Lowering of Glucose in Critical Care (LOGIC): pilot trial in medical–surgical ICU patients

Background

Similar to cardiac surgery patients, medical–surgical critically ill patients may benefit from avoiding hyperglycemia. Our objective was to evaluate the feasibility of a randomized trial of target glucose ranges 5–7 mmol/l or 8–10 mmol/l.

Setting

A 15-bed medical–surgical ICU.

Methods

We randomized patients ages >18 years if they were expected to be in the ICU for >72 hours and had serum glucose >10 mmol/l. Patients were excluded if they had diabetic ketoacidosis, hypoglycemia, liver disease, or were palliative. Patients had concealed allocation to a target glucose range of 5–7 mmol/l or 8–10 mmol/l. Arterial catheters were used for sampling. Dedicated glucometers were calibrated daily to the laboratory measurement. We used enteral nutrition as early as possible; total parenteral nutrition was used in five patients. Insulin SC and oral hypoglycemic agents were prohibited. We mixed 50 U Humulin R in 50 ml 0.9% NaCl and began infusions at 0.5 U/hour; infusions were titrated according to pretested algorithms. Preceding this pilot was a survey of ICU staff, a glucometer reliability study, and a multidisciplinary educational program.

Results

The feasibility sample size was 20 patients. The APACHE II score was 32 (10.2); 14 were insulin-dependent pre-ICU, and all 20 patients were medical admissions. A total of 758 of 1607 values (47.2%) were in the 5–7 mmol/l range, and 256 of 660 values (40.2%) were in the 8–10 mmol/l range. Glucose < 2.5 mmol/l developed nine times; treatment included 0.5 A 50 g dextrose/100 ml H₂O on six occasions and stopping the insulin infusion in each case. No adverse consequences were observed.

Conclusions

In ICU patients with high illness severity, glucose values were in two target ranges most of the time, using well-accepted initiation and maintenance insulin infusion algorithms. A large randomized trial of glycemic control is feasible in this population, examining clinically important outcomes. However, this will require more intensive measures to achieve target values.