中医药序列多重分配随机试验设计分析思路及统计方法应用

Analytical and statistical analysis methods to SMART design in the field of Chinese medicine

序列多重分配随机试验(sequential multiple assignment randomized trial,SMART)是一项多阶段随机试验,可将多种干预组分嵌入研究中,评估不同阶段效果,制定最佳干预序列的组合,在中医药领域应用前景广阔。但因设计新颖性和统计分析复杂性,限制了其广泛应用。本文结合医学实践中经典SMART设计实例,在文献基础上总结SMART主要的研究目的,根据研究目的分别阐述数据整理及分析阶段的统计假设和统计分析思路方法,强调SMART设计中如何规范处理缺失数据的问题,并介绍文章撰写中规范展示方法学部分的内容,以期为今后研究者开展SMART设计提供方法学依据。

Sequential multiple assignment randomized trial(SMART) is a multi-stage randomized trial,which can embed multiple intervention components into the study,assess the effects of different stages,and formulate the optimal combination of intervention sequences,and has a promising application in the field of Chinese medicine.However,the novelty and complexity of its statistical analysis have limited its wide application.This paper combines classic SMART design examples in medical practice,summarizes the main research purposes of SMART on the basis of the literature,elaborates the statistical assumptions and statistical analysis ideas and methods in the analysis phase according to the research purposes,emphasizes how to standardize the handling of missing data in SMART design,and finally,introduces how to standardize the presentation of the methodology section in the article,with a view to providing a methodological basis for researchers to carry out SMART design in the future.