Study on the predictive value of multiple prediction models on the prognosis of patients with intravenous thrombolysis in acute ischemic stroke
10.3969/j.issn.1008-9691.2024.03.008
- VernacularTitle:多种预测模型对急性缺血性脑卒中静脉溶栓患者预后的预测价值研究
- Author:
Yinglei LI
1
;
Lingyun XI
;
Bing DAI
;
Qing LIANG
;
Tao QIE
Author Information
1. 保定市第一中心医院急诊医学科,河北保定 071000
- Keywords:
Stroke;
Intravenous thrombolysis;
Predictive modeling
- From:
Chinese Journal of Integrated Traditional and Western Medicine in Intensive and Critical Care
2024;31(3):293-299
- CountryChina
- Language:Chinese
-
Abstract:
Objective To evaluate the predictive performance of the Alberta stroke program early computed tomography score(ASPECTS),the Lausanne stroke scale(ASTRAL),DRAGON,START,and the total healthy risks for patients with vascular events calculated(THRIVE-c)score prediction models for the prognosis of patients with acute ischemic stroke(AIS)in Baoding advanced stroke centers,Hebei Province.Methods Clinical data of 909 patients with AIS who were admitted for intravenous thrombolysis in the department of emergency of the advanced stroke center of the Baoding NO.1 Central Hospital from January 2016 to August 2022 were collected and scored using the ASPECTS,ASTRAL,DRAGON,START,and THRIVE-c scales.The 3-month modified Rankin scale(mRS)score was observed for each score,with a score of 0-2 as a good prognosis and 3-6 as a poor prognosis,and multivariate Logistic regression coefficients for the variables in the 5 scores were compared with the original derivation cohort.The predictive efficacy of each scoring model for 3-month poor prognosis of AIS patients was analyzed by using the receiver operator characteristic curve(ROC curve),and area under the curve(AUC)was calculated;the degree of fit of each model to the actual prognostic results was assessed by using the Hosmer-Lemeshow goodness-of-fit test;at the same time,the calibration curves and Brier scores were used as an evaluation of the model's calibration performance indicators.Results A total of 786 patients were enrolled and 340 had a poor prognosis.Comparison of the validation cohort with the original cohort for model construction showed that the mean age in the validation cohort was 65 years old,with a relatively small proportion of 33.84%females,and a median glycemic value of 8.09 mmol/L.The proportion of poor prognosis was lower in the validation cohort compared to the original ASPECTS and THRIVE-c cohort(43.68%vs.51.90%,43.68%vs.50.30%)and higher in the validation cohort compared to the original cohorts of ASTRAL,DRAGON,and START(43.68%vs.34.00%,43.68%vs.35.20%,43.68%vs.39.10%,all P<0.05).ROC curve analysis showed that ASPECTS,ASTRAL,DRAGON,START and THRIVE-c scores all had predictive value for the prognosis of patients with AIS,with AUC and 95%confidence intervals(95%CI)of 0.848(0.820-0.876),0.825(0.795-0.855),0.833(0.805-0.861),0.838(0.811-0.866),and 0.727(0.691-0.762);in the anterior circulation it was 0.852(0.821-0.883),0.817(0.782-0.853),0.815(0.781-0.849),and 0.833(0.800-0.866),0.710(0.668-0.751);and 0.841(0.781-0.902),0.850(0.792-0.908),0.874(0.816-0.931),0.854(0.800-0.908),and 0.775(0.704-0.846)in the posterior loops,respectively.Hosmer-Lemeshow goodness-of-fit test showed that the DRAGON model curve had a smaller gap from the middle diagonal and was better calibrated(χ2=1.483,P=0.993).The Brier score showed that the ASPECTS scoring model had the smallest Brier score was 0.150 and the best performance.Conclusions The ASPECTS,ASTRAL,DRAGON,START,and THRIVE-c scoring models have been shown to be effective in predicting poor prognosis in patients with AIS treated with intravenous thrombolytic therapy at 3 months,with the ASPECTS model showing the most outstanding predictive performance.For anterior infarcts,the ASPECTS score demonstrated the highest predictive efficacy,whereas for posterior infarcts,the DRAGON score had the best predictive performance.The probability of the ASPECTS score in predicting prognosis in the prediction model was in good agreement with the actual probability.
