热性惊厥患儿继发癫痫发作的危险因素及构建列线图预测模型

Risk factors of secondary epileptic seizures in children with febrile seizures and construction of nomogram prediction model

目的分析热性惊厥患儿继发癫痫发作的危险因素及构建列线图预测模型。方法采用回顾性研究的方法,选取恩施州民族医院2018年8月至2021年9月收治的热性惊厥患儿235例,根据随访6个月患儿是否继发癫痫发作分为发作组(62例)和未发作组(173例)。所有患者受试者工作特征曲线分析获取各因素的最佳截断值,多因素Cox回归分析热性惊厥患儿继发癫痫发作的独立风险因素,R软件"rms"包构建预测热性惊厥患儿继发癫痫发作高风险的列线图模型,校正曲线对列线图模型进行内部验证,决策曲线评估列线图模型的预测效能。结果发作组患者年龄低于未发作组[(14.45±1.54)个月比(21.47±2.18)个月],有癫痫家族史比例高于未发作组[56.45%(35/62)比35.84%(62/173)],围生期(异常)比例高于未发作组[59.68%(37/62)比15.61%(27/173)],惊厥发作类型(全面性)比例高于未发作组[70.97%(44/62)比36.99%(64/173)],脑电图情况(异常)比例高于未发作组[74.19%(46/62)比20.81%(36/173)],惊厥次数高于未发作组[(5.45±2.32)次比(2.04±1.02)次],惊厥持续时间高于未发作组[(18.89±4.29)min比(12.62±2.34)min]及肿瘤坏死因子-α(TNF-α)高于未发作组[(25.65±5.32)ng/L比(18.21±2.29)ng/L],差异有统计学意义(P均<0.05);年龄、惊厥次数、惊厥持续时间、TNF-α的曲线下面积分别为0.906、0.913、0.899、0.890;最佳截断值分别是3岁、4次、15 min、21 ng/L;年龄(≤3岁)、癫痫家族史(有)、惊厥发作类型(全面性)、围生期(异常)、惊厥次数(≥4次)、惊厥持续时间(≥15 min)是热性惊厥患儿继发癫痫发作的独立危险因素(P<0.05);此列线图预测模型的C-index为0.744(0.567~0.932);决策曲线显示当风险阈值>0.11时,此预测模型所提供的临床净收益均高于单个独立危险因素并且在预测热性惊厥患儿继发癫痫发作的高风险方面可以提供显著额外的临床净收益。结论基于年龄、癫痫家族史、惊厥发作类型、围生期、惊厥次数、惊厥持续时间构建热性惊厥患儿继发癫痫发作风险的列线图模型,为癫痫发作的防治提供重要策略指导。

Objective To analyze the risk factors of secondary epileptic seizures in children with febrile seizures and to construct a nomogram prediction model.Methods A total of 235 children with febrile seizures who were admitted to Enshi State Hospital for Nationalities from August 2018 to September 2021 were selected.According to whether the children had secondary epileptic seizures during the 6-month follow-up,the children were divided into the seizure group(62 cases)and no-seizure group(173 cases).The best cut-off value of each factor were obtained by the receiver operating characteristic(ROC).Multivariate Cox regression analysis was used to analyze the independent risk factors of secondary seizures in children with febrile seizures.The R software"rms"package was constructed to predict secondary seizures in children with febrile seizures.High-risk nomogram models,calibration curves was used for internal validation of nomogram models,and decision curves to assess the predictive power of nomogram models.Results The age of the patients in the seizure group was lower than that in the no-seizure group:(14.45±1.54)months vs.(21.47±2.18)months;and the proportion of family history of epilepsy,the proportion of perinatal(abnormal),the proportion of seizure type(comprehensive),the proportion of electroencephalogram(EEG)(abnormal),the number of seizures,the duration of seizure,the tumor necrosis factor-alpha(TNF-α)level in the seizure group were higher than those in the no-seizure group:56.45%(35/62)vs.35.84%(62/173),59.68%(37/62)vs.15.61%(27/173),70.97%(44/62)vs.36.99%(64/173),74.19%(46/62)vs.20.81%(36/173),(5.45±2.32)times vs.(2.04±1.02)times,(18.89±4.29)min vs.(12.62±2.34)min,(25.65±5.32)ng/L vs.(18.21±2.29)ng/L,there were statistical differences(P<0.05).The area under the curve(ACU)of age,number of convulsions,duration of convulsion,and TNF-αwere 0.906,0.913,0.899,and 0.890,respectively;the best cut-off values were 3 years,4 times,15 min,21 ng/L;age(≤3 years),family history of epilepsy(yes),type of seizures(generalized),perinatal period(abnormal),number of seizures(≥4 times),duration of seizures(≥15 min)were febrile seizures independent risk factors for secondary epileptic seizures in children(P<0.05),the C-index of this nomogram prediction model was 0.744(0.567-0.932);the decision curve showed that when the risk threshold was greater than 0.11,the clinical net benefit provided by this prediction model.The benefits were all higher than individual independent risk factors and provided a significant additional net clinical benefit in predicting a high risk of seizures secondary to febrile seizures in children with febrile seizures.Conclusions This study constructed a nomogram model of the risk of secondary seizures in children with febrile seizures based on age,family history of epilepsy,type of seizures,perinatal period,number of seizures,and duration of seizures.Important strategic guidance.