ORF26-04 - Why Hypoglycemia Occurs Despite Stable Glucose Patterns: A Novel CGM-Derived Measure of Hypoglycemia Fragility in Patients with Type1 Diabetes Mellitus

Hypoglycemia risk in type 1 diabetes is traditionally assessed using exposure-based metrics such as time below range, glycemic variability, and the Low Blood Glucose Index. While these measures quantify hypoglycemia burden, they do not capture intrinsic susceptibility to hypoglycemia occurring under minimal perturbation, a clinically common yet poorly defined phenomenon. We sought to develop a novel metric, the Hypoglycemia Fragility Index (HFI), designed to quantify the probability of hypoglycemia following periods of low glycemic stress.

Synthetic CGM data set comprising 40,000 CGM-days sampled every 5 minutes (288 readings per day) was analyzed using GNU Octave programming language. Glucose profiles resembled individuals with type 1 diabetes and were categorized into four groups based on hemoglobin A1c ranges. Hypoglycemia was defined as glucose < 70 mg/dL for at least two consecutive readings (≥10 minutes). At each time point, glucose rate of change and short-term glucose excursion over the preceding 60 minutes were computed to characterize glycemic stress. Low-stress glycemic states were defined by an absolute glucose slope ≤0.5 mg/dL/min and excursion ≤20 mg/dL. For each profile, HFI was calculated as the number of low-stress glycemic time points followed by hypoglycemia within 60 minutes divided by the total number of low-stress time points for that individual. Given the absence of established clinical thresholds for hypoglycemia fragility, HFI was categorized using cohort-based quartiles to define low, moderate, and high fragility. Associations between HFI and established CGM metrics were evaluated using Spearman correlation coefficients.

Hypoglycemia Fragility Index demonstrated a moderate positive association with established hypoglycemia exposure metrics, including Low Blood Glucose Index and time below range < 70 mg/dL. In contrast, HFI showed weak or no association with measures of glycemic variability, including coefficient of variation and standard deviation. Associations between HFI and mean glucose were weakly inverse, indicating substantial variability in hypoglycemia fragility across similar levels of average glycemic control.

HFI provides a distinct dimension of hypoglycemia risk not captured by conventional exposure- or variability-based metrics. It reflects reduced glycemic reserve and impaired stability, with potential applications in risk stratification, individualized glycemic targets, and adaptive insulin delivery strategies.