Improved Algorithm for Automated Glucose Clamps.

BACKGROUND

In glucose clamp experiments, blood glucose concentrations (BGs) are kept as close as possible to a predefined target level using variable glucose infusion rates (GIRs). In automated clamps, GIRs are calculated by algorithms implemented in the device (e.g., the Biostator). Low BG- and GIR-variability is needed for high clamp quality. We therefore tried to reduce oscillations in both BG and GIR with an improved algorithm implemented in ClampArt, a modern clamp device.

SUBJECTS AND METHODS

The Biostator algorithm was first improved by numerical simulations of glucose clamps (in silico). With the results of the simulations, we started in vitro experiments using the ClampArt device and a container with water and glucose as "test subject." After a small pilot in vivo study, a larger clinical study was performed to compare the original with the optimized algorithm.

RESULTS

With the improved algorithm, in silico, in vitro, and in vivo experiments showed reduced oscillations in both BG and GIR. In the clinical study, the coefficient of variation (CV) of BG values was lowered from 6.0% (4.6%-7.8%) [median (interquartile range)] to 4.2% (3.6%-5.0%), P < 0.0001 and the CV of GIR from 60.7% (49.6%-82.0%) to 43.5% (32.8%-57.2%), P < 0.0001. Other clamp quality parameters did not change substantially, median deviation from target slightly increased from 0.6% (0.2%-1.0%) to 1.1% (0.7%-1.5%), P = 0.0005, whereas utility did not change [97.0% (93.4%-100.0%) vs. 97.0% (94.0%-98.8%), P = 0.57].

CONCLUSIONS

With the improved algorithm, all experiments confirmed a reduction in BG- and GIR-oscillations without a major impact on other glucose clamp parameters. The optimized algorithm has been implemented in ClampArt for all future glucose clamp studies.