[Skip to Content]
[Skip to Content Landing]
Views 475
Citations 0
Comment & Response
June 20, 2017

Nonreproducibility of Preclinical Research—Reply

Author Affiliations
  • 1Meta-Research Innovation Center at Stanford (METRICS), Stanford, California
JAMA. 2017;317(23):2453. doi:10.1001/jama.2017.5987

In Reply I agree with Dr Goldschmidt that basic and preclinical research, in principle, should be easier to reproduce than clinical research. When I said that “basic and preclinical research probably have a much larger challenge of nonreproducibility,” I meant that unfortunately they show an excess of nonreproducibility problems, not that they are justified to show such an excess. There can be exceptions; eg, some basic and preclinical questions and experimental systems are very complex and may pose special difficulties to reproduce. However, on average it should be easier to replicate an experiment with cell cultures or animal models than with long-term follow-up of humans. The reason that basic and preclinical research is not reproducible is more likely related to correctable deficiencies in research practices than to inherent genuine heterogeneity. For example, the large majority of in vivo animal studies do not use randomization and blinding of investigators in reading their results.1 These are easy, straightforward practices that cost nothing to implement and that would prevent many false results and conclusions. Also, the level of statistical and methodological training of many top scientists in basic and preclinical research is lacking. Again, simple improvements in these skills should make this research far more reproducible.