Even though there is growing interest in and use of trained Diabetes Alert Dogs (DADs) as an alternative method of blood glucose (BG) monitoring in type 1 diabetes (T1D), no scientific studies have confirmed the ability of DADs to accurately detect hypo- and hyperglycemia. The proposed project would conduct the first preliminary investigation into the accuracy of Diabetes Alert Dogs (DADs), using scientifically rigorous research methods similar to those required to demonstrate accuracy in BG meters. Specifically, the primary aim of the project is to test the hypotheses that DADs are accurate at detecting hypo- and hyperglycemic blood glucose (BG) levels in adults and children with T1D. The proposed project has several other secondary aims, which include determining: 1) the glycemic thresholds and time course for hypo- and hyperglycemic DAD alerts, 2) the degree to which alert accuracy is consistent across different DADs, and 3) whether or not DAD owners appear to experience better psychosocial status and quality of life compared to normative data for the diabetes population. To achieve these primary and secondary aims, DAD alerts will be compared to time-stamped glucose readings obtained from a continuous glucose monitoring (CGM) device and a BG meter (BGM). A total of 10 adults with T1D, as well as 10 children with T1D and their parents, will participate in a 4 - week data collection protocol. During this 4-week protocol, T1D participants will wear a """"""""blinded"""""""" CGM device that does not display glucose readings on the receiver. They will also follow routine instructions for daily use of BGM, which includes obtaining a BGM reading whenever a DAD alert occurs. In addition, participants will utilize the event recorder on the CGM device to record the occurrence of DAD alerts. To collect more detailed information about DAD alerts, adults with T1D and parents of children with T1D will also use a previously developed computer framework to complete electronic Daily DAD Diaries on a study website to record information about DAD alerts. This information would include time of alert, type of alert (low/high), DAD behaviors during alert, location of the alert, whether the alert occurred during the day or nocturnally during sleep, and BGM readings obtained for the alert. At the end of the 4-week protocol, the time-stamped CGM, BGM, and Daily DAD Diary data will be matched and compared. Data analysis will utilize a Signal Detection Theory Approach, guided by Error Grid Analysis criteria, to determine the rate of Hits, Misses, and False Alarms in DAD alerts. Prior to beginning the 4-week protocol, participants will also complete a battery of questionnaires assessing diabetes-related psychosocial status and quality of life. These questionnaire scores will be compared to available normative data as a preliminary test of whether or not DAD owners show less diabetes-related emotional distress and enhanced quality of life. The findings of the proposed project will lay an important foundation for future research into and clinical use of DADs as a viable and reliable alternative to technological methods of BG monitoring.

Public Health Relevance

People living with type 1 diabetes (T1D) depend on blood glucose (BG) meters or other devices to tell them when their BG levels are too low or too high. A growing number of people with T1D are using Diabetes Alert Dogs (DADs), which are trained and reportedly able to detect low or high BG levels, then alert the person. However, there is no scientific evidence that DADs can accurately detect low and high BG, and the purpose of the proposed project is to conduct the first scientific study of DAD accuracy.

Agency
National Institute of Health (NIH)
Institute
National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
Type
Exploratory/Developmental Grants (R21)
Project #
1R21DK099697-01
Application #
8571079
Study Section
Special Emphasis Panel (ZRG1-EMNR-K (80))
Program Officer
Teff, Karen L
Project Start
2013-08-01
Project End
2015-07-31
Budget Start
2013-08-01
Budget End
2014-07-31
Support Year
1
Fiscal Year
2013
Total Cost
$237,000
Indirect Cost
$87,000
Name
University of Virginia
Department
Psychiatry
Type
Schools of Medicine
DUNS #
065391526
City
Charlottesville
State
VA
Country
United States
Zip Code
22904
Gonder-Frederick, Linda A; Grabman, Jesse H; Shepard, Jaclyn A et al. (2017) Variability of Diabetes Alert Dog Accuracy in a Real-World Setting. J Diabetes Sci Technol 11:714-719
Gonder-Frederick, Linda A; Grabman, Jesse H; Shepard, Jaclyn A (2017) Diabetes Alert Dogs (DADs): An assessment of accuracy and implications. Diabetes Res Clin Pract 134:121-130