Evaluation of a plasma insulin model for glycaemic control in intensive care

Hyperglycaemia is a common complication in the intensive care unit (ICU), and is associated with worsened outcomes. Model-based insulin therapy protocols have been shown to be safe and effective in intensive care. Such protocols rely on correct modeling of glucose-insulin dynamics. In particular, model-based control typically relies on insulin sensitivity (SI) metrics, which are heavily influenced by plasma insulin kinetics. Plasma insulin samples were taken as part of a sepsis study and compared to modeled plasma insulin. Samples were taken in septic patients at the onset of glycaemic control, and once the patient consistently met less than two of the SIRs criteria that help define sepsis. It was found that inter-patient insulin dynamics were more variable at the onset of insulin therapy, than in the later samples after sepsis abated. Overall, the model adequately captured crucial steady state dynamics. Transient dynamics in plasma insulin following a bolus were faster than modeled, indicating greater clearance of insulin than currently modeled.