Can diabetes alert dogs detect hypoglycemia in patients with type 1 diabetes?

ISRCTN	ISRCTN90927134
DOI	https://doi.org/10.1186/ISRCTN90927134
Secondary identifying numbers	OHSU IRB00010881; Jaeb Center for Health Research PPQ#10061006829

Submission date: 22/04/2016
Registration date: 26/04/2016
Last edited: 26/04/2016

Recruitment status: No longer recruiting
Overall study status: Completed
Condition category: Nutritional, Metabolic, Endocrine

Prospectively registered

Protocol

Statistical analysis plan

Results

Individual participant data

Record updated in last year

Plain English summary of protocol

Background and study aims
Trained dogs are increasingly being used to detect when people with type 1 diabetes have low blood sugar (glucose). No one knows how well these dogs work, yet patients are paying tens of thousands of dollars to purchase dogs from dog trainers. Doctors don't know what to tell their patients about the dogs because we don't know enough about them. The aim of this study is to find out how reliable trained dogs are at detecting low blood sugar levels. We will compare the dog alerts to blood sugar measurement tools that are already well-tested: fingerstick blood tests and a continuous glucose monitoring (CGM) device.

Who can participate?
Type 1 diabetes patients aged 2-80 who already use a trained dog to detect low blood sugar levels

What does the study involve?
The study lasts one week. Participants go about their usual lives while wearing a "blinded" CGM which measures glucose levels but the numbers are not visible to the participant. When their dog alerts, the participant carries out a fingerstick blood test and records any low blood sugar symptoms. Participants also complete a brief survey about low blood sugar and how well they think their dog works.

What are the possible benefits and risks of participating?
Participants will receive a copy of the CGM report at the end of their participation and will be paid for their time. Possible risks include problems with the CGM insertion including pain, bleeding or infection at the insertion site, or discomfort with extra fingerstick blood tests.

Where is the study run from?
Oregon Health & Science University (USA)

When is the study starting and how long is it expected to run for?
June 2014 to August 2015

Who is funding the study?
Jaeb Center For Health Research (USA)

Who is the main contact?
Dr Evan Los

Contact information

Dr Evan Los
Scientific

Mail Code: CDRC-P
707 SW Gaines Street
Portland
97239
United States of America

ORCID ID	0000-0002-7567-0178

Study information

Study design	Pilot study exploring the test characteristics (sensitivity, positive predictive value) of a trained dog to detect hypoglycemia under real-life conditions. The study also explores patient perceptions of dog reliability and subjective value.
Primary study design	Interventional
Secondary study design	Non randomised study
Study setting(s)	Home
Study type	Diagnostic
Participant information sheet	Not available in web format, please use the contact details below to request a patient information sheet
Scientific title	Reliability of trained dogs to detect hypoglycemia in patients with type 1 diabetes
Study objectives	1. Trained dogs will not be able to reliably detect and alert to hypoglycemia in patients with type 1 diabetes 2. Compared to a continuous glucose monitor (CGM) with established reliability data, trained dogs will provide inferior detection and alert capabilities in patients with type 1 diabetes 3. Trained dogs accurately alert to rate of change and absolute glucose values
Ethics approval(s)	Oregon Health & Science University Institutional Review Board, approved 31/03/2015, renewed 27/11/2015, IRB#00010881
Health condition(s) or problem(s) studied	Type 1 diabetes
Intervention	Use of trained dog to detect and alert to hypoglycemia events. We assess and compare accuracy to measurement tools with known accuracy - capillary glucose and continuous glucose monitoring. Continuous glucose monitors are blinded to allow for detection of unrecognized hypoglycemia (by either subject or trained dog). Detailed event diaries allow assessment of dog alerts and compare to time stamp of continuous glucose monitor measurement and capillary blood glucose.
Intervention type	Other
Primary outcome measure	1. Rate of correct identification and alert to hypoglycemia event by trained dog: 1.1. Rate of correct alert (CBG or CGM <70 mg/dL and dog alert prior to other measures) 1.2. Rate of delayed alert (CBG or CGM <70 mg/dL and dog alert after other measures) 1.3. Rate of missed alert (CBG or CGM <70 mg/dL and no dog alert) 1.4. Rate of incorrect alert (alert without the presence of hypoglycemia)
Secondary outcome measures	1. Mean and median time to alert after CGM <70 2. Rate of change of CGM value at time of dog alert 3. Total duration of time with CGM value <70 mg/dL per 24 hours) 4. Subjective confidence of dog’s master in the trained dog’s ability to detect hypoglycemia 5. Rate of hypoglycemia events for which dog is not present/not available 6. Rate of correct identification and alert to hyperglycemia event by trained dog at threshold designated by dog’s master
Overall study start date	01/06/2014
Completion date	19/08/2015

Eligibility

Participant type(s)	Patient
Age group	All
Sex	Both
Target number of participants	Originally targeted 15 subjects to gather 45 hypoglycemia events. Target number of events achieved after 8 subjects. Interim power analysis showed additional subjects would not provide additional statistical power so enrollment stopped at 8 subjects
Key inclusion criteria	Age 2-80 years with diagnosis of type 1 diabetes and current user of dog formally trained to detect hypoglycemia
Key exclusion criteria	1. Pregnancy 2. Unwilling to use blinded CGM device 3. Inability to speak, read, write and understand English language
Date of first enrolment	01/06/2015
Date of final enrolment	19/08/2015

Locations

Countries of recruitment

United States of America

Study participating centre

Oregon Health & Science University

97239
United States of America

Sponsor information

Jaeb Center for Health Research (USA)
Research organisation

15310 Amberly Drive Ste. 350
Tampa
33647
United States of America

Website	https://www.jaeb.org/
"ROR"	https://ror.org/04ezjnq35

Funders

Funder type

Research organisation

Jaeb Center For Health Research (USA)

No information available

Results and Publications

Intention to publish date	01/06/2016
Individual participant data (IPD) Intention to share	Yes
IPD sharing plan summary	Available on request
Publication and dissemination plan	Oral presentation of study findings at American Diabetes Association 76th Scientific Sessions; New Orleans, Louisiana; June 2016. Anticipate submission of manuscript of study results in May/June 2016.
IPD sharing plan