［目的/意义］ 撤稿是科学共同体实施自我纠错的重要方式，但撤稿并不意味着相关论文立即退出科学交流，其仍会形成复杂的引用关联。现有研究主要关注撤稿论文产生的影响，对于引用撤稿论文是否与施引文献撤稿风险存在统计关联仍缺乏系统研究。本研究从引用关系视角分析二者之间的统计关联，为撤稿风险识别与预警提供实证参考。［方法/过程］ 研究数据以Amend撤稿数据库为核心，结合OpenAlex数据库获取施引文献信息。首先，从引用广度、引用深度和级联规模等方面分析撤稿论文的引用特征；随后，通过构建实验组与对照组分析引用撤稿论文的文献是否具有更高的撤稿风险；进一步结合统计分析与机器学习方法，从撤稿论文自身特征及其与施引文献之间的关系特征两个维度识别影响上述统计关联的关键特征。［结果/结论］ 撤稿论文的引用级联呈现出持续时间长、引用层级深及关联范围广等特征，同时，论文在撤稿后仍保持较高可见性。实验组与对照组的比较结果表明引用撤稿论文与施引文献撤稿风险之间存在统计关联，进一步分析发现，临床试验、随机对照试验、案例研究及因论文工厂问题而撤稿的论文，其施引文献具有更高撤稿率。同时，施引文献引用撤稿论文的比例与其撤稿率之间也表现出较强的统计关联。

[Purpose/Significance] Retraction is an important self-correction mechanism of the scientific community. However, retraction does not necessarily mean that the affected articles immediately disappear from scientific communication, as they may continue to generate complex citation links. Existing studies have primarily focused on the impacts of retracted papers, while whether citations to retracted papers are statistically associated with the retraction risk of citing papers remains insufficiently investigated. Therefore, this study examines the statistical association between citations to retracted articles and the retraction risk of citing articles from the perspective of citation relationships, providing empirical evidence for retraction risk identification and early warning. [Method/Process] The data were primarily obtained from the Amend retraction database, combined with citing-paper information retrieved from the OpenAlex database. First, the citation characteristics of retracted articles were analyzed in terms of citation breadth, citation depth, and cascade size. Subsequently, an experimental group and a control group were constructed to examine whether papers citing retracted papers had a higher risk of retraction. Furthermore, statistical analysis and machine learning approaches were combined to identify key features associated with this statistical relationship from two dimensions: the intrinsic characteristics of retracted articles and the relational characteristics between retracted and citing articles. [Result/Conclusion] Retracted articles maintain relatively high visibility after retraction and exhibit citation cascade characteristics, including long persistence, deep citation levels, and broad citation scope within citation networks. Comparisons between the experimental and control groups demonstrate a statistical association between citing retracted articles and the retraction risk of citing articles. Further analysis reveals that citing articles of retracted papers involving clinical trials, randomized controlled trials, case studies, and paper mill-related retractions show higher retraction rates. Moreover, the proportion of retracted articles cited by a citing article exhibits a strong statistical association with the retraction rate of citing articles.