Krummenauer Frank
Institut für Medizinische Biometrie, Epidemiologie und Informatik der Universität Mainz.
Klin Monbl Augenheilkd. 2002 Dec;219(12):896-8. doi: 10.1055/s-2002-36941.
Both an extensive data description and an explicit assessment of a study result's statistical significance should be presented in the result section of a clinical trial report. Whereas the description illustrates the order and clinical relevance of the study findings, the statistical significance describes its generalizability to patients not included in the clinical trial: Despite the random recruitment of patients into a trial, the study results may fail to represent clinical reality (for example the trial might show falsely positive efficacy findings, whereas in "clinical reality" efficacy appears rather limited). A p value measures the statistical significance of a study result -- the smaller the p value turns out, the fewer statistical evidence for "random" findings. The clinical relevance of study findings can be measured by median differences or absolute and relative risks, for example, between two therapeutic strategies. Compact table structures for the simultaneous representation of clinical relevance and statistical significance in a paper's result and discussion section are proposed.
临床试验报告的结果部分应同时呈现详尽的数据描述和对研究结果统计学显著性的明确评估。数据描述阐明了研究结果的顺序及临床相关性，而统计学显著性则描述了其对未纳入临床试验患者的可推广性：尽管患者是随机纳入试验的，但研究结果可能无法代表临床实际情况（例如，试验可能显示出假阳性疗效结果，而在“临床实际”中疗效可能相当有限）。p值衡量研究结果的统计学显著性——p值越小，“随机”结果的统计证据就越少。例如，两种治疗策略之间的研究结果的临床相关性可以通过中位数差异、绝对风险和相对风险来衡量。本文提出了在结果和讨论部分同时呈现临床相关性和统计学显著性的紧凑表格结构。
