论文部分内容阅读
非死亡时间是指在队列研究的随访期内研究对象不可能发生结局事件的一段时间,由非死亡时间错分或剔除引起的偏倚称为非死亡时间偏倚。在基于数据库的药物流行病学队列研究中,非死亡时间偏倚较为常见,而且可能对结果产生严重影响,以致于得出完全错误的结果和结论。因此在文献阅读和研究设计时,应识别和避免非死亡时间及其引起的偏倚。本文介绍了非死亡时间引起偏倚的原理、偏倚的识别以及避免该偏倚的研究设计和数据分析方法。
Non-death time refers to the period of time during which the subjects are unlikely to have an outcome during the follow-up period of the cohort study. The bias caused by the non-death time misclassification or rejection is called the non-death time bias. In a database-based pharmacoepidemiological cohort study, non-death-time bias is more common and can have serious consequences on results, resulting in entirely erroneous findings and conclusions. Therefore, non-death time and the resulting bias should be identified and avoided in literature reading and research design. This article presents the principles of non-death-induced bias, the identification of bias, and the research design and data analysis methods to avoid this bias.