Abstract：
OBJECTIVE: Continuous glucose monitoring (CGM) may have benefits in achieving glycemic control in critically ill patients. The aim of this study was to assess the accuracy of the Freestyle Libre H (professional version of the Libre Pro). in patients with acute respiratory failure (ARF) in the intensive care unit (ICU).
METHODS: 52 adult patients with ARF were selected. The performance of CGM was evaluated using arterial blood glucose (aBG) and point-of-care (POC) glucose as reference values. Numerical accuracy was evaluated by the mean absolute relative difference (MARD), Bland-Altman analysis, and %15/15(the percentage of CGM values within 15 mg/dL or 15% of reference values <100 mg/dL or >100mg/dL, respectively), %20/20 and %30/30; Clinical accuracy was assessed by Clarke error grid analysis.
RESULTS: 519 and 1504 pairs of aBG/CGM and POC/CGM glucose values were analyzed. The MARD values were 13.8% and 14.7%, respectively. The mean deviation of the Bland‒Altman analysis was 0.82 mmol/L and 0.81 mmol/L. %15/15, %20/20 and %30/30 of aBG values were 62.6%, 75.5%, and 92.4%, respectively; %15/15, %20/20 and %30/30 of POC values were 57.1%, 72.9%, and 88.7%, respectively. The Clarke error grid analysis showed that 97.8% and 99.3% of the values located in the (A+B) zone. Additionally, accuracy of CGM is not affected by general patient factors.
CONCLUSIONS: This study demonstrated that the accuracy of CGM in patients with ARF is lower than that in most outpatients, and it is not affected by general patient factors. Whether CGM is beneficial to glucose management in ICU needs further evaluation.
METHODS: 52 adult patients with ARF were selected. The performance of CGM was evaluated using arterial blood glucose (aBG) and point-of-care (POC) glucose as reference values. Numerical accuracy was evaluated by the mean absolute relative difference (MARD), Bland-Altman analysis, and %15/15(the percentage of CGM values within 15 mg/dL or 15% of reference values <100 mg/dL or >100mg/dL, respectively), %20/20 and %30/30; Clinical accuracy was assessed by Clarke error grid analysis.
RESULTS: 519 and 1504 pairs of aBG/CGM and POC/CGM glucose values were analyzed. The MARD values were 13.8% and 14.7%, respectively. The mean deviation of the Bland‒Altman analysis was 0.82 mmol/L and 0.81 mmol/L. %15/15, %20/20 and %30/30 of aBG values were 62.6%, 75.5%, and 92.4%, respectively; %15/15, %20/20 and %30/30 of POC values were 57.1%, 72.9%, and 88.7%, respectively. The Clarke error grid analysis showed that 97.8% and 99.3% of the values located in the (A+B) zone. Additionally, accuracy of CGM is not affected by general patient factors.
CONCLUSIONS: This study demonstrated that the accuracy of CGM in patients with ARF is lower than that in most outpatients, and it is not affected by general patient factors. Whether CGM is beneficial to glucose management in ICU needs further evaluation.
摘要：
目的：连续血糖监测（CGM）可能对危重患者实现血糖控制有益。这项研究的目的是评估自由式LibreH（LibrePro的专业版）的准确性。重症监护病房（ICU）的急性呼吸衰竭（ARF）患者。
方法：选择52例成人ARF患者。使用动脉血糖（aBG）和即时（POC）葡萄糖作为参考值评估CGM的性能。通过平均绝对相对差(MARD)评估数值精度，Bland-Altman分析,和％15/15（CGM值在15mg/dL或参考值的15％以内的百分比<100mg/dL或>100mg/dL，分别)，%20/20和%30/30；通过Clarke误差网格分析评估临床准确性。
结果:分析了519和1504对aBG/CGM和POC/CGM葡萄糖值。MARD值分别为13.8%和14.7%,分别。Bland-Altman分析的平均偏差为0.82mmol/L和0.81mmol/L。aBG值的％15/15，％20/20和％30/30分别为62.6％，75.5%,92.4%,POC值的%15/15、%20/20和%30/30分别为57.1%，72.9%,和88.7%，分别。Clarke误差网格分析表明97.8%和99.3%的值位于(A+B)区。此外,CGM的准确性不受一般患者因素的影响。
结论：这项研究表明，CGM在ARF患者中的准确性低于大多数门诊患者，它不受一般患者因素的影响。CGM是否有利于ICU的血糖管理需要进一步评估。
方法：选择52例成人ARF患者。使用动脉血糖（aBG）和即时（POC）葡萄糖作为参考值评估CGM的性能。通过平均绝对相对差(MARD)评估数值精度，Bland-Altman分析,和％15/15（CGM值在15mg/dL或参考值的15％以内的百分比<100mg/dL或>100mg/dL，分别)，%20/20和%30/30；通过Clarke误差网格分析评估临床准确性。
结果:分析了519和1504对aBG/CGM和POC/CGM葡萄糖值。MARD值分别为13.8%和14.7%,分别。Bland-Altman分析的平均偏差为0.82mmol/L和0.81mmol/L。aBG值的％15/15，％20/20和％30/30分别为62.6％，75.5%,92.4%,POC值的%15/15、%20/20和%30/30分别为57.1%，72.9%,和88.7%，分别。Clarke误差网格分析表明97.8%和99.3%的值位于(A+B)区。此外,CGM的准确性不受一般患者因素的影响。
结论：这项研究表明，CGM在ARF患者中的准确性低于大多数门诊患者，它不受一般患者因素的影响。CGM是否有利于ICU的血糖管理需要进一步评估。
