How to read peptide research: evidence grades, explained

By The Peptide Informative editorial team · 3 Nov 2026 · 8 min read

Why a rat study is not a human result, what “very low certainty” means, and how we grade every peptide in the library.

Search any peptide name and you'll find confident claims. Fewer of them tell you what kind of evidence sits behind the claim. This post explains how we grade evidence across the library, and the questions to ask of any study you read.

1. Ask first: who or what was studied?

The biggest single gap in peptide information is between animal results and human results. Take BPC-157. Of the 987 study results PepBank has extracted, 92% come from animals and 7% from lab cells (PepBank, data as of 30 Oct 2026). A 2026 systematic review found only three human studies with 30 participants in total (Poolton et al., Acta Orthop Belg 2026).

Compare semaglutide. One trial alone, STEP 1, randomized 1,961 adults in a double-blind design for 68 weeks (Wilding et al., NEJM 2021). Both peptides are popular, but their evidence bases are very different.

2. Know the evidence ladder

• Lab (cells): shows a mechanism is possible. It says little about effects in a whole body.
• Animal: shows effects in another species, often at doses and by routes that don't map neatly onto people.
• Human, uncontrolled: case series, chart reviews and pilots. These are useful for generating hypotheses, but they are open to bias, placebo effects and recall error.
• Human randomized controlled trials: participants are randomly assigned to treatment or control, ideally double-blind. This is the strongest single design.
• Approved drug: a regulator has reviewed the whole package for a specific use, and the product has a label that lists its risks.

3. Read the methods, not just the conclusion

One BPC-157 knee-pain study reached patients by phone 6 to 12 months after their injection and asked them to rate their pain from memory. In the authors' words, “no specific tools were used to measure their improvement” (Lee & Padgett 2021). The headline (14 of 16 reported relief) sounds strong. The design means it can't tell a real effect apart from placebo, natural recovery or recall bias.

4. Check for one-lab literature

When most papers on a compound come from one group, independent replication matters even more. An independent count of Europe PMC found that 174 of 224 BPC-157 records (about 78%) list the same senior author (Certified Research, Sep 2026). Also read author affiliations. The best-known GHK-Cu reviews list a skin-care company's R&D department (Pickart & Margolina 2018).

5. Objective versus subjective outcomes

In a small randomized GHK-Cu study after laser skin resurfacing, blinded evaluators found no objective difference in wrinkles or skin quality, but patients using GHK-Cu rated their satisfaction higher (Miller et al. 2006). Both findings are real. They answer different questions.

6. A dose in a study is not a recommendation

Our profiles list doses only as citations: who was studied, the route, the duration and the result. A dose that appeared in a two-person pilot (Lee & Burgess 2025) tells you what that study did. It does not tell you that the dose is safe, effective or right for anyone.

Your quick checklist

1. Humans, animals or cells?
2. Was there a control group, and was it randomized and blinded?
3. How many participants, and for how long?
4. Objective or self-reported outcomes?
5. Who funded it, and has anyone else replicated it?
6. Is it registered on ClinicalTrials.gov, and were results reported?