颅脑外伤患者感染发生情况及影响因素的Logistic回归分析

Investigation on the incidence of infection in patients with craniocerebral trauma and Logistic regression analysis of related influencing factors

目的 探讨颅脑外伤患者感染发生情况及影响因素,为临床防治颅脑外伤并发感染提供参考。方法 选取2018年2月—2021年3月安阳地区医院收治的189例颅脑外伤患者,统计感染发生率及感染部位分布情况,并对感染发生情况进行单因素和多因素Logistic回归分析,采用受试者工作特征(ROC)曲线及曲线下面积(AUC)分析对Logistic回归预测模型进行验证。结果 189例颅脑外伤患者,并发感染47例,感染发生率为24.87%;感染部位以肺部最为常见,其次是颅内感染、切口感染。单因素分析显示,年龄、疾病类型、入院时GCS评分、糖尿病、气管切开、脑室外引流、慢性肺部疾病、皮下或硬膜外积液、手术次数、后颅窝手术、术后切口皮下积液、术后低蛋白血症、术后放置引流管、合并肢体/躯干损伤与感染发生有关(P<0.05)。Logistic回归分析显示,开放性损伤患者感染风险是闭合性损伤患者的3.251倍,入院时GCS评分≥8分患者感染风险是<8分的0.754倍,糖尿病患者感染风险是无糖尿病者的1.370倍,气管切开患者感染风险是未切开患者的1.961倍,脑室外引流患者感染风险是未引流者的2.489倍,慢性肺部疾病患者感染风险是无慢性肺部疾病患者的2.784倍,皮下或硬膜外积液患者感染风险是无积液者的1.438倍,手术次数≥2次患者感染风险是1次者的1.768倍,后颅窝手术患者感染风险是非后颅窝手术者的1.830倍,术后切口皮下有积液者感染风险是无积液者的2.541倍,术后低蛋白血症者感染风险是无低蛋白血症者的3.407倍,合并肢体/躯干损伤者感染风险是未合并者的1.972倍(P<0.05)。ROC曲线分析显示,Logistic回归模型预测颅脑外伤并发感染的AUC为0.886,95%CI为0.855~0.918,敏感度为77.59%,特异度为82.61%。结论 颅脑外伤患者感染率较高,以肺部感染、颅内感染最为常见,疾病类型、糖尿病、气管切开、脑室外引流、慢性肺部疾病、皮下或硬膜外积液、手术次数、后颅窝手术、术后切口皮下积液、术后低蛋白血症、合并肢体/躯干损伤是颅脑外伤感染的相关独立危险因素,入院时GCS评分是颅脑外伤的相关独立保护因素,根据上述因素构建Logistic回归模型预测价值较高,能为临床干预提供客观、可靠的依据。

Objective To explore the investigation of the incidence of infection in patients with craniocerebral trauma and related influencing factors,so as to provide reference for clinical prevention and treatment of craniocerebral trauma complicated infection.Methods A total of 189 patients with craniocerebral trauma admitted to Anyang District Hospital from February 2018 to March 2021 were selected,the incidence of infection and the distribution of infection sites were counted,and the incidence of infection was analyzed by univariate and multivariate Logistic regression.The Logistic regression prediction model was verified by the analysis of the operator's operating characteristic(ROC)curve and the area under the ROC(AUC).Results Of the 189 patients with traumatic brain injury,47 were complicated by infection.The infection rate was 24.87%.The infection site was the most common in the lungs,followed by intracranial infection and incision infection.Univariate analysis showed that age,disease type,and GCS at admission Score,diabetes,tracheotomy,ventricular drainage,chronic lung disease,subcutaneous or epidural fluid,number of operations,posterior fossa surgery,postoperative incision subcutaneous fluid,postoperative hypoproteinemia,postoperative placement Drainage tube and combined limb/trunk injury were related to the occurrence of infection(P<0.05).Logistic regression analysis showed that the risk of infection in patients with open injury was 3.251 times that of patients with closed injury,patients with a GCS score of≥8 at admission were 0.754 times that of patients with a GCS score of<8,and the risk of infection in patients with diabetes was 1.370 times that of those without diabetes.The risk of infection in patients with tracheotomy was 1.961 times that of patients without incision,the risk of infection in patients with ventricular drainage was 2.489 times that of patients without drainage,and the risk of infection in patients with chronic lung disease was 2.784 times that of patients without chronic lung disease.The risk of infection in patients with subcutaneous or hard extramembranous effusion was 1.438 times that of those without effusion.The risk of infection in patients with≥2 operations was 1.768 times that of patients with one operation.The risk of infection in patients with posterior fossa surgery was 1.830 times that of non-posterior fossa surgery.The risk of infection in patients with effusion under the skin of the posterior incision was 2.541 times that of those without effusion,the risk of infection in patients with postoperative hypoproteinemia was 3.407 times that of those without hypoproteinemia,and the risk of infection in patients with limb/trunk injury was uncombined.The ROC curve analysis showed that the logistic regression model predicted that the AUC of craniocerebral trauma complicated by infection was 0.886,the 95%CI was 0.855 to 0.918,the sensitivity was 77.59%,and the specificity was 82.61%.Conclusion The infection rate of patients with craniocerebral trauma is relatively high.Pulmonary infection and intracranial infection are the most common.Type of disease,diabetes,tracheotomy,ventricular drainage,chronic lung disease,subcutaneous or epidural effusion,number of operations,Posterior fossa surgery,postoperative incision subcutaneous fluid,postoperative hypoproteinemia,combined limb/trunk injury are related independent risk factors for craniocerebral trauma infection,GCS score at admission is an independent protective factor for craniocerebral trauma.The logistic regression model constructed based on the above factors has high predictive value and can provide an objective and reliable basis for clinical intervention.