Derivation and validation model for hospital hypoglycemia

Area of Science:

• Endocrinology
• Internal Medicine
• Clinical Prediction Models

Background:

• Hypoglycemia poses a significant risk for hospitalized patients.
• An objective prognostic model is needed to guide treatment intensity.

Purpose of the Study:

• To develop and validate a simple prognostic model for predicting hypoglycemia in hospitalized patients.
• To identify key factors associated with hypoglycemia for risk stratification.

Main Methods:

• A retrospective cohort study using nationwide data from 2014 (n=839) for derivation.
• Validation performed on a separate nationwide cohort from 2016 (n=561).
• Logistic regression analysis was used to derive the prediction rule for hypoglycemia (glucose < 70 mg/dL).

Main Results:

• The incidence of hypoglycemia was 10.3% in the derivation cohort.
• Independent predictors of hypoglycemia included prior hypoglycemia, low estimated glomerular filtration rate (eGFR), high daily insulin dose, and longer hospitalization.
• The model demonstrated fair predictive accuracy with an area under the curve (AUC) of 0.72 in the derivation cohort and 0.71 in the validation cohort.

Conclusions:

• The developed rule shows fair accuracy in predicting hypoglycemia.
• Integration into clinical decision support systems can aid in guiding initial insulin therapy for at-risk patients.