McPherson K
Department of Public Health and Policy, London School of Hygiene and Tropical Medicine.
J Epidemiol Community Health. 1994 Feb;48(1):6-15. doi: 10.1136/jech.48.1.6.
This lecture aimed to create a bridge to span the conceptual and ideological gap between randomised controlled trials and systematic observational comparisons and to reduce unwanted and unproductive polarisation. The argument, simply put, is that since randomisation alone eliminates the selection effect of therapeutic decision making, anything short of randomisation to attribute cause to consequent outcome is a waste of time. If observational comparison does have any significant part in evaluating medical outcomes, there is a grave danger of "the best", to paraphrase Voltaire, becoming "the enemy of the good". The first section aims to emphasise the advantages of randomised controlled trials. Then the nature of an essential precondition--medical uncertainty--is discussed in terms of its extent and effect. Next, the role of patient choice in medical decision making is considered, both when outcomes can safely be attributed to treatment choice and when they cannot. There may be many important situations in which choice itself affects outcome and this could mean that random comparisons give biased estimates of true therapeutic effects. In the penultimate section, the implications of this possibility both for randomised controlled trials and for outcome research is pursued and lastly there are some simple recommendations for reliable outcome research.
本次讲座旨在搭建一座桥梁，跨越随机对照试验与系统观察性比较之间的概念和思想鸿沟，并减少不必要且无成效的两极分化。简而言之，论点是由于仅随机化就能消除治疗决策中的选择效应，因此任何非随机化的归因因果关系的做法都是浪费时间。如果观察性比较在评估医疗结果中确实有任何重要作用，那么用伏尔泰的话说，就存在“最好”成为“好的敌人”的严重风险。第一部分旨在强调随机对照试验的优势。然后从其程度和影响方面讨论一个基本前提条件——医学不确定性的本质。接下来，考虑患者选择在医疗决策中的作用，包括结果可安全归因于治疗选择的情况以及不可归因的情况。可能存在许多重要情形，其中选择本身会影响结果，这可能意味着随机比较会对真实治疗效果给出有偏差的估计。在倒数第二部分，探讨了这种可能性对随机对照试验和结果研究的影响，最后给出了一些关于可靠结果研究的简单建议。
