脑出血患者微创术后预后预测模型的选择:9个预测评分量表的验证和比较

Selection of models for predicting the outcomes of patients with intracerebral hemorrhage after minimally invasive surgery: External verification and comparison of nine predictive scoring scales

目的探讨适合用于预测脑出血(Intracerebral hemorrhage, ICH)患者微创手术(minimally invasive surgery, MIS)术后结局的预测模型。方法回顾性分析2012年6月-2018年12月武汉同济医院行脑出血微创手术患者的临床资料;收集这9个评分量表的预测因子和脑出血患者的临床结局,比较这9个评分量表的辨别力、校准度和临床实用性。结果本研究共纳入440例患者,58例(13.2%)患者术后1个月内死亡,174例(39.5%)患者术后1年内预后不佳。比较现有的预测评分量表,在预测脑出血MIS术后1个月内死亡方面ICH-GS(intracerebral hemorrhage grading scale)辨别能力最强[曲线下面积(Area under curve, AUC)=0.763,95%CI=0.720~0.802],ICH-FOS(ICH functional outcome score)(AUC=0.761,95%CI=0.718~0.800)稍逊于ICH-GS,两者均具有良好的校准能力(P=0.937和P=0.907)和更大的临床获益。在预测脑出血MIS术后1年预后不良方面ICH-GS的辨别力最强(AUC=0.731,95%CI=0.687~0.772),校准度良好(P=0.457),并且具有更强的临床实用性。结论这项外部验证表明ICH-GS在预测脑出血患者MIS术后结局方面可能比其他模型更准确,并且具有更大的临床收益。

Objective To select an appropriate scale for predicting the outcomes of intracerebral hemorrhage(ICH) patients with minimally invasive surgery(MIS). Methods We retrospectively analyzed the clinical data of patients with MIS for intracerebral hemorrhage in Tongji Hospital. All the predictors of the nine models and the outcome were collected in the ICH patients with MIS. The external validation of the nine model were quantified by discrimination, calibration, and clinical utility by decision curve analysis(DCA). Results Of 440 ICH patients, 58(13.2%) died within 30 days and174(39.5%) patients had unfavorable outcome at 1 year. The intracerebral hemorrhage grading scale(ICH-GS) and the ICH Functional Outcome Score(ICH-FOS) had areas under the curve of 0.763(95%CI=0.720~0.802) and 0.761(95%CI=0.718~0.800), respectively, in predicting one-month mortality. Regarding calibration, the ICH-GS(P=0.937) and the ICH-FOS(P=0.907) were well calibrated. DCA showed that the ICH-GS and the ICH-FOS had stronger clinical utility compared with other models. The ICH-GS had the largest areas under the curve of 0.731(95%CI=0.687~0.772) in predicting one-year prognosis of ICH patients with MIS. Regarding calibration, the ICH-GS(P=0.457) were well calibrated. DCA showed that ICH-GS had stronger clinical utility compared with other models. Conclusion This external validation showed that ICH-GS might be more accurate and had greater clinical value than others for predicting the outcomes of ICH patients with MIS.