Continuous Glucose Monitoring and Bolus Calculators: A Good Combination

Introduction

The availability of reliable continuous glucose monitoring (CGM) systems not only allows CGM throughout the day but also provides information for choosing an appropriate prandial insulin dose. This applies to intensified conventional insulin therapy (ICT), in which patients with diabetes (PmD) themselves determine their insulin dose to cover basal and prandial insulin requirements, as well as to the use of automated insulin delivery (AID) systems.

The currently available AID systems cover the basal insulin requirement and, if necessary, correct existing excessively high glucose values by administering correction boluses ([advanced] hybrid AID systems). The next generation of AID systems (full AID systems) should also automatically cover the prandial insulin requirement. As long as these are not yet available, the use of bolus calculators (BCs) still makes sense, especially with hybrid AID systems. However, even with advanced hybrid AID systems, the correction bolus cannot always cover increased glucose levels if the administration of bolus insulin at mealtimes is not adequate. In contrast, in full AID systems, the BC is in principle part of the algorithm.

Bolus Calculators

In both ICT and insulin pump therapy, a significant part of which is delivered through a hybrid AID system, BCs can assist the PmD in determining the prandial insulin dose. A key function of BCs is to prevent hypoglycemia by overlapping the effect of the meal bolus with active insulin still in the body from previous insulin doses. Bolus calculators use algorithms that calculate a suggested insulin dose based on an estimate of the carbohydrate content of the current meal (entered manually by the user), the measured value of the preprandial glucose concentration, and the active insulin still in the body. The calculation should include the time of day as well as the “insulin sensitivity”, estimated via the carbohydrate-insulin factor (by how much the glucose concentration is lowered by one unit of insulin). The subcutaneous application of this insulin dose before the meal should result in glycemia being within a predefined value corridor/meeting a glucose value as well as possibly at a certain time after the meal.

A BC can be integrated either directly into the respective device (blood glucose monitoring system, CGM system, insulin pump, AID system) or as a stand-alone app on a smartphone.

Because the BC also knows the last prandial insulin doses (or all insulin doses in the case of insulin pumps), it can take into account the insulin effect still present, that is, reduce the dose appropriately. Depending on the dose used, the insulin effect may last longer than the interval between two meals. The effect of the current insulin dose is then added to the residual effect still present; it is important to avoid late postprandial hypoglycemia. Therefore, the BC algorithm basically uses the structure: “Bolus according to the amount of carbohydrate minus still effective insulin.” The insulin that is still effective is all the insulin that is still available, regardless of the reason for which it was given (meal bolus, correction bolus, also basal insulin). Although the different BCs use different algorithms for the calculation (also due to patent claims), the recommendations given in the end are similar. In addition, there are the correction boluses.

With AID systems, the relevance of the meal boluses is lower because these systems automatically deliver less insulin in the period after inadequately high boluses (adaptive basal administration) and can thus largely compensate for overlapping (currently manual) meal boluses. This adaptation of the basal insulin supply to the current need does not exist with subcutaneous insulin application; once an insulin dose has been given, it is effective in the body.

Bolus Calculators and CGM Systems

Until now, users have usually transmitted the glucose value to the BCs, which was determined by a conventional plasma glucose measurement in a capillary blood sample. With the use of CGM systems, not only are there considerably more glucose values available, almost without gaps, from the time before the meal, but these also provide clear information about the glucose trend (falling, constant, rising). The assumption is that this additional information can be used for a more “qualified” determination of the prandial insulin dose.

By using so-called artificial intelligence (AI), it is possible to evaluate whether the selected insulin dose was adequate or not by analyzing the postprandial glucose course. The question always arises as to what is to be understood by AI; in many cases, these are simple algorithms that calculate certain parameters from CGM curves. In the sense of “individualization” of the AID systems, which “learn” how their user (his glucose course) reacts to different factors, optimization of the (prandial) glucose control should be possible in the future. However, this does not affect the problem of an incorrect estimation of the carbohydrate content of the meal.

Because the advantages of this combination are obvious, it is not surprising that there are various patents on the use of CGM in BC. This should make it possible to address a serious diabetological problem. However, so far there are rather few publications about this topic.^1-5

Another approach (with a “Smart BC”) tries to take into account the rather strongly changing insulin demand of the PmD (in response to fluctuations in insulin sensitivity). ⁶ The BC automatically adjusts the insulin dose to the current individual insulin sensitivity. Fifteen PmD with type 1 diabetes who used a CGM system and an insulin pump participated in two experimental days, each lasting 24 hours, in a hotel. On the afternoons of these days, they received a standardized dinner after 45 minutes of exercise. The bolus used was calculated from a standard BC or the Smart BC. The glucose course in the four hours thereafter was compared: When the dose was determined by the BC with CGM connection, a reduction in postprandial hypoglycemia was observed, without an increase in hyperglycemia during this time.

A more recent publication describes the results of a clinical evaluation of a BC combined with a CGM system (CIBC), resulting in automatic insulin dose adjustment. ⁷ A multicentre study included PmD with type 1 diabetes aged 6 to 70 years. They used an AID system (Omnipod 5) in the first phase in manual mode without a connection to a CGM system with a conventional BC for seven days and then over seven days with such a connection. Significantly fewer low glucose values <70 mg/dl were observed in the 25 study participants in the four-hour post-bolus period (2.1% vs 2.8%, P = .03), while there were no differences in elevated values (>180 mg/dl) and values in the target range (70-180 mg/dl). This effect is a mixed one, that is, here the AID system acts paired with the BC.

Summary and Outlook

For insulin delivery in the context of ICT or insulin pump therapy, the combination of BC with a CGM system is a good one because of the better data basis. The combination of CGM and BC thus represents an important next step in optimizing glucose control of PmD. This will also improve user safety by reducing the risk of acute derailments. This can be seen as a clear medical need, which is necessary for the development of appropriate products.