Introduction: Hyperglycemia is commonly observed in patients hospitalized on intensive care units. It is has been demonstrated that normalization of blood glucose level using intensive insulin therapy significantly improves prognosis of these patients. The aim of our study was comparison of standard protocol of intensive insulin therapy used on cardiac surgery ICU in General University Hospital in Prague and computer algorithm MPC (Model Predictive Control).
Patients and methods: 20 patients with glycaemia higher than 6.7 mmol/l at the time of admission to ICU were included into the study, 10 subjects were randomized for standard treatment, 10 for treatment with MPC algorithm. Glycaemia was measured hourly during 48 hours, insulin infusion was rate was adjusted hourly in MPC algorithm or in 1-2 hours in standard protocol group.
Results: Blood glucose levels were in the target range significantly longer in MPC relative to standard protocol group (26.3 +/- 2.1 hrs vs 20.3 +/- 2.5 hrs). Mean blood glucose was also lower using MPC algorithm (6.47 +/- 0.11 vs 6.72 +/- 0.23 mmol/l). On the contrary the target range was established faster using standard protocol (8.9 +/- 1.2 vs 10.3 +/- 0.9 hrs), duration of hyperglycaemia was the same in both groups (7.3 +/- 1.9 in standard protocol vs 7.3 +/- 1.3 hrs in MPC algorithm). Average 48-hours insulin dose was higher in MPC than standard protocol group (230.2 +/- 38.8 vs 199.1 +/- 27.8 IU/48 hrs). 2 hypoglycaemic episodes occured in 2 patients in standard protocol group.
Conclusions: Our results show that the use of MPC algorithm result in more effective blood glucose control in critically ill patients than standard protocol.