临床试验中救助治疗对临床效应评价的影响研究

Research on the Impact of Rescue Therapy to the Evaluation of Clinical Effects in Clinical Trials

目的本文引入秩方法对含有救助治疗的临床试验疗效结果进行分析,并与目前国内常用方法进行比较,为分析该类数据选择合适的方法提供统计学依据。方法采用Monte Carlo模拟的方法,考虑调整患者不同时点观测值之间的相关系数和救助阈值,分析疗效评价的Ⅰ型错误和检验效能,并与目前国内常用的完整案例分析和末次观测结转法进行比较。结果Ⅰ型错误方面,当救助比例在20%左右时,不同方法差异较小,当救助比例较大时,按时间和末次观测值调整的秩方法是最佳的方法。效能方面,当救助比例小于40%时,所有方法效能均大于0.8,当救助比例较大时,按末次观测值调整的秩方法是最佳的方法。结论在实际研究过程中,需要根据实际情况选择合适的分析方法,而不是直接采用完整案例分析和末次观测结转法,以得到更有效的结果。救助治疗的分析方法可以为安慰剂对照试验提供一种新的设计和分析思路。

Objective Author introduces and improves rank method to analyze the results of clinical trials containing rescue therapy, compares it with the conventional methods and provides statistical basis for the analysis of such data to select the appropriate method. Methods Through Monte Carlo simulation, consider adjusting the correlation coefficient of the observed values of the same patient at different points and rescue threshold,compare type Ⅰ error and power of efficacy evaluation among rank method and conventional methods （complete case analysis （CC） and last observation carried forward （LOCF））. Results In terms of type Ⅰ error, when the rescue proportion is about 20%, the difference between different methods is small;when the rescue proportion is large,rank method adjusted with time and last visit （ARTLV） is the best choice. In terms of power, when the rescue proportion is less than 40%, the power of all methods is larger than 0.8, when the rescue proportion is large, rank method adjusted with last visit （ARLV） is the best choice. Conclusion Researchers should select appropriate method based on actual situation to get accurate results. It＇s ethical to set rescue therapy in placebo-controlled trials, so analysis method of rescue therapy can provide a new idea for the design and analysis of placebo-controlled trials.