par Righy Shinotsuka, Claudia 
;Brasseur, Alexandre ;Fagnoul, David ;Softley, Timothy T.P.;Vincent, Jean Louis ;Preiser, Jean-Charles
Référence Critical care, 20, 1, 380
Publication Publié, 2016-11
;Brasseur, Alexandre ;Fagnoul, David ;Softley, Timothy T.P.;Vincent, Jean Louis ;Preiser, Jean-Charles Référence Critical care, 20, 1, 380
Publication Publié, 2016-11
Article révisé par les pairs
| Résumé : | Background: Intravascular continuous glucose monitoring (CGM) may facilitate glycemic control in the intensive care unit (ICU). We compared the accuracy of a CGM device (OptiScanner®) with a standard reference method. Methods: Adult patients who had blood glucose (BG) levels >150 mg/dl and required insertion of an arterial and central venous catheter were included. The OptiScanner® was inserted into a multiple-lumen central venous catheter. Patients were treated using a dynamic-scale insulin algorithm to achieve BG values between 80 and 150 mg/dl. The BG values measured by the OptiScanner® were plotted against BG values measured using a reference analyzer. The correlation between the BG values measured using the two methods and the clinical relevance of any differences were assessed using the coefficient of determination (r 2) and the Clarke error grid, respectively; bias was assessed by the mean absolute relative difference (MARD). Three different standards of glucose monitoring were used to assess accuracy. Glycemic control was assessed using the time in range (TIR). Six indices of glycemic variability were calculated. Results: The analysis included 929 paired samples from 88 patients, monitored for a total of 2584 hours. Reference BG values ranged between 60 and 484 mg/dl. The r 2 value was 0.89. The percentage of BG values within zones A and B of the Clarke error grid was 99.9%; the MARD was 7.7%. Using the ISO 15197 standard and Food and Drug Administration and consensus standards, respectively, 80.4% of measurements were within 15 mg/dl and 88.2% within 15% of reference values, 40% of measurements were within 7 mg/dl and 72.5% within 10% of reference values, and 65.2% of measurements were within 10 mg/dl and 82.7% within 12.5% of reference values. The TIR was slightly lower with the OptiScanner® than with the reference method. The J-index, standard deviation and maximal glucose change were the indices of glycemic variability least affected by the measurement device. Conclusions: Based on the MARD, the performance of the OptiScanner® is adequate for use in ICU patients. Because recent standards for accuracy were not met, the OptiScanner® should not be used as a sole monitor. The assessment of glycemic variability is influenced by the time interval between BG determinations. Trial registration: Clinicaltrials.gov NCT01720381. Registered 31 October 2012. |
