Background: Among critically ill patients, glycemic control reduces mortality and morbidities, but the use of intravenous insulin infusion is complicated by hypoglycemia. Having a standardized algorithm increases the likelihood of effective and safe utilization of intravenous insulin therapy. A tabular dose-defining protocol for intravenous insulin infusion is described, containing design elements intended to minimize risk for hypoglycemia while seeking control in a narrow target range, and performance is evaluated among critically ill trauma service patients.
Methods: The protocol assigns insulin infusion rate (IR) for ranges of blood glucose (BG). The columns are arranged in order of increasing maintenance rate (MR) for insulin infusion. Patient column assignment is determined according to rate of change of BG. During stable column assignment, the IR is a function of column MR and BG. Within-column, the protocol formula provides that (a) for BG between 70 mg/dL and target BG, the IR increases exponentially to the column MR; and (b) for BG above upper target BG range, the IR increases linearly as an adaptation of the rule of 1800, with slope determined by the column MR. Values for IR calculated by formula are rounded to correspond to BG ranges of the table. Performance was assessed in 27 sequential runs among 24 trauma service patients admitted to a surgical intensive care unit (2004-2005).
Results: Using point-of-care measurements, mean preinfusion BG was 230.0 +/- 67.9 mg/dL. BG < 140 mg/dL was reached during all 27 runs (median time 5.0 h), and target BG was < 110 mg/dL during 25 runs (median time 11.0 h). For the group of runs attaining target before interruption of insulin infusion, the average +/- SD of the principal measure of glycemic control, the within-run mean BG, was 113.7 +/- 14.8 mg/dL (coefficient of variation 13%, n = 25 runs). After attaining target, the average within-run SD for BG was 22.9 mg/dL. The within-run frequency of hypoglycemic measurements (BG < 70 mg/dL) as a percentage of BG determinations was 2.4%. In this series, no instance of BG <50 mg/dL was seen.
Conclusions: This report describes a nurse-implemented tabular protocol for intravenous insulin infusion having the advantages of efficacy, safety, and simplicity of use. Wide variability of IR in the neighborhood of BG 110 mg/dL is associated with stable BG response, and protection against hypoglycemia is achieved by rapid decline of IR at BGs in or below the target range.