During hospitalization in the intensive care unit (ICU), the occurrence of a blood glucose imbalance is frequent and associated with increased mortality. These observations have resulted in the hypothesis that intensive insulin therapy designed to control blood glucose would improve the prognosis of patients admitted into the ICU. In a prospective, randomized, single center study in a surgical ICU during which the majority of patients had undergone cardiac surgery, intensive insulin therapy with the objective to maintain glycemia below 110 mg/dl (6.1 mmol/L) provided a significant reduction in ICU mortality and hospital mortality compared to a group with a glycemic objective of 200 mg/dl.

In a recent published article, the beneficial effect of intensive insulin therapy seems less obvious in a randomized single center study in a medical ICU. One of the potential factors limiting the impact of a therapeutic strategy like this one is the absence of achieving strict glycemic control for all patients on intensive insulin therapy. Additionally, the implementation of such a therapeutic strategy results in an increased risk of hypoglycemia, the consequences of which on morbidity remain unclear.

The aim of our study is to determine, in a mixed population of medical and surgical patients admitted to the ICU, requiring artificial ventilation with a expected duration above 48 hours, the impact of effective strict glycemic control (<6,1 mmol/l) compared to a conventional glycemic control (<11mmol/l) on hospital mortality.

In both randomization arms, continuous insulin infusion will be used via the venous route of administration. Rapid action insulin Novorapid HM (Novo Nordisk, Copenhagen, Denmark) will be used.

ICU patient management requires many intravenously administered treatments in a limited number of venous lines (catecholamines, sedation, feeding, vascular filling, antiotics…). This situation does not enable to dedicate an infusion line for the intravenous administration of insulin. Despite continuous administration of insulin infusion, the concomitant administration of other treatments in the same infusion line obviously leads to significant variations in the flow of insulin actually delivered, which can lead to variations in blood glucose and adjustments secondary to the inappropriate dose of insulin. To limit this phenomenon, an OCTOPUS (Vygon, Ecouen, France) type infusion connector will be added. The infusion connector is made of 2 infusion lines one of which will be exclusively dedicated to insulin therapy subsequently limiting the risk of variations in insulin administration flow.

The determination of the number of subjects to include was carried out by using a 45% hospital mortality hypothesis in the conventional glycemic control group. and a 32 % hospital mortality hypothesis in the strict glycemic control group.

Inclusion Criteria:

• over 18 years of age
• requiring mechanical ventilation with an expected duration above 48 hours

Exclusion Criteria:

• admission for cardiac arrest
• admission for an attempt of drug autolysis or acute drunkenness
• admission for hyperosmolar and/or ketoacidosis coma
• admission for massive cerebral hemorrhage
• admission from an another ICU
• admission after surgery without any other organ failure than respiratory support (with FiO2 below 50% and PeeP below 5cm H2O)
• inclusion in an another interventional study
• patient or next of kind refusal of study participation
• pregnant women