Making an assessment traceable.
Our assessment record should identify the precise claim, the research population, the intervention and comparator, the outcome, and the literature cut-off date. Broad statements such as “this peptide works” are too vague to receive a meaningful evidence classification.
Search and selection
Record the databases or official sources consulted, relevant names and synonyms, and the search date. Look for contradictory findings and completed trials without available results as well as supportive studies. Explain any important restrictions, such as inaccessible full text or language limits. A narrative educational search should not be described as a systematic review unless it follows and documents the required methods.
Extraction and comparison
For each relevant study, capture its design, population, formulation, comparator, follow-up, outcomes, and funding disclosures. Keep separate reports of the same trial together to avoid counting them as independent confirmation. Compare the published report with its protocol or registry where possible.
Rationale and limits
The final account should explain which findings support the claim, which limitations reduce confidence, and why the chosen classification fits the outcome. If the material is insufficient to complete the assessment, show review pending rather than suggesting a completed judgment. This is our documentation standard, not a claim that every existing article has a publicly downloadable assessment file.
An example of a carefully bounded conclusion.
A compound produces a measurable change in a laboratory assay and an animal model. No relevant controlled human outcomes are available in the material assessed. The appropriate explanation describes the laboratory and animal findings and states that a clinical benefit has not been established. It does not assign a human benefit merely because the proposed mechanism is plausible.
If a later human trial becomes available, reassess its population, comparator, outcomes, precision, and limitations. The existence of one trial does not automatically establish broad benefit, long-term safety, or applicability to a different formulation. An updated classification should include a reason readers can follow.
A classification belongs to a claim.
A peptide can have well-described pharmacology while its proposed clinical benefit remains uncertain. A profile should therefore identify the outcome attached to each evidence badge. Confidence can differ between outcomes, populations and formulations. [2]
They are not formal GRADE ratings, regulatory determinations or numerical scores. GRADE uses a different four-level certainty system. [2]
The rules below are publication standards, not a claim that a staffed review programme already exists. Independent scientific review of this methodology is pending verification.
How an assessment is made.
Define the claim
Record the population, formulation, comparator, outcome and follow-up. Separate mechanism, clinical benefit and harms; do not infer one from another.
Search and document
Search PubMed and relevant trial registries; check FDA/EMA product documents for regulatory claims. Save search terms, search date, source identifiers and reasons for excluding evidence.
Extract the findings
Record design, sample size, duration, endpoints, effect estimates, uncertainty and reported harms. Separate registered plans from actual results and avoid counting multiple reports of one trial as independent studies.
Assess the body of evidence
Document study limitations and conflicting findings. Use the quality checks below; explain which evidence is directly relevant to the claim and which is only supporting context.
Classify with a rationale
Apply the decision rules below to each claim. Keep conclusion direction separate: support, no support, mixed or uncertain. Do not upgrade a claim because a compound is popular or approved for a different use.
Verify before publication
Require a traceable source list, a written rationale and a real reviewer record for a verified assessment. If evidence or reviewer information is missing, keep the assessment pending instead of manufacturing a grade.
Our source-tracing standard draws on study protocols and NLM’s literature records; it does not treat a database listing as an endorsement. [3][4][5]
Source priority depends on the question.
For approval status, use official regulatory documentation. For treatment effects, begin with rigorous human research and its synthesis. The following is a search priority, not an automatic ranking of every paper.
Highest priority for the relevant question
Official product assessments and labels for approval claims; high-quality systematic reviews and well-conducted randomized trials for intervention effects. A review inherits limitations in its underlying studies. [6][7]
High priority
Other controlled human studies and prospective clinical research, assessed for design and relevance rather than publication prestige alone. [1]
Supporting evidence
Observational studies can illuminate real-world associations and harms. Case reports may flag unusual events but cannot establish how commonly they occur. Human pharmacokinetics informs exposure, not proof of a clinical benefit. [2][3]
Early and preclinical evidence
Animal, cellular and laboratory studies can inform mechanisms and research directions. Pure theory is not experimentally demonstrated evidence. [8][9]
Mechanism, diagnosis, treatment benefit and rare harms ask different questions. A study’s relevance to the question matters more than its place in a generic pyramid. [10]
The five classification rules.
Apply these criteria to the specified claim. If major contradictions remain unexplained, do not use Strong Evidence. Neither participant count nor the number of publications is a stand-alone threshold.
Use only when direct human evidence is precise enough for the claim, bias is low, findings are consistent or differences are explained, and material reporting gaps do not undermine the conclusion. No fixed study-count threshold.
Require directly relevant human outcome data and a defensible conclusion. Explain each remaining limitation; do not use this label for animal-only or purely theoretical claims.
Require an identifiable completed human study relevant to the claim. A registered trial without results is not enough. Pharmacokinetic findings alone do not establish a clinical outcome.
Require an identifiable relevant nonclinical experiment. Label the model and measured outcome. Unsupported speculation belongs under insufficient evidence.
Use when a documented search finds no usable evidence, or available studies cannot support a conclusion. Record the search date and reason. “Not yet reviewed” is a separate workflow state.
These are Inside Peptides’ rules. An unassessed claim receives Review pending, not Insufficient Evidence. The latter requires a documented assessment.
Human and preclinical research answer different questions.
| Evidence type | What it can tell us | What it cannot tell us |
|---|---|---|
| In vitro | A response or mechanism in a defined laboratory system. | Clinical effectiveness, whole-body effects or safety in people. |
| Animal | Biological responses and safety signals in a model. | That the same response, benefit or risk will occur in humans. |
| Observational human | Associations, real-world patterns and possible safety signals. | A causal treatment effect without addressing confounding and other biases. |
| Controlled human | A comparison with another group, reducing some alternative explanations. | Freedom from confounding if assignment is not randomized. |
| Randomized controlled trial | A stronger test of a causal intervention effect when properly designed and conducted. | Universal applicability or complete knowledge of rare or long-term harms. |
| Systematic review / meta-analysis | A structured synthesis; pooling can improve precision when studies are sufficiently comparable. | That weak, biased or incompatible studies become reliable simply by combining them. |
Study type is a starting point, not a verdict. Nonrandomized evidence can be especially relevant to rare or long-term harms. [9][1][7][2]
What we check before drawing a conclusion.
Design and comparison
Check study design, control groups, randomization, concealed assignment, blinding, missing data and selective reporting. Blinding is not always possible; explain its relevance to the measured outcome. [1]
Precision and meaning
Consider participant numbers alongside effect estimates and confidence intervals. Separate statistical significance from a benefit meaningful to patients; a non-significant result is not automatically proof of no effect. [10]
Relevance and replication
Check duration, population, formulation and clinically relevant endpoints. Examine consistency across independent studies, not repeated reports of the same participants. [3][7]
Transparency
Record funding, conflicts, publication status and corrections. Examine whether results match the planned outcomes. Peer review and journal reputation do not substitute for checking methods. [1]
For a body of research, document bias, inconsistency, indirectness, imprecision and reporting/publication bias. These domains inform our rationale; we do not imply that a formal GRADE assessment has been performed. [2]
Our claim language must match the evidence.
Preclinical
Use: “Animal studies suggest…” or “In preclinical models…”
Specify the species or model and outcome. Do not write “This peptide heals…” when the evidence is not from an appropriate human study.
Early human
Use: “Early human research has investigated…”
Describe the design and limitation. Do not imply established clinical effectiveness or complete safety.
Stronger human evidence
Use: “Human trials in [population] found [specific outcome]…”
Give magnitude and uncertainty where available. Do not expand the finding to other uses or formulations.
Uncertain or negative findings
Use: “Evidence is insufficient to determine…” or “The study did not support…”
Separate an inconclusive study from persuasive evidence of no meaningful effect.
These are our editorial wording rules. They limit extrapolation from models and keep outcome interpretation within the evidence. [9][10]
Approval and evidence remain separate.
FDA approval evaluates a particular medicine application. In the EU centralised procedure, EMA conducts scientific evaluation and the European Commission grants authorisation. [6][11]
A specified medicine has authorisation from a named regulator for a stated use. Record jurisdiction, product, formulation and indication; an ingredient name alone is not enough.
A documented clinical research programme is studying the compound or use. Cite the study and its current status; this does not mean an approved treatment.
An editorial description of research-stage use. It is not an official approval category and can overlap with investigational or unapproved status.
No relevant medicine authorisation is documented for the specified jurisdiction and use. State the scope and verification date; approval elsewhere may differ.
Investigational and experimental use can also be unapproved. These labels are not a mutually exclusive legal taxonomy. Record approval separately from research stage, with jurisdiction and source.
Approval for one use does not establish every proposed claim. Compounded drugs are not FDA-approved products, and safety monitoring continues after medicine approval. [12][13]
Make every important statement traceable.
Link a claim to the original paper, PubMed record, DOI, trial record or regulatory document. Distinguish protocols, abstracts, completed publications and regulator assessments. A trial registration alone supplies no result. [5][4]
Search for conflicting and negative findings, not just supporting papers. Preserve identifiers and explain inaccessible full text. Where only an abstract is available, disclose that limit; do not invent details absent from the source.
Peptide sellers, affiliate pages, clinics, influencers, social posts and forums are not evidence sources for our scientific claims. A secondary explanation may aid understanding, but should lead back to the underlying source.
Read the Sources Policy and Editorial Policy.
Evidence can change. So can the assessment.
Publication standards require a genuine last-reviewed date and a last-updated date. New material evidence, corrections or regulatory changes should trigger reassessment. Record the reason for a major classification change, update the wording and preserve a change log.
Safety knowledge can evolve after approval. [13] See the Corrections Policy.
Last updated: 10 October 2026. Last reviewed: independent scientific review pending verification. No reviewer identity or qualification has been invented.
References
Official source destinations checked 2026-10-06. External links are marked ↗.
- Cochrane Handbook, chapter 8: Risk of bias in randomized trials (external website)
Randomization, allocation concealment, missing outcomes and selective reporting.
Back to citation ↑ - Cochrane Handbook, chapter 14: Grading certainty of evidence (external website)
Outcome-specific certainty, five uncertainty domains; formal GRADE has four levels.
Back to citation ↑ - FDA: Step 3: Clinical research (external website)
Human study design, trial phases and investigational research.
Back to citation ↑ - NLM: About PubMed (external website)
PubMed provides literature citations and abstracts, with links to full text when available.
Back to citation ↑ - ClinicalTrials.gov: Official disclaimer (external website)
Registration is not government endorsement; sponsors supply study records.
Back to citation ↑ - FDA: Step 4: Drug review (external website)
FDA reviews submitted evidence and labeling in approval decisions.
Back to citation ↑ - Cochrane Handbook, chapter 10: Meta-analysis (external website)
Evidence synthesis, heterogeneity and limitations of pooling.
Back to citation ↑ - FDA: Step 2: Preclinical research (external website)
Laboratory and nonclinical testing; evaluation before testing in people.
Back to citation ↑ - NIH / NIEHS: Alternatives to animal testing (external website)
Research models include cells, animals and computer models; human relevance can be uncertain.
Back to citation ↑ - Cochrane Handbook, chapter 15: Interpreting results (external website)
Effect size, confidence intervals, clinical relevance and applicability.
Back to citation ↑ - EMA: Authorisation of medicines (external website)
EMA evaluates centrally authorised medicines; the European Commission grants authorisation.
Back to citation ↑ - FDA: Human drug compounding (external website)
Compounded drugs are not FDA-approved products.
Back to citation ↑ - FDA: Step 5: Post-market safety monitoring (external website)
Safety surveillance continues after approval.
Back to citation ↑