2018 年第 4 期 第 13 卷

营养学相关临床易用性检验指标对老年重症肺炎患者死亡风险的预测价值

Predictive value of nutrition related clinical easy-to-use indicators for death risk of elderly patients with severe pneumonia

作者：唐万平颜晓玉周晖曾强林

英文作者：

单位：610082成都大学附属医院呼吸内科&肺血管病诊治中心

英文单位：

关键词：重症肺炎；营养状态；易用性检验指标；受试者工作特征曲线

英文关键词：

• 摘要：

• 目的    探讨营养学相关临床易用性检验指标对老年重症肺炎患者死亡风险的预测价值。方法    回顾性分析2012年7月至2015年6月成都大学附属医院呼吸内科收治的97例老年重症肺炎患者的病历资料，依据患者治疗结局分为死亡组（38例）和未死亡组（59例）。收集患者的淋巴细胞百分比、血红蛋白、总蛋白、白蛋白、前白蛋白、血肌酐、总胆固醇等指标，采用受试者工作特征曲线分析各指标对患者死亡风险预测的曲线下面积（AUC）、敏感度及特异度。结果    血红蛋白（AUC＝0.660，敏感度＝0.568，特异度＝0.769）和总蛋白（AUC＝0.718，敏感度＝0.514，特异度＝0.808）对于老年重症肺炎患者死亡风险的预测价值有统计学意义（均P＜0.05）。多指标联合预测的价值更高：AUC＝0.782，敏感度＝0.808，特异度＝0.676（P＜0.05）。结论    营养学相关临床易用性指标血红蛋白和总蛋白对老年重症肺炎患者死亡风险具有较好的预测价值，多指标联合检测可提高预测价值。

• Objective    To analyze the predictive value of nutrition related clinical easy-to-use indicators for death risk of elderly patients with severe pneumonia. Methods    Clinical records of 97 elderly patients with severe pneumonia in Affiliated Hospital of Chengdu University were retrospectively analyzed from July 2012 to June 2015; the patients were divided into dead group(n＝38) and survival group (n＝59) according to the outcomes. Related indicators including lymphocyte ratio, hemoglobin, total protein, albumin, prealbumin, serum creatinine and cholesterol were collected; values of these indicators for predicting death risk of patients were analyzed by the receiver operating characteristic curve. Results   Area under curve of hemoglobin in predicting death risk was 0.660; the sensitivity was 0.568 and the specificity was 0.769. Area under curve of total protein in predicting death risk was 0.718; the sensitivity was 0.514 and the specificity was 0.808（both P＜0.05）. Combined detection of all indicators showed a high predictive value; area under curve was 0.782; the sensitivity was 0.808 and the specificity was 0.676（P＜0.05）. Conclusion    Hemoglobin and total protein have good values in predicting death risk of elderly patients with severe pneumonia; combined detection of multiple indicators can improve the predictive value.