Abstract: Background　Currently，there are few studies on objective nutritional assessment tools in patients with heart failure，and there is no recognized nutritional assessment method for such patients. Objective　To evaluate the prognostic value of geriatric nutritional risk index（GNRI） in hospitalized patients with chronic heart failure. Methods　A total of 293 chronic heart failure inpatients were selected from Heart Center，Hebei General Hospital between January and December 2017.Clinical data（including general demographics，admission heart rate and blood pressure，NYHA class，laboratory indices，and echocardiographic indices） collected through consulting electronic medical records were compared between patients divided by GNRI〔GNRI group 1（100 cases with GNRI≤98，having a risk of malnutrition），and GNRI group 2（193 cases with GNRI>98，and normal nutrition）〕. All cases were followed up by telephone till 2019-04-25. All-cause mortality was used as the endpoint. ROC analysis was conducted to estimate the predictive value of GNRI，BMI and serum albumin for death in chronic health failure. The Kaplan-Meier curve was plotted to estimate the survival in two GNRI groups. The Cox proportional-hazards model was adopted to examine the association of GNRI with the risk of death in chronic heart failure. Results　The median follow-up time was 509（253）days. For predicting death in chronic heart failure，GNRI〔AUC=0.702，95%CI（0.633，0.772） with the optimal cut-off value of 97.87〕 had higher accuracy than serum albumin 〔AUC=0.657，95%CI（0.586，0.728） with the optimal cut-off value of 35.65 g/L〕，and BMI〔AUC=0.645，95%CI（0.572，0.718） with the optimal cut-off value of 22.88 kg/m2〕. Kaplan-Meier survival analysis found that GNRI group 2 had a statistically higher survival rate than GNRI group 1（P<0.001）. Univariate Cox regression analysis of factors showing statistically intergroup differences and potential factors associated with death revealed that age 〔HR=1.048，95%CI（1.026，1.071）〕，GNRI level〔HR=3.283，95%CI（2.085，5.171）〕，history of diabetes 〔HR=1.955，95%CI（1.245，3.071）〕，history of old cerebral infarction 〔HR=1.779，95%CI（1.131，2.800）〕，use of ACEI/ARB〔HR=0.470，95%CI（0.300，0.763）〕，hospital stay 〔HR=1.036，95%CI（1.018，1.054）〕，diastolic blood pressure 〔HR=0.984，95%CI（0.969，0.999）〕，NYHA class 〔HR=1.546，95%CI（1.066，2.240）〕，serum hemoglobin 〔HR=0.987，95%CI（0.979，0.995）〕，serum creatinine〔HR=1.003，95%CI（1.001，1.005）〕，and glomerular filtration rate 〔HR=0.985，95%CI（0.975，0.995）〕 were associated with death in chronic heart failure（P<0.05）. Multivariate Cox regression analysis of the above-mentioned factors showed that age 〔HR=1.030，95%CI（1.006，1.054）〕，GNRI level 〔HR=2.821，95%CI（1.727，4.616）〕，and diabetes history 〔HR=2.031，95%CI（1.261，3.273）〕 were associated with the risk of death in chronic heart failure（P<0.05）. Conclusion　Malnutrition may be a risk factor for poor prognosis in hospitalized patients with chronic heart failure. The prognostic value of nutritional status assessed by GNRI may be higher than that of serum albumin or BMI，which will help to predict the prognosis of such patients.

Key words: Heart failure, Geriatric nutritional risk index, Nutritional status, Prognosis, Survival analysis

摘要： 背景　目前客观营养状态评估工具在心力衰竭患者中的研究较少，且尚无公认的心力衰竭患者营养评估方法。目的　利用老年人营养风险指数（GNRI）评价住院慢性心力衰竭患者营养状态对预后的预测价值。方法　选择2017年1—12月在河北省人民医院心脏中心住院治疗的慢性心力衰竭患者293例，根据GNRI评估的营养状态分为有营养不良风险GNRI≤98为GNRI 1组（n=100）和无营养不良风险GNRI>98为GNRI 2组（n=193）。从电子病历中收集患者的临床资料，主要包括一般资料、入院时患者心率及血压、美国纽约心脏病协会（NYHA）分级、实验室检查指标及心脏超声检查指标。对研究对象进行电话随访，随访截止日期为2019-04-25，终点事件定义为全因死亡。采用受试者工作特征（ROC）曲线分析GNRI、体质指数（BMI）、清蛋白分别对慢性心力衰竭患者死亡事件的预测价值，采用Kaplan-Meier法绘制生存曲线分析不同GNRI对患者的生存影响；采用Cox比例风险回归模型分析探讨GNRI对慢性心力衰竭患者死亡风险的影响。结果　随访中位时间是509（253）d。GNRI〔ROC曲线下面积（AUC）=0.702，95%CI（0.633，0.772），最佳截断值为97.87〕预测慢性心力衰竭患者死亡事件能力优于清蛋白〔AUC=0.657，95%CI（0.586，0.728），最佳截断值为35.65 g/L〕及BMI〔AUC=0.645，95%CI（0.572，0.718），最佳截断值为22.88 kg/m2〕。绘制Kaplan-Meier生存曲线发现，GNRI 2组生存率高于GNRI 1组（P<0.001）。对两组临床资料比较差异有统计学意义的变量及有可能影响慢性心力衰竭患者死亡率的因素进行单因素Cox比例风险回归模型分析，结果显示，年龄〔风险比（HR）=1.048，95%CI（1.026，1.071）〕、GNRI分组〔HR=3.283，95%CI（2.085，5.171）〕、糖尿病史〔HR=1.955，95%CI（1.245，3.071）〕、陈旧性脑梗死史〔HR=1.779，95%CI（1.131，2.800）〕、血管紧张素转换酶抑制剂（ACEI）/血管紧张素受体拮抗剂（ARB）〔HR=0.470，95%CI（0.300，0.763）〕、住院时间〔HR=1.036，95%CI（1.018，1.054）〕、舒张压〔HR=0.984，95%CI（0.969，0.999）〕、NYHA分级〔HR=1.546，95%CI（1.066，2.240）〕、血红蛋白〔HR=0.987，95%CI（0.979，0.995）〕、肌酐〔HR=1.003，95%CI（1.001，1.005）〕、肾小球滤过率（GFR）〔HR=0.985，95%CI（0.975，0.995）〕是慢性心力衰竭患者死亡事件的影响因素（P<0.05）。将上述指标进行多因素Cox比例风险回归模型分析，结果显示，年龄〔HR=1.030，95%CI（1.006，1.054）〕、GNRI分组〔HR=2.821，95%CI（1.727，4.616）〕、糖尿病史〔HR=2.031，95%CI（1.261，3.273）〕是慢性心力衰竭患者死亡风险的影响因素（P<0.05）。结论　营养不良是住院慢性心力衰竭患者不良预后的危险因素。GNRI评估营养状态有助于预测住院慢性心力衰竭患者的预后且其预测价值优于清蛋白、BMI。

关键词: 心力衰竭, 老年人营养风险指数, 营养状态, 预后, 生存分析