慢性心力衰竭患者营养不良风险列线图预测模型的构建与验证

目的 分析慢性心力衰竭患者营养不良的危险因素,构建营养不良风险列线图预测模型。方法 选取2020年9月至2021年10月于唐山市某三甲医院心内科住院治疗的320例慢性心力衰竭患者为研究对象,采集患者一般资料、疾病相关资料,采用单因素分析和多因素logistic回归分析慢性心力衰竭患者营养不良的危险因素,建立营养不良风险的列线图预测模型,采用受试者工作特征曲线和校准曲线评价列线图模型的诊断效能和校准度、临床决策曲线评估临床收益性。结果 多因素logistic回归分析显示,年龄≥65岁(OR=2.472,95%CI: 1.189～5.139)、水肿(OR=2.228, 95%CI: 1.190～4.171)、贫血(OR=3.266, 95%CI: 1.566～6.810)、血清尿素氮≥8.85 mmol/L(OR=3.241,95%CI: 1.639～6.411)、血清总蛋白＜60g/L(OR=9.166,95%CI: 4.083～20.576)是慢性心力衰竭患者营养不良的独立危险因素。列线图预测模型的受试者工作特征曲线下面积为0.833,灵敏度为74.04%,特异度为84.72%,H-L检验P=0.254,校准曲线内部验证的C-index为0.824,平均绝对误差为0.022。临床决策曲线的阈值概率为0.08～0.9时,临床净收益率较高。结论 基于危险因素构建的慢性心力衰竭患者营养不良风险列线图预测模型有较好的预测效率和临床适用性,有望成为营养不良的重要预测工具。

Construction and validation of anomogram prediction model for malnutrition risk in patients with chronic heart failure

Objective To analyze the risk factors of malnutrition in patients with chronic heart failure, a nomogram prediction model of malnutrition risk is constructed. Methods A total of 320 patients with chronic heart failure who were hospitalized in the Department of Cardiology of a hospital in Tangshan from September 2020 to October 2021 were selected as the research objects. The general data and clinical medical records of the patients were collected, and univariate analysis and multivariate Logistic regression were used to analyze the risk factors of malnutrition in patients with chronic heart failure, and a nomogram prediction model of malnutrition risk was established. Receiver operating characteristic curve and calibration curve were used to evaluate the diagnostic performance and calibration degree of the nomogram model, and clinical decision curve to evaluate clinical benefit. Results Multivariate logistic regression analysis showed that age greater than or equal to 65 years(OR=2.472, 95%CI: 1.189-5.139), edema(OR=2.228, 95%CI: 1.190-4.171), anemia(OR=3.266, 95%CI: 1.566-6.810), serum urea nitrogen greater than or equal to 8.85 mmol/L(OR=3.241, 95%CI: 1.639-6.411), serum total protein less than 60g/L(OR=9.166, 95%CI: 4.083-20.576)are chronic independent risk factors for malnutrition in heart failure patients. The area under the curve of the nomogram prediction model was 0.833, the sensitivity was 74.04%, the specificity was 84.72%, the H-L test P=0.254, the concordance index of the internal validation of the calibration curve was 0.824, and the mean absolute error was 0.022. When the threshold probability of the decision curve analysis is 0.08-0.9, the clinical net return rate is higher. Conclusion The nomogram prediction model of malnutrition risk in chronic heart failure patients based on risk factors has good prediction efficiency and clinical applicability, and is expected to become an important prediction tool for the occurrence of malnutrition.