de Lorgeril M, Salen P, Caillat-Vallet E, Hanauer M T, Barthelemy J C, Mamelle N
Laboratoire de Physiologie, Faculté de Médecine J. Lisfranc, Saint-Etienne, France.
Eur J Clin Nutr. 1997 Feb;51(2):116-22. doi: 10.1038/sj.ejcn.1600374.
A major limitation of dietary trials is that double blind design is not feasible. These trials are therefore prone to biases. The Lyon diet heart study is a single-blind secondary prevention trial to test the hypothesis that a Mediterranean-type of diet may prevent recurrences after a first myocardial infarction. A surprising 73% reduction of the risk of new major cardiac events was observed in the experimental group. For this reason, it is important to describe the methods used in the trial. We now report our techniques to randomize the patients, to change their diet and to control for possible bias, in particular any investigator or attending physician bias.
In this dietary trial, a specific design was used to recruit and randomize the patients without informing them and their physicians that they were participating in a comparative trial. The attending physician bias was evaluated by studying drug usage and the investigator bias by constructing a questionnaire from which specific scores were used to evaluate (1) how the patients appreciated their participation in the study and (2) whether this participation resulted in significant changes in their way of living.
605 survivors of a first myocardial infarction were randomized into either a control or a Mediterranean group.
The two randomized groups were similar for all the variables of prognosis. Drug usage was not significantly different between groups, suggesting that there was no major attending physician bias. Analyses of the appreciation scores and of the change score did not detect any significant investigator bias.
Although the study cannot be completely shielded from minor biases, the data presented here provide evidence that the dietary modifications per se were protective, not other (including psychosocial) changes resulting from the participation to the trial.
饮食试验的一个主要局限性在于双盲设计不可行。因此，这些试验容易产生偏差。里昂饮食心脏研究是一项单盲二级预防试验，旨在检验地中海式饮食可预防首次心肌梗死后复发的假设。在实验组中，新的主要心脏事件风险令人惊讶地降低了73%。因此，描述该试验中使用的方法很重要。我们现在报告我们使患者随机分组、改变其饮食以及控制可能偏差的技术，特别是任何研究者或主治医生的偏差。
在这项饮食试验中，采用了一种特定设计来招募患者并使其随机分组，而不告知他们及其医生他们参与的是一项对照试验。通过研究药物使用情况来评估主治医生的偏差，通过构建一份问卷来评估研究者的偏差，该问卷使用特定分数来评估：（1）患者对参与研究的感受如何；（2）这种参与是否导致他们生活方式的显著改变。
605名首次心肌梗死幸存者被随机分为对照组或地中海组。
两个随机分组在所有预后变量方面相似。两组之间的药物使用情况无显著差异，这表明不存在主要的主治医生偏差。对满意度评分和变化评分的分析未发现任何显著的研究者偏差。
尽管该研究无法完全避免微小偏差，但此处呈现的数据提供了证据，表明饮食改变本身具有保护作用，而非参与试验导致的其他（包括心理社会方面的）变化。
