实时动态血糖监测的准确性评估及相关因素分析

Assessment of the accuracy of real.time continuous glucose monitoring system and its correlated factors

目的评估实时动态血糖监测系统（real—timecontinuousglucosemonitoring，RT—CGM）的准确性及其影响因素。方法79例糖尿病患者应用RT—CGM连续监测3d，每天输人4～8次指尖血糖值进行校准，计算输入指血值与RT—CGM探头值的匹配率，对匹配率与日内平均血糖波动幅度（MAGE），血糖的标准差（SDBG），24h平均血糖（MBG），血糖大于10mmol／L曲线下面积（AUC10），血糖小于3．9mmol／L曲线下面积（AUC3．9），有效血糖波动次数（NGE）进行Pearson相关分析及多元线性回归分析，按MAGE、MBG对应的匹配率分组，比较组间差异。同时分别对全天、餐前、餐后、夜间指尖血糖与对应探头值做Pea阳0n相关分析。结果Pearson相关分析显示，匹配率与MAGE、SDBG、MBG、AUC10呈负相关。MAGE、MBG、NGE进入回归方程。MAGE〈3．9mmol／L，匹配率达81％，3．9≤MAGE〈7mmol／L，匹配率为67％，MAGEl〉7mmoL，匹配率为52％，3组差异有统计学意义（P〈0．01）。MBG〈7．8mmol／L，匹配率达78％，7．8mmol／L〈MBG〈10mmol／L，匹配率达68％，MBG≥10mmoL／L，匹配率为57％，3组差异有统计学意义（P〈0．01）。约登指数为0．409时，MAGE小于3．745mmol／L时，匹配率可达75％以上。约登指数为0．369时，MBG小于8．38mmol／L时．匹配率可达75％以上。全天、餐前、餐后、夜间探头值与相应指尖血糖的相关系数分别为0．958、0．963、0．944和0．965（均P〈0．01）。结论探头值与指尖血糖值有很好的一致性，可用于血糖评估。血糖波动是RT—CGM准确性的影响因素，血糖波动越小，平均血糖越小，血糖波动次数越少，准确率越高，夜间和餐前输入校正指血糖较为合适。

Objective To assess the factors that influence the accuracy of real-time continuous glucose monitoring system （RT-CGM）. Methods A total of 79 diabetic patients wore RT-CGM for three days continuously while calibrating by interphalangeal glucose values 4-8 times a day. We counted matching rate of interphalangeal glucose values and RT-CGM probe value, and analyzed correlation of the matching rate with MAGE, SDBG, MBG, AUCI0, AUC3.9, and NGE by Pearson correlation analysis and multiple linear regression analysis. Then we grouped corresponding match rate by MAGE and MBG, compared the difference among groups. Furthermore, we analyzed correlation of RT-CGM probe values of all-day, fasting-postprandial, and nighttime and the corresponding iuterphalangeal glucose values by Pearson correlation analysis. Results Pearson correlation analysis showed that the matching rate was negatively correlated with the MAGE, SDBG, MBG, and AUC10. MAGE, MBG, and NGE were in the regression equation. MAGE 〈3.9 mmol/L, matching rate was 81%, 3.9 〈 MAGE〈7 mmol/L, matching rate was 67%, MAGE 7 mmol/L, the matching rate was 52%, the difference among 3 groups was statistically significant （ P〈0.01 ）. MBG〈7.8 mmol/L, matching rate was 78% , 7.8 〈 MBG〈 10 mmol/L, matching rate was 68% , MBG 〉10 retool/L, the matching rate was 57%, the difference among 3 groups was statistically significant （P 〈 0.01 ）. When Youden index was 0. 409 and MAGE was less than 3. 745 mmol/L, the match rate was more than 75%. When Youden index was 0. 369 and MBG was less than 8.38 retool/L, the matching rate was more than 75 %. The correlation coefficients of all day, fasting, postprandial and nighttime probe values and the corresponding interphalangeal glucose values were 0. 958, 0. 963, 0. 944, and 0. 965 （ all P〈0.01 ）. Conclusions The probe values show good consistency with the interphalangeal glucose values, are able to be used for blood glucose assessment. Blood glucose fluctuations influences accuracy of RT-CGM, the smaller the blood glucose fluctuations,the smaller the mean glucose concentration, the smaller the effective blood glucose fluctuations in frequency, the higher would be the accuracy. It seems to be more effective to do calibration during fasting and nighttime.