The four gates
Every candidate protocol passes through the same four gates, in order. Most things people call protocols fail at gate one or gate two, which is the point.
1. Claim, isolated
A protocol enters as a sentence with a population, a dose, a duration, and an outcome. “Cold plunges are good for you” is not a claim. “In healthy adults, 11 minutes of weekly cold-water immersion reduces self-reported stress at 6 weeks” is. Anything that cannot be written this way is rejected before grading.
2. Evidence, weighed
We rank human randomised trials above cohorts, cohorts above mechanism, and mechanism above testimonial. For each entry we record sample size, follow-up length, who funded it, and whether an independent group reproduced it. Absence of replication is recorded as a fact, not smoothed over.
3. Cost, counted
A protocol you cannot sustain is not a protocol. Every entry carries hours per week, money per month, and the adverse effects reported in the trial arms — including dropout, which is often the most honest number in a paper.
4. Graded, then re-graded
Grades are dated, not permanent. Entries are re-reviewed on a rolling schedule and whenever a substantial new trial lands. Every movement is written into the entry’s changelog with the reason.
The A–D rubric
| Grade A | Multiple randomised controlled trials in humans, consistent direction of effect, at least one independent replication. Effect is real; size may still be debated. |
|---|---|
| Grade B | One well-conducted trial, or several consistent observational cohorts. Probably real, size uncertain, replication incomplete. |
| Grade C | Mixed or underpowered trials, short follow-up, or reliance on surrogate markers. Plausible and unproven. |
| Grade D | Mechanism, animal models, or anecdote only. Listed so the gap between the marketing and the evidence is visible. |
A low grade is not a warning and a high grade is not a recommendation. Grade A says the research is solid, not that the protocol is right for you.
Independence and funding
- No sponsored entries, and no paid placement in the index.
- No grade has ever been altered for a commercial reason. Grades are not for sale.
- No advertising, no ad networks, and nothing sold from these pages.
- Google Analytics is used to count visits. Nothing else is loaded from another origin.
Protocol Index has a financial relationship with Nalu Labs, which sells peptides. Where an article names a supplier it links there, and the link is marked as commercial at the point of use. There are no other paid links on the site.
That is a conflict of interest, and naming it is more useful than pretending otherwise. Where the wall sits: entries are selected and graded from the published literature before any commercial consideration, most of the index is protocols nobody sells anything for, and a great many graded entries cost nothing to run. If an entry's grade looks bent toward a product, send the citation and the correction gets published.
The site is funded by its author. If that changes, this section changes first and the change is dated.
Corrections policy
Errors get fixed in place, and the fix is logged in the entry’s changelog with the date and what changed. Substantive corrections — anything that moves a grade or a dose — are noted at the top of the entry itself. We do not quietly edit and move on.
If you think an entry is wrong, send the citation. Arguments without one are read but not acted on.
What this site is not
Protocol Index is a reference, not a clinician. Nothing here diagnoses, treats, or prescribes. Evidence grades describe published research in populations — they say nothing about your medications, your history, or your risk. Talk to a qualified professional before starting, stopping, or changing anything, and particularly if you are pregnant, taking medication, or managing a diagnosed condition.
Contact
Corrections, citations, and protocol suggestions: hello@protocolindex.org. Every email with a citation attached gets read.