Scholarlink

<p>Monitoring adverse events following immunization is essential for maintaining vaccine safety in real-world settings. The Vaccine Adverse Event Reporting System (VAERS) provides a large, spontaneous reporting database, but signal detection is complicated by data sparsity, correlation among adverse events (AEs), and the high dimensionality of vaccine–AE relationships. In this study, we implement an improved two-stage hierarchical testing framework to the 2024 VAERS data to improve detection of vaccine-associated AEs. Specifically, in Stage 1, we group related AEs into 27 System Organ Classes (SOCs) and test vaccine–SOC associations using score-type statistics with extended <i>p</i>-value combination methods (Fisher, Adaptive, and Optimal). In Stage&#xa0;2, we examine individual vaccine–AE pairs within the identified SOCs using Fisher’s exact test with the Benjamini–Hochberg procedure to control the false discovery rate. The analysis results recovered several well-established vaccine–AE associations and, importantly, also highlight potential new signals involving rare adverse events, illustrating its ability to detect weak or sparse signals while controlling false discoveries. Therefore, this study demonstrates that the modified two-stage hierarchical testing framework is a practical and scalable tool for safety signal detection in VAERS and can be adapted for other pharmacovigilance systems to support ongoing vaccine safety monitoring.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hierarchical Testing for Vaccine Safety Surveillance: Implementing a Two-Stage Framework in the VAERS Database

  • 期刊论文

  • 发表日期：

    2026年4月16日

查看全文
  • Erika Gibson,
  • Jianping Sun

摘要

Monitoring adverse events following immunization is essential for maintaining vaccine safety in real-world settings. The Vaccine Adverse Event Reporting System (VAERS) provides a large, spontaneous reporting database, but signal detection is complicated by data sparsity, correlation among adverse events (AEs), and the high dimensionality of vaccine–AE relationships. In this study, we implement an improved two-stage hierarchical testing framework to the 2024 VAERS data to improve detection of vaccine-associated AEs. Specifically, in Stage 1, we group related AEs into 27 System Organ Classes (SOCs) and test vaccine–SOC associations using score-type statistics with extended p-value combination methods (Fisher, Adaptive, and Optimal). In Stage 2, we examine individual vaccine–AE pairs within the identified SOCs using Fisher’s exact test with the Benjamini–Hochberg procedure to control the false discovery rate. The analysis results recovered several well-established vaccine–AE associations and, importantly, also highlight potential new signals involving rare adverse events, illustrating its ability to detect weak or sparse signals while controlling false discoveries. Therefore, this study demonstrates that the modified two-stage hierarchical testing framework is a practical and scalable tool for safety signal detection in VAERS and can be adapted for other pharmacovigilance systems to support ongoing vaccine safety monitoring.