Clinical Performance of CGMS in Type 1 Diabetic Patients Treated by Continuous Subcutaneous Insulin Infusion Using Insulin Analogs

Abstract

OBJECTIVE—Performance criteria have been established for in vitro blood glucose monitoring, particularly for the self-monitoring of blood glucose using glucose meters. Devices intended for use in the future, such as the continuous glucose monitoring system (CGMS), should satisfy similar criteria, particularly in diabetic patients under intensive therapy. RESEARCH DESIGN AND METHODS—The analysis was conducted on 18 type 1 diabetic patients (not controlled, HbA1c >7.5%) treated by external pump using insulin analogs. Each patient received a glucose sensor for 3 days during his/her hospitalization and was instructed in its operation. Medtronic criteria were used to determine the accuracy of the CGMS. In addition, the data were analyzed according to American Diabetes Association (ADA) criteria, Clarke Error Grid analysis, and method of residuals, with the glucose oxidase method using a Beckman analyzer used as the reference method. Specificity and sensitivity were evaluated from the viewpoint of accuracy in the detection of hypoglycemia. For nine patients, two glucose sensors were simultaneously inserted into an abdominal site to determine the reproducibility of the system. RESULTS—Among the 33 glucose sensors inserted, 6 (18%) were nonfunctional. The mean duration of CGMS recording was 63 ± 12 h. From all of the 692 sets of data that paired glucose readings and CGMS, the coefficients of correlation ranged from 0.87 to 0.92 and the mean absolute error ranged from 12.8 to 15.7%. The time experienced in hypoglycemia (<55 mg/dl) was reported at 86 ± 62 min/day. Only 39% of the CGMS values satisfied the ADA precision criteria to within ±10%, and 19% of these values satisfied the future ADA precision criteria of accuracy to within ±5%. The means of difference method showed that the CGMS slightly underestimated the plasma glucose values (mean = −12 mg/dl). Error grid analysis showed only 77% of the glucose sensor values were in zone A, and 98.9% were in zones A and B. Two values fell in zone C and a single value fell in zone D. The sensitivity and specificity of the CGMS to detect hypoglycemia were 33 and 96%, respectively. A total of 6,666 paired sensor values were recorded with a coefficient of correlation of 0.84 with a coefficient of variation of 8.25%. CONCLUSIONS—CGMS could be useful in routine clinical practice to provide much more information on the glucose profile than intermittent self-monitoring of blood glucose (SMBG). However, CGMS cannot be used as a replacement for glucose meters because it does not satisfy the conventional performance goals set down for in vitro glucose measurements and could therefore lead to clinically incorrect treatment decisions.