目的·探讨老年营养风险指数（geriatric nutritional risk index，GNRI）对住院老年2型糖尿病（type2 diabetes mellitus，T2DM）患者发生肺部感染的预测价值。方法·回顾性连续收集2022年6月至2024年6月期间在上海交通大学医学院附属同仁医院老年科住院治疗的老年T2DM患者。根据住院期间患者是否发生肺部感染分为感染组和非感染组，统计患者基线资料（性别、年龄、身高、体质量、糖尿病病程、合并症等）并计算GNRI。使用多变量Logistic回归模型筛选发生肺部感染的独立危险因素。应用受试者操作特征（receiver operating characteristic，ROC）曲线分析GNRI对T2DM患者发生肺部感染的预测价值。结果·共纳入264例住院老年T2DM患者，其中154例患者发生肺部感染。感染组的GNRI、血清白蛋白、白细胞、中性粒细胞比值、淋巴细胞比值、糖化血红蛋白、空腹血糖、白介素-6、C反应蛋白、降钙素原和非感染组差异有统计学意义（均P<0.05）。多变量Logistic回归分析显示，GNRI较低是发生肺部感染的独立危险因素（OR=0.798，95%CI 0.712~0.894；P<0.001）。相关性分析提示GNRI与C反应蛋白、降钙素原呈负相关。ROC曲线分析显示，GNRI预测肺部感染曲线下面积为0.828，敏感度为77.9%，特异度为76.6%。结论·GNRI较低是老年T2DM患者发生肺部感染的独立危险因素，同时对是否发生肺部感染有较好的预测价值。

Objective ·To investigate the predictive value of the geriatric nutritional risk index (GNRI) for the occurrence of lung infection in hospitalized elderly patients with type 2 diabetes mellitus (T2DM). Methods ·Elderly T2DM patients who were admitted to the Geriatric Department of Shanghai Tongren Hospital between June 2022 and June 2024 were retrospectively and consecutively enrolled. They were divided into infected and non-infected groups according to whether lung infection occurred during hospitalization. Baseline data (gender, age, height, weight, duration of diabetes, comorbidities, etc.) were collected and GNRI was calculated. A multivariate Logistic regression model was used to screen the independent risk factors for pulmonary infections,and the predictive value of GNRI for pulmonary infections in T2DM patients was analysed using receiver operating characteristic (ROC) curves. Results ·A total of 264 elderly T2DM patients were enrolled, among whom 154 developed pulmonary infections. Significant differences were observed between the infected and non-infected groups in GNRI, albumin, leukocyte count, neutrophil ratio, lymphocyte ratio, glycated hemoglobin, fasting glucose, interleukin-6, C-reactive protein, and procalcitonin levels (P<0.05). Multivariate Logistic regression analysis showed that a lower GNRI was an independent risk factor for lung infection (OR=0.798, 95%CI 0.712‒0.894; P<0.001). Correlation analysis showed that GNRI was negatively correlated with C-reactive protein and calcitoninogen. ROC curve analysis showed that GNRI predicted pulmonary infection with an area under the curve of 0.828, a sensitivity of 77.9%, and a specificity of 76.6%. Conclusion ·A lower GNRI is an independent risk factor for pulmonary infections in elderly T2DM patients, and also has a good predictive value for the occurrence of pulmonary infections.