Receiver operating characteristic (ROC) curves for GA/A1c ratio and FCPR index (FCPR/FPG×100) for predicting glycaemic variability indices.

<p>ROC analyses were repeated using serial cut-off data for GA/A1c ratio and FCPR index, and the greatest AUC of 0.7665 for GA/A1c ratio (closed triangles) and of 0.8504 for FCPR index (closed squares) was obtained when the cut-off values were set at 2.8 and 3.0, respectively. ROC analysis simultaneously using a GA/A1c ratio of 2.8 and FCPR index of 3.0 to predict glycaemic fluctuations yielded an AUC of 0.9154 (open circles).