病毒性脑炎患儿发生神经系统后遗症的危险因素及其Logistic回归预测模型的构建

Risk Factors of Neurological Sequelae in Children with Viral Encephalitis and Construction of Logistic Regression Detective Model

目的探讨病毒性脑炎患儿发生神经系统后遗症的危险因素及其联合检测因子模型构建和预测价值。方法选取2016年9月至2019年9月安徽医科大学附属六安医院收治的病毒性脑炎患儿113例,均随访6个月,根据是否发生神经后遗症分为神经后遗症组和非神经后遗症组。收集两组一般临床资料,采用Logistic回归分析构建病毒性脑炎患儿发生神经系统后遗症的联合检测因子模型,并绘制受试者工作特征曲线(ROC曲线)。结果随访6个月,其中成功随访102例,随访率为90.27%(102/113),102例患儿中39例发生神经系统后遗症。神经后遗症组意识障碍、合并脏器损伤、惊厥持续状态、头颅磁共振成像(MRI)分级、入住重症监护病房、应激性高血糖比例均高于非神经后遗症组(P<0.05);经Logistic回归分析结果显示,惊厥持续状态(OR=2.166,95%CI 1.306~3.592)、头颅MRI分级(OR=4.632,95%CI 1.127~19.033)及应激性高血糖(OR=7.862,95%CI 4.045~15.278)是影响病毒性脑炎患儿发生神经系统后遗症的独立危险因素,得出联合检测因子模型表达式为:Logit(P)=-4.617+0.773 X1(惊厥持续状态)+1.533X2(头颅MRI分级)+2.062 X3(应激性高血糖);联合检测因子模型最佳临界值为0.552,灵敏度为97.20%,特异度为87.50%,曲线下面积为0.810(0.638~0.983)。结论基于Logistic回归模型构建联合检测因子能较好地预测病毒性脑炎患儿发生神经系统后遗症。

Objective To explore the risk factors of neurological sequelae in children with viral encephalitis and the construction and predictive value of model of the combined detective factors.Methods A total of 113 children with viral encephalitis admitted to Lu′an Hospital Affiliated to Anhui Medical University from Sep.2016 to Sep.2019 were included in the study.All the patients were followed up for 6 months,and were divided into a neurological sequelae group and a non-neurological sequelae group according to the occurrence of neurological sequela.General clinical data of the two groups were collected.Logistic regression was used to construct a combined detective factor model of neurological sequelae in children with viral encephalitis,and the receiver operator characteristic(ROC)curve was drawn.Results Followed up for 6 months,102 cases were successfully followed up,the follow-up rate was 90.27%(102/113),and 39 cases had neurological sequelae.The rate of consciousness disorder,combined organ injury,convulsion status,brain magnetic resonance imaging(MRI)classification,admission to ICU and stress hyperglycemia in the neurological sequelae group were higher than those in the non-neurological sequelae group(P<0.05).Logistic regression analysis showed that the status of convulsion(OR=2.166,95%CI 1.306-3.592),MRI grade of head(OR=4.632,95%CI 1.127-19.033)and stress hyperglycemia(OR=7.862,95%CI 4.045-15.278)were independent risk factors for neurological sequelae in children with viral encephalitis.The expression of joint detective factor model was:Logit(P)=-4.617+0.773 X1(status of convulsion)+1.533 X2(MRI grade of head)+2.062 X3(stress hyperglycemia).The best critical value of the combined detective factor model was 0.552,the sensitivity was 97.20%,the specificity was 87.50%,and the area under the curve was 0.810(0.638-0.983).Conclusion The neurological system sequelae of children with viral encephalitis can be well predicted by constructing the combined detective factors based on Logistic regression model.