Spiegelhalter D J
MRC Biostatistics Unit, Cambridge.
Scand J Gastroenterol Suppl. 1988;147:40-5. doi: 10.3109/00365528809099158.
We consider intensive studies of individual response to therapy in a controlled experiment. It is helpful to distinguish between strict 'N = 1' studies, which are pragmatic trials intended to draw conclusions concerning only the patient under consideration, and 'N much greater than 1' studies, which are explanatory trials intended to make more general statements about a treatment with highly variable response, for which aggregate measures of effect on groups are inappropriate. Issues of design, measurement, and statistical significance are discussed, and the rule of permutation tests emphasised. Three published examples are used as illustrations.
我们考虑在对照实验中对个体治疗反应进行深入研究。区分严格的“N = 1”研究和“N远大于1”研究是有帮助的。“N = 1”研究是务实性试验，旨在仅就所考虑的患者得出结论；“N远大于1”研究是解释性试验，旨在对反应高度可变的治疗做出更一般性的陈述，对于这种治疗，对群体的总体效应测量并不合适。本文讨论了设计、测量和统计显著性问题，并强调了置换检验规则。文中使用了三个已发表的例子作为说明。
