Endocrine Abstracts

We studied regarding simplifying standard deviation (SD) and useful metrics for achieving target SD. In a cross-sectional study, we analyzed 24-h glucose levels measured using continuous glucose monitor (CGM) [iPro2] in 150 patients with type 2 diabetes. We analyzed 51 time in range (TIR) (reference ranges: mean glucose level (M)±20, M±21 … M±69, M±70 mg/dl [we referred to the underlined numbers as “Margin”]). We analyzed optimal cutoff values of TIR to predict SD below target, corresponding to 51 TIR × 26 target SD (25, 26 … 49, 50 mg/dl), using receiver operating characteristic analysis. Of the 51 reference ranges, the reference range for which TIR had the largest AUC for each target SD was referred to as “optimal reference range 1”. We arranged the patients in ascending order of SD, ranking from 1 to 150. Consecutive 110 patients were selected 41 times while shifting one by one (from “1–110” to “41–150”: 41 groups). Of the 51 reference ranges, the reference range for which TIR most correlated to SD for each group was referred to as “optimal reference range 2”. Mean absolute deviation (MAD) correlated to SD (r=0.995, P<0.001; MAD=0.87 × SD - 1.22) (n=150). “Margin” of “optimal reference range 1” (M1, mg/dl) tended to increase with increased target SD (mg/dl), from 42 for “target SD of 25” to 66 for “target SD of 50”. Target SD correlated to M1 (r=0.92, P<0.001; M1=1.31×target SD + 5.56) (n=26). The optimal cutoff values of TIR for “optimal reference range 1” was 86.8±4.6 in 26 target SD. “Margin” of “optimal reference range 2” (M2) tended to increase with increased mean of SD, from 35 for “mean of SD of 30.6” to 64 for “mean of SD of 47.7”. Mean of SD correlated to M2 (r=0.96, P<0.001) (n=41). SD means difference from M. If “optimal reference range 1” is used as an alarm threshold in a personal CGM and TIR for “optimal reference range 1” > 80–90% is achieved, SD below the target may be achieved. The mean of SD in study participants may affect the M2.