SAT097 Investigation Of The Minimum Duration Needed To Estimate The Percentile Values For Target Duration In Ambulatory Glucose Profile

Disclosure: S. Takeishi: None. T. Inoue: None. Ambulatory glucose profile (AGP) is effective in grasping glycemic variability and determining therapeutic strategy. In AGP for intermittently scanned continuous glucose monitoring (isCGM), it is common for patients to intervene at the time when the interquartile range of glucose levels deviates from the target range. It is important to consider the cases having glucose levels with missing values measured using isCGM. There are limited investigations regarding the “minimum duration needed to estimate percentile values of glucose levels for target duration” (MDpercentile). This is a cross-sectional study. In Analysis 1, we analyzed the glucose levels without missing values measured using CGM (FreeStyle Libre Pro) over 24 h for 13 days (from 00:00 on Day 2 to 00:00 on Day 15; CGM attachment: Day 1) for 100 outpatients with type 2 diabetes, who did not change their diabetic treatment through the CGM-wearing duration. In Analysis 2, we analyzed the glucose levels with missing values measured using isCGM (FreeStyle Libre) over 24 h for 60 days (from 00:00 on Day 31 to 00:00 on Day 91; isCGM attachment: Day 1) for 50 outpatients with type 2 diabetes, who did not change their diabetic treatment through the isCGM-wearing duration. In Analysis 1, MDpercentile for the 2-h mean glucose level for 13 days (MDpercentile2h13days) was provided by correlation coefficient analysis using R(2)=0.9 as the threshold. In Analysis 2, MDpercentile for the 24-h mean glucose level for 60 days (MDpercentile24h60days) was provided by correlation coefficient analysis using R(2)=0.9 as the threshold. The Pearson’s product-moment correlation coefficient was used for the statistical analysis. In Analysis 1, the average of 12 MDpercentile2h13days in 24 h for 5 percentile values, 25 percentile values, 50 percentile values, 75 percentile values, and 95 percentile values was 7.4, 7.3, 6.9, 6.9, and 5.9 days, respectively. In Analysis 2, the MDpercentile 24h 60 days for 5 percentile values, 25 percentile values, 50 percentile values, 75 percentile values, and 95 percentile values was 26, 17, 15, 13, and 45 days, respectively. In Analysis 1, the lower percentile values needed more days to predict the percentile values. However, in Analysis 2, the 95 percentile value needed the most days to predict the percentile value. Lower percentile values may need more days to predict the percentile values in the calculation using glucose levels without missing values. Quite high percentile values may need many days to predict the percentile value in the calculation using glucose levels with missing values. Presentation: Saturday, June 17, 2023