Analysis of risk factors and construction of a dynamic nomogram for postoperative heart failure in elderly patients with knee joint fractures

QING Qing; XUE Hua-wei; CHEN Wei

doi:10.3969/j.issn.2095-252X.2025.01.011

Abstract：Objective To analyse risk factors for postoperative heart failure (HF) in elderly knee fracture patients and construct a network calculator to predict postoperative HF risk. Methods Elderly knee fracture patients who were surgically treated between January 2022 and January 2024 in our hospital were selected. Logistic regression and random forest (RF) were used to analyse the risk factors for postoperative HF in elderly knee fracture patients. Risk factors were used as predictors for constructing nomogram. Internal data were used to validate the performance of the nomogram,and receiver operating characteristic (ROC) curves and decision curve analyses were used to assess the predictive accuracy of the nomogram. The DynNom package was used to publish the nomogram to the web to construct a web-based calculator of postoperative HF risk in older adults with knee fractures. Results The incidence of postoperative HF in 512 elderly patients with knee fractures was 18.0％ (92/512). Logistic regression analysis showed that age,diabetes mellitus,proportion of American Society of Anesthesiologists (ASA) classification (grade Ⅲ-Ⅳ),haemoglobin (Hb),brain natriuretic peptide (BNP),platelets (PLT) and operation time were independent risk variables for postoperative HF. RF algorithm results showed that the top 6 variables associated with the risk of postoperative HF were,in order,BNP,duration of surgery,PLT,Hb,alanine aminotransferase (ALT) and age. A nomogram predicting postoperative HF in elderly knee fractures was constructed based on the common risk variables age,Hb,BNP,PLT,and operative time from logistic regression and RF algorithm. The ROC curve analysis of the nomogram showed an area under the curve (AUC) of 0.943,and the calibration curve analysis showed a consistency index of 0.905,indicating the predictive accuracy of the nomogram. Conclusions A network calculator for predicting postoperative HF risk after knee fracture in the elderly,constructed on the basis of age,Hb,BNP,PLT,and operative time,can provide a practical tool for clinicians to actively perform cardiac risk assessment.