Zhao Lili, Lee Sunghun, Li Rongxia, Ong Edison, He Yongqun, Freed Gary
Department of Biostatistics, University of Michigan, Ann Arbor, Michigan, USA.
Department of Statistics, University of Michigan, Ann Arbor, Michigan, USA.
Stat Biopharm Res. 2020;12(3):303-310. doi: 10.1080/19466315.2020.1764862. Epub 2020 Jun 8.
As a national public health surveillance resource, Vaccine Adverse Event Reporting System (VAERS) is a key component in ensuring the safety of vaccines. Numerous methods have been used to conduct safety studies with the VAERS database. These efforts focus on the downstream statistical analysis of the vaccine and adverse event associations. In this paper, we primarily focus on processing the raw data in VAERS before the analysis step, which is also an important part of the signal detection process. Due to the semi-annual update in the Medical Dictionary for Regulatory Activities (MedDRA) coding system, adverse event terms that describe the same symptom might change in VAERS; therefore, we identify these terms and combine them to increase the signal detection power. We also consider the uncertainty of the vaccine and adverse event pairs that arise from reports with multiple vaccines. Finally, we discuss four commonly used statistics in assessing the vaccine and adverse event associations, and propose to use the statistics that are robust to the reporting bias in VAERS and adjust for potential confounders of the vaccine and adverse event association to increase signal detection accuracy.
作为一项国家公共卫生监测资源，疫苗不良事件报告系统（VAERS）是确保疫苗安全的关键组成部分。人们已使用多种方法对VAERS数据库进行安全性研究。这些工作侧重于疫苗与不良事件关联的下游统计分析。在本文中，我们主要关注在分析步骤之前处理VAERS中的原始数据，这也是信号检测过程的重要组成部分。由于监管活动医学词典（MedDRA）编码系统每半年更新一次，VAERS中描述相同症状的不良事件术语可能会发生变化；因此，我们识别这些术语并将它们合并，以提高信号检测能力。我们还考虑了来自多种疫苗报告的疫苗与不良事件对的不确定性。最后，我们讨论了评估疫苗与不良事件关联时常用的四种统计方法，并建议使用对VAERS中的报告偏倚具有稳健性且针对疫苗与不良事件关联的潜在混杂因素进行调整的统计方法，以提高信号检测准确性。
