>That means treatments are tested as to if they work, not how they work. Ideally in multiple well run double blinded randomised controlled trials with at least some against treatment

There are some concerning departures from this with medical approvals based not on treatment, but biomarkers.

Without getting too political, two noteworthy recent examples come to mind. The first is abelacimab for alzheimer's, approved based on plaque, not disease progression.

The other is juvenile covid vaccines, approved based on antibody presence.