Tsong Y
Division of Biometrics, Center for Drug Evaluation and Research Food and Drug Administration, Rockville, Maryland 20857, USA.
J Biopharm Stat. 1995 Mar;5(1):95-114. doi: 10.1080/10543409508835100.
The Food and Drug Administration has collected spontaneous reports on adverse events (AE) from manufacturers and individuals. These data provide useful information on the safety of marketed drugs. Due to many unique characteristics of this reporting system, the information is difficult to evaluate. Incidence rates for specific adverse events and drug combinations cannot be estimated. However, reporting rates (number of reports per market share) based on prescriptions can be computed. These reporting rates provide signals of serious adverse experience that may deserve attention. When the ratio of reporting rates is used for the comparison of two drugs of the same drug class, adjustments are needed for the marketing year and secular trends of all-drug-all-AE reporting. The Mantel-Haenszel procedure is used to combine the multiyear data into one summary statistic. Application of this analysis is illustrated on reports of upper gastrointestinal bleeding, perforation, and ulcer associated with nonsteroidal anti-inflammatory drugs, as given in Rossi et al. (12).
美国食品药品监督管理局已从制造商和个人处收集了关于不良事件（AE）的自发报告。这些数据为已上市药品的安全性提供了有用信息。由于该报告系统的许多独特特性，这些信息难以评估。无法估计特定不良事件和药物组合的发生率。然而，可以计算基于处方的报告率（每市场份额的报告数量）。这些报告率提供了可能值得关注的严重不良事件的信号。当使用报告率之比来比较同一药物类别的两种药物时，需要对营销年份和所有药物所有不良事件报告的长期趋势进行调整。Mantel-Haenszel程序用于将多年数据合并为一个汇总统计量。如Rossi等人（12）所述，该分析方法在与非甾体抗炎药相关的上消化道出血、穿孔和溃疡报告中的应用得到了说明。
