The effect of a computer-assisted insulin protocol on glycemic control in a surgical intensive care unit

Background: Computer-assisted insulin protocols (CAIPs) contain complex mathematical algorithms to assist with insulin dosing. This study compared the quality of glucose control utilizing a CAIP with a paper-based insulin protocol (PBIP). Methods: This before-after study identified consecutive patients who received continuous insulin therapy for at least 24 h. Patients were stratified into two groups (PBIP and CAIP). The target blood glucose range for both was 80-110 mg/dL. Hypoglycemia was defined as the percentage of patients with any glucose value <40 mg/dL. Variability was measured by reporting the SD for each patients mean glucose value. Results: There were 192 patients evaluated (PBIP, n=145; CAIP, n=47). More glucose readings were within target range using the CAIP protocol (49±14% vs. 40±12%, P < 0.001), but no difference in mean glucose was noted (113±11 mg/dL with CAIP vs. 116±11 mg/dL with PBIP, P=0.067). The incidence of hypoglycemia was similar between the CAIP and PBIP groups, respectively (2.1% vs. 4.1%, P=0.518). Glucose variability was lower with the CAIP (25±9 mg/dL vs. 31±11 mg/dL, P=0.001). The CAIP required more frequent blood glucose assessments (16±2 vs. 12±2 per day, P < 0.001), more insulin dosing adjustments (14±3 vs. 5±2 per day, P < 0.001), and more time per day (84±15 vs. 51±8 min per patient, P < 0.001) compared with the PBIP. Conclusions: A CAIP will lead to minor improvements in glucose control and decrease glucose variability but will not change the rate of hypoglycemia or response to insulin therapy. These differences could largely be due to more aggressive monitoring and titrations required by a CAIP.