Construction of a Risk Prediction Model and Associative Path Analysis for Impaired Awareness of Hypoglycemia in...

We aimed to develop a risk prediction model for impaired awareness of hypoglycemia (IAH) in diabetes and explore the associations with clinical variables, providing a basis for targeted screening and personalized intervention.

This cross-sectional study included 280 hospitalized people with diabetes (73.2% type 2 and 42.9% female) who experienced hypoglycemia between October 2023 and February 2024. We defined IAH as a Gold Scale score ≥4. We used binary logistic regression to construct the prediction model and structural equation modeling to analyze the associated pathways. Sensitivity analysis stratified by diabetes type verified the robustness of the results.

IAH affected 60.4% of the participants. Multivariate analysis revealed lower total cholesterol (TC) levels (aOR=0.819; 95% CI: 0.684-0.981; P=0.030) as independent correlates. The model, which incorporated BMI, TC level, serum albumin level, peripheral neuropathy (PN) incidence, glomerular filtration rate (GFR), age, sex, and insulin use, showed moderate discriminative ability (AUC=0.630, P<0.001), with high sensitivity (87.0%) and good calibration (Hosmer-Lemeshow test: P=0.743). Path analysis revealed that age indirectly influenced IAH risk (β=0.243) by decreasing the GFR (β=-0.419; P<0.001) and increasing PN risk (β=0.352; P<0.001) and that a lower GFR (β=-0.133; P=0.024) and TC level (β=-0.131; P=0.024) were directly correlated with IAH risk. The serum albumin level was maintained by preserving the GFR (β=0.203; P<0.001) and reducing PN risk (β=-0.139; P=0.012).

Higher BMI and lower TC are key correlates of IAH in hospitalized patients with diabetes. Age, GFR, serum albumin, and PN interact through direct and indirect pathways to affect the risk of IAH. This model, based on routine clinical indicators and with high sensitivity, represents a practical screening tool. Clinicians should perform early risk assessment and intervention in diabetic patients with higher BMI, lower TC, advanced age, malnutrition, renal insufficiency, or PN.