A pleasant catch by Max at Most Progress:
[A common critique] is that the NIH funds too many “small crappy trials.” That quote is from a FDA larger up, however the story has been repeated by many others…I downloaded the entire scientific trial knowledge from ClinicalTrials.gov to search out out….The median NIH funded trial has 48 individuals whereas the median trade funded trial has 67. The common NIH funded trial has 288 individuals whereas the common trade trial has 335 and the common “Different” funded trial (largely universities and the related hospitals) has 923 individuals.
By median or by common NIH trials are the smallest out of all of the funders. This appears to substantiate the “small crappy trials” narrative
…This narrative is reversed, nevertheless, while you break up up the trials by part.
Throughout all trials NIH funded ones are the smallest, however inside every part NIH trials are the most important or second largest. Their general small enrollment common is simply on account of the truth that they fund extra Part I trials than Part III. However NIH Part I trials have an even bigger pattern dimension than trade funded trials on common.
That is an instance of Simpson’s Paradox within the wild!
Arguing that the NIH ought to cease funding unusually small trials is straightforward however arguing that they need to shift from funding the Part I trials closest to primary analysis in direction of later stage trials is much less clear.
The NIH’s scientific trial technique is actually not good and bettering it’s beneficial. However a scientific bias in direction of “small crappy trials” doesn’t actually look like it’s an necessary drawback going through the NIH.
