However, these views are outdated (data tends to run to 2010), often fail to account for start-ups (i.e. by following a pre-existing cohort) and are an extrapolation of a trend.
The pictured chart is an updated graph of the number of new drugs or new molecular entities (NMEs) per $bn of R&D spend.
Interestingly it has actually been stable for much of the 2000s and can be explained in part by the industry having better information and how it is used. This article explains in depth.
