去骨瓣减压术后患者颅内感染风险预测模型的构建及验证

Construction and validation of a predictive model for the risk of intracranial infection in patients after decompressive craniectomy

目的 分析去骨瓣减压术(decompressive craniectomy,DC)后患者颅内感染的风险因素,构建DC患者颅内感染的风险预测模型,并开展模型的内部及外部验证。方法 回顾性分析2020年1月—2021年9月某三级甲等医院神经外科发生颅内感染的61例DC患者纳入感染组,按照1:2比例选择同一科室、年龄相近未发生颅内感染的DC患者122例纳入非感染组进行病例对照研究,采用单因素分析和多因素logistic回归分析法,应用R软件建立列线图模型并进行内部验证。选取2019年1—12月接受DC患者61例临床资料对模型进行外部验证。结果 建模组患者的发生颅内感染率为33.3%(61/183)。Logistic回归分析显示:手术时间(OR=1.006,P=0.039)、血肿大小(OR=1.037,P=0.004)、ICU入住时间(OR=1.092,P=0.004)、脑出血(OR=16.128,P=0.019)、创伤性脑损伤(OR=18.100,P=0.028)、脑肿瘤(OR=44.286,P=0.003)是DC患者颅内感染的独立危险因素。据此建立列线图模型,模型的AUC为0.863(95%CI:0.760~0.967),外部验证结果显示模型的AUC为0.933(95%CI:0.862~1.000)。结论 DC患者的手术时间越长、血肿越大、ICU入住时间越长、有脑出血史、创伤性脑损伤史、脑肿瘤史发生颅内感染的风险越高。本研究构建的DC患者颅内感染预测模型具有良好的准确度,有一定的临床应用价值。

Objective To analyze the risk factors for intracranial infection in patients after decompressive craniectomy(DC),to construct a risk prediction model for intracranial infection in DC patients,and to conduct internal and external validation of the model.Methods The 61 DC patients who had intracranial infections in the neurosurgery department of a tertiary care hospital from January 2020 to September 2021 were retrospectively analyzed and included in the infection group,and 122 DC patients of the same department and similar age who did not have intracranial infections were selected according to the ratio of 1:2 to be included in the non-infection group for a case-control study,and the column line graph model was established and internally validated by applying R sofware using univariate analysis and multi-factor logistic regression analysis.Sixty-one clinical data of patients receiving DC from January to December 2019 were selected for external validation of the model.Results The incidence of intracranial infection in the modeled group of patients was 33.3%(61/183).Logistic regression analysis showed that surgery time(0R=1.006,P=0.039),hematoma size(0R=1.037,P=0.004),ICU stay(0R=1.092,P=0.004),cerebral hemorrhage(0R=16.128,P=0.019),traumatic brain injury(OR=18.100,P=0.028),and brain tumor(0R=44.286,P=0.003)were independent risk factors for intracranial infection in DC patients.The column line graph model was developed accordingly,and the model AUC was 0.863(95%CI:0.760-0.967),and the external validation results showed an AUC of 0.933(95%CI:0.862-1.000)for the model.Conclusion The longer the surgery time,the larger the hematoma,the longer the ICU stay,the history of cerebral hemorrhage,the history of traumatic brain injury,and the history of brain tumor,the higher the risk of intracranial infection in DC patients.The prediction model of intracranial infection in DC patients constructed in this study has good accuracy and has some clinical application value.