So -- Somebody did a controlled experiment and the results differed from previous observational studies.

And the author wants us to believe the observational studies?

At the very least the author could provide an experimental design they feel would be valid.

I think the author needs to read more Ben Goldacre.