Real-World Clinical Study Evaluating The Accuracy Of Flash Glucose Monitoring In T2DM Patients

Objective: In the present study,we aimed to evaluate the performance of Free Style Libre Flash Glucose Monitoring System(FGM)and explore the risks that affect its stability and safety in T2 DM patients in real-world.Methods: From Jan.9,2018 to Jan.7,2020,742 adult T2 DM patients were included from the Department of Endocrinology,the Second Affiliated Hospital of Guangzhou Medical University.The glucose levels of fingertip capillary blood were tested by blood glucose meter in patients at eight time points(3 AM,7 AM,9 AM,11 AM,1 PM,5 PM,7 PM,and9 PM)as the reference blood glucose(REF)to compare FGM data at the same time.The mean absolute relative difference(MARD),Bias,Bland-Altman,Parkes error grid(PEG),surveillance Error Grid(SEG),and logistic regression analysis were used to evaluate the consistency and difference between FGM and REF.Results: A total of 25579 paired FGM/REF readings of glucose measurements within the glucose meter and FGM monitoring range was obtained.75% of paired readings showed the FGM value was lower than the REF value at the same time point.Total MARD was14.0% for all paired data,24.5% for paired data of 3 AM,and 12.6-14.4% for the rest paired data of the other seven time points.The frequency of FGM data below the read range of FGM(i.e.,LO)was highest at 3 AM point(39.1%).The difference of paired glucose value between FGM and REF(±15% or ±15 mg/dl of the ISO15197: 2013 standard)was below80% either at each time point or at the overall time.Using the error grid to analyze the paired data,we found that both PEG and SEG showed a clear linear relationship between FGM and REF values.Comparing with PEG,the paired readings with the higher risk scores appeared more in SEG.In addition,logistic regression analysis showed that hypoglycemia,infection,diabetic foot,DKA,and HBP were independent risk factors for the accuracy of FGM in the patients(OR>1,P<0.05),while Hemoglobin(Hb)and insulin were protective factors(OR<1,P<0.05).Conclusion: The Free Style Libre FGM commonly underestimates the blood glucose level and its accuracy decreases significantly in lower glucose ranges.SEG shows greater precision for quantifying the risk of paired FGM/REF reading than PEG.SEG analysis shows the appropriate accuracy of FGM,nevertheless it should be noted that FGM accuracy is affected by many clinical factors.