Sensitivity to carb-counting error in the T1D management

In type 1 diabetes (T1D) therapy, patients need to take an exogenous insulin dose at mealtime in order to compensate the rise in glycaemia caused by carbohydrates (CHO) intake. To properly tune the meal insulin dose, the patient needs to estimate the amount of CHO. Errors in CHO estimation are, however, common and can have a negative impact on the quality of glycaemic control. The aim of this work is to quantify how much different levels of carb-counting error affect the overall glycaemic control. This sensitivity study is performed in silico using the popular T1D patient decision simulator. In 100 virtual subjects simulated for 7 days, different levels of carb-counting error are generated with Gaussian distributions varying the error mean from -10% to +10% and standard deviation ranging from 0% to 50%. The effect of the error is evaluated by computing time inside (TIR), above (TAR) and below (TBR) the target glycaemic range using the absence of error as reference case. We found that random errors globally deteriorate the glycaemic control; systematic underestimation lead to, on average, up to 5.2% more of TAR than the reference case, while systematic overestimation results in up to 0.8% more of TBR. Such results could be useful to assist diabetologists in patients’ therapy adjustments or training.