基于真实世界数据的随机对照试验模拟研究设计、评价与实例分析

Design Framework,Evaluation and Case Application of Randomized Controlled Trial Emulation Research

目前开展的真实世界研究(real world study,RWS)仍存在诸多局限性,而未能在卫生技术评估方面充分发挥应有的作用。因此,有必要改进传统RWS设计,以产生高质量的医学证据。策略之一是仿照随机对照试验(randomized controlled trial,RCT)开展观察性研究(又称RCT仿真或模拟研究),既借鉴了RCT设计优势,增强了因果推断的强度,又保留了RWS的代表性,可以作为打通药品上市前与上市后证据链的桥梁,以期为医疗决策提供合理的证据支撑。仿照RCT开展观察性研究分为两步:第一步,基于临床问题构建一个目标试验,参照目标试验关键特征构建相应的RWS方案(包括纳排标准、治疗策略、分配程序、随访、结局定义、因果对比和统计分析策略),特别注意减小模拟差异和控制相关偏倚;第二步,使用真实世界数据(real world data,RWD)按照既定研究方案进行数据分析,得到相应结果。

Currently,real world studies(RWS)still have certain limitations and fail to play due role in health technology assessment.It is necessary to refine traditional RWS designs to produce high-quality medical evidence.One strategy is to carry out observational studies,known as randomized controlled trial(RCT)emulation or RCT simulation,based on the principles of RCTs.This approach combines the advantages of RCT design,enhances the intensity of causal inference,and retains the representativeness of RWS.It can bridge the gap between pre-marketing and post-marketing evidence,providing rational evidence for medical decision-making.The process of conducting observational studies based on RCT involves two steps.The first step is to construct a target trial based on a specific clinical question and then develop a corresponding RWS protocol(including key features such as inclusion and exclusion criteria,treatment strategies,assignment procedures,followup,outcome definitions,causal comparisons,and statistical analysis strategies).Special attention should be paid to reducing emulation differences and controlling related biases.The second step is to use real world data to analyze the data according to the established research protocol and obtain the corresponding results.