Intensive insulin therapy: enhanced Model Predictive Control algorithm versus standard care

Objective To investigate the effectiveness of an enhanced software Model Predictive Control (eMPC) algorithm for intravenous insulin infusion, targeted at tight glucose control in critically ill patients, over 72 h, in two intensive care units with different management protocols. Design Comparison with standard care in a two center open randomized clinical trial. Setting Two adult intensive care units in University Hospitals. Patients and participants Thirty-four critically ill patients with hyperglycaemia (glucose >120 mg/dL) or already receiving insulin infusion. Interventions Patients were randomized, within each ICU, to intravenous insulin infusion advised by eMPC algorithm or the ICU’s standard insulin infusion administration regimen. Measurements and results Arterial glucose concentration was measured at study entry and when advised by eMPC or measured as part of standard care. Time-weighted average glucose concentrations in patients receiving eMPC advised insulin infusions were similar [104 mg/dL (5.8 mmol/L)] in both ICUs. eMPC advised glucose measurement interval was significantly different between ICUs (1.1 vs. 1.8 h, P