非双盲随机临床试验中降压药物疗效组别偏倚中国高血压临床研究荟萃分析

目的了解非双盲临床试验中同一降压药物作为试验药物和对照药物时,被观察的疗效结果是否存在差别.方法按照一定的标准,通过《中国医院知识仓库》收集降压药物的随机对照试验,寻找同一药物作为试验组的研究与作为对照组的研究进行配组,在配组的用药剂量、疗程和疗效标准相同的情况下,比较两组疗效的差别.结果寻找到11个配组,涉及18项试验药物研究和19项对照药物研究.①同一降压药物作为试验药物的各项研究与作为对照药物的各项研究的总有效率的差异具有统计学意义;②试验药物与对照药物的各配组的总有效率的差异具有统计学意义;③试验药物与对照药物的各药物的总有效率的差异没有统计学意义.试验药物与对照药物的加权平均总有效率分别为85.38%和77.17%,试验药物比对照药物高约10%.结论在非双盲随机对照试验中,试验组降压药物的疗效可能被夸大,对照组被贬低.应重视这种偏倚在新药临床试验和Meta分析中的影响.

Grouping-related bias of effects of the same anti-hypertensive drugs in non-doubleblind randomized trials: a meta-analysis in Chinese hypertension clinical literature

Objective This literature review investigated whether the reported efficacy was comparable or not for the same anti-hypertensive drug when used as treatment agent or reference agent in randomized clinical trials in China. Method The randomized clinical anti-hypertensive trials for the treatment of mild to moderate hypertensive patients in Chinese-language journals between 1994 and 2004 were identified by searching the China National Knowledge Infrastructure in China Hospital Knowledge Database web (CNKI-CHKD). A clinical trial was selected for the present review if the same drug in the same dose was used either as treatment agent or reference agent in the trial with identical therapy duration and effects evaluation criteria. The anti-hypertensive efficacy of the same drug as a treatment drug or reference drug was compared.Results Six drugs (bisoprolol, benazepril, enalapril, felodipine, amlodipine and losartan) were used in 37 clinical trials and used as treatment drug in 18 trials and as reference drug in 19 trials. The rank sum test indicated that there is a significant difference in the total effective rate between the treatment groups (85.38%, n=18) and the reference groups (77.17%, n=19), P<0.05. The effective rate for the same drug was significantly higher when it was used as treatment drug than that used as reference drug (P<0.05). Conclusion The anti-hypertensive effect of the same drug is often overestimated when used as a treatment agent and underestimated when used as a reference agent in randomized clinical trials. This grouping-related efficacy bias should be considered when evaluating the effects of a tested anti-hypertensive drug in clinical trials.