On this page
Start here: evaluate a clinical-trial claim
Use this four-guide sequence to move from a headline to the registered plan, the correct development stage and the full evidence.
1. Read the result
Check population, comparator, endpoint, estimand, missing data and safety.
2. Understand the phase
Learn what Phase 1, 2, 3 and 4 can—and cannot—establish.
3. Audit the registry
Decode study status, outcomes, dates, amendments and posted results.
4. Test the headline
Compare sponsor claims with prespecified outcomes and fuller reports.
Start with the study question, not the headline. A clinical trial can establish an important result for a defined group of people under controlled conditions. It cannot, by itself, tell every reader what will happen to them, prove that an online product matches the trial material, or grant Australian approval. This guide shows how to follow a claim from a news release to a registry record, a full report and a regulator.
Editorial review: 6 October 2026. We identify source type and jurisdiction below. Study and regulatory records can change; follow the linked primary source for the latest entry.
Choose the evidence you need
- What was planned? Open the TRIUMPH-1 trial record (NCT05929066) and inspect eligibility, comparators, outcome measures and record history. Registration describes a plan; it is not a finding.
- What was observed? Read the peer-reviewed TRIUMPH-1 Phase 3 report alongside the Phase 2 report. Compare population, duration, estimand, endpoint and adverse-event reporting rather than lifting one headline number.
- What was announced before publication? Compare Lilly’s TRIUMPH-1 sponsor announcement with the later peer-reviewed report. The difference between 28.3% and 25.0% is primarily an estimand difference, not a correction to the same analysis.
- What is Australia’s position? Check the TGA’s 14 September 2026 retatrutide alert. Trial progress is distinct from approval and product identity.
Four records, four different answers
| Record | What it can establish | What it cannot establish alone |
|---|---|---|
| Trial registry | Prespecified design, population, outcomes, recruitment and record updates. | That a result is favourable, independently reviewed or approved. |
| Sponsor announcement | What the sponsor reports at a stated date, often before the complete paper. | All analyses, full safety context or independent confirmation. |
| Peer-reviewed report | Methods and reported findings available for detailed scrutiny. | Universal effectiveness, absence of rare harms or regulatory authorisation. |
| Regulatory record | Decisions, indications, conditions and warnings for a jurisdiction. | That every similarly named product is the authorised product. |
A worked example: why two retatrutide numbers should not be merged
The 2023 peer-reviewed Phase 2 report described a mean body-weight change of 24.2% in one trial group after 48 weeks, versus 2.1% with placebo. The peer-reviewed TRIUMPH-1 Phase 3 report, published online 29 September 2026, reported −25.0% at 80 weeks for the 12 mg group versus −3.9% with placebo under the treatment-regimen estimand. Lilly’s earlier sponsor release reported −28.3% versus −2.2% under an efficacy estimand. These are different analyses with different assumptions; neither number describes every participant, and the separate Phase 2 and Phase 3 trials should not be merged into one trend line.
For the detailed programme, adverse-event context and source list, use our retatrutide trial tracker. For the meaning of the “GLP-3” nickname and the three receptor systems, use the GLP-3 evidence guide.
The six checks that change how a result reads
- Population: Who was eligible, who was excluded, and were participants living with diabetes or another condition? Results from one population are not automatically transferable to another.
- Comparator: Placebo, usual care or another active medicine? Cross-trial percentages are a poor substitute for a direct randomised comparison; see our SURMOUNT-5 head-to-head reading.
- Interpretation: Record the estimand, endpoint, discontinuations and missing-data method. Use our weight-loss drug trial guide as a repeatable checklist.
- Outcome and time: Was the headline the prespecified primary endpoint, and at how many weeks? A planned completion date is not a results date.
- Analysis: Which participants and missing-data assumptions contribute to the estimate? “Efficacy” and “treatment-regimen” estimands address different questions. A discontinuation rate needs its denominator and reason.
- Harms: Were adverse events actively collected, and for long enough to detect the type of harm being discussed? A headline mean does not describe tolerability.
- Publication stage: Is this a registry entry, sponsor topline, conference abstract or full report? Label the stage beside every numerical claim.
Follow the methods, then the molecule
Read how to read a registry record, what trial phases can and cannot establish, and our six-question check for press releases. The Research Library connects these methods to individual compound profiles. We keep older research-profile URLs for continuity, while treating those pages as evidence resources rather than offers to buy a substance.
Our rule: We link the exact record behind a claim, identify who produced it, state the date and population, and separate Australian approval from study progress. We correct material errors under the Corrections Policy. This is general scientific information, not treatment or dosing advice.
Explore oral GLP-1 and amylin study records
The oral GLP-1 tracker connects FDA formulation-specific decisions with OASIS 4 and ATTAIN-1 trial identifiers, populations and analysis labels. The amylin evidence guide separates cagrilintide alone, CagriSema combinations, petrelintide and eloralintide, with links to the named human studies.
Metabolic-liver trial data
The MASH/MASLD evidence tracker connects FDA decisions with MAESTRO-NASH, ESSENCE and SYNERGY-NASH records, keeping analysis populations, biopsy endpoints and longer-term clinical outcomes separate.
Registered study or human benefit? Compare the MOTS-c human-evidence records: an exercise-measurement study, a sponsor-submitted intervention record without posted results, and a completed study of the related CB4211 analogue.