Diabetes Alert Dogs Perform Better Than Previously Thought

Liam Davenport

January 21, 2019

UK researchers have shown that in patients with type 1 diabetes who have well-trained medical alert "diabetes" dogs, the animals have greater sensitivity to changes in blood glucose than has been shown in previous studies.

Such dogs could therefore improve the quality of life for patients with type 1 diabetes, particularly those who are hypoglycemia unaware.

The authors, led by Nicola J. Rooney, PhD, from Bristol Veterinary School, University of Bristol, UK, studied 28 dogs and their human partners, and more than 4000 hypo- and hyperglycemic episodes.

Their research, published online on January 15 in PLOS One, showed that the dogs — which had gone through a rigorous training program — alerted their human partners to 83% of hypoglycemic and 67% of hyperglycemic episodes.

Four dogs even had a positive predictive value (PPV)for detecting high or low glucose episodes of 100%, and the median was 81%.

"We already know from previous studies that patients' quality of life is vastly improved by having a medical detection dog," Rooney commented in a press release from her institution.

"However, to date, evidence has come from small scale studies. Our study provides the first large-scale evaluation of using medical detection dogs to detect hypoglycemia."

"Since the usage of such dogs is growing, it's important that any dogs used for these purposes are professionally trained, matched, and monitored by professional organizations," Rooney stressed.

And she qualified: "Our research shows a dog's effectiveness is affected by the individual dog and its connection with its human partner."

Coauthor Claire M. Guest, chief executive and cofounder of Medical Detection Dogs, Milton Keynes, UK, which trained the dogs used in the study, commented: "The findings are fantastic news for all those who are living with type 1 diabetes and other conditions."

Dogs Are Another Option to Help Detect Out-of-Range Glucose Levels

The researchers say that a quarter of patients with type 1 diabetes are unaware of changes in their blood glucose levels, which increases their risk of severe hypoglycemia up to sevenfold.

Moreover, the fear of hypoglycemia, particularly during nighttime, can drive patients to manipulate their insulin levels to keep their blood glucose levels high, thus increasing the risk of complications from hyperglycemia.

Among the range of technologies available to help patients monitor their glucose levels, the glycemia alert dog has recently received attention, partly because of the noninvasive nature of the intervention.

Similar to dogs trained to detect contraband, these dogs are conditioned to respond with specific alerting behaviors when their owners' blood glucose fall outside a target range, known as an out-of-range (OOR) episode. This prompts the patient to test their blood glucose level and take appropriate action (eg, insulin administration or eating) to retain appropriate glucose levels.

Medical Detection Dogs is the only agency in the UK officially recognized to train diabetes alert dogs. To be accredited, dogs must show ≥ 75% sensitivity to hypoglycemic samples and a PPV > 85% (ie, 85% or more alerts occur when glucose levels are out of target range) over 3 months.

Bigger Study, Better Trained Dogs Show Better Results

For the study, the researchers examined the records of 27 dogs trained by the charity and their partners, patients with type 1 diabetes. There were 4197 hypo- and hyperglycemic episodes over a 6- to 12-week period.

The dogs had been accredited for a median of 1.5 years. Eight dogs were partnered with a child, and five had been the patient's own dog before training. In all, there were 16 female and 11 male participants with type 1 diabetes.

The team found that the dogs varied in performance, at a median sensitivity to OOR episodes of 70%, and 21 dogs had values ≥ 50%.

The median sensitivity to hypoglycemic episodes was 83%, while that for hyperglycemic episodes was 67%.

The median PPV was 81%, and only two dogs were incorrect for over half of their alerts. Four dogs had a PPV of 100%. There was no significant correlation between PPV and either low or high OOR episodes.

Analysis revealed that newly accredited dogs performed significantly better than unaccredited dogs and dogs that had been accredited for longer.

In addition, dogs that had previously been household pets were more sensitive to high, but not low, glucose levels than other dogs.

The current findings contrast with those of previous investigators, however, who found that dogs could not compete with, for example, continuous glucose monitoring (CGM).

Evan A. Los, MD, now at East Tennessee State University, Johnson City, conducted a study with his then colleagues at Oregon Health & Science University, Portland, in eight patients with type 1 diabetes and their dogs.

In comparing the animals' performance with CGM, they found that the device alerted the patient first in 73% of hypoglycemic events, an average of 22 minutes before the dog. Moreover, just 12% of the 16 to 20 weekly alerts from the dogs occurred when the patient actually had hypoglycemia.

Rooney and colleagues acknowledge that, in their study, the patients themselves provided the data, which means that some "less favorable" results could have been excluded, but they say they have that because the original purpose of providing data was to improve their individual dogs' training and consent for research was only given post hoc. "It is reasonable to assume that the majority of these data were authentic," they say.

Overall, they believe their better results "likely reflect the rigorous training and accreditation procedures" used by the charity that trained the dogs, as well the greater number of dogs and episodes included.

"Optimal performance of glycemic alert dogs depends not only on good initial and ongoing training, but also careful selection of dogs for the conditions in which they will be working," they conclude.

No funding for the study was declared. Lydia C.M. Swanson and Guest are full-time employees of the charity that trained the dogs. Rooney works primarily for the University of Bristol but is also a part-time paid employee of Medical Detection Dogs. Morant, who did the majority of the analysis, has no financial interest in Medical Detection Dogs.

PLoS One. 2019;14:e0210092. Full text

For more diabetes and endocrinology news, follow us on Twitter and on Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.