Bolus advisor masterclass part 2: Post-prandial correction

Cotswold Falconry Centre, April 2017

The first instalment of feedback from the recent course I attended was mostly about adjusting insulin dosage to account for fat and protein content of meals. The second half of the course left several delegates behind...

Post-prandial correction doses

This was by far the nerdiest section of the course, and took quite a bit of concentration and asking the presenters to 'just say that again more slowly.' It was all about how the bolus advisor technology built into blood glucose meters and insulin pump handsets works out correction doses of insulin when blood glucose is high following a meal.

Unless you have a fully functional pancreas, you cannot avoid your blood glucose rising after a meal, even if you have injected the 'right' amount of rapid-acting insulin. Trials have shown that for the rapid insulins currently on the market, the ideal time to inject is 15-20 minutes before a meal. This is usually impractical, because you don't know how much insulin you will need until the food is in front of you, and then you don't want to wait 15-20 minutes before eating it. So the period when the injected insulin is reaching the peak of its action ('offset time') lags behind peak glucose entering the bloodstream, and this is one reason why post-prandial blood glucose tends to rise more than for a person without diabetes ('meal rise'). But if you've worked out the right amount of insulin, your blood glucose should return to 'normal' levels within four hours, which is the 'acting time' for rapid insulin.

So if you monitor your blood glucose less than four hours after your last bolus or injection, the blood glucose level that you see may actually decrease further without any action from you, due to 'insulin on board' - active insulin still in your system. So how do you know whether to correct it or not? At any time within four hours of your last injection, how high is 'too high'?

Imagine a scenario where your blood glucose level is within the ideal range before a meal, you have counted 60g carb in your meal and your insulin to carb ratio is 1 unit for every 10g - this means you will need 6 units of insulin. Roche told us that other manufacturers' algorithms assume that all insulin injected is 'active' insulin. So if your correction ratio is 1 unit of insulin to reduce your blood glucose by 3 mmol/L, then immediately after the meal your blood glucose could be up to 18 mmol/L higher than its pre-meal level and you would not be advised to take a correction dose because of the 6 units of active insulin. Roche also told us that other manufacturers assume a linear reduction in blood glucose, so after 2 hours your blood glucose could still be up to 9 mmol/L higher than the pre-meal level and no correction would be advised.

Fig 1. Correction is not advised if blood glucose falls below the line

This is not good. Even in the worst case scenario, blood glucose should not rise this high after a meal. There's no perfect number to aim for, but (assuming the meal wasn't Frosties) I would be happier with a meal rise of no more than 4 mmol/L, and 3 mmol/L would be even better.

Roche's algorithm makes quite a different set of assumptions, the main one being that only pre-prandial correction doses (insulin injected because pre-prandial blood glucose is too high) count as active insulin after the meal. Insulin injected for carbs is accounted for, and is not available to act on a high post-prandial blood glucose level. They say they have evidence to support this assertion.

So for Roche, a correction is required if the blood glucose rises higher than the 'meal rise' setting during the 'offset time'. After that a linear decrease to the pre-prandial level by the end of the 'acting time' is assumed. If meal rise is set to 4 mmol/L, offset time is 1 hour and acting time is 4 hours, then a blood glucose rise of 9 mmol/L at 2 hours would definitely suggest a correction. The shorter the offset time and the smaller the meal rise settings in the handset, the more aggressive the correction regime. For someone frail and elderly or prone to hypos it makes sense to have a higher meal rise and a longer offset time, to minimise risk of over-correction and hypoglycaemia.

Fig 2. Correction is indicated for the same post-prandial blood glucose level as Fig 1

Blood glucose correction after snacks

This was the hardest part of the course to understand, and therefore to explain. It would be so much simpler if people with Type 1 diabetes didn't eat between meals! But given that they do, they need advice on whether to correct blood glucose after a snack. The aim of the 'snack size' setting is to determine whether to apply a 'meal rise' and 'offset time' or not.

The 'meal rise' setting doesn't change depending on the size of the meal; it is the same whether the meal is small or large. So the 'snack size' setting is the carb threshold between applying a meal rise or not. If 'snack size' is set to 20g of carb, then for snacks up to this amount correction will be indicated afterwards if blood glucose is above the line in the previous graph. If a 'snack' 2 hours after a meal contains 30g of carb then the meal rise is applied at that point and a new graph is drawn, with corrections only advised for blood glucose levels above the new line.

Fig 3. Presence of the meal rise allows post-prandial blood glucose to be higher without advising correction

This is pretty sophisticated stuff, and I'm pretty sure that none of our patients understand how these settings are used. I'm only just working it out as I write this. What it boils down to is that most people using a basic type of meter have to take a stab in the dark when correcting post-prandial blood glucose levels, but people who are using this technology should get a good indication of whether to correct and how much insulin to give, as long as the settings have been adjusted to meet their particular requirements.

I always try hard to make sure that the main insulin to carb ratio and correction factors are right, but I have been less attentive in the past about the meal rise and snack size, because up to now I didn't understand what they were for.

The last point to mention is what we can do for people using pumps and meters that use the linear algorithm in Figure 1, which don't give useful advice about post-prandial correction. The team delivering the training suggested shortening the acting time setting to 3 hours instead of 4, because then at least a few more high blood glucose levels will fall above the line. This is not ideal, but the best they could come up with.