A novel toolkit for the prediction of clinical outcomes following mechanical thrombectomy
A novel toolkit for the prediction of clinical outcomes following mechanical thrombectomy
复制标题
DOI：
10.1016/j.crad.2020.06.026
复制
发表时间：
2020-10-01
影响因子：
2.6
通讯作者：
Nayak, S.
中科院分区：
文献类型：
--
作者：
Raseta, M.;Bazarova, A.;Nayak, S.
目的：为了开发一个强大的工具包，以帮助决策的机械血栓切除术（MT）的基础上，现成的患者变量，可以准确地预测功能的结果后MT。材料和方法：从我们的MT数据库中识别出数据来自前循环卒中患者谁接受MT 2009年10月至2018年1月（n 1/4239）。患者解释变量为年龄、性别、美国国立卫生研究院卒中量表（NIHSS）、阿尔伯塔卒中项目早期CT评分（ASPECTS）、侧支评分和格拉斯哥昏迷量表。从数据中开发了五种模型来预测五种关注的结局：模型1：预测生存期：改良兰金量表（mRS）为0-5（存活）或6（死亡）;模型2：预测良好/不良结局：mRS为0-3（良好）或4-6（不良）;模型3：预测良好/不良结局：mRS为0-2（良好）或3-6（不良）;模型4：mRS类别预测：mRS 0-2（无残疾）、3（轻度残疾）、4-5（重度残疾）或6（死亡）;模型5：准确mRS评分预测（mRS作为连续变量）。每个预测模型的准确性和辨别力进行了测试。结果：生存预测准确率为87%（曲线下面积[AUC] 0.89）。模型2的良好/不良结局预测准确率为91%（AUC 0.94），模型3的准确率为95%（AUC 0.98）。mRS分类的预测准确率为76%，使用“一分法”可提高至98%。准确预测mRS值的误差为0.89。结论：该新工具包可准确估计MT的结局。(c)2020年皇家放射科医师学院。由爱思唯尔有限公司出版。保留所有权利。
AIM: To develop a robust toolkit to aid decision-making for mechanical thrombectomy (MT) based on readily available patient variables that could accurately predict functional outcome following MT.MATERIALS AND METHODS: Data from patients with anterior circulation stroke who underwent MT between October 2009 and January 2018 (n1/4239) were identified from our MT database. Patient explanatory variables were age, sex, National Institutes of Health Stroke Scale (NIHSS), Alberta Stroke Program Early CT Score (ASPECTS), collateral score, and Glasgow Coma Scale. Five models were developed from the data to predict five outcomes of interest: model 1: prediction of survival: modified Rankin Scale (mRS) of 0-5 (alive) or 6 (dead); model 2: prediction of good/poor outcome: mRS of 0-3 (good), or 4-6 (poor); model 3: prediction of good/poor outcome: mRS of 0-2 (good), or 3-6 (poor); model 4: prediction of mRS category: mRS of 0-2 (no disability), 3 (minor disability), 4-5 (severe disability) or 6 (dead); model 5: prediction of the exact mRs score (mRs as a continuous variable). The accuracy and discriminative power of each predictive model were tested.RESULTS: Prediction of survival was 87% accurate (area under the curve [AUC] 0.89). Prediction of good/poor outcome was 91% accurate (AUC 0.94) for Model 2 and 95% accurate (AUC 0.98) for Model 3. Prediction of mRS category was 76% accurate, and increased to 98% using the "one-score-out rule". Prediction of the exact mRS value was accurate to an error of 0.89.CONCLUSIONS: This novel toolkit provided accurate estimations of outcome for MT. (c) 2020 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.