Blood Glucose Measurement Traceability to the Primary Reference Measurement Procedure (ID-MS) for Noninvasive Glucose Monitoring

A process for demonstrating the traceability of a glucose measurement to a reference measurement procedure and reference material is described in the International Organization for Standardization (ISO) 15197 ''in vitro diagnostic test system-requirements for blood glucose monitoring systems for self-testing in managing diabetes mellitus.'' The accuracy (precision and trueness) of the monitoring device is commuted among the established measurement procedures and the reference materials in the process to describe the degree to which the results agree with the true glucose values. This process can be applied to the current invasive type of glucose monitoring systems. Recently, we have developed a noninvasive type of glucose monitoring system called the metabolic heat conformation (MHC) method. Because the noninvasive MHC method requires only placing a fingertip over a series of sensors, the traceability process described in ISO 15197 cannot be directly applied to the MHC system because there are no means for simultaneously measuring the reference materials. Therefore, we propose a new traceability process for the noninvasive type of glucose monitoring system. Simultaneous blood collections and noninvasive measurements for patients are proposed to bridge the traceability chain. To demonstrate the validity of this approach, we conducted an evaluation of five groups of patients using the MHC method with 127 data points under actual conditions of use. Independent measurements by the noninvasive glucose monitoring system were compared with the reference glucose values determined by the hexokinase glucose-6-phosphate dehydrogenase method, which was calibrated with reference materials (the secondary calibrators). Individual uncertainties of the three levels of secondary calibrators were 0.45, 0.80, and 1.35 mg/dL, respectively, and within-run imprecisions were 0-1.48 mg/dL. Uncertainties of the regression curve (Sy/x) were 6.8-11.7 mg/dL. Combined uncertainties were estimated to be between 7.1 and 11.9 mg/dL.