Objective: A viscometric affinity sensor has been developed to measure the interstitial glucose concentration continuously. In a pilot clinical study its performance was assessed under conditions close to everyday life. Additionally, different insertion sites were tested for their suitability to apply subcutaneous glucose sensors.
Methods: Twelve subjects, 10 of whom with type 1 diabetes, were examined for 8 h. Sensors were applied subcutaneously at the forearm and the abdomen of each subject. Capillary blood glucose references were obtained from the finger tip every 30 min. Retrospective calibration was carried out individually with Deming regression.
Results: After retrospective calibration the 95% limits of agreement in the plot of the differences between sensor signals and references versus their means were +/-60 mg/dL. The sensitivity of the sensors remained stable over the entire measuring period, without any significant differences between the sensors at forearm and abdomen. Correcting for the observed time delay of 15 min between references and sensor values the limits of agreements were reduced to +/-38 mg/dL. Furthermore, error grid analysis showed 89.3% of the paired values in zone A and 9.6% in zone B. Only 1.1% were clinically unacceptable (zone D).
Conclusions: The performance of the viscometric affinity sensor shows the potential of the measuring principle under in vivo conditions. Forearm and abdomen seem to be similarly well suited for the application of subcutaneous sensors. The signal stability over time and the absence of enzymatic, chemical, or electrode reactions are advantages of the viscometric affinity principle.