阶梯设计在随机对照试验中的应用  

Application of ＂stepped-wedge design＂ methodology in randomized controlled trials

作　　者：杨祖耀[1] 詹思延[1] 

机构地区：[1]北京大学公共卫生学院流行病与卫生统计学系,100191

出　　处：《中华流行病学杂志》2010年第1期92-95,共4页Chinese Journal of Epidemiology

基　　金：“十一五”国家科技支撑计划（2007BA124803）

摘　　要：介绍阶梯设计的基本原理和实施方法，探讨其在随机对照试验中的应用。当研究者希望在一定范围内全面推行某项通常来说“利大于弊”的措施，又想通过随机对照试验来对该措施的效果进行评价，尤其是当资源有限使得本来就只能分步实施干预的时候，阶梯设计非常适用。另外，该方法还可用来发现或控制时间趋势对效果评价的影响。但与传统设计相比，阶梯设计的试验周期更长，统计学问题也要复杂得多。因此，在设计和实施的过程中应严格把握有关注意事项，以保证研究结果的稳健性。In this article, two research cases are employed to show the rationale of the stepped-wedge design, under what situations that such a design is desirable, and how it can be implemented. Stepped-wedge design seems to suit to randomized controlled trials in which the entire study population will receive intervention programs as they would ＂provide more advantages than harm＂. When intervention can not be given to all the targets simultaneously due to limited resources, this design is particularly useful. The stepped-wedge design is also relevant when there is a hope to detect or control the time trend effect on the effectiveness of the intervention strategy. On the other hand, however, this design requires longer trial duration and presents a number of statistical challenges. Hence, careful planning and monitoring are essential to ensure that a robust evaluation is undertaken.

关 键 词：阶梯设计 随机对照试验 设计延迟 分阶段引入 分阶段实施 

分 类 号：R195[医药卫生—卫生统计学]