Analysis of continuous glucose monitoring data from non-diabetic and diabetic children: a tale of two algorithms

Abstract

Use of the Medtronic MiniMed Continuous Glucose Monitoring System (CGMS) in non-diabetic children has revealed many low and high sensor glucose (SG) values, suggesting that the original analytical algorithm (Solutions 2.0) might be overreading glycemic excursions. A revised algorithm (Solutions 3.0) was introduced in 2001. Our aim was to compare analyses of the same sensor profiles using both programs. Twenty-five lean, non-diabetic subjects (mean age 14 +/- 4 years) underwent continuous glucose monitoring with CGMS for up to 72 h. Sensor tracings were analyzed with both algorithms and compared. Separate analyses were performed for nocturnal readings (12-6 a.m.). Mean SG values were similar (103 +/- 24 mg/dL for version 2.0 vs. 100 +/- 14 for version 3.0), but the distribution was significantly different: 13.8% of total SG were <70 mg/dL by version 2.0 versus 8.2% by version 3.0 (p < 0.001), and 7.7% of total SG were >150 mg/dL by version 2.0 versus 4.7% by version 3.0 (p = 0.02). Of nocturnal SG values, 25.8% were <70 mg/dL by version 2.0 compared with 17.9% by version 3.0, and 9.4% were >150 mg/dL by version 2.0 compared with 4.0% by version 3.0. In lean non-diabetic children, Solutions 2.0 identified significantly more hypoglycemia and hyperglycemia than Solutions 3.0. Similar analyses in 40 children with type 1 diabetes revealed no significant differences. Solutions 3.0 may be a more useful algorithm for preventing over-reading of low and high SG readings in non-diabetic children, whereas both algorithms give similar results in children with diabetes.