探讨多个评分系统对ICU脓毒症患者30天预后预测价值——一项基于MIMIC-IV数据库的回顾性分析

Exploring the Predictive Value of Multiple Scoring Systems for 30-Day Prognosis in ICU Patients with Sepsis—A Retrospective Study Based on MIMIC-IV Database

摘要: 目的：评价急性生理学评分III (APS III)、Logistic器官功能障碍系统评分(LODS)、牛津急性疾病严重程度评分(OASIS)、简化急性生理学评分II (SAPS II)、全身炎症反应综合征评分(SIRS)、序贯器官衰竭评分(SOFA)及部分评分联合对ICU脓毒症患者30天内死亡风险的预警效能。方法：以美国贝斯以色列迪康医学中心重症监护室数据库(MIMIC-IV)1符合脓毒症3.0诊断标准的脓毒症患者为基础，纳入符合入选标准的研究对象。以患者多次入院情况判断存活状态，分别采用受试者工作特征(ROC)曲线分析6种评分系统对ICU脓毒症患者30天死亡风险的预警效能；采用决策曲线分析(DCA)比较各评分系统的临床应用价值。最后，联合价值较大的评分系统，再次进行ROC曲线及DCA分析。结果：最终纳入7082例符合标准的脓毒症患者进行分析，年龄为63 (52, 73)岁，男性3904例(55.1%)，入ICU后1517例(21.4%) 30天内死亡。各评分系统的AUC值：APS III (AUROC 0.775，95% 置信区间(CI) 0.765~0.785)和LODS (AUROC 0.766, 95% CI 0.756~0.776)对脓毒症患者30天死亡率的预测价值优于OASIS (AUROC 0.736, 95% CI 0.726~0.746)、SAPS II (AUROC 0.721, 95% CI 0.710~0.731)、SIRS (AUROC 0.578, 95% CI 0.566~0.590)、SOFA (AUROC 0.722, 95% CI 0.711~0.732) (P均 < 0.05)。但两评分之间AUC面积无统计学意义(P > 0.05)，而APS III与LODS联合的预测价值高于二者单独时的价值(P均 < 0.05)。结论：预测ICU脓毒症患者30天死亡时，APS III与LODS优于OASIS等其余评分，但二者联合的预测效果更佳且临床应用价值更大。

Abstract: Aim: To evaluate the early warning efficacy of acute physiology score III (APS III), logistic organ dysfunction score (LODS), Oxford acute severity of illness score (OASIS), simplified acute physiology score II (SAPS II), systemic inflammatory response syndrome score (SIRS), sepsis-related organ failure assessment (SOFA) and the combined of partial scores for 30-day mortality risk in ICU sepsis patients. Method: Based on the sepsis patients who met the diagnostic criteria of sepsis 3.0 in the intensive care unit (ICU) database of Beth Israel Deakon Medical Center (MIMIC-IV) in the United States, subjects who met the inclusion criteria were included. The survival status was judged based on the patients’ multiple admissions, and the receiver operating characteristic (ROC) curve was used to analyze the early warning efficacy of the six scoring systems on the 30-day mortality risk of ICU sepsis patients; decision curve analysis (DCA) was used to compare the scoring systems of clini-cal application value. Combined with more valuable scoring systems, the ROC curve and DCA analy-sis were performed again. Results: Finally, there were 7082 patients with sepsis who met the crite-ria were included for analysis, with a median age of 63 [interquartile range (IQR) 52~73] years, among them, 3904 (55.1%) were males and 1517 (21.4%) patients died within 30 days. AUC values of all scoring systems were as follows: the APS III [0.775, 95% (CI) 0.765~0.785] and the LODS [0.766, 95% CI 0.756~0.776)] were better than OASIS [0.736, 95% CI 0.726~0.746)], SAPS II [(0.721, 95% CI 0.710~0.731)], SIRS [(0.578, 95% CI 0.566~0.590)] and SOFA [(0.722, 95% CI 0.711~0.732)] (with all P < 0.05) in predicting the 30-day mortality of sepsis patients. However, there was no significant difference in AUC area between the two scores (P > 0.05), while the predic-tive value of APS III combined with LODs was higher than that of both alone (P < 0.05). Conclusions: The predictive value of APS III and LODS scores are superior than that of SAPS II, SIRS, OASIS and SOFA score in predicting the 30-day mortality of ICU sepsis patients, the combination of the two was of best prediction effect and greater clinical application value.