AUTHOR=Sun Jiao , Ran Mohan , Lv Shiying , Li Jiacheng , Zan Hongjing , Li Wei , Li Qingchu 

TITLE=Unraveling hypoglycemia risk during hemodialysis: a predictive model from a nested case-control study

JOURNAL=Frontiers in Physiology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2025.1660936

DOI=10.3389/fphys.2025.1660936

ISSN=1664-042X

ABSTRACT=BackgroundHemodialysis (HD) can significantly lower blood glucose levels, increasing the risk of hypoglycemia. The contributing factors are not fully understood. This study aimed to identify key risk factors for hypoglycemia during HD and develop a predictive model.MethodsA retrospective nested case-control study was conducted at the Third Hospital of Shandong Province from January 2020 to December 2023. Clinical and laboratory data were collected from electronic medical records and patient questionnaires. Univariate and multivariate analyses identified independent risk factors, and a predictive model was developed using stepwise logistic regression. Internal validation was performed using 10-fold stratified cross-validation, with model performance evaluated by mean area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity.ResultsAmong 114 HD patients (57 cases, 57 controls), six independent risk factors were identified: afternoon HD session, presence of cardiovascular disease, and low levels of albumin (<37.35 g/L), creatinine (<828.65 μmol/L), urea (<28.05 mmol/L), and pre-dialysis blood glucose (<5.75 mmol/L). The predictive model demonstrated good internal validity with mean AUC 0.79, accuracy 0.71, sensitivity 0.64, and specificity 0.78, indicating stable discriminative performance.ConclusionSix key risk factors for hypoglycemia during HD were identified, and a predictive model integrating disease status, HD timing, and laboratory markers was developed. Early identification of high-risk patients may help prevent hypoglycemic events and improve HD outcomes. Future studies should externally validate and refine this model for broader clinical application.