基于倾向性评分匹配分析老年营养风险指数与老年重症患者并发谵妄的相关性

The correlation between geriatric nutritional risk index and concurrent delirium in the elderly patients with critical illness based on a propensity score matching study

目的基于倾向性评分匹配(PSM)分析老年营养风险指数(GNRI)与老年重症患者并发谵妄的相关性。方法选取2021年1月至2023年9月在海南医学院第二附属医院重症监护病房(ICU)内治疗的577例老年重症患者作为研究对象,收集患者的一般情况及临床资料等数据,计算入ICU时GNRI。绘制GNRI预测老年重症患者并发谵妄的受试者工作特征(ROC)曲线,计算最佳截断值,依次分为低GNRI组(GNRI≤96.63)和高GNRI组(GNRI>96.63),使用PSM法均衡两组临床特征。比较PSM后两组老年重症患者谵妄发生率,采用Logistic回归分析评估PSM前和PSM后数据中GNRI与老年重症患者并发谵妄的关系。另基于PSM后数据,对GNRI的预测效能进行进一步验证。结果ROC曲线分析结果显示,GNRI预测老年重症患者并发谵妄的曲线下面积为0.748(95%CI 0.710~0.783,P<0.001),敏感度为70.29%,特异度为84.58%,最佳截断值为96.63。经PSM法处理后,168对老年重症患者匹配成功,匹配后低GNRI组和高GNRI组基线资料水平均衡可比。PSM后,低GNRI组老年重症患者谵妄发生率为63.69%(107/168),明显高于高GNRI组的17.86%(30/168),差异有统计学意义(χ^(2)=73.071,P<0.001)。Logistic回归分析结果显示,PSM前和PSM后数据中GNRI值降低均为老年重症患者并发谵妄的独立危险因素(P<0.05)。加以验证的ROC分析结果显示,GNRI的曲线下面积是0.753(95%CI 0.704~0.798,P<0.001),以96.44为最佳截断值的敏感度和特异度分别是69.64%和85.12%。结论老年重症患者入ICU时GNRI值降低是并发谵妄的独立危险因素,且GNRI作为一种容易获得的营养状态评估工具,可用于预测老年重症患者并发谵妄。

Objective To analyze the correlation between the geriatric nutritional risk index(GNRI)and concurrent delirium in the elderly patients with critical illness based on a propensity score matching(PSM)study.Methods A total of 577 elderly patients with critical illness treated in the intensive care unit(ICU)of the Second Affiliated Hospital of Hainan Medical University from January 2021 to September 2023 were selected as study subjects,and the general data,clinical data and the GNRI on ICU admission of patients were collected.Receiver operating characteristic(ROC)curve analysis was performed to estimate the predictive value and optimal cut-off value of GNRI for concurrent delirium in the elderly patients with critical illness,and the patients were classified into low GNRI group(GNRI≤96.63)and high GNRI group(GNRI>96.63).PSM method was performed to balance clinical characteristics between two groups.The incidences of delirium in the elderly patients with critical illness between the two groups were compared after PSM,and Logistic regression analysis was used to evaluate the relationship between GNRI and delirium in the elderly patients with critical illness in the data before and after PSM.According to the data after PSM,the predictive efficacy of GNRI was further verified.Results ROC analysis revealed that the area under curve of GNRI for predicting the delirium in the elderly patients with critical illness was 0.748(95%CI 0.710-0.783,P<0.001)with a sensitivity of 70.29% and a specificity of 84.58%,when the optimal cut-off point was determined as 96.63.One hundred and sixty-eight elderly patients with critical illness were successfully matched after PSM.After matching,the baseline data in the low GNRI group and the high GNRI group were equally comparable.The incidence of delirium in low GNRI group was higher than that in high GNRI group after PSM[63.69%(107/168)vs.17.86%(30/168),χ^(2)=73.071,P<0.001].Logistic regression analysis showed that the decrease of GNRI was an independent risk factor of concurrent delirium in the elderly patients with critical illness in the data before and after PSM(P<0.05).The verified ROC analysis results showed that the area under curve of GNRI was 0.753(95%CI 0.704-0.798,P<0.001),and the sensitivity and the specificity was 69.64% and 85.12% with 96.44 as the optimal cut-off value.Conclusions The decrease of GNRI on ICU admission is an independent risk factor of concurrent delirium in the elderly patients with critical illness,and GNRI as a readily available nutritional status assessment tool could be used as a predictor for concurrent delirium in the elderly patients with critical illness.