How the grades and the match are made — and what they cannot tell you.

Principles

• Structured, not generated. There is no language model anywhere in the pipeline. Reports and matches are assembled from database fields and fixed rules, so the same inputs always give the same output.
• Human evidence first. Animal and laboratory data never raise a grade above “Insufficient”. Marketing claims and anecdote are not evidence.
• Regulatory status is recorded, never inferred. If we have not confirmed a status from a regulator source, the entry says “Unclear”.
• The same grade whether or not we sell it. Grades are assigned to compounds, not pens, and a pen page shows the compound’s grade unchanged. The 24 pens are graded by the same criteria as the 11 compounds we do not sell.
• The matcher can return nothing. A match is a product of the rules, not a guarantee of one. When the rules rule everything out, the Checkup says so.
• Clinician-reviewable. Every compound record lists its sources; every flag in a report names the answer that produced it; every match lists the reasons it was made.

Each compound carries an evidence grade for each goal it has been studied for, and an overall grade for its most-researched use. The two are deliberately separate: a licensed medicine with strong evidence for one indication may have only limited evidence for the goal you care about, and the report shows both.

How a grade is assigned

We start from the highest-quality human evidence available for that goal: randomised controlled trials and regulator assessments first, then controlled human studies, then early-phase or observational human data. We consider size, duration, whether the population studied resembles people likely to use the site, whether the outcome measured is the one the goal describes, and whether findings have been replicated. Evidence for a related but different use — visceral fat when the goal is scale weight, for example — is graded “Limited” at most.

The human-evidence ladder

Alongside the grade, every compound is placed on a ladder that describes what kind of human evidence exists at all:

1. Approved medicine (for specific indications)
2. Late-stage clinical trials
3. Early-stage human studies
4. Food supplement with human trials
5. Topical cosmetic use
6. Animal / laboratory data only

The dosing studies quoted in reports and on compound pages are the ones that support the grade, with their citation, population, duration, amounts, outcomes and adverse events.

For each compound we record a status in 5 jurisdictions — United Kingdom, United States, European Union, Australia, Canada — plus a general entry for other jurisdictions. Each entry has four parts: the status, a one-sentence summary, a longer explanation of the licensed indication and any caveats, and the date it was last reviewed.

How it is maintained

• Entries are written and updated by a person reading regulator sources — marketing authorisations, product information, safety communications and trial registries — not by software.
• The last-reviewed date is shown wherever a status appears. A status is never carried over from one jurisdiction to another or inferred from a compound’s class.
• The report uses the jurisdiction for the country you enter. The United Kingdom, our home market, has its own entry (United Kingdom) and the pen page shows it first. Where the country is not one of the 5 named jurisdictions, the report uses the general entry and tells you to check with your national regulator. The pen page says the same.
• Anti-doping status under the World Anti-Doping Code is recorded separately — prohibited at all times, in competition only, or not listed — and shown on compound pages, in comparisons, in reports and on the match.

The engine takes your structured answers and each compound’s record and applies a fixed set of checks. It produces flags — each with a severity (note, caution or review required), a source (medical history, medication, pregnancy, dose, regulatory, stack and so on), a title and a plain-English detail — and then one of three suitability labels per compound.

The checks

1. Contraindications. Each condition you mark as current, previous or unsure is checked against the compound’s contraindication list, which grades every entry as absolute or caution. Unsure answers are carried into the report so a clinician can clarify them.
2. Interactions. Each medicine you enter is matched to a class (insulin, anticoagulant, thyroid hormone and so on) and checked against the compound’s documented interactions, graded major, moderate or minor.
3. Pregnancy, breastfeeding and age. Applied from the compound’s pregnancy field and the age you enter.
4. Amount comparison and blend analysis, described below.
5. Regulatory and anti-doping. Status for your country, and prohibited status where you compete in tested sport or your goal is athletic performance.
6. Expectations, history and source. A timeframe shorter than trials measured, a previous adverse reaction, or treatment stopped for adverse effects each becomes a flag.

From flags to a label

Where your answers identify factors that may make an option inappropriate, or that require professional assessment before proceeding, the label is Higher concern — and the matcher will not land on a pen containing that compound. Where your goal overlaps the compound’s researched use, the label is Potentially relevant, still with every identified factor listed. Where the record or your answers do not contain enough to assess, the label is Insufficient information. The labels are not a ranking of how “good” a compound is and are not a clinical determination.

The matcher runs after the report and reads only two things: the report and the catalogue. It is deterministic — the same answers always land on the same pen — and it knows nothing about price, margin or stock. Each of the 24 pens is scored on five inputs, in this order of weight:

The output is a primary match (the best-scoring pen that is not ruled out), up to two alternatives, and the list of pens that were ruled out with the reason for each. The result routes you to the pen’s page, where the “Your match” panel shows the reasons, the evidence grade for your goal, the cautions to review and the price and lot — and adds nothing to your cart until you ask it to. Where nothing survives the safety verdict, or nothing in the range is listed for your goal, there is no primary match and the panel says so.

Dose comparison

Dosing on this site is research information, not a recommendation. We report what published human studies used — population, duration, route, dose escalation, outcomes and adverse events — because that is the only honest basis for a conversation about amounts.

If you enter an amount you are considering, the engine converts it to the same units as the study exposures on record and returns one verdict: within range, above range, below range, different frequency, different route, no human data to compare against, or not comparable. Where the verdict is above range or there are no human data, the report asks you to seek professional review before making a decision. We do not publish “protocols”, and pen pages state the amount in the pen, not what to dial.

Blends and stacks

Wolverine, Klow, Glow and CagriSema are blends. The report treats each component as a compound in its own right — its own record, grade, contraindications and interactions — and then checks each pair against the database’s stack notes: whether the combination has been studied in humans (studied, limited or none), whether two components act on the same pathway, and any overlapping considerations. It counts the evidence gaps and returns an uncertainty rating — low, moderate, moderate–high or high — with a plain-English summary. No blend in the range has human data as the combination sold here, and the report says so rather than averaging the components into a grade that looks better than any of them.

The store and the grades

We sell 24 pens containing 21 compounds from this database, and we have an interest in selling them. Three things keep that interest away from the grades.

• Grades attach to compounds, not pens. A pen page reads the compound record; it cannot carry a different grade, and the compound page is linked from every pen.
• Neither the engine nor the matcher knows what is in stock. Suitability labels are computed from your answers and the record; the match is computed from the labels and the catalogue’s goal and problem fields. Price, margin and stock are not inputs.
• The database is larger than the range. 11 of the 32 compounds are ones we do not sell, graded by the same criteria, and the report names them where they are researched for your goal.

Limitations

• The assessment is only as complete as your answers. Categories are easier to evaluate than essays, but they cannot capture everything a clinician would ask. The completeness section tells you what was skipped or missing.
• Medicines are matched by class. A medicine we cannot match to a class is listed in your report for a clinician to review, but the engine cannot check it for interactions.
• Evidence grades are judgements, made consistently against written criteria, but judgements nonetheless. Two reviewers could reasonably differ at the boundary between grades.
• A match is a fit to the record, not a prediction. The matcher says which pen the evidence and your answers point to. It does not say the pen will do anything.
• Regulatory status has a review date. It may have changed since.
• Testing describes the pen, not the outcome. A certificate confirms identity, purity and endotoxin for a lot. It says nothing about whether the compound works or is safe in people; only the evidence grade speaks to that, and for most of the range it says “Insufficient” or “Limited”.
• It is not a substitute for a consultation. Nothing here diagnoses, treats or replaces professional advice.

Update cadence

Compound records and regulatory entries are reviewed on a rolling basis, and immediately when a regulator issues a safety communication, a licensing decision or a shortage notice that affects a compound in the range. Each record and each regulatory entry shows its own last-reviewed date.

Corrections

If you believe a grade, a regulatory entry, a study citation, a rule or a match is wrong, email hello@comparepeptide.co.uk with the compound or pen, the field and, ideally, the source you are relying on. Corrections are made to the database, so they reach the compound pages, pen pages, comparisons, matches and future reports at the same time. You can also read who we are and how to use the site safely.