自适应随机在具有响应延迟的临床试验中的应用评价

中国临床药理学与治疗学 ›› 2010, Vol. 15 ›› Issue (9): 1040-1045.

自适应随机在具有响应延迟的临床试验中的应用评价

魏永越, 荀鹏程, 周敏林, 赵杨, 朱晶晶, 于浩, 陈峰

南京医科大学公共卫生学院流行病与卫生统计学系,南京 210029,江苏

收稿日期:2010-07-14 修回日期:2010-08-12 出版日期:2010-09-26 发布日期:2020-09-17
通讯作者: 陈峰,男,教授,博导,研究方向:医学研究中的统计理论与方法。Tel: 025-86862754, E-mail: FengChen@njmu.edu.cn
作者简介:魏永越,男,博士研究生,研究方向:新药临床试验中的统计理论、方法与规范。E-mail: weiyongyue@126.com
基金资助:
江苏省自然科学基金资助(BK2008449)

Adaptive randomization for clinical trials with delayed response

WEI Yong-yue, XUN Peng-cheng, ZHOU Min-lin, ZHAO Yang, ZHU Jin-jin, YU Hao, CHEN Feng

Department of Epidemiology & Biostatistics, School of Public Health,Nanjing Medical University,Nanjing 210029,Jiangsu,China

Received:2010-07-14 Revised:2010-08-12 Online:2010-09-26 Published:2020-09-17

摘要/Abstract

摘要： 目的: 探讨在试验结果具有延迟特征的临床试验中,自适应随机的应用条件及效果评价。方法: 通过模拟试验,考察不同有效率水平、率差水平、不同延迟程度下,从较优组(A组)的样本构成比、平均减少失败人数及检验效能等角度,考察改良的随机化胜者优先(MRPW)原则的应用条件及随机效果。结果: 当两组率差较小时:若有效率较低,不同延迟程度对MRPW的效果影响甚微,随着率的增大,延迟效应对MRPW的效果的影响逐渐显现;当两组率差较大时:随着延迟时间的增大,MRPW的效果衰减趋势明显,且成功率越高,下降趋势越明显。若以MRPW的效果下降不超过20%为限,则当成功率及率差较低时,对延迟时间的限制较为宽松,率差及成功率越大,对延迟时间限制越严格。结论: 在具有延迟效应的临床试验中,MRPW仍有一定的应用价值。若延迟时间满足一定的条件,则可保证其随机效果,从而扩展了其应用条件。实际临床试验中,应根据具体的参数情况,探讨延迟效应对自适应随机效果的影响,从而决定MRPW是否适用于该试验。

关键词: 自适应随机, 随机化胜者优先原则, 延迟效应

Abstract: AIM: To discuss the application condition and evaluate the effect of the adaptive randomization in clinical trials with delayed response.METHODS: The statistical simulations were conducted to evaluate the modified randomized play-the-winner (MRPW) rule and to explore its application condition in clinical trials with delayed response. All the tests were done under circumstances of different effective rates, different levels of rate differences between groups and different delayed levels of response. The proportion of the sample size of the superior group(Group A), the average reduced failure number and the power were also considered during the evaluation.RESULTS: When the rate difference between two groups was small, with relatively low effective rate, different delayed levels had little impact on the effect of MRPW. As effective rate went up, delayed effect on MRPW gradually emerge. When the rate difference between two groups was relatively high, with the increase of delayed time, the weakening tendency of effects on MRPW was obvious. This trend was more obvious in condition of higher effective rate. If the application criterion was set as that the reduction of average reduced failure number should be in the upper limit of 20%, the restriction of delay was less stringent in the condition of lower effective rate, while this restriction was rigorous when effective rates and the difference between effective rates were higher.CONCLUSION: The study extends the application of MRPW in clinical trials with delayed response, which shows that the results of MRPW fit the demand when the delay degree is in some region. According to specific parameters and delayed levels of response in actual clinical trial, we should explore the effect of MRPW to discuss whether to conduct the MRPW in such trial.

Key words: Adaptive randomization, Randomized play-the-winner role, Delayed response

中图分类号:

R311

魏永越, 荀鹏程, 周敏林, 赵杨, 朱晶晶, 于浩, 陈峰. 自适应随机在具有响应延迟的临床试验中的应用评价[J]. 中国临床药理学与治疗学, 2010, 15(9): 1040-1045.

WEI Yong-yue, XUN Peng-cheng, ZHOU Min-lin, ZHAO Yang, ZHU Jin-jin, YU Hao, CHEN Feng. Adaptive randomization for clinical trials with delayed response[J]. Chinese Journal of Clinical Pharmacology and Therapeutics, 2010, 15(9): 1040-1045.

参考文献

[1] Thompson WR. On the likelihood that one unknown probability exceeds another in the revises of the evidence of the two samples[J]. Biometrika, 1933, 25(3/4): 285-294.
[2] ROBBINS H. Some aspects of the sequential design of experiments[J]. Bull Amer Math Soc, 1952, 58(5):527-535.
[3] ANSCOMBE F. Sequential medical trials[J]. J Am Stat Assoc, 1963, 58(302):365-384.
[4] Colton T. A model for selecting one of two medical treatments[J]. J Am Stat Assoc, 1963, 58(302):388-401.
[5] Cornfield J, Halperin M, Greenhouse SW. An adaptive procedure for sequential clinical trials[J]. J Amer Statist Assoc, 1969, 64(325):759-770.
[6] FDA. Guidance for Industry: Adaptive Design Clinical Trials for Drugs and Biologics (Draft Guidance). Available at http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/
[7] Chow SC, Chang M. Adaptive design methods in clinical trials-a review[J]. Orphanet J Rare Disease,2008, 3(11):1-13.
[8] Paulo Guimara ES, Palesch YK. Power and sample size simulations for randomized play-the-winner rules[J]. Contemporary Clinical Trials, 2007, 28(4): 487-499.
[9] Hu F, Rosenberger. The theory of response-adaptive randomization in clinical trials[M]. Wiley 2006.
[10] Zelen M. Play the winner rule and the controlled clinical trials[J]. J Am Stat Assoc, 1969, 64(325):131 -146.
[11] Athreya KB, Karlin S. Embedding of urn schemes into continuous time Markov branching processes and related limit theorems[J]. Ann Math Statist, 1968, 39(6):1801-1817.
[12] Wei LJ, Durham S. The randomized play-the-winner rule in medical trials[J]. J Am Stat Assoc, 1978, 73(364):840-843.
[13] William F, Rosenberger WF. New directions in adaptive designs[J]. Statistical Science, 1996, 11(2):137-149.
[14] Caroline C, Morgan D, Stephen C. A comparison of adptive allocation rules for group-sequential binary response clinical trials[J]. Statistics in Medcine, 2007, 26 (9) :1937-1954.
[15] 刘晓燕,陈峰,于浩,等.响应-自适应随机化分组方法[J]. 临床药理学,2008,13(6):684-689.
[16] Feifang HU, Zhang LX, Xuming HE. Efficient randomized-adaptive designs[J]. The Annals of Statistics, 2009, 37(5): 2543-2560.
[17] Treanor JJ, Hayden FG, Vrooman PS, et al. Efficacy and safety of the oral neuraminidase inhibitor oseltamivir in treating acute influenza: a randomized controlled trial[J]. JAMA, 2000, 283(8):1016-1024.
[18] Wei LJ. Constructing exact two-sample permutational tests with the randomized play-the-winner rule[J]. Biometrika, 1988, 75(3): 603-606.
[19] Bandyopadhyay U, Biswas A. Delayed response in randomized play-the-winner rule: a decision theoretic approach[J]. Calcutta Statist Assoc Bull,1996, 46(1): 69-88.
[20] Bai ZD, Hu F, Rosenberger WF. Asymptotic properties of adaptive designs for clinical trials with delayed response[J]. Ann Statist, 2002, 30(1): 122-139.
[21] Hu F, Zang LX. Asymptotic normality of urn models for clinical trials with delayed response[J]. Bernoulli, 2004, 10(3): 447-463.
[22] Fleiss JL. Statistical methods for rates and proportions. 3nd Edition[M]. New York: John Wiley. 1981:64-83.
[23] Dupont WD, Plummer WD. Power and sample size calculations for studies Involving linear regression[J]. Control Clin Trials, 1998, 19(6):589-601. Software is available at:http://biostat.mc.vanderbilt.edu/wiki/Main/PowerSampleSize
[24] Zhang LX, Chan WS, Cheung SH, et al. A Generalized drop-the-loser urn for clinical trials with delayed responses[J]. Statist Sin, 2007, 17(1): 387-409.