Prediction of lowest nocturnal blood glucose level based on self-monitoring of blood glucose in Japanese patients with type 2 diabetes

Continuous glucose monitoring (CGM) is not available for all patients with type 2 diabetes (T2D) at risk of nocturnal hypoglycemia (NH). This study was performed to predict the lowest nocturnal blood glucose (LNBG) levels. An LNBG prediction formula was developed by multivariate analysis using the data including self-monitoring of blood glucose from a formula making (FM) group of 29 insulin-treated T2D patients with CGM. The validity of the formula was assessed by nonparametric regression analysis of actual and predicted values in a formula validation group consisting of 21 other insulin-treated patients. The clinical impact on prediction was evaluated using a Parkes error grid. In the FM group with a median age of 64.0, the following formula was established: Predicted LNBG (mg/dL) = 127.4–0.836 × Age (y) + 0.119 × Self-monitored fasting blood glucose (mg/dL) + 0.717 × Basal insulin dose (U/day) (standard error of calibration 17.2 mg/dL). Based on the validation results, standard error of prediction was 31.0 mg/dL. All predicted values fell within zones A (no effect on clinical action) and B (little or no effect on clinical outcome) on the grid. LNBG could be predicted, and may be helpful for NH prevention.