重症病毒性肺炎并发肺部侵袭性霉菌感染的危险因素及预测模型构建
该回顾性观察研究纳入新疆医科大学第一附属医院呼吸重症监护室(RICU)2022年8月至2024年10月收治的201例重症病毒性肺炎(SVP)患者,按7:3随机分为训练集(140例,其中非PIMI组97例、PIMI组43例)和验证集(61例)。多因素Logistic回归显示糖尿病、慢性肺病、实体器官肿瘤、广谱抗菌药物使用≥14天、胸膜心包积液是肺部侵袭性霉菌感染(PIMI)的独立危险因素(OR均>1,P均<0.05)。基于上述5项指标构建的预测模型训练集AUC为0.849(95%CI 0.785-0.914),验证集AUC为0.773(95%CI 0.658-0.889),校准曲线拟合良好,DCA提示临床适用价值良好。
要点
- SVP患者中,糖尿病、慢性肺病、实体器官肿瘤、广谱抗菌药物使用≥14天、胸膜心包积液为PIMI独立危险因素(OR均>1,P均<0.05)。
- 基于5项独立危险因素构建的预测模型训练集AUC 0.849(95%CI 0.785-0.914),验证集AUC 0.773(95%CI 0.658-0.889)。
- 训练集单因素分析中,PIMI组与非PIMI组在CURB-65评分、肺实变、胸腔及心包积液、入RICU时PaO2/FiO2、CRP>50 mg/L、合并肺部细菌感染、脓毒性休克、广谱抗菌药物使用≥14天等方面差异有统计学意义(P均<0.05)。
- 模型校准曲线在训练集与验证集均提示拟合良好,DCA曲线提示临床适用价值良好。
- 该模型为回顾性单中心研究构建,尚需进一步验证。
适用人群与场景
- 2022年8月至2024年10月就诊于新疆医科大学第一附属医院呼吸重症监护室(RICU)的重症病毒性肺炎(SVP)患者,共201例。
- 训练集140例(非PIMI组97例、PIMI组43例),验证集61例。
诊断标准与关键阈值
- 采用英国胸科协会改良肺炎评分(CURB-65)、急性生理学与慢性健康状况评分II(APACHE II)评估病情。
- PIMI分组依据是否发生肺部侵袭性霉菌感染。
研究设计与关键结果
- 研究类型:回顾性观察研究;样本量201例SVP患者;训练集140例(非PIMI组97例、PIMI组43例),验证集61例。
- 单因素分析(训练集):PIMI组与非PIMI组在糖尿病、慢性肺病、实体器官肿瘤、CURB-65评分、肺实变、胸腔及心包积液、入RICU时PaO2/FiO2、CRP>50 mg/L、合并肺部细菌感染、脓毒性休克、广谱抗菌药物使用≥14天方面差异均有统计学意义(P均<0.05)。
- 多因素Logistic回归:糖尿病、慢性肺病、实体器官肿瘤、广谱抗菌药物使用≥14天、胸膜心包积液为PIMI独立危险因素(OR均>1,P均<0.05)。
- 预测模型ROC曲线:训练集AUC 0.849(95%CI 0.785-0.914),验证集AUC 0.773(95%CI 0.658-0.889)。
- 校准曲线:训练集与验证集均提示模型拟合良好;DCA曲线提示模型临床适用价值良好。
证据类型、等级与局限
- 证据类型:回顾性观察研究。
- 局限:原文未报告局限性。
对床旁实践的改变
- 对SVP患者,若合并糖尿病、慢性肺病、实体器官肿瘤、广谱抗菌药物使用≥14天或胸膜心包积液,作者认为应关注PIMI发生风险。
- 作者认为基于上述5项因素构建的预测模型对PIMI发生具有良好预测价值和临床效用。
提炼说明
- 本条依据论文摘要整理;摘要未涉及「处置流程与药物剂量」「红旗征与不能漏的情况」「指南或共识出处」,故未列出这些栏目;全文请点「阅读原文」。
英文原文(摘要)· 点击展开
OBJECTIVE: To analyze the risk factors for pulmonary invasive mold infection (PIMI) in patients with severe viral pneumonia (SVP), construct a predictive model, and evaluate its efficacy.
METHODS: A retrospective observational study was conducted, and patients with SVP admitted to the respiratory intensive care unit (RICU) of the First Affiliated Hospital of Xinjiang Medical University from August 2022 to October 2024 were enrolled. Patients were divided into a training set and a validation set at a ratio of 7 : 3 using the random number table method. Then, SVP patients in the training set were divided into a non-PIMI group and a PIMI group based on whether PIMI occurred or not. The data of patients in each group were collected and compared, including gender, age, underlying diseases, and British Thoracic Society's modified pneumonia score (CURB-65), Acute Physiology And Chronic Health Evaluation II (APACHE II) score, chest CT findings, oxygenation index (PaO2/FiO2), and laboratory indicators at RICU admission. Variables with statistically significant differences in univariate analysis were included in multivariate Logistic regression analysis to identify the risk factors for PIMI and construct a risk prediction model. The effectiveness and application value of the prediction model were evaluated using the receiver operator characteristic curve (ROC curve), calibration curve, and decision curve analysis (DCA). The model was further validated in the validation set.
RESULTS: A total of 201 patients with SVP were enrolled. There were 140 patients in the training set (97 cases in the non-PIMI group and 43 cases in the PIMI group) and 61 patients in the validation set. Univariate analysis showed that in the training set, there were statistically significant differences between the PIMI group and the non-PIMI group in terms of diabetes mellitus, chronic lung disease, solid organ tumor, CURB-65 score, pulmonary consolidation, pleural and pericardial effusion, PaO2/FiO2 at RICU admission, C-reactive protein (CRP) >50 mg/L, concurrent pulmonary bacterial infection, septic shock, and duration of broad-spectrum antimicrobial use≥14 days (all P<0.05). Multivariate Logistic regression analysis revealed that diabetes mellitus, chronic lung disease, solid organ tumor, duration of broad-spectrum antimicrobial use≥14 days, and pleural and pericardial effusion were independent risk factors for PIMI in SVP patients [all odds ratio (OR) >1, all P<0.05]. A risk prediction model was constructed using the above five indicators. ROC curve showed that the area under the ROC curve (AUC) of the prediction model in the training set and validation set were 0.849 [95% confidence interval (95%CI) was 0.785-0.914] and 0.773 (95%CI was 0.658-0.889), respectively. The calibration curves of both the training set and the validation set indicated good fitting performance of the model, and the DCA curves suggested that the model had good clinical application value.
CONCLUSIONS: Diabetes mellitus, chronic lung disease, solid organ tumor, duration of broad-spectrum antimicrobial use≥14 days, and pleural and pericardial effusion are risk factors for PIMI in patients with SVP. The constructed predictive model based on the above factors has good predictive value and clinical utility for the occurrence of PIMI.
参考来源
- Wei C, Shen J, Wu X, et al. [Risk factors and construction of predictive model for severe viral pneumonia complicated with pulmonary invasive mold infection]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2026. PMID: 41833347. doi:10.3760/cma.j.cn121430-20250803-00420
供执业医师抢救与诊疗参考,不替代临床判断;剂量请结合体重、肝肾功能和本院药品规格复核,以指南原文为准。