The point is that this qualifier - that research must be "worth publishing in a reputable journal" - imposes significant constraints on what kind of research may be attempted to begin with.

An hypothesis that is too far outside of mainstream accepted thinking may not be studied, if it seems (to the people proposing the study or the people considering funding it) like it will be too hard to get it through peer review or editorial review.

I'm not suggesting this is necessarily a flaw in the system; it is healthy that there are constraints on what kind of research gets funded, and for things that are way outside of established thinking, the bar should be higher [1].

But it does mean there will inevitably be a gap - quite possibly a very big gap - between what has been "proved to work" and what can be effective for those who want or need (like myself) to take matters into their own hands.

[1] Problems like politicisation, egotism and bullying are often said to be rife in scientific research institutions, so it's questionable whether the system is as healthy as it could be, but that's a separate discussion.

Hand wavy recollection of what a very knowledgeable cancer patient told me years ago: They were incredibly frustrated that most cancer research related to the type of cancer they had was variations on the same thing because it was the trendy thing to fund. This person was hoping for real progress, but was seeing just dozens of studies of the same thing with minor variations and nothing actually interesting or really new being done.

That could just as well be confusing cause and effect. Particular avenues of research can capture a lot of funding and become trendy because they seem most promising at the time. Hard decisions also have to be made on risk versus reward; should we shoot for the moon or target limited incremental progress? It's easy to criticize after the fact with the benefit of hindsight.

The individual in question worked for a major health organization themselves. I don't think they were confusing anything.