检索历史 清除

目的 分析影响重型颅脑损伤患者持续意识障碍的危险因素,建立风险预测模型,并评价其预测能力.方法 回顾性研究山西医科大学第一附属医院急诊科2011年7月至2013年11月收治的重型颅脑损伤患者,符合纳入/排除标准165例.按照收治顺序分组,推导组(115例),验证组(50例).纳入标准:(1)年龄＞15岁;(2)有明确颅脑外伤史;(3)行头颅CT或MRI检查确诊颅脑损伤;(4)格拉斯哥评分(glasgow coma scale,GCS)≤8分;(5)入院当日呈昏迷或意识障碍逐渐加重至昏迷状态.排除标准:(1)外伤后仅有短暂意识丧失或昏迷的患者(昏迷时间＜6h);(2)外伤后癔病或精神抑郁状态致昏迷样表现;(3)外伤后情绪诱发癫痫持续状态而呈意识障碍.对28个可能影响因素进行单因素筛选及logistic多重回归分析并建立风险模型.运用Hosmer-Lemeshow检验和受试者工作特征曲线(receiver operating characteristic,ROC)对模型进行拟合优度检验及判别.结果 入院时GCS、神经系统并发症、弥漫性轴索损伤(diffuse axonal injury,DAI)、电解质紊乱是影响颅脑损伤患者持续昏迷的独立风险因素.推导组Hosmer-Lemeshow检验显示x2 =4.380,P=0.496,预测昏迷率与实际昏迷率差异无统计学意义,风险模型的辨识度在推导组(AUC=0.87; 95％CI:0.798～0.942)及验证组(AUC=0.90;95％ CI:0.803 ～0.997)中均较好.结论 预测模型虽有一定局限性,但仍能对颅脑损伤患者持续意识障碍进行较准确的估计.

Objective To investigate the risk factors related to persistent unconsciousness in patients with severe traumatic brain injury (sTBI) by way of building a prognosis model.Methods The clinical data of 165 sTBI patients admitted from July 2011 to November 2013 were retrospectively analyzed.The eligible patients were randomly assigned to derivation cohort (n =115) and verification cohort (n =50) by treatment order.Inclusion criteria:(1) age ＞ 15 years; (2) definitive history of head injury; (3) traumatic brain injury confirmed by head computerized tomography or brain MRI; (4) initial Glasgow coma score (GCS) was less than 8; (5) patient's light come or consciousness impairment gradually deteriorating to profound coma on the day of admission.The exclusion criteria were as follows:(1) only a brief loss of consciousness or coma after trauma (coma time ＜ 6 hours) ; (2) post-injury hysteria or dementia causes appearance like coma ; (3) unconsciousness results of status epilepticus which was induced by emotion after injury.Univariate and muhivariable logistic regression were employed to determine the independent predictors of persistent unconsciousness in the derivation cohort,and then prognosis model was established.The Hosmer-Lemeshow goodness-of-fit test and the area under the receiver operating characteristic curve were used to assess the capacity of model for discrimination and calibration.Results Logistic regression analysis was used to identify GCS score,neurological complications,diffuse axonal injury (DAI),electrolytes disturbance as the most important predictors of persistent unconsciousness.The model was well calibrated in the derivation cohort (Hosmer-Lemeshow test,x2 =4.380,P =0.496).The model showed good discrimination (area under the receiver operating characteristic curve) in the derivation cohort (0.87; 95％ CI:0.798-0.942) and in the versification cohort (0.90; 95％ CI:0.803-0.997).Conclusions The prognosis model could accurately predict the persistent unconsciousness lasting in sTBI patients despite its certain limitations,and therefore,it has significantly clinical and societal value.