探究脑出血早期血肿扩大的影响因素及预测模型构建与验证

Study on the influencing factors of early hematoma enlargement in spontaneous cerebral hemorrhage

目的:探索高血压脑出血出现早期血肿扩大(Hematoma enlargement,HE)的独立预测因素,并构建模型验证。方法:纳入2017年1月—2020年12月所有符合标准的原发性脑出血患者232例。按照HE超过33%或体积超过6 mL为标准,分为HE组和非HE组;寻找早期HE相关的因素,并构建早期HE的模型并验证其预测的准确性。结果:HE 39例,未发生扩大193例。单因素分析显示GSC评分、中线移位及平均密度值有统计学差异;多因素Logistic回归分析构建GSC、中线移位、平均密度值及联合预测模型。联合模型AUC最高为0.786 8,GSC、中线移位、平均密度值的AUC值分别为0.730 5、0.662 7、0.635 0;交叉验证结果显示,联合模型预测准确率77.2%,灵敏度为64.1%,特异度为79.8%。结论:中线移位、GSC评分及平均密度值是早期HE的预测因素,而基于临床因素、实验室检查及影像特征构建的模型具相当不错的预测能力。

Objective: To investigate the independent prognostic factors of early hematoma enlargement(HE) in hypertensive cerebral hemorrhage and build according prognostic models. Methods: All the patients with primary intracerebral hemorrhage from January 2017 to December 2020 who met the criteria were enrolled in this study, with a result of 232 cases in total.We divided the cases into two groups: HE group and non-HE group. The criteria is whether the HE more than 33% or the volume exceeds 6 mL. Single-factor analysis was performed to find factors related to early HE. Multi-factor Logistic regression analysis was performed to find independent factors affecting early HE, and to verify the predictive value of the independent influencing factors of the occurrence of HE. Results: Thirty-nine cases showed HE, and 193 cases did not have HE. GSC score, midline shift and average density value were statistically significant. The areas under the GSC curve, midline shift curve and the curve of average density value were 0.730 5, 0.662 7 and 0.635 0. Leave-one-out cross-validation result showed that the model prediction accuracy rate was 77.2%, the sensitivity was 64.1%, and the specificity was 79.8%. Conclusion: Models based on clinical factors, laboratory tests and imaging features showed certain predictive capabilities. Midline shift, GSC score,and average density value are independent predictors of HE.