目的 探讨在全球领导人营养不良倡议(GLIM)标准下,患者主观整体评估(PG-SGA)、预后营养指数(PNI)和老年 营养风险指数(GNRI)在老年消化系统肿瘤患者中的适用性。 方法 选取 2021 年 3 月至 2022 年 6 月在徐州医科大学附属医院 接受消化系统肿瘤切除术的老年患者,以 GLIM 为诊断营养不良的“金标准”,应用 PG-SGA、PNI 和 GNRI 进行营养评估。 计 算并比较 3 种营养评估工具的灵敏度、特异度、阳性预测值和阴性预测值,绘制受试者操作特征(ROC)曲线并使用 Delong 检 验比较曲线下面积(AUC),探究各工具的适用价值。 结果 经 GLIM 诊断下营养不良阳性率为 29. 3%,PG-SGA、PNI 和 GNRI 评估分别有 37. 0%、44. 0%和 45. 7%的患者为营养不良。 体质指数(BMI)、肌肉质量和相位角(PA)水平显示更差,血清白蛋 白(ALB)、前白蛋白(PA)、血红蛋白(HB)水平均较低,差异存在统计学意义(P<0. 05)。 GLIM 标准下,PG-SGA 的灵敏度最 差( 44. 1%),特异度一般( 65. 9%),与 GLIM 诊断无一致性( Kappa = 0. 09,P = 0. 311);PNI 的灵敏度( 61. 8%) 和特异度 (63. 4%)均一般,与 GLIM 诊断的一致性一般(Kappa = 0. 22,P = 0. 013);GNRI 特异度(88. 2%)和灵敏度(72. 0%)均表现最 佳,与 GLIM 诊断具有良好的一致性(Kappa = 0. 52,P<0. 001)。 ROC 曲线图中,GNRI 的预测价值最高(AUC = 0. 817,95%CI = 0. 734~ 0. 882),PNI 次之(AUC= 0. 621,95%CI = 0. 526~ 0. 709)。 Delong 检验显示 GNRI 与 PG-SGA、GNRI 与 PNI 的预测价值 比较均具有统计学差异(Z= 3. 83,4. 85,P<0. 001)。 GNRI 评估为营养不良组术前、术后 1 d 及术后 3 ~ 5 d 的 PA、ALB、TP 和 HB 指标均显著低于营养良好者;此外,营养不良者 90 d 内手术部位感染发生率也更高( χ 2 = 5. 03,P = 0. 036)。 结论 GNRI 在 老年消化系统肿瘤患者的营养评估中表现出较好的适用性。 GNRI 可能预测出患者围手术期的营养状况和短期预后。

Objective To explore the applicability of patient -generated subjective nutrition assessment PG-SGA prognostic nutrition index PNI and geriatric nutrition risk index GNRI in the nutritional assessment of elderly patients with digestive system tumors based on Global Leadership Initiative on Malnutrition criteria GLIM . Method Elderly patients who underwent digestive system tumor resection in Affiliated Hospital of Xuzhou Medical University from March 2021 to June 2022 were selected. With GLIM as the " gold standard" for the diagnosis of malnutrition PG-SGA PNI and GNRI were used for nutritional assessment and the sensitivity、 specificity、positive predictive value and negative predictive value were calculated and compared. The receiver operating characteristic ROC curve was plotted and the area under the curve AUC was compared by Delong test to explore the clinical value of each tool comparison between groups was conducted. Result The positive rate of malnutrition diagnosed by GLIM was 29. 3%. The positive rates of PG-SGA PNI and GNRI were 37. 0% 44. 0% and 45. 7% respectively. The body mass index muscle loss and phase angle indexes were worse in the patients with malnutrition and the serum albumin ALB prealbumin PA and hemoglobin HB indexes were decreased with statistical significance P < 0. 05 . Compared with GLIM standard PG - SGA showed the worst sensitivity 44. 1% and middle specificity 65. 9% with no diagnostic consistency Kappa = 0. 09 P = 0. 311 The sensitivity 61. 8% and specificity 63. 4% of PNI were average and there was weak agreement with GLIM diagnosis Kappa = 0. 22 P= 0. 013 . GNRI had the best specificity 88. 2% and sensitivity 72. 0% and had a good concordance with GLIM diagnosis Kappa = 0. 52 P<0. 001 . In the ROC curve GNRI had the highest predictive value AUC= 0. 817 95%CI = 0. 734-0. 882 followed by PNI AUC = 0. 621 95%CI = 0. 526 - 0. 709 . Delong test showed that the predictive value of GNRI and PG- SGA GNRI and PNI were statistically significant Z= 3. 83 4. 85 P < 0. 001 . The indexes of PA ALB TP and HB before、1 day after and 3-5 days after the surgery of the malnourished patients were significantly lower P<0. 05 . In addition the incidence of surgical site infection SSI was also higher in malnourished patients χ 2 = 5. 03 P = 0. 036 . Conclusion GNRI has shown good applicability in the nutritional assessment of elderly patients with digestive system tumors. In addition GNRI may be able to predict a patient's perioperative nutritional status and short-term prognosis.