Blood Glucose Control by a Model Predictive Control Algorithm with Variable Sampling Rate Versus a Routine Glucose Management Protocol in Cardiac Surgery Patients: A Randomized Controlled Trial

Abstract

Elevated blood glucose levels occur frequently in the critically ill. Tight glucose control by intensive insulin treatment markedly improves clinical outcome. This is a randomized controlled trial comparing blood glucose control by a laptop-based model predictive control algorithm with a variable sampling rate [enhanced model predictive control (eMPC); version 1.04.03] against a routine glucose management protocol (RMP) during the peri- and postoperative periods. The study was performed at the Department of Cardiac Surgery, University Hospital. A total of 60 elective cardiac surgery patients were included in the study. Elective cardiac surgery and treatment with continuous insulin infusion (eMPC) or continuous insulin infusion combined with iv insulin boluses (RMP) to maintain euglycemia (target range 4.4-6.1 mmol/liter) were performed. There were 30 patients randomized for eMPC and 30 for RMP treatment. Blood glucose was measured in 1- to 4-h intervals as requested by each algorithm during surgery and postoperatively over 24 h. Mean blood glucose, percentage of time in target range, and hypoglycemia events were used. Mean blood glucose was 6.2 +/- 1.1 mmol/liter in the eMPC vs. 7.2 +/- 1.1 mmol/liter in the RMP group (P < 0.05); percentage of time in the target range was 60.4 +/- 22.8% for the eMPC vs. 27.5 +/- 16.2% for the RMP group (P < 0.05). No severe hypoglycemia (blood glucose < 2.9 mmol/liter) occurred during the study. Mean insulin infusion rate was 4.7 +/- 3.3 IU/h in the eMPC vs. 2.6 +/- 1.7 IU/h in the RMP group (P < 0.05). Mean sampling interval was 1.5 +/- 0.3 h in the eMPC vs. 2.1 +/- 0.2 h in the RMP group (P < 0.05). Compared with RMP, the eMPC algorithm was more effective and comparably safe in maintaining euglycemia in cardiac surgery patients.