How to reduce scientific irreproducibility: the 5-year reflection

摘要

We discuss in depth six causes of scientific irreproducibility and their ramifications for the clinical sciences: fraud, unfounded papers published by prominent authorities, bias, technical deficiencies, fragmented science and problems with big data. Some proposed methods to combat this problem are briefly described, including an effort to replicate results from some high impact papers and a proposal that authors include detailed preclinical data in papers with supposedly high translational value. We here advocate for a 5-year reflection on papers with seemingly high clinical/translational potential, published alongside the original paper where authors reflect on the quality, reproducibility and impact of their findings. These reflections can be used as a benchmark for credibility, and begin a virtuous cycle of improving the quality of published findings in the literature.