营养风险指数预估老年血液透析患者病死率的价值

Value of the Geriatric Nutritional Risk Index in predicting mortality of elderly patients undergoing hemodialysis

目的探讨营养风险指数（GNRI）在预估老年血液透析患者病死率中的价值。方法回顾性分析上海交通大学附属第一人民医院肾内科2007年1月至2015年4月125例老年血液透析患者的临床资料。所有患者入组前透析时间超过6个月，计算GNRI指数，根据GNRI指数四分位数分为4组，以患者死亡为终点事件，采用Kaplan–Meier分析和cox回归模型分析两者关系，用ROC曲线分析GNRI对患者病死率的预测价值。结果4组老年血液透析患者GNRI指数分别为第1组≤92.06，第2组92.07～96.15，第3组96.16～101.25，第4组≥101.26，4组全因病死率分别为68.8%、58.1%、45.2%和25.8%（P＝0.002），心血管病死率分别为50.0%、33.3%、26.1%和11.5%（P＝0.013）。Kaplan–Meier分析结果提示随着GNRI的升高，5年总体生存率升高，cox回归分析结果显示GNRI与老年血液透析患者病死率显著相关（HR＝0.940，P＝0.001，95%CI：0.907～0.974），与心血管死亡显著相关（HR＝0.906，P〈0.001，95%CI： 0.863～0.951）。ROC曲线提示GNRI预测全因病死率和心血管病死率曲线下面积分别为0.667（P＝0.001）和0.717（P＝0.001）。结论GNRI作为一个简单易行的客观评价指标，与血液透析患者的预后密切相关。

ObjectiveTo explore the reliability of the Geriatric Nutritional Risk Index （GNRI） as a mortality predictor in elderly patients undergoing hemodialysis （HD）.MethodsA total of 125 maintenance HD patients aged 〉60 years old who had received dialysis for over 6 months before entry was retrospectively examined. The values of GNRI were calculated, and death was taken as the end point. The patients were divided into 4 groups according to the quartiles of GNRI values. All-cause and cardiovascular mortality were calculated using Kaplan-Meier and cox proportional-hazards analyses, and ROC curve was adopted for analyzing the predicting value of GNRI on mortality.ResultsThe GNRI of the four groups were ≤92.06, 92.07-96.15, 96.16-101.25, ≥101.26, respectively. Kaplan-Meier survival rate was significantly different among 4 groups for all-cause and cardiovascular mortality. Cox regression model analysis demonstrated that the GNRI was a predictor for all cause （HR=0.940, P=0.001, 95%CI： 0.907-0.974） and cardiovascular mortality （HR=0.906, P〈0.001, 95%CI： 0.863-0.951）. The area under ROC curve was 0.667 （P=0.001） for all-cause mortality and 0.717 （P=0.001） for cardiovascular mortality.ConclusionThe GNRI is a reliable predictor for all-cause mortality in maintainance HD patients, but more multi-center studies with larger samples are still needed.