Admission-based predictors of in-hospital AKI in HFrEF: a single-centre retrospective study
- 1Department of Cardiology, The Fourth Affiliated Hospital of Harbin Medical University, Harbin, China
Abstract
Objective: This study aimed to construct and internally validate a concise prediction model for in-hospital acute kidney injury (AKI) among patients admitted with heart failure with reduced ejection fraction (HFrEF). The model was limited to information available at hospital presentation.
Methods: We retrospectively analyzed 1,071 HFrEF hospitalizations at the Fourth Affiliated Hospital of Harbin Medical University from September 2020 to September 2022. The outcome was incident AKI during the index admission, defined using a modified KDIGO (48-h serum creatinine) criterion, based on an increase of ≥26.5 µmol/L within 48 h. Candidate variables measured at admission were screened with least absolute shrinkage and selection operator (LASSO) logistic regression using the 1−SE rule. Bootstrap stability and clinical parsimony were then considered to preserve temporal validity and bedside usability. Internal validation was performed using 5-fold stratified out-of-fold cross-validation.
Results: In-hospital AKI occurred in 118 of the 1,071 patients (11.0%). Five key on-admission clinical parameters were identified as the strongest predictors: age, furosemide-equivalent dose, hypotension on presentation, RDW, eGFR. In internal validation, the combination of these five strictly admission-based parameters achieved an AUC-ROC of 0.81 [95% bootstrap confidence interval (CI): 0.77–0.85], a precision-recall AUC (PR-AUC) of 0.47, a Brier score of 0.08, a calibration intercept of −0.02, and a calibration slope of 0.98. Decision curve analysis demonstrated clinical utility with a positive net benefit across most clinically plausible decision thresholds.
Conclusion: A five-predictor model based entirely on admission data can provide interpretable early risk stratification for in-hospital AKI in patients hospitalized with HFrEF. By excluding post-admission treatment responses and weight change, the model reduces reverse-causality bias and may support renal safety decisions at the time of presentation.
Related
- Prognostic Value of Admission Serum Chloride and of the Composition of Metabolic Acidosis in Postoperative Intensive Care: A Single-Centre Retrospective Cohort Study
- Optimising hospital pharmacy inventory for high-value, intermittent-demand drugs using a machine-learning approach with LightGBM-Tweedie regression: a single-centre, retrospective study.
- Predictors of 90-day Readmission After Hospitalization in Heart Failure with Reduced Ejection Fraction Patients: A Single-Centre Analytical Cohort Retrospective Study
- Diagnostic delay and predictors of true stroke among patients with suspected in-hospital stroke: a single-center retrospective study
- Exploratory clinical predictors and internally validated models of in-hospital mortality after PCI in obese patients with acute myocardial infarction and documented cardiogenic shock: a single-center retrospective study.
BibTeX
@article{dai2026admissionbased,
title = {{Admission-based predictors of in-hospital AKI in HFrEF: a single-centre retrospective study}},
author = {Feng Dai and Xiaomin Li and Zihe Wang and Siyu Tan and Fengyang Han and Xiang Song},
journal = {Frontiers in Physiology},
year = {2026},
month = oct,
doi = {10.3389/fphys.2026.1928628},
url = {https://doi.org/10.3389/fphys.2026.1928628},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Admission-based predictors of in-hospital AKI in HFrEF: a single-centre retrospective study
AU - Feng Dai
AU - Xiaomin Li
AU - Zihe Wang
AU - Siyu Tan
AU - Fengyang Han
AU - Xiang Song
PY - 2026
DA - 2026/10/12
JO - Frontiers in Physiology
DO - 10.3389/fphys.2026.1928628
UR - https://doi.org/10.3389/fphys.2026.1928628
AB - Objective: This study aimed to construct and internally validate a concise prediction model for in-hospital acute kidney injury (AKI) among patients admitted with heart failure with reduced ejection fraction (HFrEF). The model was limited to information available at hospital presentation. Methods: We retrospectively analyzed 1,071 HFrEF hospitalizations at the Fourth Affiliated Hospital of Harbin Medical University from September 2020 to September 2022. The outcome was incident AKI during the index admission, defined using a modified KDIGO (48-h serum creatinine) criterion, based on an increase of ≥26.5 µmol/L within 48 h. Candidate variables measured at admission were screened with least absolute shrinkage and selection operator (LASSO) logistic regression using the 1−SE rule. Bootstrap stability and clinical parsimony were then considered to preserve temporal validity and bedside usability. Internal validation was performed using 5-fold stratified out-of-fold cross-validation. Results: In-hospital AKI occurred in 118 of the 1,071 patients (11.0%). Five key on-admission clinical parameters were identified as the strongest predictors: age, furosemide-equivalent dose, hypotension on presentation, RDW, eGFR. In internal validation, the combination of these five strictly admission-based parameters achieved an AUC-ROC of 0.81 [95% bootstrap confidence interval (CI): 0.77–0.85], a precision-recall AUC (PR-AUC) of 0.47, a Brier score of 0.08, a calibration intercept of −0.02, and a calibration slope of 0.98. Decision curve analysis demonstrated clinical utility with a positive net benefit across most clinically plausible decision thresholds. Conclusion: A five-predictor model based entirely on admission data can provide interpretable early risk stratification for in-hospital AKI in patients hospitalized with HFrEF. By excluding post-admission treatment responses and weight change, the model reduces reverse-causality bias and may support renal safety decisions at the time of presentation.
SN - 1664-042X
ER -