The study was aimed at determining whether the dawn phenomenon in type 2 diabetes (T2D) can be predicted and quantified using simple and easily accessible glucose determinations.

A total of 210 non-insulin-treated persons with T2D underwent continuous glucose monitoring (CGM). The dawn phenomenon was quantified as the absolute increment from the nocturnal glucose nadir to the pre-breakfast value (Δdawn, mg/dL). Pre-lunch (preL) and pre-dinner (preD) glucose, and their averaged values (preLD), were compared with the nocturnal nadir. These pre-meal values were subtracted from the pre-breakfast values. The differences obtained (Δpre-mealL, Δpre-meal D and Δpre-meal LD) were correlated with Δdawn values. The receiver operating characteristic (ROC) curve was used to select the optimal Δpre-meal value that best predicted a dawn phenomenon, set at a threshold of 20mg/dL.

All pre-meal glucose levels and differences from pre-breakfast values (Δpre-meal) significantly correlated (P<0.0001) with the nocturnal nadir and Δdawn values, respectively. The strongest correlations were observed for the parameters averaged at preL and preD time points: r=0.83 for preLD and r=0.58 for Δpre-meal LD. ROC curve analysis indicated that the dawn phenomenon at a threshold of 20mg/dL can be significantly predicted by a Δpre-meal LD cut off value of 10mg/dL. The relationship between Δdawn (Y, mg/dL) and Δpre-meal LD (X, mg/dL) was Y=0.49 X+15.

The self-monitoring of preprandial glucose values at the three main mealtimes can predict the presence/absence of the dawn phenomenon, and permits reliable assessment of its magnitude without requiring continuous overnight glucose monitoring.