Presentation (CTI)

OBJECTIVES: Continuous glucose monitoring (CGM) is expanding into hospital settings, where glycemic variability affects outcomes and resource use. FreeStyle Libre 2 differs from earlier Libre generations in calibration and alarm functionality, and prior CGM-in-ICU reviews are device-agnostic. This study systematically reviews and meta-analyzes FreeStyle Libre 2 accuracy and clinical safety specifically in hospitalized adults.
METHODS: A systematic review following PRISMA 2020 is underway across PubMed, Embase, Cochrane CENTRAL, and Web of Science. Eligible studies report paired FreeStyle Libre 2-reference glucose measurements in hospitalized adults. Outcomes are mean absolute relative difference (MARD), error grid concordance, and safety events. A pre-specified contingent plan applies random-effects meta-analysis if ≥4 eligible studies are confirmed; narrative synthesis otherwise.
RESULTS: Three FreeStyle Libre 2- specific studies in hospitalized adults have been confirmed to date. In an intermediate care cohort (Portugal, n=23,414 paired readings vs. capillary point-of-care glucose), MARD was 18.62±11.87%, with 99.0% in Parkes Zones A+B; accuracy improved during intravenous fluid therapy and shock (p<0.001). In a mixed diabetic/non-diabetic ICU cohort (Italy, n=40, vs. arterial blood gas), MARD was 12±10%, with vasopressor use associated with reduced accuracy. A registered US perioperative trial (NCT05937373, Libre 2 vs. Dexcom G6 vs. laboratory glucose) has posted results, not yet extracted. The two confirmed quantitative studies differ substantially in point MARD estimates (18.6% vs. 12%) and show opposite-direction hemodynamic subgroup effects, plausibly reflecting differences in reference standard (capillary vs. arterial) rather than device performance alone.
CONCLUSIONS: Preliminary findings indicate FreeStyle Libre 2 retains clinical safety in hospitalized patients despite numerical accuracy below outpatient benchmarks but suggests reference-method selection and hemodynamic status materially affect observed accuracy, a factor existing device-agnostic CGM-ICU reviews have not isolated. If confirmed across the full search, this implies hospital CGM accuracy claims should be reference-standard-specific rather than generalized, with direct relevance to inpatient monitoring protocols and technology assessment.