| 张少浩,朱勇冬,林麒.病毒性脑炎 217例症状性癫痫发生风险的预测研究[J].安徽医药,2024,28(7):1430-1434. |
| 病毒性脑炎 217例症状性癫痫发生风险的预测研究 |
| A predictive study on the risk of symptomatic epilepsy in 217 cases of viral encephalitis |
| |
| DOI：10.3969/j.issn.1009-6469.2024.07.035 |
| 中文关键词: 脑炎，病毒性 症状性癫痫 风险因素 列线图模型 临床获益率 |
| 英文关键词: Encephalitis,viral Symptomatic epilepsy Risk factors Line graph model Clinical benefit rate |
| 基金项目:汕头市科技计划医疗卫生项目( 191221115263119) |
|
| 摘要点击次数: 413 |
| 全文下载次数: 283 |
| 中文摘要: |
| 目的分析病毒性脑炎并发症状性癫痫的风险因素，据此构建列线图预测模型。方法回顾性分析 2018年 2月至 2022年 5月汕头市中心医院收治的 217例病毒性脑炎病人临床资料，抽取 70%为建模集(152例)，30%为验证集(65例)。根据病人是否合并症状性癫痫，将建模集进一步分为发生组和未发生组，比较两组病人一般资料，选择差异有统计学意义的指标用逐步向前回归法进行非条件多因素 logistic分析病毒性脑炎病人症状性癫痫发生的影响因素，并采用 R3.4.3软件包绘制基于多因素分析结果的列线图模型。采用 Bootstrap法分别对建模集和验证集进行验证，并绘制受试者操作特征曲线(ROC曲线)和决策曲线(DCA)以评估列线图模型的预测效能和临床净获益率。结果 217例病毒性脑炎病人中，共 46例病人合并症状性癫痫(21.20%)，其中建模集中有 32例合并症状性癫痫，验证集中有 14例合并症状性癫痫；发生组昏迷、大脑皮质损坏、脑电图重度异常、颅脑核磁共振成像(MRI)有责任病灶、累及颞叶或额叶、脑脊液单纯疱疹病毒(HSV)(+)占比及脑脊液压力均高于未发生组(P<0.05)； logistic多元回归分析，昏迷、大脑皮质损坏、脑电图重度异常、颅脑 MRI有责任病灶、累及颞叶或额叶、脑脊液压力、脑脊液 HSV(+)均是病毒性脑炎合并症状性癫痫的影响因素(P<0.05)；经 Bootsrap法进行验证，建模集其一致性指数( C-index)为 0.833，验证集的 Cindex则为 0.830，校正曲线和标准曲线拟合度较好。建模集 ROC曲线下面积(AUC)、灵敏度、特异度分别为 0.84［98%CI:(0.78，0.89)］、79.17%、84.04%，验证集则为 0.81［98%CI:(0.76，0.86)］，83.04%，73.64%，提示模型区分度良好。 DCA曲线显示病人根据列线图模型进行风险评估可获得满意的净收益。结论昏迷、大脑皮质损坏、脑电图重度异常、颅脑 MRI有责任病灶、累及颞叶或额叶、脑脊液压力、脑脊液 HSV(+)均是病毒性脑炎合并症状性癫痫的影响因素，综合上述因素针对病毒性脑炎病人构建的列线图预测模型可以较好地个体化预测症状性癫痫的发生，对临床防治症状性癫痫提供指导。 |
| 英文摘要: |
| Objective To explore the risk factors of viral encephalitis complicated with symptomatic epilepsy, and to build a nomogram prediction model based on this.Methods The clinical data of 217 patients with viral encephalitis admitted to Shantou CentralHospital from February 2018 to May 2022 were retrospectively analyzed, 70 % of whom were randomly selected as the modeling set (n= 152) and 30 % as the validation set (n=65). Based on whether the patients had symptomatic epilepsy, the modeling set was further assigned into occurrence group and non-occurrence group. The general information of the two groups of patients was compared, and indicators with statistical significance were selected for unconditional multivariate logistic analysis using stepforward regression method.The influencing factors of symptomatic epilepsy in patients with viral encephalitis were analyzed, and a column chart model based onthe results of multivariate analysis was drawn using R3.4.3 software package. The Bootstrap method was used to validate the modelingand validation sets, and receiver operating characteristic (ROC) and decision curve (DCA) were drawn to evaluate the predictive performance and clinical net benefit rate of the column chart model.Results Among 217 patients with viral encephalitis, a total of 46 patients had symptomatic epilepsy, with an incidence rate of 21.20%. Among them, 32 patients had symptomatic epilepsy in the modelingset and 14 patients had symptomatic epilepsy in the validation set. The proportions of coma, cerebral cortex damage, severe abnormalelectroencephalogram, responsible lesions of brain magnetic resonance imaging (MRI), temporal lobe or frontal lobe involvement, cerebrospinal fluid herpes simplex virus (HSV)(+) and cerebrospinal fluid pressure in the occurrence group were higher than those in thenon-occurrence group (P<0.05). Logistic multiple regression analysis results showed that coma, cerebral cortex damage, severe abnormal electroencephalogram, responsible lesions of brain MRI, temporal lobe or frontal lobe involvement, cerebrospinal fluid pressureand cerebrospinal fluid HSV(+) were the influencing factors of viral encephalitis combined with symptomatic epilepsy (P<0.05). After validation using the Bootrap method, the consistency index (C-index) of the modeling set was 0.833, while the C-index of the validation set was 0.830. The calibration curve and the standard curve fit well. The area under the ROC curve (AUC), sensitivity, and specificity of the modeling set were 0.84 [98%CI:(0.78,0.89)], 79.17%, and 84.04%, respectively, while those of the validation set were 0.81 [98%CI: (0.76, 0.86)], 83.04%, and 73.64%, indicating good model discrimination. The DCA curve showed that patients could achieve satisfactory net benefits through risk assessment based on the column chart model.Conclusions Coma, cerebral cortex damage, severe abnormal electroencephalogram, responsible lesions of brain MRI, temporal lobe or frontal lobe involvement, cerebrospinal fluid pressureand cerebrospinal fluid HSV(+) are the influencing factors of viral encephalitis combined with symptomatic epilepsy. The nomogramprediction model based on the above factors can better predict the occurrence of viral encephalitis combined with symptomatic epilepsy, which can provide guidance for clinical prevention and treatment of symptomatic epilepsy. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
