Continuous glucose monitoring for diabetes mellitus in the perioperative and intensive care setting: a prospective observational validation study✰.

Abstract

Background: In-hospital diabetes mellitus management currently relies on intermittent glucose monitoring with point-of-care (POC) testing, rather than continuous glucose monitoring (CGM). Most studies on CGM have focused on its benefits and accuracy in the outpatient setting, in contrast to the hospital setting, where the accuracy may be different.

Methods: In this prospective observational study (NCT06297850), we assessed the accuracy of CGM (Dexcom G7, Dexcom Inc.) compared with POC monitors in people with diabetes mellitus requiring glucose-lowering medication undergoing surgery or admission to the ICU. The primary outcome was agreement between CGM and POC measurements, defined as median absolute relative difference (MARD). Clarke error grid (CEG) and Bland-Altman analyses were used to quantify the accuracy. Secondary analyses assessed agreement during periods of hypoperfusion, defined as a mean arterial pressure of <65 mm Hg, infusions with vasopressors, or reduced peripheral perfusion index.

Results: In 58 participants (mean [standard deviation] age 71 [11]yr; 34% female), the median MARD was 10.9% (5.1-19.4%); 1186 of 1212 paired POC and CGM measurements (98%) were within clinically acceptable ranges (CEG zones A and B). MARD was 4.5% higher (95% confidence interval 2.3-6.7%) during periods of hypoperfusion. Given that hypoglycaemia (<4 mM) did not occur, no paired readings were obtained. Bland-Altman analysis showed similar CGM and POC measurements, with a mean difference of -1 mM (95% limits of agreement -5.6 to 3.4 mM).

Conclusions: For the vast majority of interstitial glucose measurements, CGM was accurate in people with diabetes mellitus. Although overall performance was good, caution may be warranted when interpreting CGM data during periods of impaired peripheral perfusion.

Clinical trial registration: NCT06297850.