高血压性脑出血并发神经源性肺水肿的风险预测模型构建与验证

Construction and validation of a risk prediction model for hypertensive cerebral hemorrhage complicated with neurogenic pulmonary edema

目的 探讨高血压性脑出血并发神经源性肺水肿的危险因素,并以此构建风险预测模型。方法 回顾性分析2019年2月—2022年10月本院收治的412例高血压性脑出血患者的临床资料,将研究对象按照7︰3比例分为模型组(n=288)和验证组(n=124)。运用单因素及多因素Logistic回归分析筛选出并发神经源性肺水肿的危险因素构建风险预测模型,并由验证组评估该风险预测模型的可行性。结果 模型组Logistic回归分析结果显示,丘脑出血、脑出血量大、心率高是高血压性脑出血并发神经源性肺水肿的危险因素(P<0.05),格拉斯哥昏迷指数(GCS)评分高、氧合指数(PaO_(2)/FiO_(2))高是高血压性脑出血并发神经源性肺水肿的保护因素(P<0.05)。基于以上5项风险因素建立高血压性脑出血并发神经源性肺水肿的风险预测模型,并对该模型进行内外部验证,校准曲线显示列线图模型的校准曲线和理想曲线的一致性良好,且该模型的C-index为0.826(95%CI=0.777~0.868);受试者工作特征(ROC)曲线显示,模型组和验证组患者神经源性肺水肿发生风险的曲线下面积(AUC)分别为0.817、0.802;决策曲线(DCA)显示,高血压性脑出血患者根据该模型预测神经源性肺水肿发生风险的净收益较高。结论 基于脑出血部位、脑出血量、GCS评分、心率及PaO_(2)/FiO_(2)构建的高血压性脑出血并发神经源性肺水肿的风险预测模型的预测效能及适用性较好,可用于筛查此类人群中神经源性肺水肿的发生风险。

Objective To explore the risk factors of neurogenic pulmonary edema in patients with hypertensive cerebral hemorrhage,and build a risk prediction model.Methods The clinical data of 412 patients with hypertensive cerebral hemorrhage who were admitted to our hospital from February 2019 to October 2022 were retrospectively analyzed,and the subjects were divided into model group(n=288)and validation group(n=124)according to the ratio of 7:3.The risk factors of neurogenic pulmonary edema were screened by single factor and multiple factor Logistic regression analysis to build a risk prediction model,and the risk prediction model was verified in the validation group.Results Logistic regression analysis of model group showed that thalamic hemorrhage,large intracerebral hemorrhage and high heart rate were risk factors for neurogenic pulmonary edema in patients with hypertensive cerebral hemorrhage(P<0.05),while high Glasgow coma index(GCS)score and high oxygenation index(PaO 2/FiO 2)were protective factors for neurogenic pulmonary edema in patients with hypertensive cerebral hemorrhage(P<0.05).Based on the above five risk factors,the risk prediction model of neurogenic pulmonary edema in patients with hypertensive cerebral hemorrhage was established,and the model was verified internally and externally.The calibration curve suggested that the calibration curve of the nomograph model was consistent with the ideal curve,and the C-index of this model was 0.826(95%CI=0.777~0.868).The receiver operating characteristic(ROC)showed that the area under the curve(AUC)of the risk prediction model predicting the risk of neurogenic pulmonary edema in the model group and the validation group were 0.817 and 0.802,respectively.The decision curve(DCA)showed that patients with hypertensive cerebral hemorrhage had higher net income in predicting the risk of neurogenic pulmonary edema according to the model.Conclusions The risk prediction model of neurogenic pulmonary edema in patients with hypertensive cerebral hemorrhage based on the bases of intracerebral hemorrhage,intracerebral hemorrhage volume,GCS score,heart rate and PaO2/FiO2 have good prediction efficiency and applicability,and it can be used to screen the risk of neurogenic pulmonary edema in this populations.