Deep Dives › Ozone and Regenerative Therapies
How to judge a therapy: building an evidence map that survives contact with marketing
A transferable method for working out what any therapy is actually supported for: the hierarchy of evidence and its real limits, relative against absolute effect, what a confidence interval says that a p-value cannot, surrogate outcomes, why procedural trials are the hardest to read, heterogeneity, small-study effects, and a three-pass checklist you can run on any claim.
Reviewed by Team SOMA
Last updated August 2026 · English
What you'll be able to do
- âś“The hierarchy of evidence, and the point at which it stops being useful
- âś“Relative against absolute effect, and why the difference is usually the whole story
- âś“What a confidence interval says that a p-value cannot
- âś“Surrogate outcomes, and when one is worth believing
- âś“Why procedural trials are the hardest of all to read
- âś“Heterogeneity, small-study effects, and a three-pass checklist to run on any claim
What's inside
Is this for you?
Who it's for
- Anyone with a result in hand who wants the reasoning, not only the conclusion
- People with no medical training - every technical term is explained the moment it appears
- Practitioners who want the thinking behind a number rather than a protocol
What you need
- No prior training. Every technical term is explained in plain words the first time it appears.
- A willingness to read something twice.
About this Deep Dive
Most claims made for a therapy are not false. They are true about something narrower than the sentence implies. Finding that gap is a method rather than a judgement, and it transfers to any therapy at all.
What this covers
- The hierarchy of evidence, and the point at which it stops being useful
- Relative against absolute effect, and why the difference is usually the whole story
- What a confidence interval says that a p-value cannot
- Surrogate outcomes, and when one is worth believing
- Why procedural trials are the hardest of all to read
- Heterogeneity, small-study effects, and a three-pass checklist to run on any claim
How it is built
Five parts, taken in order: a short orientation, the reading itself, one worked case, a check to see whether it landed, and a page on what changes once you have it. About forty minutes, and it is written to be read twice. There is no certificate here — that is what the Certified Programmes are for.
Ratings
No ratings yet.