Risk factors for heart failure in patients with hemodialysis and construction of nomogram model
10.3760/cma.j.cn115455-20221229-01150
- VernacularTitle:血液透析患者发生心力衰竭的危险因素分析及列线图模型构建
- Author:
Li TANG
1
;
Min TIAN
;
Ximin QIAO
;
Lina CAO
;
Ping WANG
Author Information
1. 咸阳市中心医院肾病风湿免疫科，咸阳 712000
- Keywords:
Dialysis;
Heart failure;
Risk factors;
Nomograms
- From:
Chinese Journal of Postgraduates of Medicine
2023;46(7):651-657
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To analyze the risk factors for heart failure in patients with hemodialysis, and to construct a nomogram model.Methods:The clinical data of 218 patients with hemodialysis in Xianyang Central Hospital from January 2021 to April 2022 were retrospectively analyzed. Among them, 83 cases developed heart failure (heart failure group), and 135 cases did not develop heart failure (control group). The relevant clinical data were recorded, including age, sex, body mass index, disease duration, concurrent infection, blood calcium, blood phosphorus, soluble CD 146 (sCD 146), soluble growth-stimulated expression gene 2 protein (sST2), N-terminal brain natriuretic peptide precursor (NT-proBNP), time-averaged urea concentration (TACurea), tumor necrosis factor α (TNF-α), blood creatinine and 24 h urine volume. Receiver operating characteristic (ROC) curve was used to analyze the efficacy of each index in predicting heart failure in patients with hemodialysis. Multivariate Logistic regression was used to analyze the independent risk factors of heart failure in patients with hemodialysis. R language software 4.0 "rms" package was used to construct the nomogram model for predicting the heart failure in patients with hemodialysis, the calibration curve was internally validated, and the decision curve was used to evaluate the predictive efficacy of the nomogram model. Results:There were no statistical difference in gender composition, age, body mass index, disease duration, 24 h urine volume and blood creatinine between the two groups ( P>0.05); the rate of concurrent infection, blood phosphorus, sCD 146, sST2, NT-proBNP, TNF-α and TACurea in heart failure group were significantly higher than those in control group: 39.76% (33/83) vs. 8.89% (12/135), (1.53 ± 0.34) mmol/L vs. (1.27 ± 0.24) mmol/L, (43.60 ± 10.24) μmol/L vs. (28.08 ± 7.99) μmol/L, (49.00 ± 9.41) μg/L vs. (34.53 ± 8.05) μg/L, (38.57 ± 6.79) μg/L vs. (29.72 ± 5.64) μg/L, (5.18 ± 0.92) μg/L vs. (4.07 ± 1.13) μg/L and (24.28 ± 4.37) mmol/L vs. (17.96 ± 2.52) mmol/L, the blood calcium was significantly lower than that in control group: (1.95 ± 0.36) mmol/L vs. (2.31 ± 0.39) mmol/L, and there were statistical differences ( P<0.01). ROC curve analysis result showed that the optimal cut-off values of blood calcium, blood phosphorus, sCD 146, sST2, NT-proBNP, TNF-α and TACurea for heart failure in patients with hemodialysis were 2.01 mmol/L, 1.42 mmol/L, 34.15 μmol/L, 40.37 μg/L, 35.37 μg/L, 4.33 μg/L and 20.74 mmol/L. Multivariate Logistic regression analysis result showed that the blood calcium (≤2.01 mmol/L), blood phosphorus (>1.42 mmol/L), sCD 146 (>34.15 μmol/L), sST2 (>40.37 μg/L), NT-proBNP (>35.37 μg/L), TNF-α (>4.33 μg/L) and TACurea (>20.74 mmol/L) were independent risk factors for heart failure in patients with hemodialysis ( OR = 1.183, 1.582, 1.915, 1.105, 1.459, 1.347 and 1.717; 95% CI 1.102 to 1.191, 1.274 to 1.868, 1.716 to 2.105, 1.072 to 1.141, 1.225 to 1.703, 1.132 to 1.574 and 1.482 to 1.935; P<0.05 or <0.01). The blood calcium, blood phosphorus, sCD 146, sST2, NT-proBNP, TNF-α and TACurea were used as predictors to construct a nomogram model for predicting heart failure in patients with hemodialysis. Internal validation result showed that the nomogram model predicted the heart failure with good concordance in patients with hemodialysis (C-index = 0.811, 95% CI 0.675 to 0.948); the nomogram model predicted the heart failure in patients with hemodialysis at a threshold>0.18, provided a net clinical benefit, and all had higher clinical net benefits than blood calcium, blood phosphorus, sCD 146, sST2, NT-proBNP, TNF-α and TACurea. Conclusions:The nomogram model constructed based on blood calcium, blood phosphorus, sCD 146, sST2, NT-proBNP, TNF-α and TACurea has better clinical value in predicting the heart failure in patients with hemodialysis.
