| Results: |
Although daily 4-point (pre-prandial and bedtime) home blood glucose monitoring (HBGM) profiles are widely used for titrating insulin, and HbA1c is used for evaluating therapeutic effectiveness, the ability of HBGM to provide efficient estimates of glycemic variability is less clear. To examine this question, we conducted a cross-sectional predictive validity analysis of paired estimates of HbA1c, FPG, 4-point HBGM daily mean and standard deviation (SD), and continuous glucose monitoring (CGMS) statistics (mean, median, range, SD, and % of values >180 and <70 mg/dL) during a 72-hour period in 58 insulin-treated type 1 and 2 DM patients (31 male / 27 female; mean±SD age = 50±11 yrs, FPG = 171±69 mg/dL, HbA1c = 7.5±0.7%). With HBGM (12 measures/patient), mean glucose = 186±44 mg/dL, within-individual SD = 66±19 mg/dL, and coefficient of variation (COV = SD/mean) = 37±11%. With CGMS (864 measures/patient), mean glucose = 177±40 mg/dL, SD = 47±13 mg/dL, COV = 28±9%, range = 183±41 mg/dL, values >180 mg/dL = 44±24%, and values <70 mg/dL = 5±9%). HBGM mean glucose correlated with FPG and HbA1c (r=0.37 and 0.34; P<0.01) while CGMS mean glucose correlated with HbA1c (r=0.34), but not FPG (r=0.18). HBGM mean glucose correlated with the CGMS mean and median glucose (r=0.84 and 0.83, explaining 71% of the variance, P<0.001), minimum and maximum (r=0.79 and 0.65), and % of values >180 and <70 mg/dL (r=0.80 and 0.35). HBGM SD was correlated with CGMS SD, maximum and range (r=0.60, 0.31, and 0.52; all P<0.01), and HBGM COV was highly correlated with CGMS COV (r=0.67, P<0.001). In conclusion, statistical measures of mean glucose and variance from 4-point HBGM over 3 days are significantly correlated with standard measures of glucose control (FPG, HbA1c) and explain a substantial portion of the variability observed during CGMS. Such summary measures may be helpful for disease monitoring and insulin titration when CGMS is not available. |