Using Abbott Freestyle Libre CGMs on a canine

Update: It's now a while since I published this post about the inherent inaccuracy I have experienced with 21 Freestyle Libre continuous glucose monitors. In this update I include my updated graph and relationship between the CGM reading and blood taken from an ear prick. Needless to say I have not had any response from Abbott...

Before I get into this article I need to qualify what it’s about.? I have a dog that suffered sudden onset Type 1 Diabetes mellitus just short of a year ago.? He is raw fed (zero carbs) so we cannot use the approach of determining his insulin doses from how many grams of carb-rich kibble we would need to feed him.? Additionally, as he is raw fed, he does not show carbohydrate-induced blood glucose ‘spikes,’ other than when he might have some inflammation, eg ear infection.

To avoid having to undertake regular ear pricks to check his blood sugar, we resorted to using a Continuous Glucose Monitor (CGM) - we use the Abbott Freestyle Libre.? These are well-worth the cost in that they provide (relatively) stress-free patient monitoring.? While I understand that Abbott does not support use on dogs or cats, these CGMs are actually very widely used by veterinarians.? I have recently been informed that Abbott sales reps are, in fact, recommending them to vets in South Africa.

However, it’s not all plain sailing as the readout from the CGM is often very different, invariably higher, than that in the blood. Overtime I noticed that for actual blood glucose readings above 7 mmol/l, the CGMs start to over-read and that the CGM readout may be significantly in excess of that in the dog’s blood.

Our doggo has had 17 CGMs fitted consecutively over the past 8 months and the graph below shows the comparisons of blood glucose levels determined by ear prick (Accuchek Instant Glucometer) vs that reported by the CGM. The equation expressing the relationship between the two is also shown.? Some of the ear prick values were confirmed by venous blood draws undertaken by our vet.? We use a standard ear prick method using a commercial pressure safety lancet which minimises impact on the ear and always produces a drop of blood for the test.

The graph’s solid line represents a theoretical 1:1 relationship between the two measurements and, with the exception of values between 3 and 7 mmol/l, only three of the 55 data pairs above 7 mmol approximate this relationship.? All the rest lie off to the right and you will notice that while there are several reported in the CGMs range 20-25 mmol, the highest ear prick value was 19.7.? So, as the blood sugar increases above 7, there is a progressively increasing and fairly predictable over-read.

At the low end, below 4 mmol, we have also occasionally detected the reverse, ie the CGM may under-read. This is much more of a problem for safe pet management. While the values from the two methods are generally closest at the lower end of the scale, we have had instances where, for example, the CGM reports 4.6 mmol/l whereas the ear blood reveals a much lower 3.2 mmol/l. So, over reliance on the CGM can be risky and best to always routinely check low values.

So, using CGMs is absolutely fine provided the discrepancy is taken into account.? We use a table of values prepared using the equation shown above. At the outset I was told by a medicine specialist to be aware of high values from these CGMs, and to take care not to ‘manage insulin dosing for the highs’ as they could be less than reported. This would increase the risk of insulin-induced hypoglycaemia. We have heeded this advice religiously since we discovered the anomaly discussed here.

I am not a medical professional of any description, and I have not been able to find a reason for the above.? I have tried contacting Abbott to gain some insights but this is a company that appears to do its utmost to prevent unsolicited communications. So, for what my data may be worth to anyone with a diabetic pet, here they are. I would be most interested to hear of other pet-owner experiences with CGM measurement accuracy.