Abstract

Pain is a subjective experience, and as such, pre-clinical models of human pain are highly simplified representations of clinical features. These models are nevertheless critical for the delivery of novel analgesics for human pain, providing pharmacodynamic measurements of activity and, where possible, on-target confirmation of that activity. It has, however, been suggested that at least 50% of all pre-clinical data, independent of discipline, cannot be replicated. Additionally, the paucity of “negative” data in the public domain indicates a publication bias, and significantly impacts the interpretation of failed attempts to replicate published findings. Evidence suggests that systematic biases in experimental design and conduct and insufficiencies in reporting play significant roles in poor reproducibility across pre-clinical studies. It then follows that recommendations on how to improve these factors are warranted.