Meta分析中缺失数据的处理方法简介

摘要

At the beginning of a clinical trial design, and even in the middle and later stages when data is tracked and followed up, partial data may be missed inevitably. However, when the missing data and research results are possibly linked, bias may be induced in a randomized controlled trial (RCT), and the bias risk will also be introduced to the results of Meta-analysis. Due to the status of missing data is very complex, processing methods for missing data should be selected according to actual situation. The aim of this paper is to present missing data mechanism including missing at random (MAR), missing completely at random (MCAR), missing not at random (MAR), and common processing methods.%临床试验设计之初，乃至中后期对数据追踪与随访，都可能无法避免部分数据丢失。然而当缺失数据与研究结果可能存在联系时，可能会导致随机对照试验（RCT）偏倚，并且偏倚风险也将被引入到Meta分析结果中。由于缺失数据的情况是非常复杂的，所以对于缺失数据处理要根据实际情况来选择合适的方法。本文就随机缺失、完全随机缺失、非随机缺失的数据缺失机制及其常见的处理方法给出相关简介。