Methodology

How I decide what counts as evidence, and how I rank what I find. This page is the audit trail for every claim in every article.

Source hierarchy

I weight sources roughly in this order, strongest first:

1. FDA review documents — NDA / BLA primary review, Medical Officer's review, sponsor briefing documents. The single richest source on safety and efficacy at the time of approval.
2. Randomized controlled trials, Phase 3 — large-N pivotal trials, ideally with active comparator and pre-registered endpoints on ClinicalTrials.gov.
3. Pre-registered Phase 2 RCTs — with attention to whether the primary endpoint was hit and what was reported as exploratory.
4. Systematic reviews and meta-analyses — useful for synthesis, but only as good as the underlying trials. I read the funnel plot.
5. Single-arm or open-label clinical studies — informative for safety signal and pharmacokinetics, weak for efficacy.
6. Mechanistic and preclinical work — explains how something might work, not whether it works in humans.
7. Case reports and surveys — hypothesis-generating only.

What I cite, and how

Every empirical claim ties to one of: a PubMed identifier (PMID), a ClinicalTrials.gov number (NCT), a DOI, or an FDA application number (NDA / BLA). Where possible the link goes directly to the document so the reader can verify without my framing in the way.

I do not treat secondary citations (a review citing a trial I haven't read) as equivalent to the trial itself. If a claim matters, I read the trial.

Conflict-of-interest handling

When I write about a sponsor-funded trial, I name the sponsor in the article. When the only published data on a compound comes from a single research group, I say so. When a study's primary endpoint is missed and a secondary endpoint is promoted, I describe the original protocol.

I do not accept compensation from manufacturers, distributors, telehealth platforms, or compounding pharmacies. The only commercial relationships involve unpaid editorial cross-publishing on PeptideHackerLab, which I write for.

What I avoid

• Dosing guidance. Dose recommendations belong with a clinician, not a writer. I describe the doses studied in trials. I do not extend that into "what to take."
• Sourcing guidance. Where to buy a compound is not a research question. I do not name vendors or compounding pharmacies.
• Off-label inference for indications without controlled human data. If the evidence base for a use case is preclinical only, that is what I say.
• Anecdotal claims as evidence. Forum reports and influencer accounts are interesting as patient-reported outcomes; they are not a substitute for trial data.

Update policy

When new controlled data contradicts an earlier piece, I update the piece in place and add an update note. I do not silently rewrite history. The git history of every published article is publicly viewable through PeptideHackerLab's repository.

What corrects an article fastest

A primary source. If you can show me a PMID or trial result that I missed or misread, I will fix the article and credit the correction. The slowest path is "I know someone who" or "I read on a forum." The methodology constraint is the same one I apply to my own work.