Aim: Our aims were (1) to design and standardize a statistical approach for data reduction in continuous glucose monitoring, allowing comparison of circadian glycemic patterns in therapeutic subcohorts of patients with type 1 diabetes, and (2) to investigate the applicability of this approach for CGMS assessment in clinical study of basal insulin replacement quality with various timings of basal injections (pre-breakfast, dinner, bedtime) of a new insulin analog.
Methods: Prospective randomized three-arm parallel study with switch over after 6 months for another 3 months of free choice injection time point (options pre-breakfast, pre-dinner and bedtime) of the new insulin analog in 16 type 1 diabetic subjects on functional insulin treatment (FIT: basal, prandial and correctional dosages). CGMS was used at the end of each follow up period of a clinical study. Representative daily profiles were off-line computed as "circadian sensor modal days" for each insulin regimen consisting of consecutive means of hourly glucose values.
Results: Although the overall quality of glycemic control (HbAIC) for different regimens did not reach statistical differences, CGMS displayed slightly divergent maximal swings in the course of glycemia (p=0.04-0.08) and allowed--with delineated data reduction procedure--a reliable between treatment comparison.
Conclusion: Off-line computation of "hourly circadian sensor modal days" for data reduction can be effectively used with CGMS for description of circadian glycemic patterns in type 1 diabetes.