Pilot evaluation of continuous subcutaneous glucose monitoring in children with multiple organ dysfunction syndrome

Abstract

To evaluate continuous subcutaneous glucose monitoring in pediatric critical illness. Prospective evaluation. Mixed university pediatric intensive care unit. Children aged 1 mo to 16 yrs requiring mechanical ventilation with at least two organ system failures. None. Blood samples obtained from an arterial line, measurements using point-of-care glucometer, and laboratory analysis were compared with continuous subcutaneous glucose monitoring. Fourteen patients yielded 11,880 continuous subcutaneous glucose monitoring measurements; 436 glucometer levels and 34 laboratory levels had mean time-paired glucose values of 108 +/- 29 mg/dL and 110 +/- 25 mg/dL, respectively. Mean continuous subcutaneous glucose monitoring glucose was 101 +/- 31 mg/dL for samples paired with glucometer and 95 +/- 40 mg/dL for samples paired with laboratory tests. Continuous subcutaneous glucose monitoring measurements correlated with glucometer (r = 0.44) and laboratory testing (r = 0.48). Mean absolute differences between continuous subcutaneous glucose monitoring measurement and glucometer and laboratory values were 18 +/- 16 mg/dL and 25 +/- 20 mg/dL, respectively. Clarke error grid analysis found 69% of the measurements to be in zone A (clinically accurate), 29% in zone B (benign errors), and 2% in zone D (failure to detect errors). The mean absolute relative difference between the continuous subcutaneous glucose monitoring measurement and glucometer and laboratory measurements were 17% and 23%, respectively. Bland-Altman analysis showed good agreement between continuous subcutaneous glucose monitoring and the other methods of glucose measurement. However, in the lower range (< or =74 mg/dL) 39% of the continuous subcutaneous glucose monitoring readings had a difference >15 mg/dL. On multiple regressions, only glucometer glucose values, continuous subcutaneous glucose monitoring levels, and base deficit were associated with the mean absolute relative difference. The performance of continuous subcutaneous glucose monitoring against point-of-care glucometer and laboratory measurements may be considered "good" using statistical definitions (Bland-Altman and Clarke error grid analysis). However, there are important limitations in children with large base deficit, being actively cooled, and with glucose in the lower range, which may limit its application.