Abstract
Background: Continuous glucose monitors (CGMs) are increasingly being used to guide glucose management in the hospital. However, uncertainty regarding their accuracy in this setting remains. Methods: We conducted a nonrandomized, open-label, clinically blinded prospective trial of the Dexcom G6 Pro (G6P) and FreeStyle Libre Pro (FLP) in the inpatient setting among critically ill hospitalized patients (n = 40) requiring continuous intravenous insulin infusion. In parallel with CGM data, reference serum (Lab) glucose and point-of-care (POC) glucose values were obtained. On completion of the study, CGM and reference values were analyzed to assess CGM accuracy. Results: A total of 1015 matched G6P-Lab pairs had a mean absolute relative difference (MARD) of 22.7%, 2369 G6P-POC pairs had an MARD of 22.9%, 1006 matched FLP-Lab pairs had an MARD of 25.2%, and 2353 FLP-POC pairs had an MARD of 27.0%. Both CGM systems demonstrated considerable inter-patient variability in sensor accuracy and tended to underestimate glucose in comparison with the reference values. Rarely were low reference values overestimated by either sensor. Conclusions: Factory-calibrated continuous glucose monitors may require accuracy validation and per-patient calibration for inpatient use in critically ill patients.
Funding Information
  • JDRF (5-ECR-2022-1177-A-N)
  • DexCom Inc. (Investigator Initiated Study)
  • National Heart, Lung, and Blood Institute (R01HL159760)
  • Division of Diabetes, Endocrinology, and Metabolic Diseases (K12DK133995)
  • Division of Diabetes, Endocrinology, and Metabolic Diseases (K23DK134880)
  • Division of Diabetes, Endocrinology, and Metabolic Diseases (P30DK020593)
  • Division of Diabetes, Endocrinology, and Metabolic Diseases (R01DK129840)

This publication has 15 references indexed in Scilit: