Green S J, Fleming T R, O'Fallon J R
J Clin Oncol. 1987 Sep;5(9):1477-84. doi: 10.1200/JCO.1987.5.9.1477.
Interim analyses of comparative trials are necessary in order to monitor for extreme therapeutic results. However, closing studies and reporting results whenever "trends" appear increases the probability of a false conclusion to well over the desired .05 level. Guidelines for early stopping of comparative trials must be carefully defined to avoid this problem. In addition, to avoid inappropriate early closure of studies due to declining accrual (as investigators draw their own conclusions from early unreliable data), it is recommended that access to interim data be limited to a multidisciplinary monitoring committee responsible for (1) performing and reviewing interim analyses, and (2) deciding when early termination should be considered. Accrual and reporting of studies from two clinical trials groups, one with a policy of limited access to interim data and one without, are compared. The group without monitoring committees had a higher incidence of accrual and reporting problems than the group with monitoring committees.
为监测极端治疗结果，比较性试验的期中分析很有必要。然而，每当出现“趋势”就终止研究并报告结果，会使得出错误结论的概率大幅超过期望的0.05水平。必须仔细界定比较性试验提前终止的准则，以避免这一问题。此外，为避免因入组人数下降导致研究过早不当终止（因为研究人员会从早期不可靠数据中得出自己的结论），建议将期中数据的获取限制在一个多学科监测委员会，该委员会负责(1)进行和审查期中分析，以及(2)决定何时应考虑提前终止。比较了两个临床试验组的研究入组和报告情况，其中一组对期中数据的获取有限制政策，另一组则没有。没有监测委员会的组比有监测委员会的组出现入组和报告问题的发生率更高。
