Quantitative Biology, Discovery Sciences, IMED Biotech Unit, AstraZeneca, Cambridge, United Kingdom.
PLoS Biol. 2018 Mar 5;16(3):e2005413. doi: 10.1371/journal.pbio.2005413. eCollection 2018 Mar.
Translational failures and replication issues of published research are undermining preclinical research and, if the outcomes are questionable, raise ethical implications over the continued use of animals. Standardization of procedures, environmental conditions, and genetic background has traditionally been proposed as the gold standard approach, as it reduces variability, thereby enhancing sensitivity and supporting reproducibility when the environment is defined precisely. An alternative view is that standardization can identify idiosyncratic effects and hence decrease reproducibility. In support of this alternative view, Voelkl and colleagues present evidence from resampling a large quantity of research data exploring a variety of treatments. They demonstrate that by implementing multi-laboratory experiments with as few as two sites, we can increase reproducibility by embracing variation without increasing the sample size.
发表研究的转化失败和复制问题正在破坏临床前研究，如果结果值得怀疑，就会对继续使用动物提出伦理问题。传统上，人们提出标准化程序、环境条件和遗传背景是黄金标准方法，因为它可以减少变异性，从而在环境被精确定义时提高敏感性和支持可重复性。另一种观点认为，标准化可以识别特殊性影响，从而降低可重复性。为了支持这一替代观点，Voelkl 及其同事从重新分析大量研究数据中提供了证据，这些数据探讨了各种治疗方法。他们表明，通过实施具有两个站点的多实验室实验，我们可以在不增加样本量的情况下通过接受变异性来提高可重复性。
