Dogs 'warn diabetics' after smelling low blood sugar

“Dogs could be trained to warn diabetic patients when their blood sugar levels are about to become low,” The Daily Telegraph reports.

The story comes from a study of 17 people with diabetes who had been given a dog trained to sniff out and alert them when their blood sugar (glucose) levels were too low (hypoglycaemia).

Hypoglycaemia is potentially serious and if left untreated could result in coma.

During interviews the owners reported the dogs had improved their lives and helped with their diabetes. Blood test results confirmed the perception that the dogs could detect glucose levels outside of a desired range in many cases, and that having a dog made the owner more likely to remain in a desired range. 

These were encouraging results but they were based on a very small sample of people and were not always consistent. So, the results should be interpreted with some caution.

Another practical consideration is that the supply of ‘diabetes-sniffing’ dogs is limited. The UK charity that trains the dogs used in the study – Medical Detection Dogs – has a waiting list of three years for dogs.

If you are living with diabetes and are concerned that your symptoms are poorly controlled there are alternative options, such as going on a diabetes course, which helps you better understand and manage your condition.

Where did the story come from?

The study was carried out by researchers from the University of Bristol and the University of Dundee in collaboration with a charity called Medical Detection Dogs based in Milton Keynes. It was funded by The Company of Animals – a pet accessories company. The funders had no role in the study design.

The study was published in the peer-reviewed Public Library of Science (PLoS) One – a science journal. The journal is open access so the study is free to read or download.

The media reporting of the study was generally accurate.

What kind of research was this?

This was an observational study. The researchers wanted to test whether specially trained dogs were effective at alerting their owners, who had diabetes, when their blood sugar levels fell outside a normal range.

Diabetes is a condition where the body cannot control its blood sugar levels adequately. Too much glucose in the blood (hyperglycaemia) or too little (hypoglycaemia) can cause a range of medical complications in the short and long term.

This research aimed to focus on the dog’s ability to detect hypoglycaemia which is a relatively common state that in more extreme cases can cause unconsciousness, coma and even death.

Consequently, some people with diabetes report significant fear of hypoglycaemia and change their lifestyles to minimise the risk.

Early detection systems may be able to provide reassurance that the risk will be caught early and enable the person to live more independently with fewer worries.

Previous research, the study authors report, suggests pet dogs can spontaneously exhibit certain behaviours when their owner’s blood glucose levels decrease, such as barking, nuzzling, licking, biting or jumping up and staring at their owner. The theory is that they can use their acute sense of smell to sniff out the blood glucose changes through changes in the owner’s sweat or breath. This study aimed to test whether these preliminary reports were accurate.

What did the research involve?

The research involved interviewing 17 people with diabetes (16 had type I, one had type II) about their experiences of glucose management before and after getting the dog trained in detecting glucose levels.

Researchers visited the people’s homes to perform a structured interview with thirty four questions collecting information on:

• clients’ experiences with diabetes
• opinions of the value of their dog
• the frequency with which they recalled hypoglycaemia-related events prior to, and after acquiring the dog

Researchers read 10 statements to each client designed to assess the impact of the dog on their life and they were asked to rate (on a five point scale) the extent to which they agreed with each. (For example “I am more independent since I obtained my dog”).

A second phase of the study involved letting the researchers have access to past blood tests given to the dog charity before they received their detection dogs. This covered blood tests approximately one month before they got their detection dog. Participants were also asked to record their dog’s alerting behaviour to see what they did when they detected a problem.

The main analysis looked to see if the dog’s alert behaviour corresponded to the periods when blood test results showed hypoglycaemia, and whether the owners reported better controlled glucose levels after they were given the detection dog.

What were the basic results?

There was a wide range of participants aged from five to 66 years old who had lived with their detection dogs for anywhere from four months to seven years. Not all 17 participants completed all aspects of the interviews or blood test monitoring, and so responses are not always out of 17.

Main results from the interviews

When asked to recall the occurrence of hypoglycaemia, currently and before having a trained dog, all participants reported a reduction in either frequency of low blood sugar, unconscious episodes or paramedic call outs, six clients believed all three had been reduced.

The majority of clients “totally agreed” that they were more independent post-dog (12/16), whilst two “somewhat agreed” and two clients were “neutral”.

Almost all the clients (15), trusted their dog to alert them when their blood sugars were low and 13 also trusted them to alert when blood sugars were high (six totally, seven somewhat).

Main results from the blood tests

Overall there was statistically significant change after dog acquisition. In eight out of nine cases, there was a shift (an improvement) in the distribution of glucose levels relative to the client’s target range following the placement of their dog. In all cases, except one, there was an increase in the percentage of samples within target range post-dog, but the pattern of change differed between clients.

Blood tests from 8 out of 10 owners (who gave information) indicated the odds of the dogs giving an alert when the blood glucose levels were outside of a target range (that is too high or too low) were statistically significantly different from those taken at random.

That is, the dogs were better than chance at detecting glucose levels outside of the target range.

There was not much information to base the estimate of the accuracy of the dogs on, and it varied a lot. It should also be noted that one of the dogs was alerting its owner at random.

When they measured HbA1C, a commonly used biological indicator of longer-term blood glucose regulation, they found it showed a small but non-statistically significant reduction following dog placement.

How did the researchers interpret the results?

The researchers summed up that the “acquisition of a trained alert dog was greatly valued by the majority of this self-selected sample of medical alert-dog users. They believed their dog to reliably alert to changes in blood sugar and hence described increased independence since obtaining the dog. The population, overall, reported reduced unconscious episodes and paramedic call outs, which if accurate, is of great importance since not only does it represent increased health and safety to the client, but also potentially significant reduced costs in health care”.

Conclusion

This small study on trained blood glucose detection dogs suggests they are highly valued by their owners. The dog’s impact on maintaining blood glucose within a desired range appeared generally positive. However, it was less clear how beneficial this was at improving longer term diabetes control and reducing risk of disease complications. Particularly given that an important measure of longer term glucose regulation (HbA1C) showed no significant improvement.

The study was also quite small and not all of the 17 participants had usable information to analyse. Hence, its results may not be totally reliable and need to be confirmed by studies with more participants.

Another limitation was the interview data which may have been subject to recall bias.

Participants were asked to recall unconscious episodes and paramedic call outs related to blood glucose control before and after the introduction of the dog, which for some people was more than five years in the past. They may not have recalled this information accurately and may have been more inclined to remember more bad episodes before they had the dog because they liked having the dog and perceived it to be beneficial.

Using objective accounts of emergency call outs of hospital visits would have been a more accurate way of assessing benefit.

However, this would still not have been perfect, as people may have had good periods and bad periods of regulating their blood glucose levels (from changes in insulin regimes, doctors, stress, maturation etc.) that may have coincided with the detection dog’s arrival, rather than being caused by it.

The results clearly showed that the majority of the dog owners valued their dog, which is perhaps not surprising as there was presumably some process of applying to get the dog which required some desire to have one in the first place (selection bias).

However, it was less clear exactly how effective the dogs were at detecting wayward glucose levels.

The results, based on only a handful of participants, seemed to suggest there was a beneficial effect overall and for most participants, but it varied from dog to dog so the results may not be wholly reliable. 

Furthermore, there was no beneficial effect on the longer term measure of blood glucose regulation (HbA1c) after the owner received the dog. So the study provides no evidence that the dog may improve longer term diabetes control and reduce risk of disease complications, despite the perceived benefit from the owners.

It could be the case that the majority of participants who reported a greater sense of independence were benefiting from the psychological effect of owning a dog (a sense of companionship, security and so on) rather than long-term improvements in their physical symptoms.

A final point is that the current supply for trained dogs cannot meet demand. The UK charity that trains some of the dogs used in the study, Medical Detection Dogs, estimates that there is a three year waiting list for trained dogs.

If you are concerned that your diabetes may be poorly controlled ask your diabetes nurse of GP for advice. There may be a number of lifestyle changes, and in some cases treatments, that could help you. Read more about Living with diabetes.