Establishment and validation of a prediction and risk stratification model for readmission in patients with coronary heart disease complicated by heart failure

LU Aifen; LIU Junyou; ZHANG Ke

doi:10.3969/j.issn.1674-4055.2026.01.016

Abstract：Objective To establish and internally validate a prediction and risk stratification model for readmission in patients with coronary heart disease(CHD)complicated by heart failure(HF).Methods A total of 212 patients with CHD complicated by HF admitted to Puyang Oilfield General Hospital were retrospectively enrolled from June 2022 to Oct.2024.They were divided into readmission group(n=65)and control group(n=147)according to whether unplanned readmission occurred within 6 months after treatment.Baseline demographic data,medical history,disease-related data,laboratory test results and other clinical information of the patients were collected.After univariate analysis,variables were screened using the least absolute shrinkage and selection operator(LASSO)regression model.The screened indicators were then incorporated into a multivariate Logistic regression model to identify independent factors associated with readmission in CHD patients complicated by HF.Rsoftware was applied to construct a Nomogram warning model for predicting patient readmission.Receiver operating characteristic(ROC)curves were plotted to evaluate the predictive efficacy of the model,and Bootstrap method was used for internal validation.Calibration curves and decision curve analysis(DCA)were performed to assess the calibration degree and clinical benefit of the model.Furthermore,the SHapley Additive exPlanations(SHAP)model was used to explore the contribution of each independent influencing factor to the warning model.A risk stratification system was established based on the risk scores derived from Nomogram warning model.Results The unplanned readmission rate was 30.66%(65/212)among 212 CHD patients complicated by HF within 6 months after treatment.Compared with control group,the readmission group had significantly higher proportions of diabetes mellitus and chronic kidney disease,higher Charlson comorbidity index(CCI)scores,more advanced New York Heart Association(NYHA)functional classes,longer length of hospital stay,higher levels of triglyceride-glucose index(TyG),N-terminal pro-B-type natriuretic peptide(NT-proBNP),serum creatinine(Cr)and uric acid(UA),whereas levels of medication adherence and hemoglobin(Hb)were notably lower(all P<0.05).The LASSO regression model identified 7 predictive variables,namely CCI score,NYHA functional class,TyG index,NT-proBNP,serum Cr,UA,and medication adherence.The results of multivariate Logistic regression analysis revealed that CCI score,NYHA functional class,TyG index,NT-proBNP,serum Cr and UA were risk factors for unplanned readmission within 6 months in CHD patients complicated by HF,while medication adherence served as a protective factor(all P<0.05).The results of receiver operating characteristic(ROC)curve analysis demonstrated that the Nomogram warning model had an area under the curve(AUC)of 0.967(95%CI:0.947～0.988)for predicting 6-month readmission.The results of calibration curves and DCA analyses indicated that the model had good calibration and clinical utility,with excellent consistency confirmed by internal validation.The results of SHAP model analysis showed that the ranking of the importance of independent factors influencing the Nomogram warning model for readmission in CHD patients complicated by HF was as follows(in descending order):NT-proBNP>TyG index>UA>serum Cr>medication adherence>NYHA functional class>CCI score.A comprehensive risk score was calculated using the Nomogram warning model,and patients were stratified into three risk levels(high,moderate and low risk).The unplanned readmission rate increased significantly with the elevation of risk stratification in CHD patients complicated by HF within 6 months(P<0.050).Conclusion The Nomogram prediction model constructed based on NT-proBNP,TyG index,UA,serum Cr,medication adherence,NYHA functional class and CCI score has good predictive performance for readmission risk in CHD patients complicated by HF.