If true, it would somewhat undermine another point of the article, which is that since almost all of our interventions are ineffective we need to invest more in understanding which ones actually work.

I'm sure that many other examples could be found to replace this one, but if a lot of the "ineffective interventions" turned out to be effective under closer scrutiny then it would be a blow to the thesis for sure.