作者：孙夕峰，唐勇，樊永忠，严朝俊

单位：

212300江苏省丹阳市人民医院神经外科 通信作者：孙夕峰，E-mail：sunrmyy@163.com

Units：

Department of Neurosurgery, People's Hospital of Danyang, Danyang 212300, China Corresponding author: SUN Xifeng, E-mail: sunrmyy@163.com

关键词：

颅脑外伤; 外伤后急性弥漫性脑肿胀; 影响因素分析; 列线图模型;

Keywords：

Traumatic brain injury; Post-traumatic acute diffuse brain swelling; Root cause analysis; Nomogram model

CLC：

DOI：

10.12114/j.issn.1008-5971.2022.00.102

Funds：

摘要：

目的 分析颅脑外伤(TBI)患者发生外伤后急性弥漫性脑肿胀(PADBS)的危险因素,并构建其列线图预测模型。方法 选取2019年4月至2021年10月丹阳市人民医院收治的TBI患者245例,根据TBI后是否发生PADBS将其分成PADBS组(n=69)与无PADBS组(n=176)。收集所有患者的临床资料,采用多因素Logistic回归分析探讨TBI患者发生PADBS的影响因素,并将危险因素引入R 3.6.3软件及rms程序包,以构建TBI患者发生PADBS的列线图预测模型;绘制ROC曲线以评估该列线图预测模型对TBI患者发生PADBS的区分度;绘制校准曲线及进行Hosmer-Lemeshow拟合优度检验以评估该列线图预测模型预测TBI患者发生PADBS的准确性。结果 PADBS组和无PADBS组患者年龄、TBI至治疗时间、误吸发生率、低血压发生率、颅内CT血肿厚度、脑疝发生率、多发伤发生率、原发性脑干损伤发生率及格拉斯哥昏迷量表(GCS)评分比较,差异有统计学意义(P<0.05)。多因素Logistic回归分析结果显示,TBI至治疗时间>3 h〔OR=8.213,95%CI(3.615,18.659)〕、误吸〔OR=8.019,95%CI(2.818,22.818)〕、原发性脑干损伤〔OR=27.736,95%CI(7.571,101.616)〕、GCS评分≤8分〔OR=8.677,95%CI(3.544,21.242)〕是TBI患者发生PADBS的危险因素(P<0.05)。以上述危险因素为指标构建TBI患者发生PADBS的列线图预测模型。ROC曲线分析结果显示,该列线图预测模型预测TBI患者发生PADBS的曲线下面积为0.831〔95%CI(0.765,0.897)〕;该列线图预测模型预测TBI患者发生PADBS的校准曲线与实际曲线基本吻合;Hosmer-Lemeshow拟合优度检验结果显示,该列线图预测模型的拟合效果良好(χ2=6.272,P=0.180)。结论 TBI至治疗时间>3 h、误吸、原发性脑干损伤、GCS评分≤8分是TBI患者发生PADBS的危险因素,基于上述危险因素构建的TBI患者发生PADBS的列线图预测模型具有良好的区分度及准确性。

Abstract：

【Abstract】　Objective To analyze the risk factors for post-traumatic acute diffuse brain swelling (PADBS) aftertraumatic brain injury (TBI) , and construct its nomogram prediction model. Methods A total of 245 TBI patients admittedto People's Hospital of Danyang from April 2019 to October 2021 were selected and divided into PADBS group (n=69) andnon-PADBS group (n=176) according to whether PADBS appeared after TBI. The clinical data of all patients were collected,the influencing factors of PADBS in TBI patients were analyzed by multivariate Logistic regression, and the risk factors wereintroduced into R 3.6.3 software and rms package to construct the nomogram prediction model of PADBS in TBI patients; the ROCcurve was drawn to evaluate the discrimination of the nomogram prediction model in predicting PADBS in patients with TBI; thecalibration curve was drawn and Hosmer-Lemeshow goodness of fit test was performed to evaluate the accuracy of the nomogramprediction model in predicting PADBS in patients with TBI. Results There were statistically significant differences in age, TBIto treatment time, incidence of aspiration, incidence of hypotension, hematoma thickness of intracranial CT, incidence of cerebralhernia, incidence of multiple injuries, incidence of primary brain stem injury and Glasgow Coma Scale (GCS) score betweenPADBS group and non-PADBS group (P <0.05) . Multivariate Logistic regression analysis showed that the TBI to treatment time> 3 h [OR=8.213, 95%CI (3.615, 18.659) ] , aspiration [OR=8.019, 95%CI (2.818, 22.818) ] , primary brain stem injury[OR=27.736, 95%CI (7.571, 101.616) ] , GCS score ≤ 8 [OR =8.677, 95%CI (3.544, 21.242) ] were the risk factors for PADBS inpatients with TBI (P <0.05) . The nomogram prediction model of PADBS in patients with TBI was constructed based on the aboverisk factors. The ROC curve analysis showed that the area under the curve of the nomogram prediction model in predicting PADBSin patients with TBI was 0.831 [95% CI (0.765, 0.897) ] ; the calibration curve of the nomogram prediction model for predictingPADBS in patients with TBI was basically consistent with the actual curve, and the Hosmer-Lemeshow goodness of fit test showedthat the fitting effect of the nomogram prediction model was good (χ2 =6.272, P=0.180) . Conclusion TBI to treatment time > 3 h,aspiration, primary brain stem injury and GCS score ≤ 8 are the risk factors of PADBS in TBI patients. The nomogram predictionmodel of PADBS in TBI patients based on the above risk factors has good discrimination and accuracy.

ReferenceList：