The individualized statistical analysis of the continuous glucose monitoring data

Background: Continuous glucose monitoring (CGM) has shown its benefits in pregnant women with diabetes. Flash glucose monitoring (FGM), as one of the CGM types, has not been well assessed in this patient group. The interpretation of a big volume of information on glycaemia obtained with various CGM devices is possible with statistical analysis according to the algorithms proposed by manufacturers. While these algorithms cannot be comprehensive, evaluation of alternative approaches to the CGM data statistical analysis and comparison of the results obtained with different devices seem reasonable. No unified algorithm for modification of antidiabetic treatment according to the CGM results has been yet developed. This study was performed in a pregnant patient with type 1 diabetes mellitus (T1DM) to demonstrate the methods to individualized analysis of the data from various devices (CGM, FGM, glucometer) that could be used in routine clinical practice.Aim: To evaluate the individual advantages and disadvantages of the simultaneous use of FGM, CGM and SMBG in a pregnant woman with type 1 diabetes.Materials and methods: This was an observational case study with a retrospective assessment of the patient's data obtained with FGM, CGM and a glucometer in a 31-year female patient with T1DM of 6-year duration and 9 weeks of gestation, who had been on pump insulin therapy for one year and had an HbA1c level of 5.4%. During the study the patient continued her pump therapy and performed blood glucose self-monitoring (BGSM) and simultaneously used FGM and CGM. The following FGM data were compared with CGM and glucometer results: measurement numbers, time in range, mean daily glucose, mean absolute difference (MAD), and mean absolute relative difference (MARD).Results: The FGM-derived mean daily glucose was lower than that measured with the glucometer: 5.1±1.9 mmol/L vs 6.4±2.2 mmol/L (p<0.001). The number of measurements with FGM was 32.0±12.9 times daily and with a glucometer 15.1±5.5 times daily (p<0.001). MAD values were minimal in the hypoglycemic range (0.5±0.3 mmol/L) and maximal in the hyperglycemic range (1.6±1.2 mmol/L, р<0.001). The MARD values were significantly smaller in the hyperglycemic than in the normoglycemic (16.6±12.6% vs 21.3±14.0%, р=0.035). The highest MAD and MARD were observed on the Day 1 of the sensor installation. The comparison of FGM and the glucometer readings with the Clarke consensus error grid showed that 82% of the FGM readings were in zone A or B. The FGM accuracy was higher from Day 2 to Day 9 (72.5% of the FGM readings in zone A). MAD between FGM and CGM readings was not different from that between FGM and the glucometer: 1.3±1.0 mmol/L and 1.2±0.9 mmol/L, respectively (p=0.09). MARD for the FGM and CGM comparison was higher than that for FGM and glucometer comparison: 24.4±23.0% and 18.8±13.5%, respectively (р<0.001). The Pearson's correlation coefficient FGM and CGM seemed lower than that between FGM and the glucometer (0.837 and 0.889, respectively). FGM has identified more hypoglycemic events compared to CGM: time below range was 29.4% and 8.8%, respectively, p<0.001).Conclusion: The FGM readings highly correlate with the glucometer. The FGM difference with the glucometer was lower in the hypo- and hyperglycemic ranges. FGM shows higher values for time below range than CGM. It is necessary to continue the study of the clinical acceptability of FGM in pregnant women and determination of its optimal regimen for the treatment of this patient category, as well as to develop an algorithm for treatment modification based on the results of FGM.