lotrevaBegin with the study question.
A trial result is attached to a question, a population and an analysis. Before copying a number into a research brief, write those three things beside it. A body-weight outcome, an imaging measurement and a symptom score are separate observations even when they appear in discussions of the same molecule.
Use the full study title and a stable identifier. A news headline may emphasize the largest result while omitting whether it was a primary endpoint, a subgroup or an exploratory analysis.
A worked example: weight and liver fat.
The 2023 obesity report and the 2024 liver-fat substudy are related, rather than two wholly independent replications. The substudy included 98 participants from the parent trial with specified liver-fat criteria; its primary question concerned relative change in liver fat at 24 weeks. A liver-fat measurement is not itself a finding that fibrosis, clinical complications or mortality improved.
| Publication | Design detail | Reading question |
|---|---|---|
| Obesity trial | 338 randomized adults; primary weight endpoint at 24 weeks | Who entered the trial, and which analysis is reported? |
| Liver-fat substudy | 98 selected participants; primary liver-fat change at 24 weeks | What does the imaging endpoint establish, and what does it leave unanswered? |
Sources: Sanyal et al. (2024): randomized retatrutide liver-fat substudy
The analysis is part of the result.
ICH E9(R1) uses the term estimand for a precise description of the treatment effect a trial intends to estimate. The framework makes events such as treatment discontinuation part of the question to be addressed. Different choices can yield different estimates without either number being a transcription error.
When two summaries disagree, inspect their population, endpoint, time point and handling of discontinuation before deciding which one is correct. Keep uncertainty intervals and the stated analysis label beside the estimate.
Sources: FDA / ICH E9(R1): estimands and sensitivity analysis (2021)
What to record when new results appear.
Date a literature-search note separately from the publication date. A sponsor announcement, conference presentation and journal article may describe the same trial at different levels of detail. Add the newer document to the same trial record rather than counting each announcement as a new study.
| Record | What to preserve |
|---|---|
| Trial identifier and population | Exact record, eligibility and sample size |
| Comparator and endpoint | What was compared and what was measured |
| Time point and analysis | Specified follow-up, estimand and missing-data approach |
| Benefit and harm reporting | Effect estimate, uncertainty, adverse events and discontinuations |
| Source status | Abstract, presentation, sponsor release or full publication |
Do not infer a head-to-head result.
Separate studies of retatrutide and semaglutide do not, by themselves, establish that one outperforms the other in the same patients. Treat cross-trial comparisons as questions for further investigation. A direct comparative claim needs an appropriate comparative design and interpretation.
These reading tools are for evaluating publications. Trial results do not authenticate independently sourced vials or provide instructions for personal treatment.