The Peptide Evidence index (PEI v0.1) is the 0 to 100 score the safety report attaches to every peptide entry. This page documents the five sub-scores, the weighting, and what the number does and does not say.
The PEI measures the strength of the human evidence base for a peptide in the indications covered by the field reference and the safety report. It does not measure efficacy in a single narrow indication, it does not measure cost, and it does not measure safety. A high PEI means the published record in humans is large, well-replicated, recent, and population-diverse. It does not mean the peptide works for you, and it does not mean it is safe.
The PEI is recalculated quarterly. When a major trial is published or a major citation is removed, the score moves with the evidence. The current version is v0.1; the methodology is being refined and the v0.1 label is honest about the limits of the system.
The PEI is the sum of five sub-scores, each scored 0 to 20. The total runs 0 to 100.
This sub-score combines the number of published randomised controlled trials with the methodological quality of the largest and most-cited ones. A peptide with three well-powered, double-blind, placebo-controlled trials with pre-registered protocols scores near the top. A peptide with one small open-label pilot scores low. Animal studies, case reports, and retrospective series do not count toward this sub-score, though they may inform the plausibility grade.
This sub-score measures whether the published findings replicate across independent groups and whether effect sizes are consistent across trials. A peptide whose effects are reported by multiple independent groups at consistent magnitudes scores well. A peptide whose single positive trial has not been replicated scores poorly. Failures to replicate pull the score down faster than confirmations pull it up, because the literature is biased toward positive publication.
This sub-score asks whether the demonstrated effect, on the published record, would be clinically meaningful to a real patient. A 30 percent improvement in a primary endpoint with a tight confidence interval scores well. A statistically significant change in a surrogate marker with no demonstrated clinical correlate scores low. Effect size is the hardest sub-score to standardise, because what counts as clinically meaningful varies by indication, and the v0.1 framework leans on the editorial team to make the call.
This sub-score asks whether the published trials cover the populations a prescriber sees in practice: both sexes, a range of ages, varied baseline health, and adequate representation of the groups most likely to use the peptide in practice. A peptide with trials in narrow demographics (one sex, one age band, one comorbidity profile) scores low even if the trial itself is high quality. A peptide with consistent findings across age, sex, and comorbidity bands scores well.
This sub-score gives more weight to recent publications, with a seven-year half-life. A peptide whose evidence base is mostly from the 1990s scores lower than a peptide with the same evidence size but with a 2020s publication record. The half-life of seven years reflects a balance: peptide science moves faster than most medical fields, and a finding that has not been re-examined in seven years is starting to look stale. The recency sub-score never drops below 4 even for an entirely pre-2000 record, because the older work still counts.
The PEI does not measure safety. A peptide can have a PEI of 80 and a serious safety concern that the safety report flags separately. The plausibility grade (tractable, plausible, speculative, experimental) is a separate editorial call that combines the PEI with the regulatory status and the safety record. The grade is not a function of the PEI alone.
The PEI does not measure practitioner consensus. A peptide that practitioners widely use can score low on the PEI if the published record is thin, and a peptide that the published record supports but that no practitioner reaches for can score high. Practitioner consensus is captured in the field reference under the "Named practitioners" line, not in the PEI.
The PEI does not measure cost, access, or supply chain. A peptide with a strong evidence base and a poor compounding pharmacy record is still high on the PEI. The supply chain risk is in the safety report, not in the PEI.
A score in the 70 to 100 range means the human evidence base is large, well-replicated, and population-diverse. A score in the 40 to 70 range means there is real evidence but it is narrower, less replicated, or older. A score below 40 means the human evidence is thin, the peptide is being used mostly on the basis of preclinical or surrogate-marker data, or both.
For comparison, semaglutide for weight loss scores in the high 80s on the basis of the STEP trial programme. Thymosin alpha-1 scores in the mid-50s after the recent corrections to the evidence base. BPC-157 scores in the 30s, reflecting strong rodent data and weak human data. Dihexa is in the high teens, reflecting preclinical work only.
The PEI v0.1 system has known limitations. The effect size sub-score is the most editorial-dependent. The population diversity sub-score is hard to score for peptides that are mostly used in narrow practitioner populations. The recency sub-score can penalise an older but still valid evidence base.
The PEI does not correct for publication bias. A peptide with a strong positive result in a small trial scores higher than it should, because the system cannot see the unpublished negative results. The v0.2 framework will attempt to adjust for this; the v0.1 label is honest about the gap.
The PEI was not developed with formal psychometric validation. The five sub-scores were chosen on editorial grounds, not on the basis of factor analysis or inter-rater reliability testing. The v0.2 framework will attempt to formalise the sub-score definitions and the weighting; the v0.1 label reflects that this is a working system, not a published instrument.
The PEI is recalculated quarterly. The cadence is in the report footer of every peptide entry as "Source data last refreshed" with a date. When a major trial is published between recalculations, the editorial team notes the new evidence but does not update the score until the next quarterly cycle. The exception is when a load-bearing citation is found to be wrong or fabricated, in which case the score is updated as part of the correction.
If you find a PEI sub-score that does not match the published evidence, flag it via the editorial contact on the about page. The editorial team reviews every correction request and either updates the score with a public note or explains why the score stands.
For corrections: corrections@thepeptideindex.info.