作者：阳珊1，2，董粜霞2，朱南希2，邓仁丽1

单位：

1.519000广东省珠海市，遵义医科大学第五附属（珠海）医院护理部　2.563000贵州省遵义市，遵义医科大学附属医院脑血管科

Units：

1.Cerebrovascular Department, Affiliated Hospital of Zunyi Medical University, Zunyi 563000, China2.Nursing Department, the Fifth Affiliated Hospital of Zunyi Medical University, Zhuhai, Zhuhai 519000, China

关键词：

缺血性卒中；复发；预测模型；系统评价

Keywords：

Ischemic stroke; Recurrence; Prediction model; Systematic reviews

CLC：

R 743.3

DOI：

10.12114/j.issn.1008-5971.2022.00.266

Funds：

贵州省科技计划项目（黔科合支撑［2021］一般446）

摘要：

目的　系统评价缺血性卒中（IS）复发风险预测模型。方法　计算机检索中国知网、万方数据知识服务平台、维普网、PubMed、Ovid、Cochrane Library、Web of Science发表的IS复发风险预测模型的相关文献。由2名研究者参考CHARMS清单提取纳入文献的基本特征，采用PROBAST独立评价纳入文献的风险偏倚和适用性。结果　共纳入20项研究。12项研究采用Logistic回归模型建模，7项研究采用Cox回归模型建模，1项研究采用其他模型建模；7项研究仅进行内部验证，6项研究仅进行外部验证，2项研究进行内部研究和外部验证，5项研究未进行模型验证；IS复发率为2%~48%。20项研究纳入的预测因子为4~20个，最常见的预测因子是年龄、短暂性脑缺血发作（TIA）/卒中史、高血压、糖尿病、心血管疾病、外周动脉疾病。16项研究报道了模型的AUC，为0.55~0.933，其中10项研究模型的AUC≥0.7；3项研究报道了一致性指数（CI ），为0.630~0.68；1项研究未报告模型的区分度。仅6项研究进行了模型校准。结论　基于现有文献，IS复发风险预测模型预测IS复发的AUC为0.55~0.933，纳入研究存在较高偏倚风险，大多数模型对IS复发的区分能力不高且未进行外部验证，导致其推广和应用受限。

Abstract：

Objective　To systematic review the recurrence risk prediction model for ischemic stroke (IS) . Methods　The related literature of IS recurrence risk prediction model published by CNKI, Wanfang Data, VIP, PubMed, Ovid, CochraneLibrary and Web of Science were searched by computer. Two researchers extracted the basic characteristics of the includedliterature by referring to the CHARMS list, and used PROBAST to independently evaluate the risk bias and applicability of theincluded literature. Results　A total of 20 studies were included. Twelve studies were modeled by Logistic regression model, 7studies were modeled by Cox regression model, and 1 studies were modeled by other models. Seven studies had internal validationonly, 6 studies had external validation only, 2 studies had internal and external validation, 5 studies had no model validation; therecurrence rate of IS was 2%-48%. Four to twenty predictors were included in the 20 studies. The most common predictors wereage, transient ischemic attack (TIA) /stroke history, hypertension, diabetes, cardiovascular disease, and peripheral arterial disease.A total of 16 studies reported the AUC of the models, ranging from 0.55 to 0.933, of which the AUC of 10 studies was ≥ 0.7. Threestudies reported consistency index (CI ) of 0.630-0.68; one study did not report the discrimination of the model. Only 6 modelswere calibrated. Conclusion　Based on the existing literature, the AUC of IS recurrence risk prediction model for predictingIS recurrence is 0.55-0.933. High risk of bias in included studies. Most of the models have low discrimination ability for the ISrecurrence and have not been externally verified, resulting in limited promotion and application of them.

ReferenceList：