be able to promote their owners’ health.
2–6
Lately, some attention has been directed toward the abil-
ity of companion animals, particularly dogs, to serve as an
“early warning system” for certain types of underlying phys-
ical ailments in humans. For example, dogs have been shown
to be capable of using olfaction to detect the presence of can-
cer. In 1989, Williams and Pembroke
7
reported a case of a
Border Collie/Doberman Pinscher crossbreed sniffing re-
peatedly at a mole on its owner’s leg that later was found to
be malignant. Similar anecdotal reports have since appeared
in newspapers and scientific journals.
8–10
More recently,
Willis and colleagues
11
successfully trained 6 dogs of mixed
breed to identify people with bladder cancer using a dis-
crimination task. As a group, the dogs correctly identified
urine samples from patients with bladder cancer on 22 of 54
occasions; a mean success rate of 41%. Although this proof
of principle study has been criticized for methodological
weakness and erroneous conclusions,
12
other work, using
different methods, has yielded similar results.
13,14
In addition to detecting cancer, there is now mounting ev-
idence to suggest that some dogs can reliably predict the on-
1
School of Psychology, Queen’s University Belfast, Belfast, Northern Ireland, United Kingdom.
2
Department of Computing and Informatics, University of Lincoln, Brayford Pool, Lincoln, United Kingdom.
3
School of Health and Social Care, University of Lincoln, Lincoln, United Kingdom.

set of epileptic seizures in their human owners, in some cases
up to 45 min in advance of an event.
15–19
This ability appears
to be inherent in some animals. However, many dogs are
now being trained by specialist organizations (e.g., Support
Dogs, UK) to monitor their owners for outward signs of an
imminent seizure and to react in an appropriate manner (e.g.,
barking or pawing) if a seizure is predicted.
15,16,18–20
Hypoglycemia is a common and potentially hazardous
complication of diabetes, usually occurring in individuals
with type 1 (and increasingly type 2) diabetes who are re-
ceiving insulin. The condition typically results in physiolog-
ical effects arising from the attempt to restore glucose levels
(sympatheticoadrenal response). These effects can include
fear, tremor, tachycardia, and/or sweating, as well as a crit-
ical lack of glucose to nervous tissue (neuroglycopenia), lead-
ing to behavioral change, seizures, or coma.
21
Despite the wide range of hypoglycemia-related signals,
people with diabetes often experience poor awareness of
their own symptoms, with accuracy of detection typically
waning over time due to the effects of advancing age or com-
plications of the condition, such as autonomic neuropathy.
22
Although numerous techniques (e.g., glucose sensors, skin
conductors, heart-rate monitors) have been developed to fa-
cilitate early detection of hypoglycemia,
23,24
none of these,
as yet, are considered reliable, practical, or suitable for use
outside clinical settings.
Anecdotal evidence now suggests that some dogs may be
able to detect a fall in their owner’s glucose levels. Lim and
associates,
25
for example, indicated that 14 of 37 dog own-
ers with diabetes recognized a response from their pet in re-
action to their hypoglycemic episodes. A small number of
case reports have also shown that dogs may be able to warn
owners of impending hypoglycemia before symptoms are
noticed by those whose awareness of the condition is mostly
intact.
26,27
Very recently, O’Connor and colleagues,
28
re-
ported the case of hypoglycemia detection by a female Cav-
alier King Charles Spaniel in a 72-year-old non-diabetic man.
Thus far, attempts to demonstrate that dogs can detect hy-
poglycemia are based on little more than brief anecdotal re-
ports. The present study was therefore designed to explore
whether there is any validity to the notion that untrained do-
mestic dogs can detect hypoglycemia in people with type 1
diabetes. This is an important issue with potentially far-
reaching consequences. If, for example, it is discovered that
dogs can detect hypoglycemia reliably, and the mechanisms
underlying this feat are established, then it may be possible
to develop noninvasive electronic sensor systems to perform
the same task. Trained dogs, and/or appropriately designed
electronic sensor systems, could then be systematically em-
ployed as an early-warning system for people with diabetes,
allowing appropriate action to be taken before the effects of
hypoglycemia become disabling.
Methods
Survey
A purpose-designed survey (available from the lead au-
thor upon request), consisting of forced-choice and open-
ended questions, was developed to assess the reported ex-
periences of patients with type 1 diabetes concerning their
pet dogs’ responses to hypoglycemia. The survey was di-
vided into four sections:
• Section 1 collected information on participant demo-
graphics, e.g., sex, age, marital status, occupation, etc.
• Section 2 concentrated specifically on the health of the par-
ticipants in relation to their diabetes. Questions were
posed to collect details on the length of time the partici-
pants had suffered from type 1 diabetes, how well con-
trolled they considered their condition to be, and, if
known, their latest HbA1c reading. Since the study was
concerned specifically with hypoglycemia detection, cate-
gorical-style questions were also constructed to assess the
frequency with which the participants experienced hypo-
glycemic episodes, the types of symptoms normally
experienced with such events, and how good the indi-
viduals considered their self-awareness of their hypo-
glycemia-related symptoms to be.
• Section 3 of the survey was concerned with dog owner-
ship, and collected demographic information on the dogs
owned by the participants, i.e. sex, age, breed, length of
pet ownership.
• Section 4 was designed to collect information on the re-
action of the participants’ dogs (if any) to their owner’s
hypoglycemic episodes. Participants were initially asked
if their dog/s had ever reacted to any of their hypo-
glycemic episodes (yes, no). Individuals reporting in the
positive, were required to answer further categorical ques-
tions designed to assess how many episodes of hypo-
glycemia their dogs had reacted to and the frequency,
speed, and reliability with which they typically responded
to drops in their owner’s sugar levels. The final part of
Section 4 allowed participants the opportunity to provide
in-depth qualitative information on the behavioral reac-
tions of their dogs to hypoglycemic events.
Procedure
The survey outlined above was made available both on-
line and in the form of a hardcopy questionnaire. A feature
advertising the research and calling for participant recruit-
ment was placed in several UK-distributed newsletters and
on diabetes-related websites. The feature provided back-
ground information on the research and gave interested par-
ties the opportunity either to complete the survey on the
website, or, if preferred, to request a hardcopy version, dis-
tributed by post. A covering letter outlining how to complete
the survey was made available both online and, for those
who opted to complete the paper version of the question-
naire. A freepost return envelope was also provided to this
latter category of participants.
It was made clear in both the advertising feature and the
covering note that individuals would only be considered el-
igible for participation if they: (1) had medically diagnosed
type 1 diabetes, (2) were 18 years of age or older, and (3)
owned one or more dogs.
Data analysis
Simple descriptive statistics were carried out for each
question on the survey. A binary logistic regression analy-
sis
29
was used to explore whether canine sex (male, female),
age (6 months, 6–12 months, 1–5 years, 5 years), pedi-
gree status (i.e., pedigree vs. crossbreed) and/or length of
pet ownership (6 months, 6–12 months, 1–5 years, 5
years) predicted hypoglycemia-response likelihood. The
WELLS ET AL.1236

qualitative information that owners provided on their dog’s
behavioral responses to their hypoglycemic episodes was
categorized to establish how many dogs were reported to
show specific types of reactions.
Ethical approval
Full ethical permission for the study was granted by the
Research Ethics Committee, School of Psychology, Queen’s
University Belfast.
Results
Participants
Two hundred and twelve people chose to participate in
the study, 206 (97.2%) via the online survey option and six
(2.8%) by way of the hardcopy postal questionnaire. Demo-
graphic information on the sample can be found in Table 1,
which shows that the majority of respondents were female,
between 18–45 years of age, married with no children, and
in full-time employment.
Participants’ health
Most of the respondents reported having suffered from
medically diagnosed type 1 diabetes for longer than 10 years,
and they considered their control of the condition to be
“quite good” (Table 2). Latest HbA1c readings varied con-
siderably (range: 3.9%–18.0%) among those people who
knew their most recent outputs (n146; 68.9%), with a mean
reading of 7.8% (95% confidence interval [CI] 7.4–8.0),
slightly above the satisfactory control range of 6.5%–7.5%.
Hypoglycemia awareness
Most respondents reported having symptoms of low
blood sugar “a few times a week,” and they considered them-
selves to be “quite good” at recognizing their own symp-
toms of hypoglycemia (Table 2). Sweating, poor concentra-
tion, and trembling were the most commonly reported
symptoms of low blood sugar, although a smaller propor-
tion of the sample also indicated experiencing headaches,
breathing problems, and seizures (Table 3).
Dog ownership
Information on the types of dogs owned by the respon-
dents is provided in Table 4. Most of the dogs were male,
over one year of age, and had been owned for over 12
months. Seventeen people (8.0%) owned more than 1 dog.
Significantly (p0.001, binomial test) more pedigree dogs
were owned than cross-breeds. The pedigree animals com-
prised 45 different breeds, ranging from large, working
breeds (Newfoundlands, German Shepherds, Border Collies)
through to smaller, toy breeds (Chihuahuas, Bichon Frisé,
Yorkshire Terriers). The most commonly kept breeds were
Labrador Retrievers (owned by 23 of the respondents), Bor-
der Collies (n14), and Golden Retrievers (n10).
Dogs and hypoglycemia awareness
Respondents were more likely than not to report that their
dogs showed hypoglycemia awareness, i.e., that they had re-
acted to a hypoglycemic episode on at least one occasion (138
[65.1%] versus 74 [34.9%]; p0.001, binomial test). Binary
logistic regression analysis showed that canine sex, age, pedi-
gree status, and length of pet ownership did not significantly
CANINE RESPONSES TO HYPOGLYCEMIA IN HUMANS 1237
T
ABLE
1. D
EMOGRAPHIC
I
NFORMATION ON
THE
R
ESPONDENTS TO THE
S
TUDY
Demographic factor Number Percentage
Sex
Male 76 35.8
Female 136 64.2
Age (years)
18–35 80 37.7
36–45 59 27.8
46–55 42 19.8
56–65 22 10.4
over 65 9 4.2
Marital status
Married 102 48.1
Single 60 28.3
Cohabiting 36 17.0
Divorced 7 3.3
Widowed 7 3.3
Children
0 108 50.9
1 34 16.0
2 47 22.2
323 10.9
Occupational status
Employed full time 96 45.3
Unemployed 33 15.6
Employed part time 30 14.2
Student 26 12.3
Retired 22 10.4
Other 5 2.4
T
ABLE
2. P
ARTICIPANTS
’ R
ESPONSES TO
Q
UESTIONS
C
ONCERNING THEIR
D
IABETES
Response Number Percentage
Duration of illness
1 year 12 5.7
1–5 years 40 18.9
6–10 years 21 9.9
10 years 139 65.6
Diabetes control
Very good 42 19.9
Quite good 126 59.7
Quite poor 40 19.0
Very poor 3 1.4
Frequency of hypoglycemia episodes
Every day 13 6.1
A few times a week 99 46.7
A few times a month 48 36.8
A few times a year 22 10.4
Hypoglycemia awareness
Very good 61 28.8
Quite good 102 48.1
Quite poor 41 19.3
Very poor 8 3.8

(p0.05) predict a dog’s propensity to respond to its
owner’s episodes of hypoglycemia (Table 5). The Hosmer
and Lemeshow test demonstrated that the model adequately
fitted the data (
2
3.86, df 7, p0.79).
Most of the dogs who were reported to have responded
to their owner’s “lows” had done so several times, and in re-
sponse to most of the known hypoglycemic episodes (Table
6). The majority of dogs were also reported to respond to
these episodes “quite early on,” i.e., in the early stages of the
person realizing they were hypoglycemic. Just over a third
of the dogs were believed to have shown behavioral re-
sponses before the owners themselves were aware they were
hypoglycemic. Owners indicated that their dogs showed
“alert” behavior in a wide variety of contexts, ranging from
exercising to sleeping (Table 6).
The nature of the dogs’ behavioral responses to their own-
ers’ hypoglycemic episodes varied between animals. Most of
the dogs were perceived by their owners to engage in be-
haviors aimed at attracting the owner’s attention. Thus,
many were reported to nudge or nuzzle their owners with
their noses (n56; 40.6%), jump on top of them (n42;
30.4%) and/or lick at their hands (n30; 21.7%) and faces
(n38; 27.5%), particularly around the mouth.
Vocalizations, including whining (n45; 32.6%), barking
(n34; 24.6%), and howling (n6; 4.3%) were commonly
reported. Many participants (n53; 38.4%) also indicated
that their dogs became agitated, with some pacing back and
forth or jumping about. Other dogs appeared calmer, either
sitting directly in front of the owner (n14; 10.1%) or fol-
lowing the owner, almost like a shadow (n54; 39.1%). Re-
ports of dogs staring intently at the person’s face were rela-
tively common (n57; 41.3%).
Twenty-eight people (20.3%) indicated specifically that
their dogs woke them up from sleep (usually by displaying
one or more of the preceding behaviors); in most of these
cases (89.3%) the animal slept in an entirely different room,
and some dogs (n5; 17.8%) had to scratch or paw at a
closed door before gaining entry to the bedroom.
A smaller number of dogs showed behaviors somewhat
suggestive of fear, including trembling (n10; 7.2%), run-
ning into a different room (n7; 5.1%), and/or panting
(n3; 2.2%). Some owners (n15; 10.9%) also indicated
that their dogs tried to attract the attention of, and/or fetch,
someone else in the house. One woman indicated that her
dog had bitten her on the face after she had fallen into a di-
abetic coma.
None of the people who owned two or more dogs (n
17), indicated that more than one of their pets showed be-
havioral responses to their hypoglycemic episodes.
Discussion
The findings from this study suggest that behavioral re-
actions to hypoglycemic episodes in pet owners with type 1
diabetes commonly occur in untrained domestic dogs. Most
of the respondents to this survey reported owning a dog that
showed behavioral responses to hypoglycemia, and the fre-
quency and reliability with which these animals alerted their
owners to hypoglycemic episodes was relatively high. Over
a third (36.0%) of the owners in the sample believed that their
dogs reacted most times they went low, with the majority of
animals (40.8%) showing behavioral responses at roughly the
same time as the owners themselves were aware they were
hypoglycemic. Moreover, just over a third (33.6%) of the
dogs showed reactions before their owners had realized they
were low, prompting appropriate action on the part of the
owner (e.g., glucose readings, food consumption).
A wide variety of dogs were represented by the partici-
pants to the survey in terms of breed, sex, and age. Analy-
sis revealed no significant relationships, however, between
canine demographics and hypoglycemia response likeli-
hood. Thus, owners of Poodles and Bichon Frisés were just
as likely to report behavioral reactions to their hypoglycemic
episodes as the keepers of German Shepherds and Labrador
Retrievers. This is a positive finding in light of how difficult
WELLS ET AL.1238
T
ABLE
3. S
YMPTOMS OF
H
YPOGLYCEMIA
R
EPORTED BY THE
R
ESPONDENTS
Symptom Number Percentage
Sweating 143 67.5
Poor concentration 124 58.5
Trembling 113 53.3
Confusion 109 51.4
Irritability 106 50.0
Disorientation 97 45.8
Poor coordination 89 42.0
Hunger 85 40.1
Blurred vision 81 38.2
Weakness 78 36.8
Tiredness 78 36.8
Heart palpitations 71 33.5
Dizziness 67 31.6
Speech problems 54 25.5
Anxiety 47 22.2
Nervousness 38 17.9
Headaches 32 15.1
Loss of consciousness 22 10.4
Breathing problems 14 6.6
Seizures/fits 14 6.6
Coma 1 0.5
T
ABLE
4. D
EMOGRAPHIC
I
NFORMATION ON
THE
R
ESPONDENTS
’ D
OGS
Demographic factor Number Percentage
Sex
Male 118 55.7
Female 94 44.3
Age
6 months 3 1.4
7–12 months 15 7.1
1–5 years 99 46.7
5 years 95 44.8
Length of time owned
6 months 11 5.5
7–12 months 11 5.2
1–5 years 95 44.8
5 years 95 44.8
Breed status
Pedigree 146 70.2
Mongrel 62 29.8

the selection and recruitment of dogs for training purposes
can be.
16,20
The reported behavioral reactions of dogs to their
owner’s hypoglycemic episodes differed between animals.
Most dogs (64.5%) were reported to behave in a manner
suggestive of attracting their owners’ attention, for exam-
ple, by jumping up toward the owner’s face, licking the
owner, and/or barking. Many of these dogs acted in ways
that reduced the physical distance between themselves
and their owners, with some animals attempting to sit on
top of their caregivers or following them around like a
shadow; increased eye contact, with dogs looking at their
owners more frequently, also appeared to be relatively
common.
A smaller proportion of dogs were perceived by their own-
ers to be quite frightened by their blood sugar drops, mov-
ing themselves into another room, trembling, and/or hy-
perventilating. Interestingly, all of the multiple dog-owning
participants, reported that only one of their pets showed
“alert” behavior. This implies that not all dogs are reactive
to the cues that signal hypoglycemia, even when emitted by
the same individual.
CANINE RESPONSES TO HYPOGLYCEMIA IN HUMANS 1239
T
ABLE
5. R
ESULTS OF
B
INARY
L
OGISTIC
R
EGRESSION
A
NALYSIS FOR
H
YPOGLYCEMIA
-R
ESPONSE
L
IKELIHOOD
,
WITH
P
REDICTOR
V
ARIABLES OF
D
OG
S
EX
, A
GE
, B
REED
S
TATUS
,
AND
L
ENGTH OF
P
ET
O
WNERSHIP
Exp (B)
Variable Odds ratio 95% CI p Value
Dog sex 0.91 0.50–1.65 0.76
Dog age 0.61
Dog age 0.64 0.03–13.76 0.77
Dog age 1.93 0.40–9.30 0.41
Dog age 1.67 0.65–4.29 0.28
Breed status 1.64 0.83–3.27 0.16
Length of ownership 0.50
Length of ownership 2.13 0.33–13.57 0.42
Length of ownership 1.76 0.36–8.69 0.48
Length of ownership 0.73 0.28–1.86 0.51
Constant 0.32 0.003
CI, confidence interval.
T
ABLE
6. P
ARTICIPANTS
’ R
ESPONSES TO
Q
UESTIONS
C
ONCERNING
T
HEIR
D
OGS
’
R
EACTIONS TO
T
HEIR
O
WNER
’
S
H
YPOGLYCEMIC
E
PISODES
Response Number Percentage
Number of hypoglycemic episodes
1 35 25.4
2–5 42 30.4
6–10 17 12.3
1144 31.9
Frequency of reactions
Every time 22 16.2
Most times 49 36.0
Only occastionally 32 23.5
Hardly ever 33 24.3
Speed of detection
Very early on 42 33.6
(before the owner is aware he/she is
becoming hypoglycemic)
Quite early on 51 40.8
(in the early stages of the owner’s realizing
that he/she is hypoglycemic)
Quite late on 24 19.2
(after the owner has taken action to raise
sugar levels)
Very late on 8 6.4
(after the owner has lost consciousness)
Activities engaged in by owner
Resting 55 31.4
Moving about 52 29.7
Sleeping 49 28.0
Exercising 19 10.9

This study was purely concerned with elucidating dogs’
reactions to hypoglycemia in people with type 1 diabetes.
That said, a few people with type 2 diabetes (n14) also
contacted the investigators, indicating that their dogs reli-
ably showed “alert” behavior. Likewise, five owners of cats
reported the same phenomenon, in all cases waking their
owners up from nocturnal episodes of hypoglycemia. Fur-
ther work is needed to explore the range of animal species
that might be responsive to changes in the diabetic state.
The findings from this investigation lead one to question
what signals dogs might be detecting in their hypoglycemic
owners. Odor cues appear, at first glance, to be the most
plausible explanation. Chen and colleagues,
26
for example,
found that one dog reportedly exhibited “alert” behavior
when its owner was asleep, and possibly emitting no cues
other than olfactory ones. Similar findings were shown by
this survey. Indeed 23.1% of the participants reported that
their dog had, on at least one occasion, woken them up from
a nocturnal “low.” In most of these cases the dog slept in an-
other room of the house, and in some situations the door of
the owner’s bedroom was closed, forcing the animal to
scratch or paw at the obstacle.
Increases in sweating have been repeatedly noted in hy-
poglycemic individuals.
21
As in the case of cancer, it is pos-
sible that dogs can detect changes in the chemical composi-
tion of their owners’ sweat, using their acute sense of smell.
However, it cannot be ignored that dogs may respond to
other cues besides olfactory ones, including, for example,
subtle changes in their owner’s mood (with people often be-
coming more irritable as their sugar levels drop) or visual
signals related to their owner’s behavior (with some people
trembling, becoming disorientated, losing consciousness, or
having seizures). The possibility that individual animals em-
ploy multiple signals, or that different dogs use entirely dif-
ferent cues equally cannot be dismissed. Research is required
to determine what types of cue/s dogs employ to detect hy-
poglycemia. By establishing what it is that dogs are reacting
to, it will it be possible to train such animals to alert in the
correct manner.
Although not eligible for participation in this investiga-
tion, a female owner of a trained hypoglycemia alert dog, re-
ported an inability of the animal to detect decreases in her
sugar levels while she slept, presumably because the animal
had only been trained to alert to subtle visual changes in be-
havior associated with her sugar level’s dropping; in this case
rubbing of her hands over her face was the visual cue.
This investigation benefits from having used a relatively
large cohort of people with type 1 diabetes, all with first-
hand experience of their pets’ reactions to their hypo-
glycemic episodes. The participants, however, represented a
self-selected sample. People owning dogs with a propensity
to show overt “alert” behaviors, may thus have been more
inclined to take part in the study than those whose pets
showed little or nothing in the way of a behavioral response.
That said, a published abstract of a questionnaire-based
study conducted in Australia indicated that 72 of 106 peo-
ple with diabetes whose dogs had witnessed a hypoglycemic
event, reported behavioral reactions in their pets to their
“lows.”
30
Although the methods from this abstract are un-
clear, taken together, these two studies provide convincing
evidence for a proportion of dogs being able to sponta-
neously detect hypoglycemia-related cues in their owners.
Some dog owners may not be attuned to their pets’ reactions
to their hypoglycemic episodes, which may have underesti-
mated the number of dogs that can detect low blood sugar.
In light of this, it is difficult to draw finite conclusions on
what proportion of the canine population may be reactive to
changes in their owners’ diabetic states, except to say that
this is a commonly reported phenomenon.
The present study provides a useful basis for further, more
controlled experimentation in this area. A prospective, lon-
gitudinal approach, that uses owners’ blood glucose read-
ings as confirmation of hypoglycemia detection by dogs,
could yield objective and valuable data and would comple-
ment the more cross-sectional, retrospective design adopted
here. The prospect of continued work in this area will ad-
vance our understanding of the value of domestic dogs as
an alerting mechanism for people suffering from diabetes
and other chronic conditions, e.g., epilepsy.
Conclusions
The findings from this study suggest that many dogs can
detect hypoglycemia, often without the use of visual cues
and before the animals’ caregivers are aware of their own
symptoms. Although it was not the goal of this project to ex-
plore how dogs detect hypoglycemia, the results hint at an
odor cue, although other signals (e.g., changes in owner be-
havior due to impaired cognitive functioning) cannot be dis-
missed. Research is required to elucidate what mechanism/s
might underlie the ability of dogs to detect hypoglycemia
and to determine whether animals can be trained to consis-
tently alert their owners to the onset of hypoglycemia.
Acknowledgments
The authors thank Diabetes UK for funding this research,
and all of those individuals who participated in the study.
Thanks also to Geoff Caves, School of Psychology, Queen’s
University Belfast, for his help with the online survey. The
corresponding author is an employee of Queen’s University
Belfast; both of the co-authors are employees of the Univer-
sity of Lincoln. None of the authors have any competing in-
terests to declare.
Disclosure Statement
No competing financial interests exist.
References
1. Pet Food Manufacturer’s Association. Profile 2007. PFMA
2008.
2. Friedmann E, Thomas SA, Eddy TJ. Companion animals and
human health: Physical and cardiovascular influences. In:
Podberscek A, Paul ES, Serpell JA, eds. Companion Animals
and Us. Cambridge: Cambridge University Press,
2000:125–142.
3. Hart LA. Community context and psychosocial benefits of
animal companionship. In: Fine A, ed. Handbook of Ani-
mal-Assisted Therapy. London: Academic Press, 2006:73–94.
4. McNicholas J, Gilbey A, Rennie A, Ahmedzai S, et al. Pet
ownership and human health: A brief review of evidence
and issues. BMJ 2005;331:1252–1255.
5. Wells DL. Domestic dogs and human health: An overview.
Br J Health Psychol 2007;12:145–156.
WELLS ET AL.1240

6. Wilson CC, Turner DC. Companion Animals in Human
Health. London: Sage Publications, 1998.
7. Williams H, Pembroke A. Sniffer dogs in the melanoma
clinic? Lancet 1989;i:734.
8. Church J, Williams H. Another sniffer dog for the clinic?
Lancet 2001(9285);358:930.
9. Dobson R. Dogs can sniff out first signs of men’s cancer.
Sunday Times 2003;27th April.
10. Fraser L. Scientists put sniffer dogs on the scent of men with
cancer. Sunday Telegraph 2002;2nd June.
11. Willis CM, Church SM, Guest CM, Cook WA, et al. Olfac-
tory detection of human bladder cancer by dogs: Proof of
principle study. BMJ 2004;329:712–716.
12. Leahy M. Olfactory detection of human bladder cancer by
dogs: Cause or association? BMJ 2004;329:1286.
13. Pickel D, Manucy GP, Walker DB, Hall SB, et al. Evidence
for canine olfactory detection of melanoma. Appl Anim Be-
hav Sci 2004;89:107–116.
14. McCulloch M, Jezierski T, Broffman M, Hubbard A, et al.
Diagnostic accuracy of canine scent detection in early- and
late-stage lung and breast cancers. Integr Cancer Ther
2006;5:30–39.
15. Brown SW, Strong V. The use of seizure-alert dogs. Seizure
2001;10:39–41.
16. Dalziel DJ, Uthman BM, McGorray SP, Reep RL, et al. Sei-
zure-alert dogs: A review and preliminary study. Seizure
2003;12:115–120.
17. Kirton A, Wirrell E, Zhang J, Hamiwka L, et al. Seizure-alert-
ing and -response behaviors in dogs living with epileptic
children. Neurology 2004;62:2303–2305.
18. Strong V, Brown SW, Walker R. Seizure-alert dogs—fact or
fiction. Seizure 1999;8:62–65.
19. Strong V, Brown S, Huyton M, Coyle H. Effect of trained
Seizure Alert Dogs on frequency of tonic-clonic seizures. Sei-
zure 2002;11:402–405.
20. Strong V, Brown SW. Should people with epilepsy have un-
trained dogs as pets? Seizure 2000;9:427–430.
21. McAulay V, Deary IJ, Frier BM. Symptoms of hypogly-
caemia in people with diabetes. Diabet Med 2001;18:690–705.
22. Smith D, Amiel SA. Hypoglycaemia unawareness and the
brain. Diabetologia 2002;45:949–958.
23. Aussedat B, Thome-Duret V, Reach G, Lemonnier F, et al.
A user-friendly method for calibrating a subcutaneous glu-
cose sensor-based hypoglycaemic alarm. Biosens Bioelectron
1997;12:1061–1071.
24. Harris ND, Ireland RH, Marques JLB, Hudson S, et al. Can
changes in QT interval be used to predict the onset of hy-
poglycemia in type 1 diabetes. Comput Cardiol 2000;27:
375–378.
25. Lim K, Wilcox A, Fisher M, Burns-Cox CJ. Type 1 diabetics
and their pets. Diabet Med 1991;9:S3–S4.
26. Chen M, Daly M, Williams N, Williams S, et al. Non-inva-
sive detection of hypoglycaemia using a novel, fully bio-
compatible and patient friendly alarm system. BMJ 2000;321:
1565–1566.
27. Tauveron I, Delcourt I, Desbiez F, Somda F, et al. Canine de-
tection of hypoglycaemic episodes whilst driving. Diabet
Med 2006;23:335.
28. O’Connor MB, O’Connor C, Walsh CH. A dog’s detection
of low blood sugar: A case report. Ir J Med Sci 177:155–157.
29. Brace N, Kemp R, Snelgar R. SPSS for Psychologists. Hamp-
shire: Palgrave Macmillan, 2003.
30. Stocks AE. Can dogs help patients with hypoglycaemia? Di-
abetologia 2002;45:A79.
Address reprint requests to:
Deborah L. Wells, Ph.D.
School of Psychology
Queen’s University Belfast
Belfast BT7 1NN
Northern Ireland
United Kingdom
E-mail: d.wells@qub.ac.uk
CANINE RESPONSES TO HYPOGLYCEMIA IN HUMANS 1241