作者：徐佳佳1 ，左绍敏1 ，董泽钦1 ，薛红飞2 ，刘慧勤2 ，庞瑞2 ，李玮1

单位：

1.450003河南省郑州市，河南大学人民医院神经内科　2.450003河南省郑州市，郑州大学人民医院神经内科

Units：

1.Department of Neurology, People's Hospital of Henan University, Zhengzhou 450003, China 2.Department of Neurology, People's Hospital of Zhengzhou University, Zhengzhou 450003, China

关键词：

脑炎，病毒性；预后；影响因素分析；列线图

Keywords：

Encephalitis, viral; Prognosis; Root cause analysis; Nomograms

CLC：

R 512.3

DOI：

10.12114/j.issn.1008-5971.2024.00.103

Funds：

河南省医学科技攻关计划项目（SBGJ2018077）

摘要：

目的　分析病毒性脑炎（VE）患者预后不良的影响因素，并构建其风险预测列线图模型。方法　回 顾性选取2018年1月—2022年6月河南大学人民医院收治的VE患者218例为研究对象。通过医院临床电子病历系统及医 院信息系统收集患者的临床资料及出院后1年预后情况。将格拉斯哥预后量表（GOS）1～3级定义为预后不良，GOS 4～5级定义为预后良好，根据预后情况将患者分为预后良好组（150例）和预后不良组（68例）。采用LASSO回归、 多因素Logistic回归分析探讨VE患者预后不良的影响因素；基于多因素Logistic回归分析结果，使用R软件的glmpath 包构建VE患者预后不良的风险预测列线图模型；采用ROC曲线评估该列线图模型的区分度，采用校准曲线分析该列 线图模型的准确性，采用决策曲线分析该列线图模型的临床适用性。结果　预后不良组年龄大于预后良好组，有合 并症者占比、伴有意识障碍者占比、伴有癫痫者占比、使用糖皮质激素者占比、直接胆红素、肌酸激酶、乳酸脱氢 酶、高密度脂蛋白胆固醇、中性粒细胞计数、中性粒细胞/淋巴细胞比值（NLR）、血小板/淋巴细胞比值（PLR）、 全身性免疫炎症指数（SII）、系统性炎症反应指数（SIRI）、颅脑磁共振检查结果异常者占比高于预后良好组，白 蛋白、低密度脂蛋白胆固醇、淋巴细胞计数、淋巴细胞/单核细胞比值（LMR）低于预后良好组（P＜0.05）。LASSO 回归分析结果显示，年龄、伴有癫痫、淋巴细胞计数、LMR、NLR、颅脑磁共振检查结果可能是VE患者预后不良的 影响因素（P＜0.05）。多因素Logistic回归分析结果显示，年龄、LMR、NLR和颅脑磁共振检查结果是VE患者预后不 良的独立影响因素（P＜0.05）。基于多因素Logistic回归分析结果，构建VE患者预后不良的风险预测列线图模型。从 218例患者中随机选取66例作为验证集。ROC曲线分析结果显示，该列线图模型预测验证集VE患者预后不良的AUC为 0.871〔95%CI（0.776～0.966）〕。校准曲线分析结果显示，该列线图模型预测验证集VE患者预后不良的发生率与预 后不良的实际发生率一致。决策曲线分析结果显示，当阈值概率＜80%时，该列线图模型在验证集中的临床净获益 率＞0。结论　年龄、LMR、NLR和颅脑磁共振检查结果是VE患者预后不良的独立影响因素，基于上述影响因素构建 的VE患者预后不良的风险预测列线图模型具有较好的区分度、准确性及临床适用性。

Abstract：

　Objective To analyze the influencing factors of poor prognosis in patients with viral encephalitis (VE) , and construct the nomogram model for risk prediction of it. Methods A total of 218 VE patients admitted to People's Hospital of Henan University from January 2018 to June 2022 were retrospectively selected as the study objects. Clinical data and 1-year prognosis after discharge of patients were collected through hospital electronic medical record system and hospital information system. Glasgow Prognosis Scale (GOS) grades 1-3 were defined as poor prognosis, GOS grades 4-5 were defined as good prognosis, and patients were divided into good prognosis group (150 cases) and poor prognosis group (68 cases) according to the prognosis. LASSO regression and multivariate Logistic regression analysis were used to explore the influencing factors of poor prognosis in patients with VE. Based on the results of multivariate Logistic regression analysis, glmpath package of R software was used to construct the nomogram model for risk prediction of poor prognosis in patients with VE. ROC curve was used to evaluate the differentiation of the nomogram model, calibration curve was used to analyze the accuracy of the nomogram model, and decision curve was used to analyze the clinical applicability of the nomogram model. Results The age of the poor prognosis group was older than that of the good prognosis group, the proportion of patients with complications, the proportion of patients with consciousness disorder, the proportion of patients with epilepsy, the proportion of patients using glucocorticoids, direct bilirubin, creatine kinase, lactate dehydroase, high density lipoprotein cholesterol, neutrophil count, neutrophil-to-lymphocyte ratio (NLR) , platelet-to-lymphocyte ratio (PLR) , systemic immune-inflammation index (SII) , systemic inflammation response index (SIRI) and the proportion of patients with abnormal brain MRI results were higher than those in the good prognosis group, and albumin, low density lipoprotein cholesterol, lymphocyte count and lymphocyte-to-monocyte ratio (LMR) were lower than those in the good prognosis group (P < 0.05) . LASSO regression analysis showed that age, epilepsy, lymphocyte count, LMR, NLR and brain magnetic resonance examination results may be the influencing factors of poor prognosis in patients with VE (P < 0.05) . Multivariate Logistic regression analysis showed that age, LMR, NLR and brain magnetic resonance examination results were independent influencing factors of poor prognosis in patients with VE (P < 0.05) . Based on the results of multivariate Logistic regression analysis, the nomogram model for risk prediction of poor prognosis in patients with VE was constructed. A total of 66 patients were randomly selected from 218 patients and were as the validation set. ROC curve analysis showed that the AUC of the nomogram model in predicting poor prognosis in patients with VE in the validation set was 0.871 [95%CI (0.776-0.966) ] . Calibration curve analysis showed that the incidence of poor prognosis predicted by the nomogram model was consistent with the actual incidence of poor prognosis in patients with VE in the validation set. The results of decision curve analysis showed that when the threshold probability was < 80 %, the clinical net benefit rate of the nomogram model in the validation set was > Conclusion Age, LMR, NLR and brain magnetic resonance examination results are independent influencing factors of poor prognosis in patients with VE, and the nomogram model for risk prediction of poor prognosis in patients with VE constructed based on the above influencing factors has good differentiation, accuracy and clinical applicability.

ReferenceList：