Libre in Inpatient Covid Units

Abstract Continuous glucose monitors (CGM) have demonstrated accuracy in outpatients and were accurate with pilot data using blinded systems in inpatients. Libre CGMs were approved and donated to hospitalsfor use during the COVID-19 to minimize exposure and PPE use. We made a prospective plan to assess accuracy of the Libre on inpatients admitted to COVID units during an initial “validation phase”. Fingerstick blood glucose (FSBG) was checked for the first three days after placing the sensor and compared to Libre values within the next 1–15 minutes. Patients were instructed to scan the sensor after each FSBG, and at other times as clinically warranted. FSBG values were recorded from the medical record (CPRS) and compared to Libre values downloaded to LibreView. The mean absolute relative difference (MARD) between FSBG and Libre was calculated for each patient for one to three days. The average MARD across all patients was calculated. Accuracy was further assessed using the Bland-Altman Plot and error grid analysis using web-based tool. Of the 21 patients assessed in the validation phase, 19 had at least one day of data and 11 had at least three days. The mean MARD was 11.2% after one day, and 12.5% after three days. Four patients after one day and three patients after three days had a MARD of 15–20% where use of the Libre was continued with confirmatory FSBG. In 15/19 (78%) patients with one day of data and 8/11 (73%) of patients with 3 days of data had a MARD <15% and continued using the Libre without further FSBG. One patient had a MARD >20% that did not improve with changing the sensor, and Libre was discontinued. In 16 of 19 patients, the Libre values were lower than FSBG. No adverse events relating to Libre use were identified. Error grid analysis showed that most patients had no values outside the A and B ranges, and very few values outside of the clinically accurate range, occurring in 2 of the 19 patients (9% of values in the “slight risk” zone in 9% in one and 6% in the other). The Libre was well-accepted by patient and nursing staff, but did not have measurable effect on glycemic control, hypoglycemic events, or hospitalization measures. The libre was deemed a useful intervention in inpatients, but it can be inaccurate or only moderately accurate compared to FSBG in enough patients to require checking accuracy for at least 1–3 days. Inaccurate Libre values were mostly lower than FSBG requiring confirmatory measures of low Libre values with FSBG in our experience.