外周动脉疾病腔内治疗后再狭窄风险预测模型:系统综述与Meta分析
该系统综述与Meta分析检索8个数据库至2024年1月1日,纳入7项观察性研究共7个预测模型,评估外周动脉疾病患者腔内治疗后再狭窄风险预测模型。模型再狭窄发生率为21.8%至39.7%,汇总AUC为0.80(95%CI 0.74-0.86),全部研究存在高偏倚风险,模型仍处于发展阶段。
要点
- 7个模型再狭窄发生率21.8%至39.7%,样本量137至1578例,仅2个模型经外部验证。
- 汇总AUC为0.80(95%CI 0.74-0.86),预测区间约0.80(95%CI 0.62-0.91)。
- 全部研究存在高偏倚风险,主要因数据来源不当和分析域报告不充分。
核心结论
- 外周动脉疾病患者腔内治疗后再狭窄预测模型总体预测性能良好,但仍处于发展阶段且偏倚风险较高。
- 未来研究应遵循TRIPOD声明,聚焦更大样本、严谨设计及多中心外部验证的新模型开发。
研究设计与样本
- 系统综述与Meta分析,纳入观察性研究的预测模型构建与验证。
- 检索中国知网、中国科技期刊数据库、万方数据库、SinoMed、PubMed、Web of Science、Embase和Cochrane Library,检索时间从建库至2024年1月1日。
- 共检索4275项研究,最终纳入7篇文章,包含7个外周动脉疾病患者腔内治疗后再狭窄预测模型。
- 纳入模型的总样本量为137至1578例,最常用建模方法为逻辑回归分析,所有模型均使用R软件构建。
主要发现
- 再狭窄发生率为21.8%至39.7%。
- 仅2个模型进行了外部验证,报告的曲线下面积范围为0.728至0.864。
- 汇总曲线下面积统计量为0.80(95%CI 0.74-0.86),近似预测区间为0.80(95%CI 0.62-0.91)。
- 纳入预测因子数量为3至10个,最常见因子包括年龄、跨大西洋协作组织共识Ⅱ分类、高血压、糖尿病、超敏C反应蛋白和手术方式。
- 全部研究均存在高偏倚风险,主要归因于数据来源不当和分析域报告不佳。
关键数值与推荐
- 使用预测模型偏倚风险评估工具(PROBAST)进行模型偏倚风险和适用性评估。
- 所有研究均存在高偏倚风险。
- 汇总AUC为0.80(95%CI 0.74-0.86),近似预测区间为0.80(95%CI 0.62-0.91)。
临床怎么用
- 现有预测模型总体预测性能良好但偏倚风险高,临床使用需谨慎对待其适用性。
- 未来研究应遵循TRIPOD声明,开发更大样本、严谨设计及多中心外部验证的新模型。
局限性
- 全部研究均存在高偏倚风险,主要归因于数据来源不当和分析域报告不佳。
- 仅2个模型进行了外部验证。
- 预测模型仍处于发展阶段且偏倚风险较高。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: Peripheral artery disease (PAD) patients after endovascular treatment (EVT) have a relatively high restenosis rate. However, this risk can be mitigated through precise risk assessment and individualized self-management intervention plans. Moreover, the number of predictive models for restenosis risk in PAD patients after EVT is gradually increasing, yet these results of study exhibit certain discrepancies, raising uncertainties regarding the quality and applicability in clinical practice and future research. OBJECTIVE: The objective of this study was to systematically evaluate risk-predictive models for restenosis in patients with PAD after EVT. METHODS: A systematic review and meta-analysis of predictive model construction and validation using observational studies was undertaken. The China National Knowledge Infrastructure, China Science and Technology Journal Database, Wanfang Database, SinoMed, PubMed, Web of Science, Embase, and the Cochrane Library were searched from inception to January 1, 2024. Two researchers independently conducted literature screening and data extraction, encompassing study design, data sources, outcome definition, sample size, predictive factors, model development, and performance. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used for risk of bias and applicability assessment of the models. RESULTS: A total of 4275 studies were retrieved, ultimately resulting in the inclusion of 7 articles comprising 7 predictive models for restenosis in PAD patients after EVT, with a restenosis incidence ranging from 21.8% to 39.7%. The total sample size of the included models ranged from 137 to 1578 cases, with logistic regression analysis being the most commonly used modeling method. All models were built using R software. Only 2 models underwent external validation, and the reported area under the curve ranged from 0.728 to 0.864. The summary area-under-the-curve statistic was 0.80 (95% confidence interval [CI], 0.74-0.86), with an approximate prediction interval of 0.80 (95% CI, 0.62-0.91) . The number of included predictive factors ranged from 3 to 10, with the most common factors being age, Trans-Atlantic Inter-Society Consensus Ⅱ classification, hypertension, diabetes, high-sensitivity C-reactive protein, and surgical approach. All studies exhibited high risk of bias, primarily attributed to inappropriate sources of data and poor reporting of the analysis domain. CONCLUSION: Predictive models for restenosis after EVT in PAD patients demonstrate overall good predictive performance but are still in the developmental stage with higher risk of bias. Future studies should follow the TRIPOD statement, focusing on the development of new models with larger samples, rigorous study designs, and multicenter external validation.Clinical ImpactThis systematic review adheres to the PRISMA 2020 statement, offering the most recent systematic assessment of risk prediction models for restenosis following endovascular treatment in peripheral arterial disease.The newly developed PROBAST tool was employed to assess the risk of bias and the applicability of the existing evidence.This review emphasizes the practical utility, limitations of the current evidence, and recommendations for future research, with the goal of providing valuable information for clinicians and patients in their decision-making process, while also supporting the advancement of future research endeavors.
引用
- Quan X, Liu Y, Xiong H, et al. Risk-Prediction Model of Restenosis after Endovascular Treatment for Peripheral Arterial Disease: A Systematic Review and Meta-analysis. J Endovasc Ther. 2024. PMID: 39513458. doi:10.1177/15266028241289083
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。